OpenCloudOS-Kernel/arch/powerpc/kernel/exceptions-64s.S

2421 lines
70 KiB
ArmAsm
Raw Normal View History

License cleanup: add SPDX GPL-2.0 license identifier to files with no license Many source files in the tree are missing licensing information, which makes it harder for compliance tools to determine the correct license. By default all files without license information are under the default license of the kernel, which is GPL version 2. Update the files which contain no license information with the 'GPL-2.0' SPDX license identifier. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This patch is based on work done by Thomas Gleixner and Kate Stewart and Philippe Ombredanne. How this work was done: Patches were generated and checked against linux-4.14-rc6 for a subset of the use cases: - file had no licensing information it it. - file was a */uapi/* one with no licensing information in it, - file was a */uapi/* one with existing licensing information, Further patches will be generated in subsequent months to fix up cases where non-standard license headers were used, and references to license had to be inferred by heuristics based on keywords. The analysis to determine which SPDX License Identifier to be applied to a file was done in a spreadsheet of side by side results from of the output of two independent scanners (ScanCode & Windriver) producing SPDX tag:value files created by Philippe Ombredanne. Philippe prepared the base worksheet, and did an initial spot review of a few 1000 files. The 4.13 kernel was the starting point of the analysis with 60,537 files assessed. Kate Stewart did a file by file comparison of the scanner results in the spreadsheet to determine which SPDX license identifier(s) to be applied to the file. She confirmed any determination that was not immediately clear with lawyers working with the Linux Foundation. Criteria used to select files for SPDX license identifier tagging was: - Files considered eligible had to be source code files. - Make and config files were included as candidates if they contained >5 lines of source - File already had some variant of a license header in it (even if <5 lines). All documentation files were explicitly excluded. The following heuristics were used to determine which SPDX license identifiers to apply. - when both scanners couldn't find any license traces, file was considered to have no license information in it, and the top level COPYING file license applied. For non */uapi/* files that summary was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 11139 and resulted in the first patch in this series. If that file was a */uapi/* path one, it was "GPL-2.0 WITH Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 WITH Linux-syscall-note 930 and resulted in the second patch in this series. - if a file had some form of licensing information in it, and was one of the */uapi/* ones, it was denoted with the Linux-syscall-note if any GPL family license was found in the file or had no licensing in it (per prior point). Results summary: SPDX license identifier # files ---------------------------------------------------|------ GPL-2.0 WITH Linux-syscall-note 270 GPL-2.0+ WITH Linux-syscall-note 169 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17 LGPL-2.1+ WITH Linux-syscall-note 15 GPL-1.0+ WITH Linux-syscall-note 14 ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5 LGPL-2.0+ WITH Linux-syscall-note 4 LGPL-2.1 WITH Linux-syscall-note 3 ((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3 ((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1 and that resulted in the third patch in this series. - when the two scanners agreed on the detected license(s), that became the concluded license(s). - when there was disagreement between the two scanners (one detected a license but the other didn't, or they both detected different licenses) a manual inspection of the file occurred. - In most cases a manual inspection of the information in the file resulted in a clear resolution of the license that should apply (and which scanner probably needed to revisit its heuristics). - When it was not immediately clear, the license identifier was confirmed with lawyers working with the Linux Foundation. - If there was any question as to the appropriate license identifier, the file was flagged for further research and to be revisited later in time. In total, over 70 hours of logged manual review was done on the spreadsheet to determine the SPDX license identifiers to apply to the source files by Kate, Philippe, Thomas and, in some cases, confirmation by lawyers working with the Linux Foundation. Kate also obtained a third independent scan of the 4.13 code base from FOSSology, and compared selected files where the other two scanners disagreed against that SPDX file, to see if there was new insights. The Windriver scanner is based on an older version of FOSSology in part, so they are related. Thomas did random spot checks in about 500 files from the spreadsheets for the uapi headers and agreed with SPDX license identifier in the files he inspected. For the non-uapi files Thomas did random spot checks in about 15000 files. In initial set of patches against 4.14-rc6, 3 files were found to have copy/paste license identifier errors, and have been fixed to reflect the correct identifier. Additionally Philippe spent 10 hours this week doing a detailed manual inspection and review of the 12,461 patched files from the initial patch version early this week with: - a full scancode scan run, collecting the matched texts, detected license ids and scores - reviewing anything where there was a license detected (about 500+ files) to ensure that the applied SPDX license was correct - reviewing anything where there was no detection but the patch license was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied SPDX license was correct This produced a worksheet with 20 files needing minor correction. This worksheet was then exported into 3 different .csv files for the different types of files to be modified. These .csv files were then reviewed by Greg. Thomas wrote a script to parse the csv files and add the proper SPDX tag to the file, in the format that the file expected. This script was further refined by Greg based on the output to detect more types of files automatically and to distinguish between header and source .c files (which need different comment types.) Finally Greg ran the script using the .csv files to generate the patches. Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 22:07:57 +08:00
/* SPDX-License-Identifier: GPL-2.0 */
/*
* This file contains the 64-bit "server" PowerPC variant
* of the low level exception handling including exception
* vectors, exception return, part of the slb and stab
* handling and other fixed offset specific things.
*
* This file is meant to be #included from head_64.S due to
* position dependent assembly.
*
* Most of this originates from head_64.S and thus has the same
* copyright history.
*
*/
powerpc: Rework lazy-interrupt handling The current implementation of lazy interrupts handling has some issues that this tries to address. We don't do the various workarounds we need to do when re-enabling interrupts in some cases such as when returning from an interrupt and thus we may still lose or get delayed decrementer or doorbell interrupts. The current scheme also makes it much harder to handle the external "edge" interrupts provided by some BookE processors when using the EPR facility (External Proxy) and the Freescale Hypervisor. Additionally, we tend to keep interrupts hard disabled in a number of cases, such as decrementer interrupts, external interrupts, or when a masked decrementer interrupt is pending. This is sub-optimal. This is an attempt at fixing it all in one go by reworking the way we do the lazy interrupt disabling from the ground up. The base idea is to replace the "hard_enabled" field with a "irq_happened" field in which we store a bit mask of what interrupt occurred while soft-disabled. When re-enabling, either via arch_local_irq_restore() or when returning from an interrupt, we can now decide what to do by testing bits in that field. We then implement replaying of the missed interrupts either by re-using the existing exception frame (in exception exit case) or via the creation of a new one from an assembly trampoline (in the arch_local_irq_enable case). This removes the need to play with the decrementer to try to create fake interrupts, among others. In addition, this adds a few refinements: - We no longer hard disable decrementer interrupts that occur while soft-disabled. We now simply bump the decrementer back to max (on BookS) or leave it stopped (on BookE) and continue with hard interrupts enabled, which means that we'll potentially get better sample quality from performance monitor interrupts. - Timer, decrementer and doorbell interrupts now hard-enable shortly after removing the source of the interrupt, which means they no longer run entirely hard disabled. Again, this will improve perf sample quality. - On Book3E 64-bit, we now make the performance monitor interrupt act as an NMI like Book3S (the necessary C code for that to work appear to already be present in the FSL perf code, notably calling nmi_enter instead of irq_enter). (This also fixes a bug where BookE perfmon interrupts could clobber r14 ... oops) - We could make "masked" decrementer interrupts act as NMIs when doing timer-based perf sampling to improve the sample quality. Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org> --- v2: - Add hard-enable to decrementer, timer and doorbells - Fix CR clobber in masked irq handling on BookE - Make embedded perf interrupt act as an NMI - Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want to retrigger an interrupt without preventing hard-enable v3: - Fix or vs. ori bug on Book3E - Fix enabling of interrupts for some exceptions on Book3E v4: - Fix resend of doorbells on return from interrupt on Book3E v5: - Rebased on top of my latest series, which involves some significant rework of some aspects of the patch. v6: - 32-bit compile fix - more compile fixes with various .config combos - factor out the asm code to soft-disable interrupts - remove the C wrapper around preempt_schedule_irq v7: - Fix a bug with hard irq state tracking on native power7
2012-03-06 15:27:59 +08:00
#include <asm/hw_irq.h>
#include <asm/exception-64s.h>
#include <asm/ptrace.h>
#include <asm/cpuidle.h>
#include <asm/head-64.h>
#include <asm/feature-fixups.h>
#include <asm/kup.h>
/* PACA save area offsets (exgen, exmc, etc) */
#define EX_R9 0
#define EX_R10 8
#define EX_R11 16
#define EX_R12 24
#define EX_R13 32
#define EX_DAR 40
#define EX_DSISR 48
#define EX_CCR 52
#define EX_CFAR 56
#define EX_PPR 64
#if defined(CONFIG_RELOCATABLE)
#define EX_CTR 72
.if EX_SIZE != 10
.error "EX_SIZE is wrong"
.endif
#else
.if EX_SIZE != 9
.error "EX_SIZE is wrong"
.endif
#endif
/*
* Following are fixed section helper macros.
*
* EXC_REAL_BEGIN/END - real, unrelocated exception vectors
* EXC_VIRT_BEGIN/END - virt (AIL), unrelocated exception vectors
* TRAMP_REAL_BEGIN - real, unrelocated helpers (virt may call these)
* TRAMP_VIRT_BEGIN - virt, unreloc helpers (in practice, real can use)
* TRAMP_KVM_BEGIN - KVM handlers, these are put into real, unrelocated
* EXC_COMMON - After switching to virtual, relocated mode.
*/
#define EXC_REAL_BEGIN(name, start, size) \
FIXED_SECTION_ENTRY_BEGIN_LOCATION(real_vectors, exc_real_##start##_##name, start, size)
#define EXC_REAL_END(name, start, size) \
FIXED_SECTION_ENTRY_END_LOCATION(real_vectors, exc_real_##start##_##name, start, size)
#define EXC_VIRT_BEGIN(name, start, size) \
FIXED_SECTION_ENTRY_BEGIN_LOCATION(virt_vectors, exc_virt_##start##_##name, start, size)
#define EXC_VIRT_END(name, start, size) \
FIXED_SECTION_ENTRY_END_LOCATION(virt_vectors, exc_virt_##start##_##name, start, size)
#define EXC_COMMON_BEGIN(name) \
USE_TEXT_SECTION(); \
.balign IFETCH_ALIGN_BYTES; \
.global name; \
_ASM_NOKPROBE_SYMBOL(name); \
DEFINE_FIXED_SYMBOL(name); \
name:
#define TRAMP_REAL_BEGIN(name) \
FIXED_SECTION_ENTRY_BEGIN(real_trampolines, name)
#define TRAMP_VIRT_BEGIN(name) \
FIXED_SECTION_ENTRY_BEGIN(virt_trampolines, name)
#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
#define TRAMP_KVM_BEGIN(name) \
TRAMP_VIRT_BEGIN(name)
#else
#define TRAMP_KVM_BEGIN(name)
#endif
#define EXC_REAL_NONE(start, size) \
FIXED_SECTION_ENTRY_BEGIN_LOCATION(real_vectors, exc_real_##start##_##unused, start, size); \
FIXED_SECTION_ENTRY_END_LOCATION(real_vectors, exc_real_##start##_##unused, start, size)
#define EXC_VIRT_NONE(start, size) \
FIXED_SECTION_ENTRY_BEGIN_LOCATION(virt_vectors, exc_virt_##start##_##unused, start, size); \
FIXED_SECTION_ENTRY_END_LOCATION(virt_vectors, exc_virt_##start##_##unused, start, size)
/*
* We're short on space and time in the exception prolog, so we can't
* use the normal LOAD_REG_IMMEDIATE macro to load the address of label.
* Instead we get the base of the kernel from paca->kernelbase and or in the low
* part of label. This requires that the label be within 64KB of kernelbase, and
* that kernelbase be 64K aligned.
*/
#define LOAD_HANDLER(reg, label) \
ld reg,PACAKBASE(r13); /* get high part of &label */ \
ori reg,reg,FIXED_SYMBOL_ABS_ADDR(label)
#define __LOAD_HANDLER(reg, label) \
ld reg,PACAKBASE(r13); \
ori reg,reg,(ABS_ADDR(label))@l
/*
* Branches from unrelocated code (e.g., interrupts) to labels outside
* head-y require >64K offsets.
*/
#define __LOAD_FAR_HANDLER(reg, label) \
ld reg,PACAKBASE(r13); \
ori reg,reg,(ABS_ADDR(label))@l; \
addis reg,reg,(ABS_ADDR(label))@h
/* Exception register prefixes */
#define EXC_HV_OR_STD 2 /* depends on HVMODE */
#define EXC_HV 1
#define EXC_STD 0
#if defined(CONFIG_RELOCATABLE)
/*
* If we support interrupts with relocation on AND we're a relocatable kernel,
* we need to use CTR to get to the 2nd level handler. So, save/restore it
* when required.
*/
#define SAVE_CTR(reg, area) mfctr reg ; std reg,area+EX_CTR(r13)
#define GET_CTR(reg, area) ld reg,area+EX_CTR(r13)
#define RESTORE_CTR(reg, area) ld reg,area+EX_CTR(r13) ; mtctr reg
#else
/* ...else CTR is unused and in register. */
#define SAVE_CTR(reg, area)
#define GET_CTR(reg, area) mfctr reg
#define RESTORE_CTR(reg, area)
#endif
/*
* PPR save/restore macros used in exceptions-64s.S
* Used for P7 or later processors
*/
#define SAVE_PPR(area, ra) \
BEGIN_FTR_SECTION_NESTED(940) \
ld ra,area+EX_PPR(r13); /* Read PPR from paca */ \
std ra,_PPR(r1); \
END_FTR_SECTION_NESTED(CPU_FTR_HAS_PPR,CPU_FTR_HAS_PPR,940)
#define RESTORE_PPR_PACA(area, ra) \
BEGIN_FTR_SECTION_NESTED(941) \
ld ra,area+EX_PPR(r13); \
mtspr SPRN_PPR,ra; \
END_FTR_SECTION_NESTED(CPU_FTR_HAS_PPR,CPU_FTR_HAS_PPR,941)
/*
* Get an SPR into a register if the CPU has the given feature
*/
#define OPT_GET_SPR(ra, spr, ftr) \
BEGIN_FTR_SECTION_NESTED(943) \
mfspr ra,spr; \
END_FTR_SECTION_NESTED(ftr,ftr,943)
/*
* Set an SPR from a register if the CPU has the given feature
*/
#define OPT_SET_SPR(ra, spr, ftr) \
BEGIN_FTR_SECTION_NESTED(943) \
mtspr spr,ra; \
END_FTR_SECTION_NESTED(ftr,ftr,943)
/*
* Save a register to the PACA if the CPU has the given feature
*/
#define OPT_SAVE_REG_TO_PACA(offset, ra, ftr) \
BEGIN_FTR_SECTION_NESTED(943) \
std ra,offset(r13); \
END_FTR_SECTION_NESTED(ftr,ftr,943)
/*
* Branch to label using its 0xC000 address. This results in instruction
* address suitable for MSR[IR]=0 or 1, which allows relocation to be turned
* on using mtmsr rather than rfid.
*
* This could set the 0xc bits for !RELOCATABLE as an immediate, rather than
* load KBASE for a slight optimisation.
*/
#define BRANCH_TO_C000(reg, label) \
__LOAD_FAR_HANDLER(reg, label); \
mtctr reg; \
bctr
.macro INT_KVM_HANDLER name, vec, hsrr, area, skip
TRAMP_KVM_BEGIN(\name\()_kvm)
KVM_HANDLER \vec, \hsrr, \area, \skip
.endm
#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
/*
* If hv is possible, interrupts come into to the hv version
* of the kvmppc_interrupt code, which then jumps to the PR handler,
* kvmppc_interrupt_pr, if the guest is a PR guest.
