OpenCloudOS-Kernel/drivers/gpu/drm/drm_ioc32.c

1001 lines
27 KiB
C
Raw Normal View History

/*
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
* \file drm_ioc32.c
*
* 32-bit ioctl compatibility routines for the DRM.
*
* \author Paul Mackerras <paulus@samba.org>
*
* Copyright (C) Paul Mackerras 2005.
* All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
* WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
* IN THE SOFTWARE.
*/
#include <linux/compat.h>
#include <linux/ratelimit.h>
#include <linux/export.h>
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
#include <drm/drm_agpsupport.h>
#include <drm/drm_file.h>
#include <drm/drm_print.h>
#include "drm_crtc_internal.h"
#include "drm_internal.h"
#include "drm_legacy.h"
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
#define DRM_IOCTL_VERSION32 DRM_IOWR(0x00, drm_version32_t)
#define DRM_IOCTL_GET_UNIQUE32 DRM_IOWR(0x01, drm_unique32_t)
#define DRM_IOCTL_GET_MAP32 DRM_IOWR(0x04, drm_map32_t)
#define DRM_IOCTL_GET_CLIENT32 DRM_IOWR(0x05, drm_client32_t)
#define DRM_IOCTL_GET_STATS32 DRM_IOR( 0x06, drm_stats32_t)
#define DRM_IOCTL_SET_UNIQUE32 DRM_IOW( 0x10, drm_unique32_t)
#define DRM_IOCTL_ADD_MAP32 DRM_IOWR(0x15, drm_map32_t)
#define DRM_IOCTL_ADD_BUFS32 DRM_IOWR(0x16, drm_buf_desc32_t)
#define DRM_IOCTL_MARK_BUFS32 DRM_IOW( 0x17, drm_buf_desc32_t)
#define DRM_IOCTL_INFO_BUFS32 DRM_IOWR(0x18, drm_buf_info32_t)
#define DRM_IOCTL_MAP_BUFS32 DRM_IOWR(0x19, drm_buf_map32_t)
#define DRM_IOCTL_FREE_BUFS32 DRM_IOW( 0x1a, drm_buf_free32_t)
#define DRM_IOCTL_RM_MAP32 DRM_IOW( 0x1b, drm_map32_t)
#define DRM_IOCTL_SET_SAREA_CTX32 DRM_IOW( 0x1c, drm_ctx_priv_map32_t)
#define DRM_IOCTL_GET_SAREA_CTX32 DRM_IOWR(0x1d, drm_ctx_priv_map32_t)
#define DRM_IOCTL_RES_CTX32 DRM_IOWR(0x26, drm_ctx_res32_t)
#define DRM_IOCTL_DMA32 DRM_IOWR(0x29, drm_dma32_t)
#define DRM_IOCTL_AGP_ENABLE32 DRM_IOW( 0x32, drm_agp_mode32_t)
#define DRM_IOCTL_AGP_INFO32 DRM_IOR( 0x33, drm_agp_info32_t)
#define DRM_IOCTL_AGP_ALLOC32 DRM_IOWR(0x34, drm_agp_buffer32_t)
#define DRM_IOCTL_AGP_FREE32 DRM_IOW( 0x35, drm_agp_buffer32_t)
#define DRM_IOCTL_AGP_BIND32 DRM_IOW( 0x36, drm_agp_binding32_t)
#define DRM_IOCTL_AGP_UNBIND32 DRM_IOW( 0x37, drm_agp_binding32_t)
#define DRM_IOCTL_SG_ALLOC32 DRM_IOW( 0x38, drm_scatter_gather32_t)
#define DRM_IOCTL_SG_FREE32 DRM_IOW( 0x39, drm_scatter_gather32_t)
#define DRM_IOCTL_UPDATE_DRAW32 DRM_IOW( 0x3f, drm_update_draw32_t)
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
#define DRM_IOCTL_WAIT_VBLANK32 DRM_IOWR(0x3a, drm_wait_vblank32_t)
#define DRM_IOCTL_MODE_ADDFB232 DRM_IOWR(0xb8, drm_mode_fb_cmd232_t)
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
typedef struct drm_version_32 {
int version_major; /* Major version */
int version_minor; /* Minor version */
int version_patchlevel; /* Patch level */
u32 name_len; /* Length of name buffer */
u32 name; /* Name of driver */
u32 date_len; /* Length of date buffer */
u32 date; /* User-space buffer to hold date */
u32 desc_len; /* Length of desc buffer */
u32 desc; /* User-space buffer to hold desc */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_version32_t;
static int compat_drm_version(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_version32_t v32;
struct drm_version v;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (copy_from_user(&v32, (void __user *)arg, sizeof(v32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
v = (struct drm_version) {
.name_len = v32.name_len,
.name = compat_ptr(v32.name),
.date_len = v32.date_len,
.date = compat_ptr(v32.date),
.desc_len = v32.desc_len,
.desc = compat_ptr(v32.desc),
};
err = drm_ioctl_kernel(file, drm_version, &v,
drm/ioctl: Ditch DRM_UNLOCKED except for the legacy vblank ioctl This completes Emil's series of removing DRM_UNLOCKED from modern drivers. It's entirely cargo-culted since we ignore it on non-DRIVER_LEGACY drivers since: commit ea487835e8876abf7ad909636e308c801a2bcda6 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Sep 28 21:42:40 2015 +0200 drm: Enforce unlocked ioctl operation for kms driver ioctls Now justifying why we can do this for legacy drives too (and hence close the source of all the bogus copypasting) is a bit more involved. DRM_UNLOCKED was introduced in: commit ed8b67040965e4fe695db333d5914e18ea5f146f Author: Arnd Bergmann <arnd@arndb.de> Date: Wed Dec 16 22:17:09 2009 +0000 drm: convert drm_ioctl to unlocked_ioctl As a immediate hack to keep i810 happy, which would have deadlocked without this trickery. The old BKL is automatically dropped in schedule(), and hence the i810 vs. mmap_sem deadlock didn't actually cause a real deadlock. But with a mutex it would. The solution was to annotate these as DRM_UNLOCKED and mark i810 unsafe on SMP machines. This conversion caused a regression, because unlike the BKL a mutex isn't dropped over schedule (that thing again), which caused a vblank wait in one thread to block the entire desktop and all its apps. Back then we did vblank scheduling by blocking in the client, awesome isn't it. This was fixed quickly in (ok not so quickly, took 2 years): commit 8f4ff2b06afcd6f151868474a432c603057eaf56 Author: Ilija Hadzic <ihadzic@research.bell-labs.com> Date: Mon Oct 31 17:46:18 2011 -0400 drm: do not sleep on vblank while holding a mutex All the other DRM_UNLOCKED annotations for all the core ioctls was work to reach finer-grained locking for modern drivers. This took years, and culminated in: commit fdd5b877e9ebc2029e1373b4a3cd057329a9ab7a Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Sat Dec 10 22:52:54 2016 +0100 drm: Enforce BKL-less ioctls for modern drivers DRM_UNLOCKED was never required by any legacy drivers, except for the vblank_wait IOCTL. Therefore we will not regress these old drivers by going back to where we've been in 2011. For all modern drivers nothing will change. To make this perfectly clear, also add a comment to DRM_UNLOCKED. v2: Don't forget about drm_ioc32.c (Michel). Cc: Michel Dänzer <michel@daenzer.net> Cc: Emil Velikov <emil.l.velikov@gmail.com> Acked-by: Emil Velikov <emil.velikov@collabora.com> Acked-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190605120835.2798-1-daniel.vetter@ffwll.ch
