EEH and pci_dlpar #undef DEBUG, but I think they were added before the
ability to control this from Kconfig. It's really annoying to only get
some of the debug messages from these files. Leave the lpar.c #undef
alone as it produces so much output as to make the kernel unusable.
Update the Kconfig text to indicate this particular quirk :)
Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Acked-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The code is missing a fix that went into the main kernel variant
(we should try to share that code again at some stage)
Reported-by: Albert Cahalan <acahalan@gmail.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
While booting OMAP4 ES2.0 boards, cards on MMC1 and MMC2 controllers
are not getting detected sometimes.
During reset of command/data line, wrong pointer to base address
was passed while read operation to SYSCTL register, thus impacting
the updated reset logic.
Passing the correct base address fixes the issue.
Signed-off-by: Kishore Kadiyala <kishore.kadiyala@ti.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Acked-by: Madhusudhan Chikkature <madhu.cr@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
In jbd2_journal_init_dev(), we need make sure the journal structure is
fully initialzied before calling jbd2_stats_proc_init().
Reviewed-by: Andreas Dilger <andreas.dilger@oracle.com>
Signed-off-by: yangsheng <sheng.yang@oracle.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If the the li_request_list was empty then it returned with the lock
held. Instead of adding a "goto unlock" I just removed that special
case and let it go past the empty list_for_each_safe().
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
ext4_end_bio calls put_page and kmem_cache_free before calling
SetPageUpdate(). This can result in setting the PageUptodate bit on
random pages and causes the following BUG:
BUG: Bad page state in process rm pfn:52e54
page:ffffea0001222260 count:0 mapcount:0 mapping: (null) index:0x0
arch kernel: page flags: 0x4000000000000008(uptodate)
Fix the problem by moving put_io_page() after the SetPageUpdate() call.
Thanks to Hugh Dickins for analyzing this problem.
Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Tested-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Signed-off-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Commit 451a3c24b0 ("BKL: remove extraneous #include <smp_lock.h>")
removed the #include line that was the only thing that was surrounded by
the #ifdef/#endif.
So now that #ifdef is guarding nothing at all. Just remove it.
Reported-by: Byeong-ryeol Kim <brofkims@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Arnd Bergmann did an automated scripting run to find left-over instances
of <linux/smp_lock.h>, and had made it trigger it on the normal BKL use
of lock_kernel and unlock_lernel (and apparently release_kernel_lock and
reacquire_kernel_lock too, used by the scheduler).
That resulted in commit 451a3c24b0 ("BKL: remove extraneous #include
<smp_lock.h>").
However, hardirq.h was the only remaining user of the old
'kernel_locked()' interface, and Arnd's script hadn't checked for that.
So depending on your configuration and what header files had been
included, you would get errors like "implicit declaration of function
'kernel_locked'" during the build.
The right fix is not to just re-instate the smp_lock.h include - it is
to just remove 'kernel_locked()' entirely, since the only use was this
one special low-level detail. Just make hardirq.h do it directly.
In fact this simplifies and clarifies the code, because some trivial
analysis makes it clear that hardirq.h only ever used _one_ of the two
definitions of kernel_locked(), so we can remove the other one entirely.
Reported-by: Zimny Lech <napohybelskurwysynom2010@gmail.com>
Reported-and-acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We now use load_gs_index() to load gs safely; unfortunately this also
changes MSR_KERNEL_GS_BASE, which we managed separately. This resulted
in confusion and breakage running 32-bit host userspace on a 64-bit kernel.
Fix by
- saving guest MSR_KERNEL_GS_BASE before we we reload the host's gs
- doing the host save/load unconditionally, instead of only when in guest
long mode
Things can be cleaned up further, but this is the minmal fix for now.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
If fs or gs refer to the ldt, they must be reloaded after the ldt. Reorder
the code to that effect.
Userspace code that uses the ldt with kvm is nonexistent, so this doesn't fix
a user-visible bug.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
irq_of_parse_and_map() has an unsigned return type.
Testing for a negative error value doesn't work here.
Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
bnx2x_init_one() should return negative value on error.
By mistake it returns ENODEV instead of -ENODEV.
Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
I am not family with RealTek RTL-8139C+ series 10/100 PCI Ethernet driver.
I try to guess the meaning of RxProtoIP and IPFail.
RxProtoIP stands for received IPv4 packet that upper protocol is not tcp and udp.
!(status & IPFail) is true means that driver correctly to check checksum in IPv4 header.