*/
#define kvmppc_interrupt kvmppc_interrupt_hv
#else
#define kvmppc_interrupt kvmppc_interrupt_pr
#endif
.macro KVMTEST name, hsrr, n
lbz r10,HSTATE_IN_GUEST(r13)
cmpwi r10,0
bne \name\()_kvm
.endm
.macro KVM_HANDLER vec, hsrr, area, skip
.if \skip
cmpwi r10,KVM_GUEST_MODE_SKIP
beq 89f
.else
BEGIN_FTR_SECTION_NESTED(947)
ld r10,\area+EX_CFAR(r13)
std r10,HSTATE_CFAR(r13)
END_FTR_SECTION_NESTED(CPU_FTR_CFAR,CPU_FTR_CFAR,947)
.endif
BEGIN_FTR_SECTION_NESTED(948)
ld r10,\area+EX_PPR(r13)
std r10,HSTATE_PPR(r13)
END_FTR_SECTION_NESTED(CPU_FTR_HAS_PPR,CPU_FTR_HAS_PPR,948)
ld r10,\area+EX_R10(r13)
std r12,HSTATE_SCRATCH0(r13)
sldi r12,r9,32
/* HSRR variants have the 0x2 bit added to their trap number */
.if \hsrr == EXC_HV_OR_STD
BEGIN_FTR_SECTION
ori r12,r12,(\vec + 0x2)
FTR_SECTION_ELSE
ori r12,r12,(\vec)
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
.elseif \hsrr
ori r12,r12,(\vec + 0x2)
.else
ori r12,r12,(\vec)
.endif
#ifdef CONFIG_RELOCATABLE
/*
* KVM requires __LOAD_FAR_HANDLER beause kvmppc_interrupt lives
* outside the head section. CONFIG_RELOCATABLE KVM expects CTR
* to be saved in HSTATE_SCRATCH1.
*/
mfctr r9
std r9,HSTATE_SCRATCH1(r13)
__LOAD_FAR_HANDLER(r9, kvmppc_interrupt)
mtctr r9
ld r9,\area+EX_R9(r13)
bctr
#else
ld r9,\area+EX_R9(r13)
b kvmppc_interrupt
#endif
.if \skip
89: mtocrf 0x80,r9
ld r9,\area+EX_R9(r13)
ld r10,\area+EX_R10(r13)
.if \hsrr == EXC_HV_OR_STD
BEGIN_FTR_SECTION
b kvmppc_skip_Hinterrupt
FTR_SECTION_ELSE
b kvmppc_skip_interrupt
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
.elseif \hsrr
b kvmppc_skip_Hinterrupt
.else
b kvmppc_skip_interrupt
.endif
.endif
.endm
#else
.macro KVMTEST name, hsrr, n
.endm
.macro KVM_HANDLER name, vec, hsrr, area, skip
.endm
#endif
.macro INT_SAVE_SRR_AND_JUMP label, hsrr, set_ri
ld r10,PACAKMSR(r13) /* get MSR value for kernel */
.if ! \set_ri
xori r10,r10,MSR_RI /* Clear MSR_RI */
.endif
.if \hsrr == EXC_HV_OR_STD
BEGIN_FTR_SECTION
mfspr r11,SPRN_HSRR0 /* save HSRR0 */
mfspr r12,SPRN_HSRR1 /* and HSRR1 */
mtspr SPRN_HSRR1,r10
FTR_SECTION_ELSE
mfspr r11,SPRN_SRR0 /* save SRR0 */
mfspr r12,SPRN_SRR1 /* and SRR1 */
mtspr SPRN_SRR1,r10
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
.elseif \hsrr
mfspr r11,SPRN_HSRR0 /* save HSRR0 */
mfspr r12,SPRN_HSRR1 /* and HSRR1 */
mtspr SPRN_HSRR1,r10
.else
mfspr r11,SPRN_SRR0 /* save SRR0 */
mfspr r12,SPRN_SRR1 /* and SRR1 */
mtspr SPRN_SRR1,r10
.endif
LOAD_HANDLER(r10, \label\())
.if \hsrr == EXC_HV_OR_STD
BEGIN_FTR_SECTION
mtspr SPRN_HSRR0,r10
HRFI_TO_KERNEL
FTR_SECTION_ELSE
mtspr SPRN_SRR0,r10
RFI_TO_KERNEL
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
.elseif \hsrr
mtspr SPRN_HSRR0,r10
HRFI_TO_KERNEL
.else
mtspr SPRN_SRR0,r10
RFI_TO_KERNEL
.endif
b . /* prevent speculative execution */
.endm
/* INT_SAVE_SRR_AND_JUMP works for real or virt, this is faster but virt only */
.macro INT_VIRT_SAVE_SRR_AND_JUMP label, hsrr
#ifdef CONFIG_RELOCATABLE
.if \hsrr == EXC_HV_OR_STD
BEGIN_FTR_SECTION
mfspr r11,SPRN_HSRR0 /* save HSRR0 */
FTR_SECTION_ELSE
mfspr r11,SPRN_SRR0 /* save SRR0 */
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
.elseif \hsrr
mfspr r11,SPRN_HSRR0 /* save HSRR0 */
.else
mfspr r11,SPRN_SRR0 /* save SRR0 */
.endif
LOAD_HANDLER(r12, \label\())
mtctr r12
.if \hsrr == EXC_HV_OR_STD
BEGIN_FTR_SECTION
mfspr r12,SPRN_HSRR1 /* and HSRR1 */
FTR_SECTION_ELSE
mfspr r12,SPRN_SRR1 /* and HSRR1 */
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
.elseif \hsrr
mfspr r12,SPRN_HSRR1 /* and HSRR1 */
.else
mfspr r12,SPRN_SRR1 /* and HSRR1 */
.endif
li r10,MSR_RI
mtmsrd r10,1 /* Set RI (EE=0) */
bctr
#else
.if \hsrr == EXC_HV_OR_STD
BEGIN_FTR_SECTION
mfspr r11,SPRN_HSRR0 /* save HSRR0 */
mfspr r12,SPRN_HSRR1 /* and HSRR1 */
FTR_SECTION_ELSE
mfspr r11,SPRN_SRR0 /* save SRR0 */
mfspr r12,SPRN_SRR1 /* and SRR1 */
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
.elseif \hsrr
mfspr r11,SPRN_HSRR0 /* save HSRR0 */
mfspr r12,SPRN_HSRR1 /* and HSRR1 */
.else
mfspr r11,SPRN_SRR0 /* save SRR0 */
mfspr r12,SPRN_SRR1 /* and SRR1 */
.endif
li r10,MSR_RI
mtmsrd r10,1 /* Set RI (EE=0) */
b \label
#endif
.endm
/*
* This is the BOOK3S interrupt entry code macro.
*
* This can result in one of several things happening:
* - Branch to the _common handler, relocated, in virtual mode.
* These are normal interrupts (synchronous and asynchronous) handled by
* the kernel.
* - Branch to KVM, relocated but real mode interrupts remain in real mode.
* These occur when HSTATE_IN_GUEST is set. The interrupt may be caused by
* / intended for host or guest kernel, but KVM must always be involved
* because the machine state is set for guest execution.
* - Branch to the masked handler, unrelocated.
* These occur when maskable asynchronous interrupts are taken with the
* irq_soft_mask set.
* - Branch to an "early" handler in real mode but relocated.
* This is done if early=1. MCE and HMI use these to handle errors in real
* mode.
* - Fall through and continue executing in real, unrelocated mode.
* This is done if early=2.
*/
.macro INT_HANDLER name, vec, ool=0, early=0, virt=0, hsrr=0, area=PACA_EXGEN, ri=1, dar=0, dsisr=0, bitmask=0, kvm=0
SET_SCRATCH0(r13) /* save r13 */
GET_PACA(r13)
std r9,\area\()+EX_R9(r13) /* save r9 */
OPT_GET_SPR(r9, SPRN_PPR, CPU_FTR_HAS_PPR)
HMT_MEDIUM
std r10,\area\()+EX_R10(r13) /* save r10 - r12 */
OPT_GET_SPR(r10, SPRN_CFAR, CPU_FTR_CFAR)
.if \ool
.if !\virt
b tramp_real_\name
.pushsection .text
TRAMP_REAL_BEGIN(tramp_real_\name)
.else
b tramp_virt_\name
.pushsection .text
TRAMP_VIRT_BEGIN(tramp_virt_\name)
.endif
.endif
OPT_SAVE_REG_TO_PACA(\area\()+EX_PPR, r9, CPU_FTR_HAS_PPR)
OPT_SAVE_REG_TO_PACA(\area\()+EX_CFAR, r10, CPU_FTR_CFAR)
INTERRUPT_TO_KERNEL
SAVE_CTR(r10, \area\())
mfcr r9
.if \kvm
KVMTEST \name \hsrr \vec
.endif
.if \bitmask
lbz r10,PACAIRQSOFTMASK(r13)
andi. r10,r10,\bitmask
/* Associate vector numbers with bits in paca->irq_happened */
.if \vec == 0x500 || \vec == 0xea0
li r10,PACA_IRQ_EE
.elseif \vec == 0x900
li r10,PACA_IRQ_DEC
.elseif \vec == 0xa00 || \vec == 0xe80
li r10,PACA_IRQ_DBELL
.elseif \vec == 0xe60
li r10,PACA_IRQ_HMI
.elseif \vec == 0xf00
li r10,PACA_IRQ_PMI
.else
.abort "Bad maskable vector"
.endif
.if \hsrr == EXC_HV_OR_STD
BEGIN_FTR_SECTION
bne masked_Hinterrupt
FTR_SECTION_ELSE
bne masked_interrupt
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
.elseif \hsrr
bne masked_Hinterrupt
.else
bne masked_interrupt
.endif
.endif
std r11,\area\()+EX_R11(r13)
std r12,\area\()+EX_R12(r13)
/*
* DAR/DSISR, SCRATCH0 must be read before setting MSR[RI],
* because a d-side MCE will clobber those registers so is
* not recoverable if they are live.
*/
GET_SCRATCH0(r10)
std r10,\area\()+EX_R13(r13)
.if \dar
.if \hsrr
mfspr r10,SPRN_HDAR
.else
mfspr r10,SPRN_DAR
.endif
std r10,\area\()+EX_DAR(r13)
.endif
.if \dsisr
.if \hsrr
mfspr r10,SPRN_HDSISR
.else
mfspr r10,SPRN_DSISR
.endif
stw r10,\area\()+EX_DSISR(r13)
.endif
.if \early == 2
/* nothing more */
.elseif \early
mfctr r10 /* save ctr, even for !RELOCATABLE */
BRANCH_TO_C000(r11, \name\()_early_common)
.elseif !\virt
INT_SAVE_SRR_AND_JUMP \name\()_common, \hsrr, \ri
.else
INT_VIRT_SAVE_SRR_AND_JUMP \name\()_common, \hsrr
.endif
.if \ool
.popsection
.endif
.endm
/*
* On entry r13 points to the paca, r9-r13 are saved in the paca,
* r9 contains the saved CR, r11 and r12 contain the saved SRR0 and
* SRR1, and relocation is on.
*
* If stack=0, then the stack is already set in r1, and r1 is saved in r10.
* PPR save and CPU accounting is not done for the !stack case (XXX why not?)
*/
.macro INT_COMMON vec, area, stack, kuap, reconcile, dar, dsisr
.if \stack
andi. r10,r12,MSR_PR /* See if coming from user */
mr r10,r1 /* Save r1 */
subi r1,r1,INT_FRAME_SIZE /* alloc frame on kernel stack */
beq- 100f
ld r1,PACAKSAVE(r13) /* kernel stack to use */
100: tdgei r1,-INT_FRAME_SIZE /* trap if r1 is in userspace */
EMIT_BUG_ENTRY 100b,__FILE__,__LINE__,0
.endif
std r9,_CCR(r1) /* save CR in stackframe */
std r11,_NIP(r1) /* save SRR0 in stackframe */
std r12,_MSR(r1) /* save SRR1 in stackframe */
std r10,0(r1) /* make stack chain pointer */
std r0,GPR0(r1) /* save r0 in stackframe */
std r10,GPR1(r1) /* save r1 in stackframe */
.if \stack
.if \kuap
kuap_save_amr_and_lock r9, r10, cr1, cr0
.endif
beq 101f /* if from kernel mode */
ACCOUNT_CPU_USER_ENTRY(r13, r9, r10)
SAVE_PPR(\area, r9)
101:
.else
.if \kuap
kuap_save_amr_and_lock r9, r10, cr1
.endif
.endif
/* Save original regs values from save area to stack frame. */
ld r9,\area+EX_R9(r13) /* move r9, r10 to stackframe */
ld r10,\area+EX_R10(r13)
std r9,GPR9(r1)
std r10,GPR10(r1)
ld r9,\area+EX_R11(r13) /* move r11 - r13 to stackframe */
ld r10,\area+EX_R12(r13)
ld r11,\area+EX_R13(r13)
std r9,GPR11(r1)
std r10,GPR12(r1)
std r11,GPR13(r1)
.if \dar
.if \dar == 2
ld r10,_NIP(r1)
.else
ld r10,\area+EX_DAR(r13)
.endif
std r10,_DAR(r1)
.endif
.if \dsisr
.if \dsisr == 2
ld r10,_MSR(r1)
lis r11,DSISR_SRR1_MATCH_64S@h
and r10,r10,r11
.else
lwz r10,\area+EX_DSISR(r13)
.endif
std r10,_DSISR(r1)
.endif
BEGIN_FTR_SECTION_NESTED(66)
ld r10,\area+EX_CFAR(r13)
std r10,ORIG_GPR3(r1)
END_FTR_SECTION_NESTED(CPU_FTR_CFAR, CPU_FTR_CFAR, 66)
GET_CTR(r10, \area)
std r10,_CTR(r1)
std r2,GPR2(r1) /* save r2 in stackframe */
SAVE_4GPRS(3, r1) /* save r3 - r6 in stackframe */
SAVE_2GPRS(7, r1) /* save r7, r8 in stackframe */
mflr r9 /* Get LR, later save to stack */
ld r2,PACATOC(r13) /* get kernel TOC into r2 */
std r9,_LINK(r1)
lbz r10,PACAIRQSOFTMASK(r13)
mfspr r11,SPRN_XER /* save XER in stackframe */
std r10,SOFTE(r1)
std r11,_XER(r1)
li r9,(\vec)+1
std r9,_TRAP(r1) /* set trap number */
li r10,0
ld r11,exception_marker@toc(r2)
std r10,RESULT(r1) /* clear regs->result */
std r11,STACK_FRAME_OVERHEAD-16(r1) /* mark the frame */
.if \stack
ACCOUNT_STOLEN_TIME
.endif
.if \reconcile
RECONCILE_IRQ_STATE(r10, r11)
.endif
.endm
/*
* Restore all registers including H/SRR0/1 saved in a stack frame of a
* standard exception.