2019-06-05 20:08:35 +08:00
DRM_RENDER_ALLOW);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
v32.version_major = v.version_major;
v32.version_minor = v.version_minor;
v32.version_patchlevel = v.version_patchlevel;
v32.name_len = v.name_len;
v32.date_len = v.date_len;
v32.desc_len = v.desc_len;
if (copy_to_user((void __user *)arg, &v32, sizeof(v32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
typedef struct drm_unique32 {
u32 unique_len; /* Length of unique */
u32 unique; /* Unique name for driver instantiation */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_unique32_t;
static int compat_drm_getunique(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_unique32_t uq32;
struct drm_unique uq;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (copy_from_user(&uq32, (void __user *)arg, sizeof(uq32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
uq = (struct drm_unique){
.unique_len = uq32.unique_len,
.unique = compat_ptr(uq32.unique),
};
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
drm/ioctl: Ditch DRM_UNLOCKED except for the legacy vblank ioctl This completes Emil's series of removing DRM_UNLOCKED from modern drivers. It's entirely cargo-culted since we ignore it on non-DRIVER_LEGACY drivers since: commit ea487835e8876abf7ad909636e308c801a2bcda6 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Sep 28 21:42:40 2015 +0200 drm: Enforce unlocked ioctl operation for kms driver ioctls Now justifying why we can do this for legacy drives too (and hence close the source of all the bogus copypasting) is a bit more involved. DRM_UNLOCKED was introduced in: commit ed8b67040965e4fe695db333d5914e18ea5f146f Author: Arnd Bergmann <arnd@arndb.de> Date: Wed Dec 16 22:17:09 2009 +0000 drm: convert drm_ioctl to unlocked_ioctl As a immediate hack to keep i810 happy, which would have deadlocked without this trickery. The old BKL is automatically dropped in schedule(), and hence the i810 vs. mmap_sem deadlock didn't actually cause a real deadlock. But with a mutex it would. The solution was to annotate these as DRM_UNLOCKED and mark i810 unsafe on SMP machines. This conversion caused a regression, because unlike the BKL a mutex isn't dropped over schedule (that thing again), which caused a vblank wait in one thread to block the entire desktop and all its apps. Back then we did vblank scheduling by blocking in the client, awesome isn't it. This was fixed quickly in (ok not so quickly, took 2 years): commit 8f4ff2b06afcd6f151868474a432c603057eaf56 Author: Ilija Hadzic <ihadzic@research.bell-labs.com> Date: Mon Oct 31 17:46:18 2011 -0400 drm: do not sleep on vblank while holding a mutex All the other DRM_UNLOCKED annotations for all the core ioctls was work to reach finer-grained locking for modern drivers. This took years, and culminated in: commit fdd5b877e9ebc2029e1373b4a3cd057329a9ab7a Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Sat Dec 10 22:52:54 2016 +0100 drm: Enforce BKL-less ioctls for modern drivers DRM_UNLOCKED was never required by any legacy drivers, except for the vblank_wait IOCTL. Therefore we will not regress these old drivers by going back to where we've been in 2011. For all modern drivers nothing will change. To make this perfectly clear, also add a comment to DRM_UNLOCKED. v2: Don't forget about drm_ioc32.c (Michel). Cc: Michel Dänzer <michel@daenzer.net> Cc: Emil Velikov <emil.l.velikov@gmail.com> Acked-by: Emil Velikov <emil.velikov@collabora.com> Acked-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190605120835.2798-1-daniel.vetter@ffwll.ch
2019-06-05 20:08:35 +08:00
err = drm_ioctl_kernel(file, drm_getunique, &uq, 0);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
uq32.unique_len = uq.unique_len;
if (copy_to_user((void __user *)arg, &uq32, sizeof(uq32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
static int compat_drm_setunique(struct file *file, unsigned int cmd,
unsigned long arg)
{
/* it's dead */
return -EINVAL;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
#if IS_ENABLED(CONFIG_DRM_LEGACY)
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
typedef struct drm_map32 {
u32 offset; /* Requested physical address (0 for SAREA) */
u32 size; /* Requested physical size (bytes) */
enum drm_map_type type; /* Type of memory to map */
enum drm_map_flags flags; /* Flags */
u32 handle; /* User-space: "Handle" to pass to mmap() */
int mtrr; /* MTRR slot used */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_map32_t;
static int compat_drm_getmap(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_map32_t __user *argp = (void __user *)arg;
drm_map32_t m32;
struct drm_map map;
int err;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_from_user(&m32, argp, sizeof(m32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
map.offset = m32.offset;