If these are right, driver will set ip_summed with CHECKSUM_UNNECESSARY for other
upper protocol, e.g. sctp, igmp protocol. This will cause protocol stack ignores
checksum check for packets with invalid checksum.
This patch is only compile-test.
Signed-off-by: Shan Wei <shanwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If r8196 received packets with invalid sctp/igmp(not tcp, udp) checksum, r8196 set skb->ip_summed
wit CHECKSUM_UNNECESSARY. This cause that upper protocol don't check checksum field.
I am not family with r8196 driver. I try to guess the meaning of RxProtoIP and IPFail.
RxProtoIP stands for received IPv4 packet that upper protocol is not tcp and udp.
!(opts1 & IPFail) is true means that driver correctly to check checksum in IPv4 header.
If it's right, I think we should not set ip_summed wit CHECKSUM_UNNECESSARY for my sctp packets
with invalid checksum.
If it's not right, please tell me.
Signed-off-by: Shan Wei <shanwei@cn.fujitsu.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In rds_cmsg_rdma_args(), the user-provided args->nr_local value is
restricted to less than UINT_MAX. This seems to need a tighter upper
bound, since the calculation of total iov_size can overflow, resulting
in a small sock_kmalloc() allocation. This would probably just result
in walking off the heap and crashing when calling rds_rdma_pages() with
a high count value. If it somehow doesn't crash here, then memory
corruption could occur soon after.
Signed-off-by: Dan Rosenberg <drosenberg@vsecurity.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit ff10b88b5a (kgdb,ppc: Individual
register get/set for ppc) introduced a problem where memcpy was used
incorrectly to read and write the evr registers with a kernel that
has:
CONFIG_FSL_BOOKE=y
CONFIG_SPE=y
CONFIG_KGDB=y
This patch also fixes the following compilation problems:
arch/powerpc/kernel/kgdb.c: In function 'dbg_get_reg':
arch/powerpc/kernel/kgdb.c:341: error: passing argument 2 of 'memcpy' makes pointer from integer without a cast
arch/powerpc/kernel/kgdb.c: In function 'dbg_set_reg':
arch/powerpc/kernel/kgdb.c:366: error: passing argument 1 of 'memcpy' makes pointer from integer without a cast
[jason.wessel@windriver.com: Remove void * casts and fix patch header]
Reported-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Dongdong Deng <dongdong.deng@windriver.com>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
CC: linuxppc-dev@lists.ozlabs.org
The fix from ba773f7c51
(x86,kgdb: Fix hw breakpoint regression) was not entirely complete.
The kgdb_remove_all_hw_break() function also needs to call the
hw_break_release_slot() or else a breakpoint can get activated again
after the debugger has detached.
The kgdb test suite exposes the behavior in the form of either a hang
or repetitive failure. The kernel config that exposes the problem
contains all of the following:
CONFIG_DEBUG_RODATA=y
CONFIG_KGDB_TESTS=y
CONFIG_KGDB_TESTS_ON_BOOT=y
CONFIG_KGDB_TESTS_BOOT_STRING="V1F100"
Reported-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Tested-by: Frederic Weisbecker <fweisbec@gmail.com>
When the number of dyanmic kdb commands exceeds KDB_BASE_CMD_MAX, the
kernel will fault.
Signed-off-by: Jovi Zhang <bookjovi@gmail.com>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Call kfree in the error path as well as the success path in kdb_ll().
Signed-off-by: Jovi Zhang <bookjovi@gmail.com>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Multi-component commit f0fba2ad broke a few things which this patch should
fix. Tested on the DM355 EVM. I've been as careful as I can, but it would be
good if those with access to other Davinci boards could test.
--
The multi-component commit put the initialisation of
snd_soc_dai.[capture|playback]_dma_data into snd_soc_dai_ops.hw_params of the
McBSP, McASP & VCIF drivers (davinci-i2s.c, davinci-mcasp.c & davinci-vcif.c).
The initialisation had to be moved from the probe function in these drivers
because davinci_*_dai changed from snd_soc_dai to snd_soc_dai_driver.
Unfortunately, the DMA params pointer is needed by davinci_pcm_open (in
davinci-pcm.c) before hw_params is called. I have moved the initialisation to
a new snd_soc_dai_ops.startup function in each of these drivers. This fix
indicates that all platforms that use davinci-pcm must have been broken and
need to test with this fix.