*/
.macro EXCEPTION_RESTORE_REGS hsrr
/* Move original SRR0 and SRR1 into the respective regs */
ld r9,_MSR(r1)
.if \hsrr == EXC_HV_OR_STD
.error "EXC_HV_OR_STD Not implemented for EXCEPTION_RESTORE_REGS"
.endif
.if \hsrr
mtspr SPRN_HSRR1,r9
.else
mtspr SPRN_SRR1,r9
.endif
ld r9,_NIP(r1)
.if \hsrr
mtspr SPRN_HSRR0,r9
.else
mtspr SPRN_SRR0,r9
.endif
ld r9,_CTR(r1)
mtctr r9
ld r9,_XER(r1)
mtxer r9
ld r9,_LINK(r1)
mtlr r9
ld r9,_CCR(r1)
mtcr r9
REST_8GPRS(2, r1)
REST_4GPRS(10, r1)
REST_GPR(0, r1)
/* restore original r1. */
ld r1,GPR1(r1)
.endm
#define RUNLATCH_ON \
BEGIN_FTR_SECTION \
ld r3, PACA_THREAD_INFO(r13); \
ld r4,TI_LOCAL_FLAGS(r3); \
andi. r0,r4,_TLF_RUNLATCH; \
beql ppc64_runlatch_on_trampoline; \
END_FTR_SECTION_IFSET(CPU_FTR_CTRL)
/*
* When the idle code in power4_idle puts the CPU into NAP mode,
* it has to do so in a loop, and relies on the external interrupt
* and decrementer interrupt entry code to get it out of the loop.
* It sets the _TLF_NAPPING bit in current_thread_info()->local_flags
* to signal that it is in the loop and needs help to get out.
*/
#ifdef CONFIG_PPC_970_NAP
#define FINISH_NAP \
BEGIN_FTR_SECTION \
ld r11, PACA_THREAD_INFO(r13); \
ld r9,TI_LOCAL_FLAGS(r11); \
andi. r10,r9,_TLF_NAPPING; \
bnel power4_fixup_nap; \
END_FTR_SECTION_IFSET(CPU_FTR_CAN_NAP)
#else
#define FINISH_NAP
#endif
#define EXC_COMMON(name, realvec, hdlr) \
EXC_COMMON_BEGIN(name); \
INT_COMMON realvec, PACA_EXGEN, 1, 1, 1, 0, 0 ; \
bl save_nvgprs; \
addi r3,r1,STACK_FRAME_OVERHEAD; \
bl hdlr; \
b ret_from_except
/*
* Like EXC_COMMON, but for exceptions that can occur in the idle task and
* therefore need the special idle handling (finish nap and runlatch)
*/
#define EXC_COMMON_ASYNC(name, realvec, hdlr) \
EXC_COMMON_BEGIN(name); \
INT_COMMON realvec, PACA_EXGEN, 1, 1, 1, 0, 0 ; \
FINISH_NAP; \
RUNLATCH_ON; \
addi r3,r1,STACK_FRAME_OVERHEAD; \
bl hdlr; \
b ret_from_except_lite
/*
* There are a few constraints to be concerned with.
* - Real mode exceptions code/data must be located at their physical location.
* - Virtual mode exceptions must be mapped at their 0xc000... location.
* - Fixed location code must not call directly beyond the __end_interrupts
* area when built with CONFIG_RELOCATABLE. LOAD_HANDLER / bctr sequence
* must be used.
* - LOAD_HANDLER targets must be within first 64K of physical 0 /
* virtual 0xc00...
* - Conditional branch targets must be within +/-32K of caller.
*
* "Virtual exceptions" run with relocation on (MSR_IR=1, MSR_DR=1), and
* therefore don't have to run in physically located code or rfid to
* virtual mode kernel code. However on relocatable kernels they do have
* to branch to KERNELBASE offset because the rest of the kernel (outside
* the exception vectors) may be located elsewhere.
*
* Virtual exceptions correspond with physical, except their entry points
* are offset by 0xc000000000000000 and also tend to get an added 0x4000
* offset applied. Virtual exceptions are enabled with the Alternate
* Interrupt Location (AIL) bit set in the LPCR. However this does not
* guarantee they will be delivered virtually. Some conditions (see the ISA)
* cause exceptions to be delivered in real mode.
*
* It's impossible to receive interrupts below 0x300 via AIL.
*
* KVM: None of the virtual exceptions are from the guest. Anything that
* escalated to HV=1 from HV=0 is delivered via real mode handlers.
*
*
* We layout physical memory as follows:
* 0x0000 - 0x00ff : Secondary processor spin code
* 0x0100 - 0x18ff : Real mode pSeries interrupt vectors
* 0x1900 - 0x3fff : Real mode trampolines
* 0x4000 - 0x58ff : Relon (IR=1,DR=1) mode pSeries interrupt vectors
* 0x5900 - 0x6fff : Relon mode trampolines
* 0x7000 - 0x7fff : FWNMI data area
* 0x8000 - .... : Common interrupt handlers, remaining early
* setup code, rest of kernel.
*
* We could reclaim 0x4000-0x42ff for real mode trampolines if the space
* is necessary. Until then it's more consistent to explicitly put VIRT_NONE
* vectors there.
*/
OPEN_FIXED_SECTION(real_vectors, 0x0100, 0x1900)
OPEN_FIXED_SECTION(real_trampolines, 0x1900, 0x4000)
OPEN_FIXED_SECTION(virt_vectors, 0x4000, 0x5900)
OPEN_FIXED_SECTION(virt_trampolines, 0x5900, 0x7000)
#ifdef CONFIG_PPC_POWERNV
.globl start_real_trampolines
.globl end_real_trampolines
.globl start_virt_trampolines
.globl end_virt_trampolines
#endif
#if defined(CONFIG_PPC_PSERIES) || defined(CONFIG_PPC_POWERNV)
/*
* Data area reserved for FWNMI option.
* This address (0x7000) is fixed by the RPA.
* pseries and powernv need to keep the whole page from
* 0x7000 to 0x8000 free for use by the firmware
*/
ZERO_FIXED_SECTION(fwnmi_page, 0x7000, 0x8000)
OPEN_TEXT_SECTION(0x8000)
#else
OPEN_TEXT_SECTION(0x7000)
#endif
USE_FIXED_SECTION(real_vectors)
/*
* This is the start of the interrupt handlers for pSeries
* This code runs with relocation off.
* Code from here to __end_interrupts gets copied down to real
* address 0x100 when we are running a relocatable kernel.
* Therefore any relative branches in this section must only
* branch to labels in this section.
*/
.globl __start_interrupts
__start_interrupts:
/* No virt vectors corresponding with 0x0..0x100 */
EXC_VIRT_NONE(0x4000, 0x100)
EXC_REAL_BEGIN(system_reset, 0x100, 0x100)
#ifdef CONFIG_PPC_P7_NAP
/*
* If running native on arch 2.06 or later, check if we are waking up
* from nap/sleep/winkle, and branch to idle handler. This tests SRR1
* bits 46:47. A non-0 value indicates that we are coming from a power
* saving state. The idle wakeup handler initially runs in real mode,
* but we branch to the 0xc000... address so we can turn on relocation
* with mtmsrd later, after SPRs are restored.
*
* Careful to minimise cost for the fast path (idle wakeup) while
* also avoiding clobbering CFAR for the debug path (non-idle).
*
* For the idle wake case volatile registers can be clobbered, which
* is why we use those initially. If it turns out to not be an idle
* wake, carefully put everything back the way it was, so we can use
* common exception macros to handle it.
*/
BEGIN_FTR_SECTION
SET_SCRATCH0(r13)
GET_PACA(r13)
std r3,PACA_EXNMI+0*8(r13)
std r4,PACA_EXNMI+1*8(r13)
std r5,PACA_EXNMI+2*8(r13)
mfspr r3,SPRN_SRR1
mfocrf r4,0x80
rlwinm. r5,r3,47-31,30,31
bne+ system_reset_idle_wake
/* Not powersave wakeup. Restore regs for regular interrupt handler. */
mtocrf 0x80,r4
ld r3,PACA_EXNMI+0*8(r13)
ld r4,PACA_EXNMI+1*8(r13)
ld r5,PACA_EXNMI+2*8(r13)
GET_SCRATCH0(r13)
END_FTR_SECTION_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
#endif
KVM: PPC: Allow book3s_hv guests to use SMT processor modes This lifts the restriction that book3s_hv guests can only run one hardware thread per core, and allows them to use up to 4 threads per core on POWER7. The host still has to run single-threaded. This capability is advertised to qemu through a new KVM_CAP_PPC_SMT capability. The return value of the ioctl querying this capability is the number of vcpus per virtual CPU core (vcore), currently 4. To use this, the host kernel should be booted with all threads active, and then all the secondary threads should be offlined. This will put the secondary threads into nap mode. KVM will then wake them from nap mode and use them for running guest code (while they are still offline). To wake the secondary threads, we send them an IPI using a new xics_wake_cpu() function, implemented in arch/powerpc/sysdev/xics/icp-native.c. In other words, at this stage we assume that the platform has a XICS interrupt controller and we are using icp-native.c to drive it. Since the woken thread will need to acknowledge and clear the IPI, we also export the base physical address of the XICS registers using kvmppc_set_xics_phys() for use in the low-level KVM book3s code. When a vcpu is created, it is assigned to a virtual CPU core. The vcore number is obtained by dividing the vcpu number by the number of threads per core in the host. This number is exported to userspace via the KVM_CAP_PPC_SMT capability. If qemu wishes to run the guest in single-threaded mode, it should make all vcpu numbers be multiples of the number of threads per core. We distinguish three states of a vcpu: runnable (i.e., ready to execute the guest), blocked (that is, idle), and busy in host. We currently implement a policy that the vcore can run only when all its threads are runnable or blocked. This way, if a vcpu needs to execute elsewhere in the kernel or in qemu, it can do so without being starved of CPU by the other vcpus. When a vcore starts to run, it executes in the context of one of the vcpu threads. The other vcpu threads all go to sleep and stay asleep until something happens requiring the vcpu thread to return to qemu, or to wake up to run the vcore (this can happen when another vcpu thread goes from busy in host state to blocked). It can happen that a vcpu goes from blocked to runnable state (e.g. because of an interrupt), and the vcore it belongs to is already running. In that case it can start to run immediately as long as the none of the vcpus in the vcore have started to exit the guest. We send the next free thread in the vcore an IPI to get it to start to execute the guest. It synchronizes with the other threads via the vcore->entry_exit_count field to make sure that it doesn't go into the guest if the other vcpus are exiting by the time that it is ready to actually enter the guest. Note that there is no fixed relationship between the hardware thread number and the vcpu number. Hardware threads are assigned to vcpus as they become runnable, so we will always use the lower-numbered hardware threads in preference to higher-numbered threads if not all the vcpus in the vcore are runnable, regardless of which vcpus are runnable. Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Alexander Graf <agraf@suse.de>
2011-06-29 08:23:08 +08:00
INT_HANDLER system_reset, 0x100, area=PACA_EXNMI, ri=0, kvm=1
/*
* MSR_RI is not enabled, because PACA_EXNMI and nmi stack is
* being used, so a nested NMI exception would corrupt it.
*
* In theory, we should not enable relocation here if it was disabled
* in SRR1, because the MMU may not be configured to support it (e.g.,
* SLB may have been cleared). In practice, there should only be a few
* small windows where that's the case, and sreset is considered to
* be dangerous anyway.
*/
EXC_REAL_END(system_reset, 0x100, 0x100)
EXC_VIRT_NONE(0x4100, 0x100)
INT_KVM_HANDLER system_reset 0x100, EXC_STD, PACA_EXNMI, 0
#ifdef CONFIG_PPC_P7_NAP
TRAMP_REAL_BEGIN(system_reset_idle_wake)
/* We are waking up from idle, so may clobber any volatile register */
cmpwi cr1,r5,2
bltlr cr1 /* no state loss, return to idle caller with r3=SRR1 */
BRANCH_TO_C000(r12, DOTSYM(idle_return_gpr_loss))
KVM: PPC: Allow book3s_hv guests to use SMT processor modes This lifts the restriction that book3s_hv guests can only run one hardware thread per core, and allows them to use up to 4 threads per core on POWER7. The host still has to run single-threaded. This capability is advertised to qemu through a new KVM_CAP_PPC_SMT capability. The return value of the ioctl querying this capability is the number of vcpus per virtual CPU core (vcore), currently 4. To use this, the host kernel should be booted with all threads active, and then all the secondary threads should be offlined. This will put the secondary threads into nap mode. KVM will then wake them from nap mode and use them for running guest code (while they are still offline). To wake the secondary threads, we send them an IPI using a new xics_wake_cpu() function, implemented in arch/powerpc/sysdev/xics/icp-native.c. In other words, at this stage we assume that the platform has a XICS interrupt controller and we are using icp-native.c to drive it. Since the woken thread will need to acknowledge and clear the IPI, we also export the base physical address of the XICS registers using kvmppc_set_xics_phys() for use in the low-level KVM book3s code. When a vcpu is created, it is assigned to a virtual CPU core. The vcore number is obtained by dividing the vcpu number by the number of threads per core in the host. This number is exported to userspace via the KVM_CAP_PPC_SMT capability. If qemu wishes to run the guest in single-threaded mode, it should make all vcpu numbers be multiples of the number of threads per core. We distinguish three states of a vcpu: runnable (i.e., ready to execute the guest), blocked (that is, idle), and busy in host. We currently implement a policy that the vcore can run only when all its threads are runnable or blocked. This way, if a vcpu needs to execute elsewhere in the kernel or in qemu, it can do so without being starved of CPU by the other vcpus. When a vcore starts to run, it executes in the context of one of the vcpu threads. The other vcpu threads all go to sleep and stay asleep until something happens requiring the vcpu thread to return to qemu, or to wake up to run the vcore (this can happen when another vcpu thread goes from busy in host state to blocked). It can happen that a vcpu goes from blocked to runnable state (e.g. because of an interrupt), and the vcore it belongs to is already running. In that case it can start to run immediately as long as the none of the vcpus in the vcore have started to exit the guest. We send the next free thread in the vcore an IPI to get it to start to execute the guest. It synchronizes with the other threads via the vcore->entry_exit_count field to make sure that it doesn't go into the guest if the other vcpus are exiting by the time that it is ready to actually enter the guest. Note that there is no fixed relationship between the hardware thread number and the vcpu number. Hardware threads are assigned to vcpus as they become runnable, so we will always use the lower-numbered hardware threads in preference to higher-numbered threads if not all the vcpus in the vcore are runnable, regardless of which vcpus are runnable. Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Alexander Graf <agraf@suse.de>
2011-06-29 08:23:08 +08:00
#endif
#ifdef CONFIG_PPC_PSERIES
/*
* Vectors for the FWNMI option. Share common code.
*/
TRAMP_REAL_BEGIN(system_reset_fwnmi)
/* See comment at system_reset exception, don't turn on RI */
INT_HANDLER system_reset, 0x100, area=PACA_EXNMI, ri=0
#endif /* CONFIG_PPC_PSERIES */
EXC_COMMON_BEGIN(system_reset_common)
/*
* Increment paca->in_nmi then enable MSR_RI. SLB or MCE will be able
* to recover, but nested NMI will notice in_nmi and not recover
* because of the use of the NMI stack. in_nmi reentrancy is tested in
* system_reset_exception.
*/
lhz r10,PACA_IN_NMI(r13)
addi r10,r10,1
sth r10,PACA_IN_NMI(r13)
li r10,MSR_RI
mtmsrd r10,1
mr r10,r1
ld r1,PACA_NMI_EMERG_SP(r13)
subi r1,r1,INT_FRAME_SIZE
INT_COMMON 0x100, PACA_EXNMI, 0, 1, 0, 0, 0
bl save_nvgprs
/*
* Set IRQS_ALL_DISABLED unconditionally so arch_irqs_disabled does
* the right thing. We do not want to reconcile because that goes
* through irq tracing which we don't want in NMI.
*
* Save PACAIRQHAPPENED because some code will do a hard disable
* (e.g., xmon). So we want to restore this back to where it was
* when we return. DAR is unused in the stack, so save it there.