drm/ioctl: Ditch DRM_UNLOCKED except for the legacy vblank ioctl This completes Emil's series of removing DRM_UNLOCKED from modern drivers. It's entirely cargo-culted since we ignore it on non-DRIVER_LEGACY drivers since: commit ea487835e8876abf7ad909636e308c801a2bcda6 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Sep 28 21:42:40 2015 +0200 drm: Enforce unlocked ioctl operation for kms driver ioctls Now justifying why we can do this for legacy drives too (and hence close the source of all the bogus copypasting) is a bit more involved. DRM_UNLOCKED was introduced in: commit ed8b67040965e4fe695db333d5914e18ea5f146f Author: Arnd Bergmann <arnd@arndb.de> Date: Wed Dec 16 22:17:09 2009 +0000 drm: convert drm_ioctl to unlocked_ioctl As a immediate hack to keep i810 happy, which would have deadlocked without this trickery. The old BKL is automatically dropped in schedule(), and hence the i810 vs. mmap_sem deadlock didn't actually cause a real deadlock. But with a mutex it would. The solution was to annotate these as DRM_UNLOCKED and mark i810 unsafe on SMP machines. This conversion caused a regression, because unlike the BKL a mutex isn't dropped over schedule (that thing again), which caused a vblank wait in one thread to block the entire desktop and all its apps. Back then we did vblank scheduling by blocking in the client, awesome isn't it. This was fixed quickly in (ok not so quickly, took 2 years): commit 8f4ff2b06afcd6f151868474a432c603057eaf56 Author: Ilija Hadzic <ihadzic@research.bell-labs.com> Date: Mon Oct 31 17:46:18 2011 -0400 drm: do not sleep on vblank while holding a mutex All the other DRM_UNLOCKED annotations for all the core ioctls was work to reach finer-grained locking for modern drivers. This took years, and culminated in: commit fdd5b877e9ebc2029e1373b4a3cd057329a9ab7a Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Sat Dec 10 22:52:54 2016 +0100 drm: Enforce BKL-less ioctls for modern drivers DRM_UNLOCKED was never required by any legacy drivers, except for the vblank_wait IOCTL. Therefore we will not regress these old drivers by going back to where we've been in 2011. For all modern drivers nothing will change. To make this perfectly clear, also add a comment to DRM_UNLOCKED. v2: Don't forget about drm_ioc32.c (Michel). Cc: Michel Dänzer <michel@daenzer.net> Cc: Emil Velikov <emil.l.velikov@gmail.com> Acked-by: Emil Velikov <emil.velikov@collabora.com> Acked-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190605120835.2798-1-daniel.vetter@ffwll.ch
2019-06-05 20:08:35 +08:00
err = drm_ioctl_kernel(file, drm_legacy_getmap_ioctl, &map, 0);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
m32.offset = map.offset;
m32.size = map.size;
m32.type = map.type;
m32.flags = map.flags;
m32.handle = ptr_to_compat((void __user *)map.handle);
m32.mtrr = map.mtrr;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_to_user(argp, &m32, sizeof(m32)))
return -EFAULT;
return 0;
}
static int compat_drm_addmap(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_map32_t __user *argp = (void __user *)arg;
drm_map32_t m32;
struct drm_map map;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (copy_from_user(&m32, argp, sizeof(m32)))
return -EFAULT;
map.offset = m32.offset;
map.size = m32.size;
map.type = m32.type;
map.flags = m32.flags;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
err = drm_ioctl_kernel(file, drm_legacy_addmap_ioctl, &map,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
m32.offset = map.offset;
m32.mtrr = map.mtrr;
m32.handle = ptr_to_compat((void __user *)map.handle);
if (map.handle != compat_ptr(m32.handle))
pr_err_ratelimited("compat_drm_addmap truncated handle %p for type %d offset %x\n",
map.handle, m32.type, m32.offset);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_to_user(argp, &m32, sizeof(m32)))
return -EFAULT;
return 0;
}
static int compat_drm_rmmap(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_map32_t __user *argp = (void __user *)arg;
struct drm_map map;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
u32 handle;
if (get_user(handle, &argp->handle))
return -EFAULT;
map.handle = compat_ptr(handle);
return drm_ioctl_kernel(file, drm_legacy_rmmap_ioctl, &map, DRM_AUTH);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
#endif
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
typedef struct drm_client32 {
int idx; /* Which client desired? */
int auth; /* Is client authenticated? */
u32 pid; /* Process ID */
u32 uid; /* User ID */
u32 magic; /* Magic */
u32 iocs; /* Ioctl count */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_client32_t;
static int compat_drm_getclient(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_client32_t c32;
drm_client32_t __user *argp = (void __user *)arg;
struct drm_client client;
int err;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_from_user(&c32, argp, sizeof(c32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
client.idx = c32.idx;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
drm/ioctl: Ditch DRM_UNLOCKED except for the legacy vblank ioctl This completes Emil's series of removing DRM_UNLOCKED from modern drivers. It's entirely cargo-culted since we ignore it on non-DRIVER_LEGACY drivers since: commit ea487835e8876abf7ad909636e308c801a2bcda6 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Sep 28 21:42:40 2015 +0200 drm: Enforce unlocked ioctl operation for kms driver ioctls Now justifying why we can do this for legacy drives too (and hence close the source of all the bogus copypasting) is a bit more involved. DRM_UNLOCKED was introduced in: commit ed8b67040965e4fe695db333d5914e18ea5f146f Author: Arnd Bergmann <arnd@arndb.de> Date: Wed Dec 16 22:17:09 2009 +0000 drm: convert drm_ioctl to unlocked_ioctl As a immediate hack to keep i810 happy, which would have deadlocked without this trickery. The old BKL is automatically dropped in schedule(), and hence the i810 vs. mmap_sem deadlock didn't actually cause a real deadlock. But with a mutex it would. The solution was to annotate these as DRM_UNLOCKED and mark i810 unsafe on SMP machines. This conversion caused a regression, because unlike the BKL a mutex isn't dropped over schedule (that thing again), which caused a vblank wait in one thread to block