--
The multi-component commit also changed the McBSP driver name from
"davinci-asp" to "davinci-i2s" in davinci-i2s.c without updating the board
level references to the driver name. This change is understandable, as there
is a similarly named "davinci-mcasp" driver in davinci-mcasp.c.
There is probably no 'correct' name for this driver. The DM6446 datasheet
calls it the "ASP" and describes it as a "specialised McBSP". The DM355
datasheet calls it the "ASP" and describes it as a "specialised ASP". The
DM365 datasheet calls it the "McBSP". Rather than fix this problem by
reverting to "davinci-asp", I've elected to avoid future confusion with the
"davinci-mcasp" driver by changing it to "davinci-mcbsp", which is also
consistent with the names of the functions in the driver. There are other
fixes required, so it was never going to be as simple as a revert anyway.
--
The DM365 only has one McBSP port (of the McBSP platforms, only the DM355 has
2 ports), so I've changed the the id of the platform_device from 0 to -1.
--
In davinci-evm.c, the DM6446 EVM can no longer share a snd_soc_dai_link
structure with the DM355 EVM as they use different cpu DAI names (the DM355
has 2 ports and the EVM uses the second port, but the DM6446 only has 1 port).
This also means that the 2 boards need different snd_soc_card structures.
--
The codec_name entries in davinci-evm.c didn't match the i2c ids in the board
files. I have only checked and fixed the details of the names used for the
McBSP based platforms. Someone with a McASP based platform (eg DA8xx) should
check the others.
Signed-off-by: Chris Paulson-Ellis <chris@edesix.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
During the multi-component conversion the WM8994 register cache init
got lost.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
cmd->serial_number is never tested in any path we reach; therefore we may
remove the call to scsi_cmd_get_serial() inside DEF_SCSI_QCMD, the SCSI
host_lock acquisition surrounding it, and our own SCSI host_lock
unlock+relock cycle.
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
The stradis driver is on its way out, but it should still be marked
correctly as depending on the big kernel lock. It could easily be
changed to not require it if someone decides to revive the driver and
port it to v4l2 in the process.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reported-by: Sedat Dilek <sedat.dilek@googlemail.com>
Cc: Nathan Laredo <laredo@gnu.org>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Lock_kernel is gone from the code, so the comments should be updated,
too. nfsd now uses lock_flocks instead of lock_kernel to protect
against posix file locks.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: J. Bruce Fields <bfields@redhat.com>
Cc: linux-nfs@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The big kernel lock has been removed from all these files at some point,
leaving only the #include.
Remove this too as a cleanup.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Removed mux: and added the __func__ macro to make debugging
easier.
Signed-off-by: Dan Murphy <dmurphy@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
The mux framework allows the change of pad configuration by drivers
when needed. Prior to this the mux framework has to be initialised
with all the mux parameters specific to the board. The mux init is
already present in the board file for SDP. Adding the mux init
for panda boards.
Signed-off-by: sricharan <r.sricharan@ti.com>
Acked-by: Anand Gadiyar <gadiyar@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Select the CBS package if SDP4430 is enabled during config.
Use the proper package (CBL or CBS) based on chip revision.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Anand Gadiyar <gadiyar@ti.com>
Please note that the full muxmodes are re-defined for ES2 instead
of using the subset. There are 81 differences among 204 pins.
The subset fixup will have to iterate over the whole list for each
subset entry, which can lead to an important number of iteration.
On the other hand, it will take much more memory at boot time.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Anand Gadiyar <gadiyar@ti.com>
Select the CBL package if SDP4430 is enabled during config.
Initialize the mux framework during the board init.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Add data for OMAP4430 generated from HW pinout & register database.
The data set is split in two partitions for both core and wkup.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Starting on OMAP4, the pin mux configuration is located in two
different partitions of the control module (CODE_PAD and WKUP_PAD).
The first one is inside the core power domain whereas the second
one is inside the wakeup.
- Add the capability to add any number of partition during board init
time depending of Soc partitioning.
- Add some init flags as well in order to avoid explicit Soc version
check inside the mux core code.
- Add a comment with mux0 mode on top of omap_mux/board/<partition>
if the current mux mode is not the default one.
Thanks to Tony Lindgren <tony@atomide.com> for the following improvements:
- Add omap_mux_get for getting the partition data so platform level
device code can use it.
- Fix the rx51 board code to use the new API.
- Do not store the partition for each mux entry. Look up the partition
for debugfs instead.