*/
li r10,IRQS_ALL_DISABLED
stb r10,PACAIRQSOFTMASK(r13)
lbz r10,PACAIRQHAPPENED(r13)
std r10,_DAR(r1)
addi r3,r1,STACK_FRAME_OVERHEAD
bl system_reset_exception
/* Clear MSR_RI before setting SRR0 and SRR1. */
li r9,0
mtmsrd r9,1
/*
* MSR_RI is clear, now we can decrement paca->in_nmi.
*/
lhz r10,PACA_IN_NMI(r13)
subi r10,r10,1
sth r10,PACA_IN_NMI(r13)
/*
* Restore soft mask settings.
*/
ld r10,_DAR(r1)
stb r10,PACAIRQHAPPENED(r13)
ld r10,SOFTE(r1)
stb r10,PACAIRQSOFTMASK(r13)
EXCEPTION_RESTORE_REGS EXC_STD
RFI_TO_USER_OR_KERNEL
EXC_REAL_BEGIN(machine_check, 0x200, 0x100)
INT_HANDLER machine_check, 0x200, early=1, area=PACA_EXMC, dar=1, dsisr=1
/*
* MSR_RI is not enabled, because PACA_EXMC is being used, so a
* nested machine check corrupts it. machine_check_common enables
* MSR_RI.
*/
EXC_REAL_END(machine_check, 0x200, 0x100)
EXC_VIRT_NONE(0x4200, 0x100)
#ifdef CONFIG_PPC_PSERIES
TRAMP_REAL_BEGIN(machine_check_fwnmi)
/* See comment at machine_check exception, don't turn on RI */
INT_HANDLER machine_check, 0x200, early=1, area=PACA_EXMC, dar=1, dsisr=1
#endif
INT_KVM_HANDLER machine_check 0x200, EXC_STD, PACA_EXMC, 1
#define MACHINE_CHECK_HANDLER_WINDUP \
/* Clear MSR_RI before setting SRR0 and SRR1. */\
li r9,0; \
mtmsrd r9,1; /* Clear MSR_RI */ \
/* Decrement paca->in_mce now RI is clear. */ \
lhz r12,PACA_IN_MCE(r13); \
subi r12,r12,1; \
sth r12,PACA_IN_MCE(r13); \
EXCEPTION_RESTORE_REGS EXC_STD
EXC_COMMON_BEGIN(machine_check_early_common)
mtctr r10 /* Restore ctr */
mfspr r11,SPRN_SRR0
mfspr r12,SPRN_SRR1
/*
* Switch to mc_emergency stack and handle re-entrancy (we limit
* the nested MCE upto level 4 to avoid stack overflow).
* Save MCE registers srr1, srr0, dar and dsisr and then set ME=1
*
* We use paca->in_mce to check whether this is the first entry or
* nested machine check. We increment paca->in_mce to track nested
* machine checks.
*
* If this is the first entry then set stack pointer to
* paca->mc_emergency_sp, otherwise r1 is already pointing to
* stack frame on mc_emergency stack.
*
* NOTE: We are here with MSR_ME=0 (off), which means we risk a
* checkstop if we get another machine check exception before we do
* rfid with MSR_ME=1.
2017-04-19 21:05:47 +08:00
*
* This interrupt can wake directly from idle. If that is the case,
* the machine check is handled then the idle wakeup code is called
* to restore state.
*/
lhz r10,PACA_IN_MCE(r13)
cmpwi r10,0 /* Are we in nested machine check */
cmpwi cr1,r10,MAX_MCE_DEPTH /* Are we at maximum nesting */
addi r10,r10,1 /* increment paca->in_mce */
sth r10,PACA_IN_MCE(r13)
mr r10,r1 /* Save r1 */
bne 1f
/* First machine check entry */
ld r1,PACAMCEMERGSP(r13) /* Use MC emergency stack */
1: /* Limit nested MCE to level 4 to avoid stack overflow */
bgt cr1,unrecoverable_mce /* Check if we hit limit of 4 */
subi r1,r1,INT_FRAME_SIZE /* alloc stack frame */
/* We don't touch AMR here, we never go to virtual mode */
INT_COMMON 0x200, PACA_EXMC, 0, 0, 0, 1, 1
BEGIN_FTR_SECTION
bl enable_machine_check
END_FTR_SECTION_IFSET(CPU_FTR_HVMODE)
li r10,MSR_RI
mtmsrd r10,1
bl save_nvgprs
addi r3,r1,STACK_FRAME_OVERHEAD
bl machine_check_early
std r3,RESULT(r1) /* Save result */
ld r12,_MSR(r1)
2017-04-19 21:05:47 +08:00
#ifdef CONFIG_PPC_P7_NAP
/*
* Check if thread was in power saving mode. We come here when any
* of the following is true:
* a. thread wasn't in power saving mode
* b. thread was in power saving mode with no state loss,
* supervisor state loss or hypervisor state loss.
*
* Go back to nap/sleep/winkle mode again if (b) is true.
*/
BEGIN_FTR_SECTION
2017-04-19 21:05:47 +08:00
rlwinm. r11,r12,47-31,30,31
bne machine_check_idle_common
END_FTR_SECTION_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
#endif
2017-04-19 21:05:47 +08:00
#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
/*
* Check if we are coming from guest. If yes, then run the normal
* exception handler which will take the
* machine_check_kvm->kvmppc_interrupt branch to deliver the MC event
* to guest.
*/
lbz r11,HSTATE_IN_GUEST(r13)
cmpwi r11,0 /* Check if coming from guest */
bne mce_deliver /* continue if we are. */
#endif
/*
* Check if we are coming from userspace. If yes, then run the normal
* exception handler which will deliver the MC event to this kernel.
*/
andi. r11,r12,MSR_PR /* See if coming from user. */
bne mce_deliver /* continue in V mode if we are. */
/*
* At this point we are coming from kernel context.
* Queue up the MCE event and return from the interrupt.
* But before that, check if this is an un-recoverable exception.
* If yes, then stay on emergency stack and panic.
*/
andi. r11,r12,MSR_RI
beq unrecoverable_mce
/*
* Check if we have successfully handled/recovered from error, if not
* then stay on emergency stack and panic.
*/
ld r3,RESULT(r1) /* Load result */
cmpdi r3,0 /* see if we handled MCE successfully */
beq unrecoverable_mce /* if !handled then panic */
/*
* Return from MC interrupt.
* Queue up the MCE event so that we can log it later, while
* returning from kernel or opal call.
*/
bl machine_check_queue_event
MACHINE_CHECK_HANDLER_WINDUP
RFI_TO_KERNEL
mce_deliver:
/*
* This is a host user or guest MCE. Restore all registers, then
* run the "late" handler. For host user, this will run the
* machine_check_exception handler in virtual mode like a normal
* interrupt handler. For guest, this will trigger the KVM test
* and branch to the KVM interrupt similarly to other interrupts.
*/
BEGIN_FTR_SECTION
ld r10,ORIG_GPR3(r1)
mtspr SPRN_CFAR,r10
END_FTR_SECTION_IFSET(CPU_FTR_CFAR)
MACHINE_CHECK_HANDLER_WINDUP
/* See comment at machine_check exception, don't turn on RI */
INT_HANDLER machine_check, 0x200, area=PACA_EXMC, ri=0, dar=1, dsisr=1, kvm=1
EXC_COMMON_BEGIN(machine_check_common)
/*
* Machine check is different because we use a different
* save area: PACA_EXMC instead of PACA_EXGEN.
*/
INT_COMMON 0x200, PACA_EXMC, 1, 1, 1, 1, 1
FINISH_NAP
/* Enable MSR_RI when finished with PACA_EXMC */
li r10,MSR_RI
mtmsrd r10,1
bl save_nvgprs
addi r3,r1,STACK_FRAME_OVERHEAD
bl machine_check_exception
b ret_from_except
#ifdef CONFIG_PPC_P7_NAP
/*
* This is an idle wakeup. Low level machine check has already been
* done. Queue the event then call the idle code to do the wake up.
*/
EXC_COMMON_BEGIN(machine_check_idle_common)
bl machine_check_queue_event
/*
* We have not used any non-volatile GPRs here, and as a rule
* most exception code including machine check does not.
* Therefore PACA_NAPSTATELOST does not need to be set. Idle
* wakeup will restore volatile registers.
*
* Load the original SRR1 into r3 for pnv_powersave_wakeup_mce.
*
* Then decrement MCE nesting after finishing with the stack.
*/
ld r3,_MSR(r1)
ld r4,_LINK(r1)
lhz r11,PACA_IN_MCE(r13)
subi r11,r11,1
sth r11,PACA_IN_MCE(r13)
mtlr r4
rlwinm r10,r3,47-31,30,31
cmpwi cr1,r10,2
bltlr cr1 /* no state loss, return to idle caller */
b idle_return_gpr_loss
#endif
EXC_COMMON_BEGIN(unrecoverable_mce)
/*
* We are going down. But there are chances that we might get hit by
* another MCE during panic path and we may run into unstable state
* with no way out. Hence, turn ME bit off while going down, so that
* when another MCE is hit during panic path, system will checkstop
* and hypervisor will get restarted cleanly by SP.
*/
BEGIN_FTR_SECTION
li r10,0 /* clear MSR_RI */
mtmsrd r10,1
bl disable_machine_check
END_FTR_SECTION_IFSET(CPU_FTR_HVMODE)
ld r10,PACAKMSR(r13)
li r3,MSR_ME
andc r10,r10,r3
mtmsrd r10
/* Invoke machine_check_exception to print MCE event and panic. */
addi r3,r1,STACK_FRAME_OVERHEAD
bl machine_check_exception
/*
* We will not reach here. Even if we did, there is no way out.
* Call unrecoverable_exception and die.
*/
addi r3,r1,STACK_FRAME_OVERHEAD
bl unrecoverable_exception
b .
EXC_REAL_BEGIN(data_access, 0x300, 0x80)
INT_HANDLER data_access, 0x300, ool=1, dar=1, dsisr=1, kvm=1
EXC_REAL_END(data_access, 0x300, 0x80)
EXC_VIRT_BEGIN(data_access, 0x4300, 0x80)
INT_HANDLER data_access, 0x300, virt=1, dar=1, dsisr=1
EXC_VIRT_END(data_access, 0x4300, 0x80)
INT_KVM_HANDLER data_access, 0x300, EXC_STD, PACA_EXGEN, 1
EXC_COMMON_BEGIN(data_access_common)
/*
* Here r13 points to the paca, r9 contains the saved CR,
* SRR0 and SRR1 are saved in r11 and r12,
* r9 - r13 are saved in paca->exgen.
* EX_DAR and EX_DSISR have saved DAR/DSISR
*/
INT_COMMON 0x300, PACA_EXGEN, 1, 1, 1, 1, 1
ld r4,_DAR(r1)
ld r5,_DSISR(r1)
BEGIN_MMU_FTR_SECTION
ld r6,_MSR(r1)
li r3,0x300
b do_hash_page /* Try to handle as hpte fault */
MMU_FTR_SECTION_ELSE
b handle_page_fault
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_TYPE_RADIX)
EXC_REAL_BEGIN(data_access_slb, 0x380, 0x80)
INT_HANDLER data_access_slb, 0x380, ool=1, area=PACA_EXSLB, dar=1, kvm=1
EXC_REAL_END(data_access_slb, 0x380, 0x80)
EXC_VIRT_BEGIN(data_access_slb, 0x4380, 0x80)
INT_HANDLER data_access_slb, 0x380, virt=1, area=PACA_EXSLB, dar=1
EXC_VIRT_END(data_access_slb, 0x4380, 0x80)
INT_KVM_HANDLER data_access_slb, 0x380, EXC_STD, PACA_EXSLB, 1
EXC_COMMON_BEGIN(data_access_slb_common)
INT_COMMON 0x380, PACA_EXSLB, 1, 1, 0, 1, 0
ld r4,_DAR(r1)
addi r3,r1,STACK_FRAME_OVERHEAD
BEGIN_MMU_FTR_SECTION
/* HPT case, do SLB fault */
bl do_slb_fault
cmpdi r3,0
bne- 1f
b fast_exception_return
1: /* Error case */
MMU_FTR_SECTION_ELSE
/* Radix case, access is outside page table range */
li r3,-EFAULT
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_TYPE_RADIX)
std r3,RESULT(r1)
bl save_nvgprs
RECONCILE_IRQ_STATE(r10, r11)
ld r4,_DAR(r1)
ld r5,RESULT(r1)
addi r3,r1,STACK_FRAME_OVERHEAD
bl do_bad_slb_fault
b ret_from_except
EXC_REAL_BEGIN(instruction_access, 0x400, 0x80)
INT_HANDLER instruction_access, 0x400, kvm=1
EXC_REAL_END(instruction_access, 0x400, 0x80)
EXC_VIRT_BEGIN(instruction_access, 0x4400, 0x80)
INT_HANDLER instruction_access, 0x400, virt=1
EXC_VIRT_END(instruction_access, 0x4400, 0x80)
INT_KVM_HANDLER instruction_access, 0x400, EXC_STD, PACA_EXGEN, 0
EXC_COMMON_BEGIN(instruction_access_common)
INT_COMMON 0x400, PACA_EXGEN, 1, 1, 1, 2, 2
ld r4,_DAR(r1)
ld r5,_DSISR(r1)
BEGIN_MMU_FTR_SECTION
ld r6,_MSR(r1)
li r3,0x400
b do_hash_page /* Try to handle as hpte fault */
MMU_FTR_SECTION_ELSE
b handle_page_fault
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_TYPE_RADIX)
EXC_REAL_BEGIN(instruction_access_slb, 0x480, 0x80)
INT_HANDLER instruction_access_slb, 0x480, area=PACA_EXSLB, kvm=1
EXC_REAL_END(instruction_access_slb, 0x480, 0x80)
EXC_VIRT_BEGIN(instruction_access_slb, 0x4480, 0x80)
INT_HANDLER instruction_access_slb, 0x480, virt=1, area=PACA_EXSLB
EXC_VIRT_END(instruction_access_slb, 0x4480, 0x80)
INT_KVM_HANDLER instruction_access_slb, 0x480, EXC_STD, PACA_EXSLB, 0
EXC_COMMON_BEGIN(instruction_access_slb_common)
INT_COMMON 0x480, PACA_EXSLB, 1, 1, 0, 2, 0
ld r4,_DAR(r1)
addi r3,r1,STACK_FRAME_OVERHEAD
BEGIN_MMU_FTR_SECTION
/* HPT case, do SLB fault */
bl do_slb_fault
cmpdi r3,0
bne- 1f
b fast_exception_return
1: /* Error case */
MMU_FTR_SECTION_ELSE
/* Radix case, access is outside page table range */
li r3,-EFAULT
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_TYPE_RADIX)
std r3,RESULT(r1)
bl save_nvgprs
RECONCILE_IRQ_STATE(r10, r11)
ld r4,_DAR(r1)
ld r5,RESULT(r1)
addi r3,r1,STACK_FRAME_OVERHEAD
bl do_bad_slb_fault
b ret_from_except
EXC_REAL_BEGIN(hardware_interrupt, 0x500, 0x100)
INT_HANDLER hardware_interrupt, 0x500, hsrr=EXC_HV_OR_STD, bitmask=IRQS_DISABLED, kvm=1
EXC_REAL_END(hardware_interrupt, 0x500, 0x100)
EXC_VIRT_BEGIN(hardware_interrupt, 0x4500, 0x100)
INT_HANDLER hardware_interrupt, 0x500, virt=1, hsrr=EXC_HV_OR_STD, bitmask=IRQS_DISABLED, kvm=1
EXC_VIRT_END(hardware_interrupt, 0x4500, 0x100)
INT_KVM_HANDLER hardware_interrupt, 0x500, EXC_HV_OR_STD, PACA_EXGEN, 0
EXC_COMMON_ASYNC(hardware_interrupt_common, 0x500, do_IRQ)
EXC_REAL_BEGIN(alignment, 0x600, 0x100)
INT_HANDLER alignment, 0x600, dar=1, dsisr=1, kvm=1
EXC_REAL_END(alignment, 0x600, 0x100)
EXC_VIRT_BEGIN(alignment, 0x4600, 0x100)
INT_HANDLER alignment, 0x600, virt=1, dar=1, dsisr=1
EXC_VIRT_END(alignment, 0x4600, 0x100)
INT_KVM_HANDLER alignment, 0x600, EXC_STD, PACA_EXGEN, 0
EXC_COMMON_BEGIN(alignment_common)
INT_COMMON 0x600, PACA_EXGEN, 1, 1, 1, 1, 1
bl save_nvgprs
addi r3,r1,STACK_FRAME_OVERHEAD
bl alignment_exception
b ret_from_except
EXC_REAL_BEGIN(program_check, 0x700, 0x100)
INT_HANDLER program_check, 0x700, kvm=1
EXC_REAL_END(program_check, 0x700, 0x100)
EXC_VIRT_BEGIN(program_check, 0x4700, 0x100)
INT_HANDLER program_check, 0x700, virt=1
EXC_VIRT_END(program_check, 0x4700, 0x100)
INT_KVM_HANDLER program_check, 0x700, EXC_STD, PACA_EXGEN, 0
EXC_COMMON_BEGIN(program_check_common)
powerpc/64s: Use emergency stack for kernel TM Bad Thing program checks When using transactional memory (TM), the CPU can be in one of six states as far as TM is concerned, encoded in the Machine State Register (MSR). Certain state transitions are illegal and if attempted trigger a "TM Bad Thing" type program check exception. If we ever hit one of these exceptions it's treated as a bug, ie. we oops, and kill the process and/or panic, depending on configuration. One case where we can trigger a TM Bad Thing, is when returning to userspace after a system call or interrupt, using RFID. When this happens the CPU first restores the user register state, in particular r1 (the stack pointer) and then attempts to update the MSR. However the MSR update is not allowed and so we take the program check with the user register state, but the kernel MSR. This tricks the exception entry code into thinking we have a bad kernel stack pointer, because the MSR says we're coming from the kernel, but r1 is pointing to userspace. To avoid this we instead always switch to the emergency stack if we take a TM Bad Thing from the kernel. That way none of the user register values are used, other than for printing in the oops message. This is the fix for CVE-2017-1000255. Fixes: 5d176f751ee3 ("powerpc: tm: Enable transactional memory (TM) lazily for userspace") Cc: stable@vger.kernel.org # v4.9+ Signed-off-by: Cyril Bur <cyrilbur@gmail.com> [mpe: Rewrite change log & comments, tweak asm slightly] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2017-08-17 18:42:26 +08:00
/*
* It's possible to receive a TM Bad Thing type program check with
* userspace register values (in particular r1), but with SRR1 reporting
* that we came from the kernel. Normally that would confuse the bad
* stack logic, and we would report a bad kernel stack pointer. Instead
* we switch to the emergency stack if we're taking a TM Bad Thing from
* the kernel.