the entire desktop and all its apps. Back then we did vblank scheduling by blocking in the client, awesome isn't it. This was fixed quickly in (ok not so quickly, took 2 years): commit 8f4ff2b06afcd6f151868474a432c603057eaf56 Author: Ilija Hadzic <ihadzic@research.bell-labs.com> Date: Mon Oct 31 17:46:18 2011 -0400 drm: do not sleep on vblank while holding a mutex All the other DRM_UNLOCKED annotations for all the core ioctls was work to reach finer-grained locking for modern drivers. This took years, and culminated in: commit fdd5b877e9ebc2029e1373b4a3cd057329a9ab7a Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Sat Dec 10 22:52:54 2016 +0100 drm: Enforce BKL-less ioctls for modern drivers DRM_UNLOCKED was never required by any legacy drivers, except for the vblank_wait IOCTL. Therefore we will not regress these old drivers by going back to where we've been in 2011. For all modern drivers nothing will change. To make this perfectly clear, also add a comment to DRM_UNLOCKED. v2: Don't forget about drm_ioc32.c (Michel). Cc: Michel Dänzer <michel@daenzer.net> Cc: Emil Velikov <emil.l.velikov@gmail.com> Acked-by: Emil Velikov <emil.velikov@collabora.com> Acked-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190605120835.2798-1-daniel.vetter@ffwll.ch
2019-06-05 20:08:35 +08:00
err = drm_ioctl_kernel(file, drm_getclient, &client, 0);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
c32.idx = client.idx;
c32.auth = client.auth;
c32.pid = client.pid;
c32.uid = client.uid;
c32.magic = client.magic;
c32.iocs = client.iocs;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_to_user(argp, &c32, sizeof(c32)))
return -EFAULT;
return 0;
}
typedef struct drm_stats32 {
u32 count;
struct {
u32 value;
enum drm_stat_type type;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} data[15];
} drm_stats32_t;
static int compat_drm_getstats(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_stats32_t __user *argp = (void __user *)arg;
int err;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
drm/ioctl: Ditch DRM_UNLOCKED except for the legacy vblank ioctl This completes Emil's series of removing DRM_UNLOCKED from modern drivers. It's entirely cargo-culted since we ignore it on non-DRIVER_LEGACY drivers since: commit ea487835e8876abf7ad909636e308c801a2bcda6 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Sep 28 21:42:40 2015 +0200 drm: Enforce unlocked ioctl operation for kms driver ioctls Now justifying why we can do this for legacy drives too (and hence close the source of all the bogus copypasting) is a bit more involved. DRM_UNLOCKED was introduced in: commit ed8b67040965e4fe695db333d5914e18ea5f146f Author: Arnd Bergmann <arnd@arndb.de> Date: Wed Dec 16 22:17:09 2009 +0000 drm: convert drm_ioctl to unlocked_ioctl As a immediate hack to keep i810 happy, which would have deadlocked without this trickery. The old BKL is automatically dropped in schedule(), and hence the i810 vs. mmap_sem deadlock didn't actually cause a real deadlock. But with a mutex it would. The solution was to annotate these as DRM_UNLOCKED and mark i810 unsafe on SMP machines. This conversion caused a regression, because unlike the BKL a mutex isn't dropped over schedule (that thing again), which caused a vblank wait in one thread to block the entire desktop and all its apps. Back then we did vblank scheduling by blocking in the client, awesome isn't it. This was fixed quickly in (ok not so quickly, took 2 years): commit 8f4ff2b06afcd6f151868474a432c603057eaf56 Author: Ilija Hadzic <ihadzic@research.bell-labs.com> Date: Mon Oct 31 17:46:18 2011 -0400 drm: do not sleep on vblank while holding a mutex All the other DRM_UNLOCKED annotations for all the core ioctls was work to reach finer-grained locking for modern drivers. This took years, and culminated in: commit fdd5b877e9ebc2029e1373b4a3cd057329a9ab7a Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Sat Dec 10 22:52:54 2016 +0100 drm: Enforce BKL-less ioctls for modern drivers DRM_UNLOCKED was never required by any legacy drivers, except for the vblank_wait IOCTL. Therefore we will not regress these old drivers by going back to where we've been in 2011. For all modern drivers nothing will change. To make this perfectly clear, also add a comment to DRM_UNLOCKED. v2: Don't forget about drm_ioc32.c (Michel). Cc: Michel Dänzer <michel@daenzer.net> Cc: Emil Velikov <emil.l.velikov@gmail.com> Acked-by: Emil Velikov <emil.velikov@collabora.com> Acked-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190605120835.2798-1-daniel.vetter@ffwll.ch
2019-06-05 20:08:35 +08:00
err = drm_ioctl_kernel(file, drm_noop, NULL, 0);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
if (clear_user(argp, sizeof(drm_stats32_t)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
#if IS_ENABLED(CONFIG_DRM_LEGACY)
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
typedef struct drm_buf_desc32 {
int count; /* Number of buffers of this size */
int size; /* Size in bytes */
int low_mark; /* Low water mark */
int high_mark; /* High water mark */
int flags;
u32 agp_start; /* Start address in the AGP aperture */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_buf_desc32_t;
static int compat_drm_addbufs(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_buf_desc32_t __user *argp = (void __user *)arg;
drm_buf_desc32_t desc32;
struct drm_buf_desc desc;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (copy_from_user(&desc32, argp, sizeof(drm_buf_desc32_t)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
desc = (struct drm_buf_desc){
desc32.count, desc32.size, desc32.low_mark, desc32.high_mark,
desc32.flags, desc32.agp_start
};
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