Thanks to Dan Murphy <dmurphy@ti.com> for testing on OMAP4 and reporting
a couple of bugs.
Thanks to Anand Gadiyar <gadiyar@ti.com> for testing on OMAP3 zoom and
bug report.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Tested-by: Murphy Dan <dmurphy@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Anand Gadiyar <gadiyar@ti.com>
Replace all the printk(KERN_XXX... with pr_xxx macros.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
The recent patch to fix the removal of a non-existing proc
directory introduced this build problem for !CONFIG_PROC_FS:
drivers/block/cciss.c:4929: error: 'proc_cciss' undeclared (first use in this function)
Fix it by moving proc_cciss outside of the CONFIG_PROC_FS scope.
Reported-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
This patch fixes the following error obtained when
compile the Kernel with the VSYSCALL support enabled:
SYSCALL arch/sh/kernel/vsyscall/vsyscall-trapa.so
sh4-linux/bin/ld: error in arch/sh/kernel/vsyscall/vsyscall-trapa.o(.eh_frame);
no .eh_frame_hdr table will be created.
AS arch/sh/kernel/vsyscall/vsyscall-syscall.o
Disassembling the vsyscall-trapa.o code, with this fix applied,
we get the __kernel_vsyscall symbol defined; otherwise it was
missing.
Disassembly of section .text:
00000000 <__kernel_vsyscall>:
0: 10 00 .word 0x0010
2: 00 00 .word 0x0000
[snip]
0000040 <__kernel_sigreturn>:
40: 05 93 mov.w 4e <__kernel_sigreturn+0xe>,r3 ! 77
42: 10 c3 trapa #16
[snip]
00000060 <__kernel_rt_sigreturn>:
60: 05 93 mov.w 6e <__kernel_rt_sigreturn+0xe>,r3 ! ad
62: 10 c3 trapa #16
[snip]
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Reviewed-by: Carmelo Amoroso <carmelo.amoroso@st.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Calling sys_cacheflush with ICACHE we can direclty flush
the icache without invoking the flush_cache_all function.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Commit b5c26f97ec introduced some breakage for the OLPC XO-1 laptop,
differences in the output video signal after the patch caused some problems
with the XO's display controller chip.
Reviewing of that commit against the AMD Geode LX Data Book, it seems
that these bits were being set inversely. In both cases, active high
output is denoted by a value of 0. See section 6.8.3.44 of the databook
from February 2009 (Publication ID: 33234H)
Signed-off-by: Daniel Drake <dsd@laptop.org>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
There is an integer overflow in fb_set_user_cmap() because cmap->len * 2
can wrap. It's basically harmless. Your terminal will be messed up
until you type reset.
This patch does three things to fix the bug.
First, it checks the return value of fb_copy_cmap() in fb_alloc_cmap().
That is enough to fix address the overflow.
Second it checks for the integer overflow in fb_set_user_cmap().
Lastly I wanted to cap "cmap->len" in fb_set_user_cmap() much lower
because it gets used to determine the size of allocation. Unfortunately
no one knows what the limit should be. Instead what this patch does
is makes the allocation happen with GFP_KERNEL instead of GFP_ATOMIC
and lets the kmalloc() decide what values of cmap->len are reasonable.
To do this, the patch introduces a function called fb_alloc_cmap_gfp()
which is like fb_alloc_cmap() except that it takes a GFP flag.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
checkpatch.pl and Andrew Morton both complained about the indenting in
fb_alloc_cmap()
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Making /proc/kallsyms readable only for root by default makes it
slightly harder for attackers to write generic kernel exploits by
removing one source of knowledge where things are in the kernel.
This is the second submit, discussion happened on this on first submit
and mostly concerned that this is just one hole of the sieve ... but
one of the bigger ones.
Changing the permissions of at least System.map and vmlinux is also
required to fix the same set, but a packaging issue.
Target of this starter patch and follow ups is removing any kind of
kernel space address information leak from the kernel.
[ Side note: the default of root-only reading is the "safe" value, and
it's easy enough to then override at any time after boot. The /proc
filesystem allows root to change the permissions with a regular
chmod, so you can "revert" this at run-time by simply doing
chmod og+r /proc/kallsyms
as root if you really want regular users to see the kernel symbols.