*/
powerpc/64s/exception: remove bad stack branch The bad stack test in interrupt handlers has a few problems. For performance it is taken in the common case, which is a fetch bubble and a waste of i-cache. For code development and maintainence, it requires yet another stack frame setup routine, and that constrains all exception handlers to follow the same register save pattern which inhibits future optimisation. Remove the test/branch and replace it with a trap. Teach the program check handler to use the emergency stack for this case. This does not result in quite so nice a message, however the SRR0 and SRR1 of the crashed interrupt can be seen in r11 and r12, as is the original r1 (adjusted by INT_FRAME_SIZE). These are the most important parts to debugging the issue. The original r9-12 and cr0 is lost, which is the main downside. kernel BUG at linux/arch/powerpc/kernel/exceptions-64s.S:847! Oops: Exception in kernel mode, sig: 5 [#1] BE SMP NR_CPUS=2048 NUMA PowerNV Modules linked in: CPU: 0 PID: 1 Comm: swapper/0 Not tainted NIP: c000000000009108 LR: c000000000cadbcc CTR: c0000000000090f0 REGS: c0000000fffcbd70 TRAP: 0700 Not tainted MSR: 9000000000021032 <SF,HV,ME,IR,DR,RI> CR: 28222448 XER: 20040000 CFAR: c000000000009100 IRQMASK: 0 GPR00: 000000000000003d fffffffffffffd00 c0000000018cfb00 c0000000f02b3166 GPR04: fffffffffffffffd 0000000000000007 fffffffffffffffb 0000000000000030 GPR08: 0000000000000037 0000000028222448 0000000000000000 c000000000ca8de0 GPR12: 9000000002009032 c000000001ae0000 c000000000010a00 0000000000000000 GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000 GPR20: c0000000f00322c0 c000000000f85200 0000000000000004 ffffffffffffffff GPR24: fffffffffffffffe 0000000000000000 0000000000000000 000000000000000a GPR28: 0000000000000000 0000000000000000 c0000000f02b391c c0000000f02b3167 NIP [c000000000009108] decrementer_common+0x18/0x160 LR [c000000000cadbcc] .vsnprintf+0x3ec/0x4f0 Call Trace: Instruction dump: 996d098a 994d098b 38610070 480246ed 48005518 60000000 38200000 718a4000 7c2a0b78 3821fd00 41c20008 e82d0970 <0981fd00> f92101a0 f9610170 f9810178 Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2019-06-28 14:33:18 +08:00
andi. r10,r12,MSR_PR
bne 2f /* If userspace, go normal path */
andis. r10,r12,(SRR1_PROGTM)@h
bne 1f /* If TM, emergency */
cmpdi r1,-INT_FRAME_SIZE /* check if r1 is in userspace */
blt 2f /* normal path if not */
/* Use the emergency stack */
1: andi. r10,r12,MSR_PR /* Set CR0 correctly for label */
powerpc/64s: Use emergency stack for kernel TM Bad Thing program checks When using transactional memory (TM), the CPU can be in one of six states as far as TM is concerned, encoded in the Machine State Register (MSR). Certain state transitions are illegal and if attempted trigger a "TM Bad Thing" type program check exception. If we ever hit one of these exceptions it's treated as a bug, ie. we oops, and kill the process and/or panic, depending on configuration. One case where we can trigger a TM Bad Thing, is when returning to userspace after a system call or interrupt, using RFID. When this happens the CPU first restores the user register state, in particular r1 (the stack pointer) and then attempts to update the MSR. However the MSR update is not allowed and so we take the program check with the user register state, but the kernel MSR. This tricks the exception entry code into thinking we have a bad kernel stack pointer, because the MSR says we're coming from the kernel, but r1 is pointing to userspace. To avoid this we instead always switch to the emergency stack if we take a TM Bad Thing from the kernel. That way none of the user register values are used, other than for printing in the oops message. This is the fix for CVE-2017-1000255. Fixes: 5d176f751ee3 ("powerpc: tm: Enable transactional memory (TM) lazily for userspace") Cc: stable@vger.kernel.org # v4.9+ Signed-off-by: Cyril Bur <cyrilbur@gmail.com> [mpe: Rewrite change log & comments, tweak asm slightly] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2017-08-17 18:42:26 +08:00
/* 3 in EXCEPTION_PROLOG_COMMON */
mr r10,r1 /* Save r1 */
ld r1,PACAEMERGSP(r13) /* Use emergency stack */
subi r1,r1,INT_FRAME_SIZE /* alloc stack frame */
INT_COMMON 0x700, PACA_EXGEN, 0, 1, 1, 0, 0
b 3f
powerpc/64s/exception: remove bad stack branch The bad stack test in interrupt handlers has a few problems. For performance it is taken in the common case, which is a fetch bubble and a waste of i-cache. For code development and maintainence, it requires yet another stack frame setup routine, and that constrains all exception handlers to follow the same register save pattern which inhibits future optimisation. Remove the test/branch and replace it with a trap. Teach the program check handler to use the emergency stack for this case. This does not result in quite so nice a message, however the SRR0 and SRR1 of the crashed interrupt can be seen in r11 and r12, as is the original r1 (adjusted by INT_FRAME_SIZE). These are the most important parts to debugging the issue. The original r9-12 and cr0 is lost, which is the main downside. kernel BUG at linux/arch/powerpc/kernel/exceptions-64s.S:847! Oops: Exception in kernel mode, sig: 5 [#1] BE SMP NR_CPUS=2048 NUMA PowerNV Modules linked in: CPU: 0 PID: 1 Comm: swapper/0 Not tainted NIP: c000000000009108 LR: c000000000cadbcc CTR: c0000000000090f0 REGS: c0000000fffcbd70 TRAP: 0700 Not tainted MSR: 9000000000021032 <SF,HV,ME,IR,DR,RI> CR: 28222448 XER: 20040000 CFAR: c000000000009100 IRQMASK: 0 GPR00: 000000000000003d fffffffffffffd00 c0000000018cfb00 c0000000f02b3166 GPR04: fffffffffffffffd 0000000000000007 fffffffffffffffb 0000000000000030 GPR08: 0000000000000037 0000000028222448 0000000000000000 c000000000ca8de0 GPR12: 9000000002009032 c000000001ae0000 c000000000010a00 0000000000000000 GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000 GPR20: c0000000f00322c0 c000000000f85200 0000000000000004 ffffffffffffffff GPR24: fffffffffffffffe 0000000000000000 0000000000000000 000000000000000a GPR28: 0000000000000000 0000000000000000 c0000000f02b391c c0000000f02b3167 NIP [c000000000009108] decrementer_common+0x18/0x160 LR [c000000000cadbcc] .vsnprintf+0x3ec/0x4f0 Call Trace: Instruction dump: 996d098a 994d098b 38610070 480246ed 48005518 60000000 38200000 718a4000 7c2a0b78 3821fd00 41c20008 e82d0970 <0981fd00> f92101a0 f9610170 f9810178 Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2019-06-28 14:33:18 +08:00
2:
INT_COMMON 0x700, PACA_EXGEN, 1, 1, 1, 0, 0
3:
bl save_nvgprs
addi r3,r1,STACK_FRAME_OVERHEAD
bl program_check_exception
b ret_from_except
EXC_REAL_BEGIN(fp_unavailable, 0x800, 0x100)
INT_HANDLER fp_unavailable, 0x800, kvm=1
EXC_REAL_END(fp_unavailable, 0x800, 0x100)
EXC_VIRT_BEGIN(fp_unavailable, 0x4800, 0x100)
INT_HANDLER fp_unavailable, 0x800, virt=1
EXC_VIRT_END(fp_unavailable, 0x4800, 0x100)
INT_KVM_HANDLER fp_unavailable, 0x800, EXC_STD, PACA_EXGEN, 0
EXC_COMMON_BEGIN(fp_unavailable_common)
INT_COMMON 0x800, PACA_EXGEN, 1, 1, 0, 0, 0
bne 1f /* if from user, just load it up */
bl save_nvgprs
RECONCILE_IRQ_STATE(r10, r11)
addi r3,r1,STACK_FRAME_OVERHEAD
bl kernel_fp_unavailable_exception
0: trap
EMIT_BUG_ENTRY 0b, __FILE__, __LINE__, 0
1:
#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
BEGIN_FTR_SECTION
/* Test if 2 TM state bits are zero. If non-zero (ie. userspace was in
* transaction), go do TM stuff
*/
rldicl. r0, r12, (64-MSR_TS_LG), (64-2)
bne- 2f
END_FTR_SECTION_IFSET(CPU_FTR_TM)
#endif
bl load_up_fpu
b fast_exception_return
#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
2: /* User process was in a transaction */
bl save_nvgprs
RECONCILE_IRQ_STATE(r10, r11)
addi r3,r1,STACK_FRAME_OVERHEAD
bl fp_unavailable_tm
b ret_from_except
#endif
EXC_REAL_BEGIN(decrementer, 0x900, 0x80)
INT_HANDLER decrementer, 0x900, ool=1, bitmask=IRQS_DISABLED, kvm=1
EXC_REAL_END(decrementer, 0x900, 0x80)
EXC_VIRT_BEGIN(decrementer, 0x4900, 0x80)
INT_HANDLER decrementer, 0x900, virt=1, bitmask=IRQS_DISABLED
EXC_VIRT_END(decrementer, 0x4900, 0x80)
INT_KVM_HANDLER decrementer, 0x900, EXC_STD, PACA_EXGEN, 0
EXC_COMMON_ASYNC(decrementer_common, 0x900, timer_interrupt)
powerpc: Fix "attempt to move .org backwards" error Building a 64-bit powerpc kernel with PR KVM enabled currently gives this error: AS arch/powerpc/kernel/head_64.o arch/powerpc/kernel/exceptions-64s.S: Assembler messages: arch/powerpc/kernel/exceptions-64s.S:258: Error: attempt to move .org backwards make[2]: *** [arch/powerpc/kernel/head_64.o] Error 1 This happens because the MASKABLE_EXCEPTION_PSERIES macro turns into 33 instructions, but we only have space for 32 at the decrementer interrupt vector (from 0x900 to 0x980). In the code generated by the MASKABLE_EXCEPTION_PSERIES macro, we currently have two instances of the HMT_MEDIUM macro, which has the effect of setting the SMT thread priority to medium. One is the first instruction, and is overwritten by a no-op on processors where we save the PPR (processor priority register), that is, POWER7 or later. The other is after we have saved the PPR. In order to reduce the code at 0x900 by one instruction, we omit the first HMT_MEDIUM. On processors without SMT this will have no effect since HMT_MEDIUM is a no-op there. On POWER5 and RS64 machines this will mean that the first few instructions take a little longer in the case where a decrementer interrupt occurs when the hardware thread is running at low SMT priority. On POWER6 and later machines, the hardware automatically boosts the thread priority when a decrementer interrupt is taken if the thread priority was below medium, so this change won't make any difference. The alternative would be to branch out of line after saving the CFAR. However, that would incur an extra overhead on all processors, whereas the approach adopted here only adds overhead on older threaded processors. Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-04-26 01:51:40 +08:00
EXC_REAL_BEGIN(hdecrementer, 0x980, 0x80)
INT_HANDLER hdecrementer, 0x980, hsrr=EXC_HV, kvm=1
EXC_REAL_END(hdecrementer, 0x980, 0x80)
EXC_VIRT_BEGIN(hdecrementer, 0x4980, 0x80)
INT_HANDLER hdecrementer, 0x980, virt=1, hsrr=EXC_HV, kvm=1
EXC_VIRT_END(hdecrementer, 0x4980, 0x80)
INT_KVM_HANDLER hdecrementer, 0x980, EXC_HV, PACA_EXGEN, 0
EXC_COMMON(hdecrementer_common, 0x980, hdec_interrupt)
EXC_REAL_BEGIN(doorbell_super, 0xa00, 0x100)
INT_HANDLER doorbell_super, 0xa00, bitmask=IRQS_DISABLED, kvm=1
EXC_REAL_END(doorbell_super, 0xa00, 0x100)
EXC_VIRT_BEGIN(doorbell_super, 0x4a00, 0x100)
INT_HANDLER doorbell_super, 0xa00, virt=1, bitmask=IRQS_DISABLED
EXC_VIRT_END(doorbell_super, 0x4a00, 0x100)
INT_KVM_HANDLER doorbell_super, 0xa00, EXC_STD, PACA_EXGEN, 0
#ifdef CONFIG_PPC_DOORBELL
EXC_COMMON_ASYNC(doorbell_super_common, 0xa00, doorbell_exception)
#else
EXC_COMMON_ASYNC(doorbell_super_common, 0xa00, unknown_exception)
#endif
EXC_REAL_NONE(0xb00, 0x100)
EXC_VIRT_NONE(0x4b00, 0x100)
/*
* system call / hypercall (0xc00, 0x4c00)
*
* The system call exception is invoked with "sc 0" and does not alter HV bit.