err = drm_ioctl_kernel(file, drm_legacy_addbufs, &desc,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
desc32 = (drm_buf_desc32_t){
desc.count, desc.size, desc.low_mark, desc.high_mark,
desc.flags, desc.agp_start
};
if (copy_to_user(argp, &desc32, sizeof(drm_buf_desc32_t)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
static int compat_drm_markbufs(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_buf_desc32_t b32;
drm_buf_desc32_t __user *argp = (void __user *)arg;
struct drm_buf_desc buf;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_from_user(&b32, argp, sizeof(b32)))
return -EFAULT;
buf.size = b32.size;
buf.low_mark = b32.low_mark;
buf.high_mark = b32.high_mark;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return drm_ioctl_kernel(file, drm_legacy_markbufs, &buf,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
typedef struct drm_buf_info32 {
int count; /**< Entries in list */
u32 list;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_buf_info32_t;
static int copy_one_buf32(void *data, int count, struct drm_buf_entry *from)
{
drm_buf_info32_t *request = data;
drm_buf_desc32_t __user *to = compat_ptr(request->list);
drm_buf_desc32_t v = {.count = from->buf_count,
.size = from->buf_size,
.low_mark = from->low_mark,
.high_mark = from->high_mark};
if (copy_to_user(to + count, &v, offsetof(drm_buf_desc32_t, flags)))
return -EFAULT;
return 0;
}
static int drm_legacy_infobufs32(struct drm_device *dev, void *data,
struct drm_file *file_priv)
{
drm_buf_info32_t *request = data;
return __drm_legacy_infobufs(dev, data, &request->count, copy_one_buf32);
}
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
static int compat_drm_infobufs(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_buf_info32_t req32;
drm_buf_info32_t __user *argp = (void __user *)arg;
int err;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_from_user(&req32, argp, sizeof(req32)))
return -EFAULT;
if (req32.count < 0)
req32.count = 0;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
err = drm_ioctl_kernel(file, drm_legacy_infobufs32, &req32, DRM_AUTH);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
if (put_user(req32.count, &argp->count))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
typedef struct drm_buf_pub32 {
int idx; /**< Index into the master buffer list */
int total; /**< Buffer size */
int used; /**< Amount of buffer in use (for DMA) */
u32 address; /**< Address of buffer */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_buf_pub32_t;
typedef struct drm_buf_map32 {
int count; /**< Length of the buffer list */
u32 virtual; /**< Mmap'd area in user-virtual */
u32 list; /**< Buffer information */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_buf_map32_t;
static int map_one_buf32(void *data, int idx, unsigned long virtual,
struct drm_buf *buf)
{
drm_buf_map32_t *request = data;
drm_buf_pub32_t __user *to = compat_ptr(request->list) + idx;
drm_buf_pub32_t v;
v.idx = buf->idx;
v.total = buf->total;
v.used = 0;
v.address = virtual + buf->offset;
if (copy_to_user(to, &v, sizeof(v)))
return -EFAULT;
return 0;
}
static int drm_legacy_mapbufs32(struct drm_device *dev, void *data,
struct drm_file *file_priv)
{
drm_buf_map32_t *request = data;
void __user *v;
int err = __drm_legacy_mapbufs(dev, data, &request->count,
&v, map_one_buf32,
file_priv);
request->virtual = ptr_to_compat(v);
return err;
}
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
static int compat_drm_mapbufs(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_buf_map32_t __user *argp = (void __user *)arg;
drm_buf_map32_t req32;
int err;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_from_user(&req32, argp, sizeof(req32)))
return -EFAULT;
if (req32.count < 0)
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EINVAL;
err = drm_ioctl_kernel(file, drm_legacy_mapbufs32, &req32, DRM_AUTH);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
if (put_user(req32.count, &argp->count)
|| put_user(req32.virtual, &argp->virtual))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
typedef struct drm_buf_free32 {
int count;
u32 list;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_buf_free32_t;
static int compat_drm_freebufs(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_buf_free32_t req32;
struct drm_buf_free request;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
drm_buf_free32_t __user *argp = (void __user *)arg;
if (copy_from_user(&req32, argp, sizeof(req32)))
return -EFAULT;
request.count = req32.count;
request.list = compat_ptr(req32.list);
return drm_ioctl_kernel(file, drm_legacy_freebufs, &request, DRM_AUTH);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
typedef struct drm_ctx_priv_map32 {
unsigned int ctx_id; /**< Context requesting private mapping */
u32 handle; /**< Handle of map */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_ctx_priv_map32_t;
static int compat_drm_setsareactx(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_ctx_priv_map32_t req32;
struct drm_ctx_priv_map request;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
drm_ctx_priv_map32_t __user *argp = (void __user *)arg;
if (copy_from_user(&req32, argp, sizeof(req32)))
return -EFAULT;
request.ctx_id = req32.ctx_id;
request.handle = compat_ptr(req32.handle);
return drm_ioctl_kernel(file, drm_legacy_setsareactx, &request,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
static int compat_drm_getsareactx(struct file *file, unsigned int cmd,
unsigned long arg)
{
struct drm_ctx_priv_map req;
drm_ctx_priv_map32_t req32;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
drm_ctx_priv_map32_t __user *argp = (void __user *)arg;
int err;
if (copy_from_user(&req32, argp, sizeof(req32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
req.ctx_id = req32.ctx_id;