It does help some tools like "perf" figure them out without any
setup, so it may well make sense in some situations. - Linus ]
Signed-off-by: Marcus Meissner <meissner@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Acked-by: Eugene Teo <eugeneteo@kernel.org>
Reviewed-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6:
nfs: Ignore kmemleak false positive in nfs_readdir_make_qstr
SUNRPC: Simplify rpc_alloc_iostats by removing pointless local variable
nfs: trivial: remove unused nfs_wait_event macro
NFS: readdir shouldn't read beyond the reply returned by the server
NFS: Fix a couple of regressions in readdir.
Revert "NFSv4: Fall back to ordinary lookup if nfs4_atomic_open() returns EISDIR"
Regression: fix mounting NFS when NFSv3 support is not compiled
NLM: Fix a regression in lockd
* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
sched: Fix cross-sched-class wakeup preemption
sched: Fix runnable condition for stoptask
sched: Use group weight, idle cpu metrics to fix imbalances during idle
* 'pm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/suspend-2.6:
PM / PM QoS: Fix reversed min and max
PM / OPP: Hide OPP configuration when SoCs do not provide an implementation
PM: Allow devices to be removed during late suspend and early resume
This prevents firewire-net from submitting write requests in fast
succession until failure due to all 64 transaction labels were used up
for unfinished split transactions. The netif_stop/wake_queue API is
used for this purpose.
Without this stop/wake mechanism, datagrams were simply lost whenever
the tlabel pool was exhausted. Plus, tlabel exhaustion by firewire-net
also prevented other unrelated outbound transactions to be initiated.
The chosen queue depth was checked by me to hit the maximum possible
throughput with an OS X peer whose receive DMA is good enough to never
reject requests due to busy inbound request FIFO. Current Linux peers
show a mixed picture of -5%...+15% change in bandwidth; their current
bottleneck are RCODE_BUSY situations (fewer or more, depending on TX
queue depth) due to too small AR buffer in firewire-ohci.
Maxim Levitsky tested this change with similar watermarks with a Linux
peer and some pending firewire-ohci improvements that address the
RCODE_BUSY problem and confirmed that these TX queue limits are good.
Note: This removes some netif_wake_queue from reception code paths.
They were apparently copy&paste artefacts from a nonsensical
netif_wake_queue use in the older eth1394 driver. This belongs only
into the transmit path.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Tested-by: Maxim Levitsky <maximlevitsky@gmail.com>
The current transmit code does not at all make use of
- fwnet_device.packet_list
and only very limited use of
- fwnet_device.broadcasted_list,
- fwnet_device.queued_packets.
Their current function is to track whether the TX soft-IRQ finished
dealing with an skb when the AT-req tasklet takes over, and to discard
pending tx datagrams (if there are any) when the local node is removed.
The latter does actually contain a race condition bug with TX soft-IRQ
and AT-req tasklet.
Instead of these lists and the corresponding link in fwnet_packet_task,
- a flag in fwnet_packet_task to track whether fwnet_tx is done,
- a counter of queued datagrams in fwnet_device
do the job as well.
The above mentioned theoretic race condition is resolved by letting
fwnet_remove sleep until all datagrams were flushed. It may sleep
almost arbitrarily long since fwnet_remove is executed in the context of
a multithreaded (concurrency managed) workqueue.
The type of max_payload is changed to u16 here to avoid waste in struct
fwnet_packet_task. This value cannot exceed 4096 per IEEE 1394:2008
table 16-18 (or 32678 per specification of packet headers, if there is
ever going to be something else than beta mode).
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
a) fwnet_transmit_packet_done used to poison ptask->pt_link by list_del.
If fwnet_send_packet checked later whether it was responsible to clean
up (in the border case that the TX soft IRQ was outpaced by the AT-req
tasklet on another CPU), it missed this because ptask->pt_link was no
longer shown as empty.
b) If fwnet_write_complete got an rcode other than RCODE_COMPLETE, we
missed to free the skb and ptask entirely.
Also, count stats.tx_dropped and stats.tx_errors when rcode != 0.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
The per-cpu event channel masks can be updated unlocked from multiple
CPUs, so use the locked variant.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
To bind all event channels to CPU#0, it is not sufficient to set all
of its cpu_evtchn_mask[] bits; all other CPUs also need to get their
bits cleared. Otherwise, evtchn_do_upcall() will start handling
interrupts on CPUs they're not intended to run on, which can be
particularly bad for per-CPU ones.
[ linux-2.6.18-xen.hg 7de7453dee36 ]
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Nishanth Aravamudan