*
* The hypercall is invoked with "sc 1" and sets HV=1.
*
* In HPT, sc 1 always goes to 0xc00 real mode. In RADIX, sc 1 can go to
* 0x4c00 virtual mode.
*
* Call convention:
*
* syscall register convention is in Documentation/powerpc/syscall64-abi.rst
*
* For hypercalls, the register convention is as follows:
* r0 volatile
* r1-2 nonvolatile
* r3 volatile parameter and return value for status
* r4-r10 volatile input and output value
* r11 volatile hypercall number and output value
* r12 volatile input and output value
* r13-r31 nonvolatile
* LR nonvolatile
* CTR volatile
* XER volatile
* CR0-1 CR5-7 volatile
* CR2-4 nonvolatile
* Other registers nonvolatile
*
* The intersection of volatile registers that don't contain possible
* inputs is: cr0, xer, ctr. We may use these as scratch regs upon entry
* without saving, though xer is not a good idea to use, as hardware may
* interpret some bits so it may be costly to change them.
*/
.macro SYSTEM_CALL virt
#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
/*
* There is a little bit of juggling to get syscall and hcall
* working well. Save r13 in ctr to avoid using SPRG scratch
* register.
*
* Userspace syscalls have already saved the PPR, hcalls must save
* it before setting HMT_MEDIUM.
*/
mtctr r13
GET_PACA(r13)
std r10,PACA_EXGEN+EX_R10(r13)
INTERRUPT_TO_KERNEL
KVMTEST system_call EXC_STD 0xc00 /* uses r10, branch to system_call_kvm */
mfctr r9
#else
mr r9,r13
GET_PACA(r13)
INTERRUPT_TO_KERNEL
#endif
#ifdef CONFIG_PPC_FAST_ENDIAN_SWITCH
BEGIN_FTR_SECTION
cmpdi r0,0x1ebe
beq- 1f
END_FTR_SECTION_IFSET(CPU_FTR_REAL_LE)
#endif
/* We reach here with PACA in r13, r13 in r9. */
mfspr r11,SPRN_SRR0
mfspr r12,SPRN_SRR1
HMT_MEDIUM
.if ! \virt
__LOAD_HANDLER(r10, system_call_common)
mtspr SPRN_SRR0,r10
ld r10,PACAKMSR(r13)
mtspr SPRN_SRR1,r10
RFI_TO_KERNEL
b . /* prevent speculative execution */
.else
li r10,MSR_RI
mtmsrd r10,1 /* Set RI (EE=0) */
#ifdef CONFIG_RELOCATABLE
__LOAD_HANDLER(r10, system_call_common)
mtctr r10
bctr
#else
b system_call_common
#endif
.endif
#ifdef CONFIG_PPC_FAST_ENDIAN_SWITCH
/* Fast LE/BE switch system call */
1: mfspr r12,SPRN_SRR1
xori r12,r12,MSR_LE
mtspr SPRN_SRR1,r12
mr r13,r9
RFI_TO_USER /* return to userspace */
b . /* prevent speculative execution */
#endif
.endm
EXC_REAL_BEGIN(system_call, 0xc00, 0x100)
SYSTEM_CALL 0
EXC_REAL_END(system_call, 0xc00, 0x100)
EXC_VIRT_BEGIN(system_call, 0x4c00, 0x100)
SYSTEM_CALL 1
EXC_VIRT_END(system_call, 0x4c00, 0x100)
#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
/*
* This is a hcall, so register convention is as above, with these
* differences:
* r13 = PACA
* ctr = orig r13
* orig r10 saved in PACA
*/
TRAMP_KVM_BEGIN(system_call_kvm)
/*
* Save the PPR (on systems that support it) before changing to
* HMT_MEDIUM. That allows the KVM code to save that value into the
* guest state (it is the guest's PPR value).
*/
OPT_GET_SPR(r10, SPRN_PPR, CPU_FTR_HAS_PPR)
HMT_MEDIUM
OPT_SAVE_REG_TO_PACA(PACA_EXGEN+EX_PPR, r10, CPU_FTR_HAS_PPR)
mfctr r10
SET_SCRATCH0(r10)
std r9,PACA_EXGEN+EX_R9(r13)
mfcr r9
KVM_HANDLER 0xc00, EXC_STD, PACA_EXGEN, 0
#endif
EXC_REAL_BEGIN(single_step, 0xd00, 0x100)
INT_HANDLER single_step, 0xd00, kvm=1
EXC_REAL_END(single_step, 0xd00, 0x100)
EXC_VIRT_BEGIN(single_step, 0x4d00, 0x100)
INT_HANDLER single_step, 0xd00, virt=1
EXC_VIRT_END(single_step, 0x4d00, 0x100)
INT_KVM_HANDLER single_step, 0xd00, EXC_STD, PACA_EXGEN, 0
EXC_COMMON(single_step_common, 0xd00, single_step_exception)
EXC_REAL_BEGIN(h_data_storage, 0xe00, 0x20)
INT_HANDLER h_data_storage, 0xe00, ool=1, hsrr=EXC_HV, dar=1, dsisr=1, kvm=1
EXC_REAL_END(h_data_storage, 0xe00, 0x20)
EXC_VIRT_BEGIN(h_data_storage, 0x4e00, 0x20)
INT_HANDLER h_data_storage, 0xe00, ool=1, virt=1, hsrr=EXC_HV, dar=1, dsisr=1, kvm=1
EXC_VIRT_END(h_data_storage, 0x4e00, 0x20)
INT_KVM_HANDLER h_data_storage, 0xe00, EXC_HV, PACA_EXGEN, 1
EXC_COMMON_BEGIN(h_data_storage_common)
INT_COMMON 0xe00, PACA_EXGEN, 1, 1, 1, 1, 1
bl save_nvgprs
addi r3,r1,STACK_FRAME_OVERHEAD
KVM: PPC: Book3S HV: Implement functions to access quadrants 1 & 2 The POWER9 radix mmu has the concept of quadrants. The quadrant number is the two high bits of the effective address and determines the fully qualified address to be used for the translation. The fully qualified address consists of the effective lpid, the effective pid and the effective address. This gives then 4 possible quadrants 0, 1, 2, and 3. When accessing these quadrants the fully qualified address is obtained as follows: Quadrant | Hypervisor | Guest -------------------------------------------------------------------------- | EA[0:1] = 0b00 | EA[0:1] = 0b00 0 | effLPID = 0 | effLPID = LPIDR | effPID = PIDR | effPID = PIDR -------------------------------------------------------------------------- | EA[0:1] = 0b01 | 1 | effLPID = LPIDR | Invalid Access | effPID = PIDR | -------------------------------------------------------------------------- | EA[0:1] = 0b10 | 2 | effLPID = LPIDR | Invalid Access | effPID = 0 | -------------------------------------------------------------------------- | EA[0:1] = 0b11 | EA[0:1] = 0b11 3 | effLPID = 0 | effLPID = LPIDR | effPID = 0 | effPID = 0 -------------------------------------------------------------------------- In the Guest; Quadrant 3 is normally used to address the operating system since this uses effPID=0 and effLPID=LPIDR, meaning the PID register doesn't need to be switched. Quadrant 0 is normally used to address user space since the effLPID and effPID are taken from the corresponding registers. In the Host; Quadrant 0 and 3 are used as above, however the effLPID is always 0 to address the host. Quadrants 1 and 2 can be used by the host to address guest memory using a guest effective address. Since the effLPID comes from the LPID register, the host loads the LPID of the guest it would like to access (and the PID of the process) and can perform accesses to a guest effective address. This means quadrant 1 can be used to address the guest user space and quadrant 2 can be used to address the guest operating system from the hypervisor, using a guest effective address. Access to the quadrants can cause a Hypervisor Data Storage Interrupt (HDSI) due to being unable to perform partition scoped translation. Previously this could only be generated from a guest and so the code path expects us to take the KVM trampoline in the interrupt handler. This is no longer the case so we modify the handler to call bad_page_fault() to check if we were expecting this fault so we can handle it gracefully and just return with an error code. In the hash mmu case we still raise an unknown exception since quadrants aren't defined for the hash mmu. Signed-off-by: Suraj Jitindar Singh <sjitindarsingh@gmail.com> Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
2018-12-14 13:29:05 +08:00
BEGIN_MMU_FTR_SECTION
ld r4,_DAR(r1)
KVM: PPC: Book3S HV: Implement functions to access quadrants 1 & 2 The POWER9 radix mmu has the concept of quadrants. The quadrant number is the two high bits of the effective address and determines the fully qualified address to be used for the translation. The fully qualified address consists of the effective lpid, the effective pid and the effective address. This gives then 4 possible quadrants 0, 1, 2, and 3. When accessing these quadrants the fully qualified address is obtained as follows: Quadrant | Hypervisor | Guest -------------------------------------------------------------------------- | EA[0:1] = 0b00 | EA[0:1] = 0b00 0 | effLPID = 0 | effLPID = LPIDR | effPID = PIDR | effPID = PIDR -------------------------------------------------------------------------- | EA[0:1] = 0b01 | 1 | effLPID = LPIDR | Invalid Access | effPID = PIDR | -------------------------------------------------------------------------- | EA[0:1] = 0b10 | 2 | effLPID = LPIDR | Invalid Access | effPID = 0 | -------------------------------------------------------------------------- | EA[0:1] = 0b11 | EA[0:1] = 0b11 3 | effLPID = 0 | effLPID = LPIDR | effPID = 0 | effPID = 0 -------------------------------------------------------------------------- In the Guest; Quadrant 3 is normally used to address the operating system since this uses effPID=0 and effLPID=LPIDR, meaning the PID register doesn't need to be switched. Quadrant 0 is normally used to address user space since the effLPID and effPID are taken from the corresponding registers. In the Host; Quadrant 0 and 3 are used as above, however the effLPID is always 0 to address the host. Quadrants 1 and 2 can be used by the host to address guest memory using a guest effective address. Since the effLPID comes from the LPID register, the host loads the LPID of the guest it would like to access (and the PID of the process) and can perform accesses to a guest effective address. This means quadrant 1 can be used to address the guest user space and quadrant 2 can be used to address the guest operating system from the hypervisor, using a guest effective address. Access to the quadrants can cause a Hypervisor Data Storage Interrupt (HDSI) due to being unable to perform partition scoped translation. Previously this could only be generated from a guest and so the code path expects us to take the KVM trampoline in the interrupt handler. This is no longer the case so we modify the handler to call bad_page_fault() to check if we were expecting this fault so we can handle it gracefully and just return with an error code. In the hash mmu case we still raise an unknown exception since quadrants aren't defined for the hash mmu. Signed-off-by: Suraj Jitindar Singh <sjitindarsingh@gmail.com> Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
2018-12-14 13:29:05 +08:00
li r5,SIGSEGV
bl bad_page_fault
MMU_FTR_SECTION_ELSE
bl unknown_exception
KVM: PPC: Book3S HV: Implement functions to access quadrants 1 & 2 The POWER9 radix mmu has the concept of quadrants. The quadrant number is the two high bits of the effective address and determines the fully qualified address to be used for the translation. The fully qualified address consists of the effective lpid, the effective pid and the effective address. This gives then 4 possible quadrants 0, 1, 2, and 3. When accessing these quadrants the fully qualified address is obtained as follows: Quadrant | Hypervisor | Guest -------------------------------------------------------------------------- | EA[0:1] = 0b00 | EA[0:1] = 0b00 0 | effLPID = 0 | effLPID = LPIDR | effPID = PIDR | effPID = PIDR -------------------------------------------------------------------------- | EA[0:1] = 0b01 | 1 | effLPID = LPIDR | Invalid Access | effPID = PIDR | -------------------------------------------------------------------------- | EA[0:1] = 0b10 | 2 | effLPID = LPIDR | Invalid Access | effPID = 0 | -------------------------------------------------------------------------- | EA[0:1] = 0b11 | EA[0:1] = 0b11 3 | effLPID = 0 | effLPID = LPIDR | effPID = 0 | effPID = 0 -------------------------------------------------------------------------- In the Guest; Quadrant 3 is normally used to address the operating system since this uses effPID=0 and effLPID=LPIDR, meaning the PID register doesn't need to be switched. Quadrant 0 is normally used to address user space since the effLPID and effPID are taken from the corresponding registers. In the Host; Quadrant 0 and 3 are used as above, however the effLPID is always 0 to address the host. Quadrants 1 and 2 can be used by the host to address guest memory using a guest effective address. Since the effLPID comes from the LPID register, the host loads the LPID of the guest it would like to access (and the PID of the process) and can perform accesses to a guest effective address. This means quadrant 1 can be used to address the guest user space and quadrant 2 can be used to address the guest operating system from the hypervisor, using a guest effective address. Access to the quadrants can cause a Hypervisor Data Storage Interrupt (HDSI) due to being unable to perform partition scoped translation. Previously this could only be generated from a guest and so the code path expects us to take the KVM trampoline in the interrupt handler. This is no longer the case so we modify the handler to call bad_page_fault() to check if we were expecting this fault so we can handle it gracefully and just return with an error code. In the hash mmu case we still raise an unknown exception since quadrants aren't defined for the hash mmu. Signed-off-by: Suraj Jitindar Singh <sjitindarsingh@gmail.com> Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
2018-12-14 13:29:05 +08:00
ALT_MMU_FTR_SECTION_END_IFSET(MMU_FTR_TYPE_RADIX)
b ret_from_except