err = drm_ioctl_kernel(file, drm_legacy_getsareactx, &req, DRM_AUTH);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
req32.handle = ptr_to_compat((void __user *)req.handle);
if (copy_to_user(argp, &req32, sizeof(req32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
typedef struct drm_ctx_res32 {
int count;
u32 contexts;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_ctx_res32_t;
static int compat_drm_resctx(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_ctx_res32_t __user *argp = (void __user *)arg;
drm_ctx_res32_t res32;
struct drm_ctx_res res;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (copy_from_user(&res32, argp, sizeof(res32)))
return -EFAULT;
res.count = res32.count;
res.contexts = compat_ptr(res32.contexts);
err = drm_ioctl_kernel(file, drm_legacy_resctx, &res, DRM_AUTH);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
res32.count = res.count;
if (copy_to_user(argp, &res32, sizeof(res32)))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
typedef struct drm_dma32 {
int context; /**< Context handle */
int send_count; /**< Number of buffers to send */
u32 send_indices; /**< List of handles to buffers */
u32 send_sizes; /**< Lengths of data to send */
enum drm_dma_flags flags; /**< Flags */
int request_count; /**< Number of buffers requested */
int request_size; /**< Desired size for buffers */
u32 request_indices; /**< Buffer information */
u32 request_sizes;
int granted_count; /**< Number of buffers granted */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_dma32_t;
static int compat_drm_dma(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_dma32_t d32;
drm_dma32_t __user *argp = (void __user *)arg;
struct drm_dma d;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (copy_from_user(&d32, argp, sizeof(d32)))
return -EFAULT;
d.context = d32.context;
d.send_count = d32.send_count;
d.send_indices = compat_ptr(d32.send_indices);
d.send_sizes = compat_ptr(d32.send_sizes);
d.flags = d32.flags;
d.request_count = d32.request_count;
d.request_indices = compat_ptr(d32.request_indices);
d.request_sizes = compat_ptr(d32.request_sizes);
err = drm_ioctl_kernel(file, drm_legacy_dma_ioctl, &d, DRM_AUTH);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
if (put_user(d.request_size, &argp->request_size)
|| put_user(d.granted_count, &argp->granted_count))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
#endif
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
#if IS_ENABLED(CONFIG_AGP)
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
typedef struct drm_agp_mode32 {
u32 mode; /**< AGP mode */
} drm_agp_mode32_t;
static int compat_drm_agp_enable(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_agp_mode32_t __user *argp = (void __user *)arg;
struct drm_agp_mode mode;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (get_user(mode.mode, &argp->mode))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return drm_ioctl_kernel(file, drm_agp_enable_ioctl, &mode,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
typedef struct drm_agp_info32 {
int agp_version_major;
int agp_version_minor;
u32 mode;
u32 aperture_base; /* physical address */
u32 aperture_size; /* bytes */
u32 memory_allowed; /* bytes */
u32 memory_used;
/* PCI information */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
unsigned short id_vendor;
unsigned short id_device;
} drm_agp_info32_t;
static int compat_drm_agp_info(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_agp_info32_t __user *argp = (void __user *)arg;
drm_agp_info32_t i32;
struct drm_agp_info info;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
err = drm_ioctl_kernel(file, drm_agp_info_ioctl, &info, DRM_AUTH);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
i32.agp_version_major = info.agp_version_major;
i32.agp_version_minor = info.agp_version_minor;
i32.mode = info.mode;
i32.aperture_base = info.aperture_base;
i32.aperture_size = info.aperture_size;
i32.memory_allowed = info.memory_allowed;
i32.memory_used = info.memory_used;
i32.id_vendor = info.id_vendor;
i32.id_device = info.id_device;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_to_user(argp, &i32, sizeof(i32)))
return -EFAULT;
return 0;
}
typedef struct drm_agp_buffer32 {
u32 size; /**< In bytes -- will round to page boundary */
u32 handle; /**< Used for binding / unbinding */
u32 type; /**< Type of memory to allocate */
u32 physical; /**< Physical used by i810 */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
} drm_agp_buffer32_t;
static int compat_drm_agp_alloc(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_agp_buffer32_t __user *argp = (void __user *)arg;
drm_agp_buffer32_t req32;
struct drm_agp_buffer request;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (copy_from_user(&req32, argp, sizeof(req32)))
return -EFAULT;
request.size = req32.size;
request.type = req32.type;
err = drm_ioctl_kernel(file, drm_agp_alloc_ioctl, &request,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
req32.handle = request.handle;
req32.physical = request.physical;
if (copy_to_user(argp, &req32, sizeof(req32))) {
drm_ioctl_kernel(file, drm_agp_free_ioctl, &request,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
}
return 0;
}
static int compat_drm_agp_free(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_agp_buffer32_t __user *argp = (void __user *)arg;
struct drm_agp_buffer request;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (get_user(request.handle, &argp->handle))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return drm_ioctl_kernel(file, drm_agp_free_ioctl, &request,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
typedef struct drm_agp_binding32 {
u32 handle; /**< From drm_agp_buffer */
u32 offset; /**< In bytes -- will round to page boundary */