powerpc: Save CFAR before branching in interrupt entry paths Some of the interrupt vectors on 64-bit POWER server processors are only 32 bytes long, which is not enough for the full first-level interrupt handler. For these we currently just have a branch to an out-of-line handler. However, this means that we corrupt the CFAR (come-from address register) on POWER7 and later processors. To fix this, we split the EXCEPTION_PROLOG_1 macro into two pieces: EXCEPTION_PROLOG_0 contains the part up to the point where the CFAR is saved in the PACA, and EXCEPTION_PROLOG_1 contains the rest. We then put EXCEPTION_PROLOG_0 in the short interrupt vectors before we branch to the out-of-line handler, which contains the rest of the first-level interrupt handler. To facilitate this, we define new _OOL (out of line) variants of STD_EXCEPTION_PSERIES, etc. In order to get EXCEPTION_PROLOG_0 to be short enough, i.e., no more than 6 instructions, it was necessary to move the stores that move the PPR and CFAR values into the PACA into __EXCEPTION_PROLOG_1 and to get rid of one of the two HMT_MEDIUM instructions. Previously there was a HMT_MEDIUM_PPR_DISCARD before the prolog, which was nop'd out on processors with the PPR (POWER7 and later), and then another HMT_MEDIUM inside the HMT_MEDIUM_PPR_SAVE macro call inside __EXCEPTION_PROLOG_1, which was nop'd out on processors without PPR. Now the HMT_MEDIUM inside EXCEPTION_PROLOG_0 is there unconditionally and the HMT_MEDIUM_PPR_DISCARD is not strictly necessary, although this leaves it in for the interrupt vectors where there is room for it. Previously we had a handler for hypervisor maintenance interrupts at 0xe50, which doesn't leave enough room for the vector for hypervisor emulation assist interrupts at 0xe40, since we need 8 instructions. The 0xe50 vector was only used on POWER6, as the HMI vector was moved to 0xe60 on POWER7. Since we don't support running in hypervisor mode on POWER6, we just remove the handler at 0xe50. This also changes denorm_exception_hv to use EXCEPTION_PROLOG_0 instead of open-coding it, and removes the HMT_MEDIUM_PPR_DISCARD from the relocation-on vectors (since any CPU that supports relocation-on interrupts also has the PPR). Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-02-05 02:10:15 +08:00
EXC_REAL_BEGIN(h_instr_storage, 0xe20, 0x20)
INT_HANDLER h_instr_storage, 0xe20, ool=1, hsrr=EXC_HV, kvm=1
EXC_REAL_END(h_instr_storage, 0xe20, 0x20)
EXC_VIRT_BEGIN(h_instr_storage, 0x4e20, 0x20)
INT_HANDLER h_instr_storage, 0xe20, ool=1, virt=1, hsrr=EXC_HV, kvm=1
EXC_VIRT_END(h_instr_storage, 0x4e20, 0x20)
INT_KVM_HANDLER h_instr_storage, 0xe20, EXC_HV, PACA_EXGEN, 0
EXC_COMMON(h_instr_storage_common, 0xe20, unknown_exception)
powerpc: Save CFAR before branching in interrupt entry paths Some of the interrupt vectors on 64-bit POWER server processors are only 32 bytes long, which is not enough for the full first-level interrupt handler. For these we currently just have a branch to an out-of-line handler. However, this means that we corrupt the CFAR (come-from address register) on POWER7 and later processors. To fix this, we split the EXCEPTION_PROLOG_1 macro into two pieces: EXCEPTION_PROLOG_0 contains the part up to the point where the CFAR is saved in the PACA, and EXCEPTION_PROLOG_1 contains the rest. We then put EXCEPTION_PROLOG_0 in the short interrupt vectors before we branch to the out-of-line handler, which contains the rest of the first-level interrupt handler. To facilitate this, we define new _OOL (out of line) variants of STD_EXCEPTION_PSERIES, etc. In order to get EXCEPTION_PROLOG_0 to be short enough, i.e., no more than 6 instructions, it was necessary to move the stores that move the PPR and CFAR values into the PACA into __EXCEPTION_PROLOG_1 and to get rid of one of the two HMT_MEDIUM instructions. Previously there was a HMT_MEDIUM_PPR_DISCARD before the prolog, which was nop'd out on processors with the PPR (POWER7 and later), and then another HMT_MEDIUM inside the HMT_MEDIUM_PPR_SAVE macro call inside __EXCEPTION_PROLOG_1, which was nop'd out on processors without PPR. Now the HMT_MEDIUM inside EXCEPTION_PROLOG_0 is there unconditionally and the HMT_MEDIUM_PPR_DISCARD is not strictly necessary, although this leaves it in for the interrupt vectors where there is room for it. Previously we had a handler for hypervisor maintenance interrupts at 0xe50, which doesn't leave enough room for the vector for hypervisor emulation assist interrupts at 0xe40, since we need 8 instructions. The 0xe50 vector was only used on POWER6, as the HMI vector was moved to 0xe60 on POWER7. Since we don't support running in hypervisor mode on POWER6, we just remove the handler at 0xe50. This also changes denorm_exception_hv to use EXCEPTION_PROLOG_0 instead of open-coding it, and removes the HMT_MEDIUM_PPR_DISCARD from the relocation-on vectors (since any CPU that supports relocation-on interrupts also has the PPR). Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-02-05 02:10:15 +08:00
EXC_REAL_BEGIN(emulation_assist, 0xe40, 0x20)
INT_HANDLER emulation_assist, 0xe40, ool=1, hsrr=EXC_HV, kvm=1
EXC_REAL_END(emulation_assist, 0xe40, 0x20)
EXC_VIRT_BEGIN(emulation_assist, 0x4e40, 0x20)
INT_HANDLER emulation_assist, 0xe40, ool=1, virt=1, hsrr=EXC_HV, kvm=1
EXC_VIRT_END(emulation_assist, 0x4e40, 0x20)
INT_KVM_HANDLER emulation_assist, 0xe40, EXC_HV, PACA_EXGEN, 0
EXC_COMMON(emulation_assist_common, 0xe40, emulation_assist_interrupt)
powerpc: Save CFAR before branching in interrupt entry paths Some of the interrupt vectors on 64-bit POWER server processors are only 32 bytes long, which is not enough for the full first-level interrupt handler. For these we currently just have a branch to an out-of-line handler. However, this means that we corrupt the CFAR (come-from address register) on POWER7 and later processors. To fix this, we split the EXCEPTION_PROLOG_1 macro into two pieces: EXCEPTION_PROLOG_0 contains the part up to the point where the CFAR is saved in the PACA, and EXCEPTION_PROLOG_1 contains the rest. We then put EXCEPTION_PROLOG_0 in the short interrupt vectors before we branch to the out-of-line handler, which contains the rest of the first-level interrupt handler. To facilitate this, we define new _OOL (out of line) variants of STD_EXCEPTION_PSERIES, etc. In order to get EXCEPTION_PROLOG_0 to be short enough, i.e., no more than 6 instructions, it was necessary to move the stores that move the PPR and CFAR values into the PACA into __EXCEPTION_PROLOG_1 and to get rid of one of the two HMT_MEDIUM instructions. Previously there was a HMT_MEDIUM_PPR_DISCARD before the prolog, which was nop'd out on processors with the PPR (POWER7 and later), and then another HMT_MEDIUM inside the HMT_MEDIUM_PPR_SAVE macro call inside __EXCEPTION_PROLOG_1, which was nop'd out on processors without PPR. Now the HMT_MEDIUM inside EXCEPTION_PROLOG_0 is there unconditionally and the HMT_MEDIUM_PPR_DISCARD is not strictly necessary, although this leaves it in for the interrupt vectors where there is room for it. Previously we had a handler for hypervisor maintenance interrupts at 0xe50, which doesn't leave enough room for the vector for hypervisor emulation assist interrupts at 0xe40, since we need 8 instructions. The 0xe50 vector was only used on POWER6, as the HMI vector was moved to 0xe60 on POWER7. Since we don't support running in hypervisor mode on POWER6, we just remove the handler at 0xe50. This also changes denorm_exception_hv to use EXCEPTION_PROLOG_0 instead of open-coding it, and removes the HMT_MEDIUM_PPR_DISCARD from the relocation-on vectors (since any CPU that supports relocation-on interrupts also has the PPR). Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-02-05 02:10:15 +08:00
/*
* hmi_exception trampoline is a special case. It jumps to hmi_exception_early
* first, and then eventaully from there to the trampoline to get into virtual
* mode.
*/
EXC_REAL_BEGIN(hmi_exception, 0xe60, 0x20)
INT_HANDLER hmi_exception, 0xe60, ool=1, early=1, hsrr=EXC_HV, ri=0, kvm=1
EXC_REAL_END(hmi_exception, 0xe60, 0x20)
EXC_VIRT_NONE(0x4e60, 0x20)
INT_KVM_HANDLER hmi_exception, 0xe60, EXC_HV, PACA_EXGEN, 0
EXC_COMMON_BEGIN(hmi_exception_early_common)
mtctr r10 /* Restore ctr */
mfspr r11,SPRN_HSRR0 /* Save HSRR0 */
mfspr r12,SPRN_HSRR1 /* Save HSRR1 */
mr r10,r1 /* Save r1 */
ld r1,PACAEMERGSP(r13) /* Use emergency stack for realmode */
subi r1,r1,INT_FRAME_SIZE /* alloc stack frame */
/* We don't touch AMR here, we never go to virtual mode */
INT_COMMON 0xe60, PACA_EXGEN, 0, 0, 0, 0, 0
addi r3,r1,STACK_FRAME_OVERHEAD
bl hmi_exception_realmode
cmpdi cr0,r3,0
bne 1f
EXCEPTION_RESTORE_REGS EXC_HV
HRFI_TO_USER_OR_KERNEL
1:
/*
* Go to virtual mode and pull the HMI event information from
* firmware.
*/
EXCEPTION_RESTORE_REGS EXC_HV
INT_HANDLER hmi_exception, 0xe60, hsrr=EXC_HV, bitmask=IRQS_DISABLED, kvm=1
EXC_COMMON_BEGIN(hmi_exception_common)
INT_COMMON 0xe60, PACA_EXGEN, 1, 1, 1, 0, 0
FINISH_NAP
RUNLATCH_ON
bl save_nvgprs
addi r3,r1,STACK_FRAME_OVERHEAD
bl handle_hmi_exception
b ret_from_except
powerpc: Save CFAR before branching in interrupt entry paths Some of the interrupt vectors on 64-bit POWER server processors are only 32 bytes long, which is not enough for the full first-level interrupt handler. For these we currently just have a branch to an out-of-line handler. However, this means that we corrupt the CFAR (come-from address register) on POWER7 and later processors. To fix this, we split the EXCEPTION_PROLOG_1 macro into two pieces: EXCEPTION_PROLOG_0 contains the part up to the point where the CFAR is saved in the PACA, and EXCEPTION_PROLOG_1 contains the rest. We then put EXCEPTION_PROLOG_0 in the short interrupt vectors before we branch to the out-of-line handler, which contains the rest of the first-level interrupt handler. To facilitate this, we define new _OOL (out of line) variants of STD_EXCEPTION_PSERIES, etc. In order to get EXCEPTION_PROLOG_0 to be short enough, i.e., no more than 6 instructions, it was necessary to move the stores that move the PPR and CFAR values into the PACA into __EXCEPTION_PROLOG_1 and to get rid of one of the two HMT_MEDIUM instructions. Previously there was a HMT_MEDIUM_PPR_DISCARD before the prolog, which was nop'd out on processors with the PPR (POWER7 and later), and then another HMT_MEDIUM inside the HMT_MEDIUM_PPR_SAVE macro call inside __EXCEPTION_PROLOG_1, which was nop'd out on processors without PPR. Now the HMT_MEDIUM inside EXCEPTION_PROLOG_0 is there unconditionally and the HMT_MEDIUM_PPR_DISCARD is not strictly necessary, although this leaves it in for the interrupt vectors where there is room for it. Previously we had a handler for hypervisor maintenance interrupts at 0xe50, which doesn't leave enough room for the vector for hypervisor emulation assist interrupts at 0xe40, since we need 8 instructions. The 0xe50 vector was only used on POWER6, as the HMI vector was moved to 0xe60 on POWER7. Since we don't support running in hypervisor mode on POWER6, we just remove the handler at 0xe50. This also changes denorm_exception_hv to use EXCEPTION_PROLOG_0 instead of open-coding it, and removes the HMT_MEDIUM_PPR_DISCARD from the relocation-on vectors (since any CPU that supports relocation-on interrupts also has the PPR). Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-02-05 02:10:15 +08:00
EXC_REAL_BEGIN(h_doorbell, 0xe80, 0x20)
INT_HANDLER h_doorbell, 0xe80, ool=1, hsrr=EXC_HV, bitmask=IRQS_DISABLED, kvm=1
EXC_REAL_END(h_doorbell, 0xe80, 0x20)
EXC_VIRT_BEGIN(h_doorbell, 0x4e80, 0x20)
INT_HANDLER h_doorbell, 0xe80, ool=1, virt=1, hsrr=EXC_HV, bitmask=IRQS_DISABLED, kvm=1
EXC_VIRT_END(h_doorbell, 0x4e80, 0x20)
INT_KVM_HANDLER h_doorbell, 0xe80, EXC_HV, PACA_EXGEN, 0
#ifdef CONFIG_PPC_DOORBELL
EXC_COMMON_ASYNC(h_doorbell_common, 0xe80, doorbell_exception)
#else
EXC_COMMON_ASYNC(h_doorbell_common, 0xe80, unknown_exception)
#endif
EXC_REAL_BEGIN(h_virt_irq, 0xea0, 0x20)
INT_HANDLER h_virt_irq, 0xea0, ool=1, hsrr=EXC_HV, bitmask=IRQS_DISABLED, kvm=1
EXC_REAL_END(h_virt_irq, 0xea0, 0x20)
EXC_VIRT_BEGIN(h_virt_irq, 0x4ea0, 0x20)
INT_HANDLER h_virt_irq, 0xea0, ool=1, virt=1, hsrr=EXC_HV, bitmask=IRQS_DISABLED, kvm=1
EXC_VIRT_END(h_virt_irq, 0x4ea0, 0x20)
INT_KVM_HANDLER h_virt_irq, 0xea0, EXC_HV, PACA_EXGEN, 0
EXC_COMMON_ASYNC(h_virt_irq_common, 0xea0, do_IRQ)
EXC_REAL_NONE(0xec0, 0x20)
EXC_VIRT_NONE(0x4ec0, 0x20)
EXC_REAL_NONE(0xee0, 0x20)
EXC_VIRT_NONE(0x4ee0, 0x20)
EXC_REAL_BEGIN(performance_monitor, 0xf00, 0x20)
INT_HANDLER performance_monitor, 0xf00, ool=1, bitmask=IRQS_PMI_DISABLED, kvm=1
EXC_REAL_END(performance_monitor, 0xf00, 0x20)
EXC_VIRT_BEGIN(performance_monitor, 0x4f00, 0x20)
INT_HANDLER performance_monitor, 0xf00, ool=1, virt=1, bitmask=IRQS_PMI_DISABLED
EXC_VIRT_END(performance_monitor, 0x4f00, 0x20)
INT_KVM_HANDLER performance_monitor, 0xf00, EXC_STD, PACA_EXGEN, 0
EXC_COMMON_ASYNC(performance_monitor_common, 0xf00, performance_monitor_exception)
EXC_REAL_BEGIN(altivec_unavailable, 0xf20, 0x20)
INT_HANDLER altivec_unavailable, 0xf20, ool=1, kvm=1
EXC_REAL_END(altivec_unavailable, 0xf20, 0x20)