} drm_agp_binding32_t;
static int compat_drm_agp_bind(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_agp_binding32_t __user *argp = (void __user *)arg;
drm_agp_binding32_t req32;
struct drm_agp_binding request;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_from_user(&req32, argp, sizeof(req32)))
return -EFAULT;
request.handle = req32.handle;
request.offset = req32.offset;
return drm_ioctl_kernel(file, drm_agp_bind_ioctl, &request,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
static int compat_drm_agp_unbind(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_agp_binding32_t __user *argp = (void __user *)arg;
struct drm_agp_binding request;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (get_user(request.handle, &argp->handle))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return drm_ioctl_kernel(file, drm_agp_unbind_ioctl, &request,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
#endif /* CONFIG_AGP */
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
#if IS_ENABLED(CONFIG_DRM_LEGACY)
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
typedef struct drm_scatter_gather32 {
u32 size; /**< In bytes -- will round to page boundary */
u32 handle; /**< Used for mapping / unmapping */
} drm_scatter_gather32_t;
static int compat_drm_sg_alloc(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_scatter_gather32_t __user *argp = (void __user *)arg;
struct drm_scatter_gather request;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (get_user(request.size, &argp->size))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
err = drm_ioctl_kernel(file, drm_legacy_sg_alloc, &request,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
/* XXX not sure about the handle conversion here... */
if (put_user(request.handle >> PAGE_SHIFT, &argp->handle))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
return 0;
}
static int compat_drm_sg_free(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_scatter_gather32_t __user *argp = (void __user *)arg;
struct drm_scatter_gather request;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
unsigned long x;
if (get_user(x, &argp->handle))
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return -EFAULT;
request.handle = x << PAGE_SHIFT;
return drm_ioctl_kernel(file, drm_legacy_sg_free, &request,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
}
#endif
#if defined(CONFIG_X86)
typedef struct drm_update_draw32 {
drm_drawable_t handle;
unsigned int type;
unsigned int num;
/* 64-bit version has a 32-bit pad here */
u64 data; /**< Pointer */
} __attribute__((packed)) drm_update_draw32_t;
static int compat_drm_update_draw(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_update_draw32_t update32;
if (copy_from_user(&update32, (void __user *)arg, sizeof(update32)))
return -EFAULT;
return drm_ioctl_kernel(file, drm_noop, NULL,
DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY);
}
#endif
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
struct drm_wait_vblank_request32 {
enum drm_vblank_seq_type type;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
unsigned int sequence;
u32 signal;
};
struct drm_wait_vblank_reply32 {
enum drm_vblank_seq_type type;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
unsigned int sequence;
s32 tval_sec;
s32 tval_usec;
};
typedef union drm_wait_vblank32 {
struct drm_wait_vblank_request32 request;
struct drm_wait_vblank_reply32 reply;
} drm_wait_vblank32_t;
static int compat_drm_wait_vblank(struct file *file, unsigned int cmd,
unsigned long arg)
{
drm_wait_vblank32_t __user *argp = (void __user *)arg;
drm_wait_vblank32_t req32;
union drm_wait_vblank req;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
int err;
if (copy_from_user(&req32, argp, sizeof(req32)))
return -EFAULT;
req.request.type = req32.request.type;
req.request.sequence = req32.request.sequence;
req.request.signal = req32.request.signal;
err = drm_ioctl_kernel(file, drm_wait_vblank_ioctl, &req, DRM_UNLOCKED);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (err)
return err;
req32.reply.type = req.reply.type;
req32.reply.sequence = req.reply.sequence;
req32.reply.tval_sec = req.reply.tval_sec;
req32.reply.tval_usec = req.reply.tval_usec;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
if (copy_to_user(argp, &req32, sizeof(req32)))
return -EFAULT;
return 0;
}
#if defined(CONFIG_X86)
typedef struct drm_mode_fb_cmd232 {
u32 fb_id;
u32 width;
u32 height;
u32 pixel_format;
u32 flags;
u32 handles[4];
u32 pitches[4];
u32 offsets[4];
u64 modifier[4];
} __attribute__((packed)) drm_mode_fb_cmd232_t;
static int compat_drm_mode_addfb2(struct file *file, unsigned int cmd,
unsigned long arg)
{
struct drm_mode_fb_cmd232 __user *argp = (void __user *)arg;
struct drm_mode_fb_cmd2 req64;
int err;
if (copy_from_user(&req64, argp,
offsetof(drm_mode_fb_cmd232_t, modifier)))
return -EFAULT;
if (copy_from_user(&req64.modifier, &argp->modifier,
sizeof(req64.modifier)))
return -EFAULT;
drm/ioctl: Ditch DRM_UNLOCKED except for the legacy vblank ioctl This completes Emil's series of removing DRM_UNLOCKED from modern drivers. It's entirely cargo-culted since we ignore it on non-DRIVER_LEGACY drivers since: commit ea487835e8876abf7ad909636e308c801a2bcda6 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Sep 28 21:42:40 2015 +0200 drm: Enforce unlocked ioctl operation for kms driver ioctls Now justifying why we can do this for legacy drives too (and hence close the source of all the bogus copypasting) is a bit more involved. DRM_UNLOCKED was introduced in: commit ed8b67040965e4fe695db333d5914e18ea5f146f Author: Arnd Bergmann <arnd@arndb.de> Date: Wed Dec 16 22:17:09 2009 +0000 drm: convert drm_ioctl to unlocked_ioctl As a immediate hack to keep i810 happy, which would have deadlocked without this trickery. The old BKL is automatically dropped in schedule(), and hence the i810 vs. mmap_sem deadlock didn't actually cause a real deadlock. But with a mutex it would. The solution was to annotate these as DRM_UNLOCKED and mark i810 unsafe on SMP machines. This conversion caused a regression, because unlike the BKL a mutex isn't dropped over schedule (that thing again), which caused a vblank wait in one thread to block the entire desktop and all its apps. Back then we did vblank scheduling by blocking in the client, awesome isn't it. This was fixed quickly in (ok not so quickly, took 2 years): commit 8f4ff2b06afcd6f151868474a432c603057eaf56 Author: Ilija Hadzic <ihadzic@research.bell-labs.com> Date: Mon Oct 31 17:46:18 2011 -0400 drm: do not sleep on vblank while holding a mutex All the other DRM_UNLOCKED annotations for all the core ioctls was work to reach finer-grained locking for modern drivers. This took years, and culminated in: commit fdd5b877e9ebc2029e1373b4a3cd057329a9ab7a Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Sat Dec 10 22:52:54 2016 +0100 drm: Enforce BKL-less ioctls for modern drivers DRM_UNLOCKED was never required by any legacy drivers, except for the vblank_wait IOCTL. Therefore we will not regress these old drivers by going back to where we've been in 2011. For all modern drivers nothing will change. To make this perfectly clear, also add a comment to DRM_UNLOCKED. v2: Don't forget about drm_ioc32.c (Michel). Cc: Michel Dänzer <michel@daenzer.net> Cc: Emil Velikov <emil.l.velikov@gmail.com> Acked-by: Emil Velikov <emil.velikov@collabora.com> Acked-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190605120835.2798-1-daniel.vetter@ffwll.ch
2019-06-05 20:08:35 +08:00
err = drm_ioctl_kernel(file, drm_mode_addfb2, &req64, 0);
if (err)
return err;
if (put_user(req64.fb_id, &argp->fb_id))
return -EFAULT;
return 0;
}
#endif
static struct {
drm_ioctl_compat_t *fn;
char *name;
} drm_compat_ioctls[] = {
#define DRM_IOCTL32_DEF(n, f) [DRM_IOCTL_NR(n##32)] = {.fn = f, .name = #n}
DRM_IOCTL32_DEF(DRM_IOCTL_VERSION, compat_drm_version),
DRM_IOCTL32_DEF(DRM_IOCTL_GET_UNIQUE, compat_drm_getunique),
#if IS_ENABLED(CONFIG_DRM_LEGACY)
DRM_IOCTL32_DEF(DRM_IOCTL_GET_MAP, compat_drm_getmap),
#endif
DRM_IOCTL32_DEF(DRM_IOCTL_GET_CLIENT, compat_drm_getclient),
DRM_IOCTL32_DEF(DRM_IOCTL_GET_STATS, compat_drm_getstats),
DRM_IOCTL32_DEF(DRM_IOCTL_SET_UNIQUE, compat_drm_setunique),
#if IS_ENABLED(CONFIG_DRM_LEGACY)
DRM_IOCTL32_DEF(DRM_IOCTL_ADD_MAP, compat_drm_addmap),
DRM_IOCTL32_DEF(DRM_IOCTL_ADD_BUFS, compat_drm_addbufs),
DRM_IOCTL32_DEF(DRM_IOCTL_MARK_BUFS, compat_drm_markbufs),
DRM_IOCTL32_DEF(DRM_IOCTL_INFO_BUFS, compat_drm_infobufs),
DRM_IOCTL32_DEF(DRM_IOCTL_MAP_BUFS, compat_drm_mapbufs),
DRM_IOCTL32_DEF(DRM_IOCTL_FREE_BUFS, compat_drm_freebufs),
DRM_IOCTL32_DEF(DRM_IOCTL_RM_MAP, compat_drm_rmmap),
DRM_IOCTL32_DEF(DRM_IOCTL_SET_SAREA_CTX, compat_drm_setsareactx),
DRM_IOCTL32_DEF(DRM_IOCTL_GET_SAREA_CTX, compat_drm_getsareactx),
DRM_IOCTL32_DEF(DRM_IOCTL_RES_CTX, compat_drm_resctx),
DRM_IOCTL32_DEF(DRM_IOCTL_DMA, compat_drm_dma),
#endif
#if IS_ENABLED(CONFIG_AGP)
DRM_IOCTL32_DEF(DRM_IOCTL_AGP_ENABLE, compat_drm_agp_enable),
DRM_IOCTL32_DEF(DRM_IOCTL_AGP_INFO, compat_drm_agp_info),
DRM_IOCTL32_DEF(DRM_IOCTL_AGP_ALLOC, compat_drm_agp_alloc),
DRM_IOCTL32_DEF(DRM_IOCTL_AGP_FREE, compat_drm_agp_free),
DRM_IOCTL32_DEF(DRM_IOCTL_AGP_BIND, compat_drm_agp_bind),
DRM_IOCTL32_DEF(DRM_IOCTL_AGP_UNBIND, compat_drm_agp_unbind),
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
#endif
#if IS_ENABLED(CONFIG_DRM_LEGACY)
DRM_IOCTL32_DEF(DRM_IOCTL_SG_ALLOC, compat_drm_sg_alloc),
DRM_IOCTL32_DEF(DRM_IOCTL_SG_FREE, compat_drm_sg_free),
#endif
#if defined(CONFIG_X86) || defined(CONFIG_IA64)
DRM_IOCTL32_DEF(DRM_IOCTL_UPDATE_DRAW, compat_drm_update_draw),
#endif
DRM_IOCTL32_DEF(DRM_IOCTL_WAIT_VBLANK, compat_drm_wait_vblank),
#if defined(CONFIG_X86) || defined(CONFIG_IA64)
DRM_IOCTL32_DEF(DRM_IOCTL_MODE_ADDFB2, compat_drm_mode_addfb2),
#endif
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
};
/**
* drm_compat_ioctl - 32bit IOCTL compatibility handler for DRM drivers
* @filp: file this ioctl is called on
* @cmd: ioctl cmd number
* @arg: user argument
*
* Compatibility handler for 32 bit userspace running on 64 kernels. All actual
* IOCTL handling is forwarded to drm_ioctl(), while marshalling structures as
* appropriate. Note that this only handles DRM core IOCTLs, if the driver has
* botched IOCTL itself, it must handle those by wrapping this function.
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
*
* Returns:
* Zero on success, negative error code on failure.
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
*/
long drm_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
{
unsigned int nr = DRM_IOCTL_NR(cmd);
struct drm_file *file_priv = filp->private_data;
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
drm_ioctl_compat_t *fn;
int ret;
/* Assume that ioctls without an explicit compat routine will just
* work. This may not always be a good assumption, but it's better
* than always failing.
*/
if (nr >= ARRAY_SIZE(drm_compat_ioctls))
return drm_ioctl(filp, cmd, arg);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
fn = drm_compat_ioctls[nr].fn;
if (!fn)
return drm_ioctl(filp, cmd, arg);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
DRM_DEBUG("comm=\"%s\", pid=%d, dev=0x%lx, auth=%d, %s\n",
current->comm, task_pid_nr(current),
(long)old_encode_dev(file_priv->minor->kdev->devt),
file_priv->authenticated,
drm_compat_ioctls[nr].name);
ret = (*fn)(filp, cmd, arg);
if (ret)
DRM_DEBUG("ret = %d\n", ret);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
return ret;
}
EXPORT_SYMBOL(drm_compat_ioctl);