EXC_VIRT_BEGIN(altivec_unavailable, 0x4f20, 0x20)
INT_HANDLER altivec_unavailable, 0xf20, ool=1, virt=1
EXC_VIRT_END(altivec_unavailable, 0x4f20, 0x20)
INT_KVM_HANDLER altivec_unavailable, 0xf20, EXC_STD, PACA_EXGEN, 0
EXC_COMMON_BEGIN(altivec_unavailable_common)
INT_COMMON 0xf20, PACA_EXGEN, 1, 1, 0, 0, 0
#ifdef CONFIG_ALTIVEC
BEGIN_FTR_SECTION
beq 1f
#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
BEGIN_FTR_SECTION_NESTED(69)
/* Test if 2 TM state bits are zero. If non-zero (ie. userspace was in
* transaction), go do TM stuff
*/
rldicl. r0, r12, (64-MSR_TS_LG), (64-2)
bne- 2f
END_FTR_SECTION_NESTED(CPU_FTR_TM, CPU_FTR_TM, 69)
#endif
bl load_up_altivec
b fast_exception_return
#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
2: /* User process was in a transaction */
bl save_nvgprs
RECONCILE_IRQ_STATE(r10, r11)
addi r3,r1,STACK_FRAME_OVERHEAD
bl altivec_unavailable_tm
b ret_from_except
#endif
1:
END_FTR_SECTION_IFSET(CPU_FTR_ALTIVEC)
#endif
bl save_nvgprs
RECONCILE_IRQ_STATE(r10, r11)
addi r3,r1,STACK_FRAME_OVERHEAD
bl altivec_unavailable_exception
b ret_from_except
EXC_REAL_BEGIN(vsx_unavailable, 0xf40, 0x20)
INT_HANDLER vsx_unavailable, 0xf40, ool=1, kvm=1
EXC_REAL_END(vsx_unavailable, 0xf40, 0x20)
EXC_VIRT_BEGIN(vsx_unavailable, 0x4f40, 0x20)
INT_HANDLER vsx_unavailable, 0xf40, ool=1, virt=1
EXC_VIRT_END(vsx_unavailable, 0x4f40, 0x20)
INT_KVM_HANDLER vsx_unavailable, 0xf40, EXC_STD, PACA_EXGEN, 0
EXC_COMMON_BEGIN(vsx_unavailable_common)
INT_COMMON 0xf40, PACA_EXGEN, 1, 1, 0, 0, 0
#ifdef CONFIG_VSX
BEGIN_FTR_SECTION
beq 1f
#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
BEGIN_FTR_SECTION_NESTED(69)
/* Test if 2 TM state bits are zero. If non-zero (ie. userspace was in
* transaction), go do TM stuff
*/
rldicl. r0, r12, (64-MSR_TS_LG), (64-2)
bne- 2f
END_FTR_SECTION_NESTED(CPU_FTR_TM, CPU_FTR_TM, 69)
#endif
b load_up_vsx
#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
2: /* User process was in a transaction */
bl save_nvgprs
RECONCILE_IRQ_STATE(r10, r11)
addi r3,r1,STACK_FRAME_OVERHEAD
bl vsx_unavailable_tm
b ret_from_except
#endif
1:
END_FTR_SECTION_IFSET(CPU_FTR_VSX)
#endif
bl save_nvgprs
RECONCILE_IRQ_STATE(r10, r11)
addi r3,r1,STACK_FRAME_OVERHEAD
bl vsx_unavailable_exception
b ret_from_except
EXC_REAL_BEGIN(facility_unavailable, 0xf60, 0x20)
INT_HANDLER facility_unavailable, 0xf60, ool=1, kvm=1
EXC_REAL_END(facility_unavailable, 0xf60, 0x20)
EXC_VIRT_BEGIN(facility_unavailable, 0x4f60, 0x20)
INT_HANDLER facility_unavailable, 0xf60, ool=1, virt=1
EXC_VIRT_END(facility_unavailable, 0x4f60, 0x20)
INT_KVM_HANDLER facility_unavailable, 0xf60, EXC_STD, PACA_EXGEN, 0
EXC_COMMON(facility_unavailable_common, 0xf60, facility_unavailable_exception)
EXC_REAL_BEGIN(h_facility_unavailable, 0xf80, 0x20)
INT_HANDLER h_facility_unavailable, 0xf80, ool=1, hsrr=EXC_HV, kvm=1
EXC_REAL_END(h_facility_unavailable, 0xf80, 0x20)
EXC_VIRT_BEGIN(h_facility_unavailable, 0x4f80, 0x20)
INT_HANDLER h_facility_unavailable, 0xf80, ool=1, virt=1, hsrr=EXC_HV, kvm=1
EXC_VIRT_END(h_facility_unavailable, 0x4f80, 0x20)
INT_KVM_HANDLER h_facility_unavailable, 0xf80, EXC_HV, PACA_EXGEN, 0
EXC_COMMON(h_facility_unavailable_common, 0xf80, facility_unavailable_exception)
EXC_REAL_NONE(0xfa0, 0x20)
EXC_VIRT_NONE(0x4fa0, 0x20)
EXC_REAL_NONE(0xfc0, 0x20)
EXC_VIRT_NONE(0x4fc0, 0x20)
EXC_REAL_NONE(0xfe0, 0x20)
EXC_VIRT_NONE(0x4fe0, 0x20)
EXC_REAL_NONE(0x1000, 0x100)
EXC_VIRT_NONE(0x5000, 0x100)
EXC_REAL_NONE(0x1100, 0x100)
EXC_VIRT_NONE(0x5100, 0x100)
#ifdef CONFIG_CBE_RAS
EXC_REAL_BEGIN(cbe_system_error, 0x1200, 0x100)
INT_HANDLER cbe_system_error, 0x1200, ool=1, hsrr=EXC_HV, kvm=1
EXC_REAL_END(cbe_system_error, 0x1200, 0x100)
EXC_VIRT_NONE(0x5200, 0x100)
INT_KVM_HANDLER cbe_system_error, 0x1200, EXC_HV, PACA_EXGEN, 1
EXC_COMMON(cbe_system_error_common, 0x1200, cbe_system_error_exception)
#else /* CONFIG_CBE_RAS */
EXC_REAL_NONE(0x1200, 0x100)
EXC_VIRT_NONE(0x5200, 0x100)
#endif
EXC_REAL_BEGIN(instruction_breakpoint, 0x1300, 0x100)
INT_HANDLER instruction_breakpoint, 0x1300, kvm=1
EXC_REAL_END(instruction_breakpoint, 0x1300, 0x100)
EXC_VIRT_BEGIN(instruction_breakpoint, 0x5300, 0x100)
INT_HANDLER instruction_breakpoint, 0x1300, virt=1
EXC_VIRT_END(instruction_breakpoint, 0x5300, 0x100)
INT_KVM_HANDLER instruction_breakpoint, 0x1300, EXC_STD, PACA_EXGEN, 1
EXC_COMMON(instruction_breakpoint_common, 0x1300, instruction_breakpoint_exception)
EXC_REAL_NONE(0x1400, 0x100)
EXC_VIRT_NONE(0x5400, 0x100)
EXC_REAL_BEGIN(denorm_exception_hv, 0x1500, 0x100)
INT_HANDLER denorm_exception_hv, 0x1500, early=2, hsrr=EXC_HV
#ifdef CONFIG_PPC_DENORMALISATION
mfspr r10,SPRN_HSRR1
andis. r10,r10,(HSRR1_DENORM)@h /* denorm? */
bne+ denorm_assist
#endif
KVMTEST denorm_exception_hv, EXC_HV 0x1500
INT_SAVE_SRR_AND_JUMP denorm_common, EXC_HV, 1
EXC_REAL_END(denorm_exception_hv, 0x1500, 0x100)
#ifdef CONFIG_PPC_DENORMALISATION
EXC_VIRT_BEGIN(denorm_exception, 0x5500, 0x100)
INT_HANDLER denorm_exception, 0x1500, 0, 2, 1, EXC_HV, PACA_EXGEN, 1, 0, 0, 0, 0
mfspr r10,SPRN_HSRR1
andis. r10,r10,(HSRR1_DENORM)@h /* denorm? */
bne+ denorm_assist
INT_VIRT_SAVE_SRR_AND_JUMP denorm_common, EXC_HV
EXC_VIRT_END(denorm_exception, 0x5500, 0x100)
#else
EXC_VIRT_NONE(0x5500, 0x100)
#endif
INT_KVM_HANDLER denorm_exception_hv, 0x1500, EXC_HV, PACA_EXGEN, 0
#ifdef CONFIG_PPC_DENORMALISATION
TRAMP_REAL_BEGIN(denorm_assist)
BEGIN_FTR_SECTION
/*
* To denormalise we need to move a copy of the register to itself.
* For POWER6 do that here for all FP regs.
*/
mfmsr r10
ori r10,r10,(MSR_FP|MSR_FE0|MSR_FE1)
xori r10,r10,(MSR_FE0|MSR_FE1)
mtmsrd r10
sync
.Lreg=0
.rept 32
fmr .Lreg,.Lreg
.Lreg=.Lreg+1
.endr
FTR_SECTION_ELSE
/*
* To denormalise we need to move a copy of the register to itself.
* For POWER7 do that here for the first 32 VSX registers only.
*/
mfmsr r10
oris r10,r10,MSR_VSX@h
mtmsrd r10
sync
.Lreg=0
.rept 32
XVCPSGNDP(.Lreg,.Lreg,.Lreg)
.Lreg=.Lreg+1
.endr
ALT_FTR_SECTION_END_IFCLR(CPU_FTR_ARCH_206)
BEGIN_FTR_SECTION
b denorm_done
END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_207S)
/*
* To denormalise we need to move a copy of the register to itself.
* For POWER8 we need to do that for all 64 VSX registers
*/
.Lreg=32
.rept 32
XVCPSGNDP(.Lreg,.Lreg,.Lreg)
.Lreg=.Lreg+1
.endr
denorm_done:
mfspr r11,SPRN_HSRR0
subi r11,r11,4
mtspr SPRN_HSRR0,r11
mtcrf 0x80,r9
ld r9,PACA_EXGEN+EX_R9(r13)
RESTORE_PPR_PACA(PACA_EXGEN, r10)
BEGIN_FTR_SECTION
ld r10,PACA_EXGEN+EX_CFAR(r13)
mtspr SPRN_CFAR,r10
END_FTR_SECTION_IFSET(CPU_FTR_CFAR)
ld r10,PACA_EXGEN+EX_R10(r13)
ld r11,PACA_EXGEN+EX_R11(r13)
ld r12,PACA_EXGEN+EX_R12(r13)
ld r13,PACA_EXGEN+EX_R13(r13)
HRFI_TO_UNKNOWN
b .
#endif
EXC_COMMON(denorm_common, 0x1500, unknown_exception)
#ifdef CONFIG_CBE_RAS
EXC_REAL_BEGIN(cbe_maintenance, 0x1600, 0x100)
INT_HANDLER cbe_maintenance, 0x1600, ool=1, hsrr=EXC_HV, kvm=1
EXC_REAL_END(cbe_maintenance, 0x1600, 0x100)
EXC_VIRT_NONE(0x5600, 0x100)
INT_KVM_HANDLER cbe_maintenance, 0x1600, EXC_HV, PACA_EXGEN, 1
EXC_COMMON(cbe_maintenance_common, 0x1600, cbe_maintenance_exception)
#else /* CONFIG_CBE_RAS */
EXC_REAL_NONE(0x1600, 0x100)
EXC_VIRT_NONE(0x5600, 0x100)
#endif
EXC_REAL_BEGIN(altivec_assist, 0x1700, 0x100)
INT_HANDLER altivec_assist, 0x1700, kvm=1
EXC_REAL_END(altivec_assist, 0x1700, 0x100)
EXC_VIRT_BEGIN(altivec_assist, 0x5700, 0x100)
INT_HANDLER altivec_assist, 0x1700, virt=1
EXC_VIRT_END(altivec_assist, 0x5700, 0x100)
INT_KVM_HANDLER altivec_assist, 0x1700, EXC_STD, PACA_EXGEN, 0
#ifdef CONFIG_ALTIVEC
EXC_COMMON(altivec_assist_common, 0x1700, altivec_assist_exception)
#else
EXC_COMMON(altivec_assist_common, 0x1700, unknown_exception)
#endif
#ifdef CONFIG_CBE_RAS
EXC_REAL_BEGIN(cbe_thermal, 0x1800, 0x100)
INT_HANDLER cbe_thermal, 0x1800, ool=1, hsrr=EXC_HV, kvm=1
EXC_REAL_END(cbe_thermal, 0x1800, 0x100)
EXC_VIRT_NONE(0x5800, 0x100)
INT_KVM_HANDLER cbe_thermal, 0x1800, EXC_HV, PACA_EXGEN, 1
EXC_COMMON(cbe_thermal_common, 0x1800, cbe_thermal_exception)
#else /* CONFIG_CBE_RAS */
EXC_REAL_NONE(0x1800, 0x100)
EXC_VIRT_NONE(0x5800, 0x100)
#endif
#ifdef CONFIG_PPC_WATCHDOG
powerpc/64s: implement arch-specific hardlockup watchdog Implement an arch-speicfic watchdog rather than use the perf-based hardlockup detector. The new watchdog takes the soft-NMI directly, rather than going through perf. Perf interrupts are to be made maskable in future, so that would prevent the perf detector from working in those regions. Additionally, implement a SMP based detector where all CPUs watch one another by pinging a shared cpumask. This is because powerpc Book3S does not have a true periodic local NMI, but some platforms do implement a true NMI IPI. If a CPU is stuck with interrupts hard disabled, the soft-NMI watchdog does not work, but the SMP watchdog will. Even on platforms without a true NMI IPI to get a good trace from the stuck CPU, other CPUs will notice the lockup sufficiently to report it and panic. [npiggin@gmail.com: honor watchdog disable at boot/hotplug] Link: http://lkml.kernel.org/r/20170621001346.5bb337c9@roar.ozlabs.ibm.com [npiggin@gmail.com: fix false positive warning at CPU unplug] Link: http://lkml.kernel.org/r/20170630080740.20766-1-npiggin@gmail.com [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20170616065715.18390-6-npiggin@gmail.com Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Reviewed-by: Don Zickus <dzickus@redhat.com> Tested-by: Babu Moger <babu.moger@oracle.com> [sparc] Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-13 05:35:52 +08:00
#define MASKED_DEC_HANDLER_LABEL 3f
#define MASKED_DEC_HANDLER(_H) \
3: /* soft-nmi */ \
std r12,PACA_EXGEN+EX_R12(r13); \
GET_SCRATCH0(r10); \
std r10,PACA_EXGEN+EX_R13(r13); \
INT_SAVE_SRR_AND_JUMP soft_nmi_common, _H, 1
powerpc/64s: implement arch-specific hardlockup watchdog Implement an arch-speicfic watchdog rather than use the perf-based hardlockup detector. The new watchdog takes the soft-NMI directly, rather than going through perf. Perf interrupts are to be made maskable in future, so that would prevent the perf detector from working in those regions. Additionally, implement a SMP based detector where all CPUs watch one another by pinging a shared cpumask. This is because powerpc Book3S does not have a true periodic local NMI, but some platforms do implement a true NMI IPI. If a CPU is stuck with interrupts hard disabled, the soft-NMI watchdog does not work, but the SMP watchdog will. Even on platforms without a true NMI IPI to get a good trace from the stuck CPU, other CPUs will notice the lockup sufficiently to report it and panic. [npiggin@gmail.com: honor watchdog disable at boot/hotplug] Link: http://lkml.kernel.org/r/20170621001346.5bb337c9@roar.ozlabs.ibm.com [npiggin@gmail.com: fix false positive warning at CPU unplug] Link: http://lkml.kernel.org/r/20170630080740.20766-1-npiggin@gmail.com [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20170616065715.18390-6-npiggin@gmail.com Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Reviewed-by: Don Zickus <dzickus@redhat.com> Tested-by: Babu Moger <babu.moger@oracle.com> [sparc] Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-13 05:35:52 +08:00
/*
* Branch to soft_nmi_interrupt using the emergency stack. The emergency
* stack is one that is usable by maskable interrupts so long as MSR_EE
* remains off. It is used for recovery when something has corrupted the
* normal kernel stack, for example. The "soft NMI" must not use the process
* stack because we want irq disabled sections to avoid touching the stack
* at all (other than PMU interrupts), so use the emergency stack for this,
* and run it entirely with interrupts hard disabled.
*/
powerpc/64s: implement arch-specific hardlockup watchdog Implement an arch-speicfic watchdog rather than use the perf-based hardlockup detector. The new watchdog takes the soft-NMI directly, rather than going through perf. Perf interrupts are to be made maskable in future, so that would prevent the perf detector from working in those regions. Additionally, implement a SMP based detector where all CPUs watch one another by pinging a shared cpumask. This is because powerpc Book3S does not have a true periodic local NMI, but some platforms do implement a true NMI IPI. If a CPU is stuck with interrupts hard disabled, the soft-NMI watchdog does not work, but the SMP watchdog will. Even on platforms without a true NMI IPI to get a good trace from the stuck CPU, other CPUs will notice the lockup sufficiently to report it and panic. [npiggin@gmail.com: honor watchdog disable at boot/hotplug] Link: http://lkml.kernel.org/r/20170621001346.5bb337c9@roar.ozlabs.ibm.com [npiggin@gmail.com: fix false positive warning at CPU unplug] Link: http://lkml.kernel.org/r/20170630080740.20766-1-npiggin@gmail.com [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20170616065715.18390-6-npiggin@gmail.com Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Reviewed-by: Don Zickus <dzickus@redhat.com> Tested-by: Babu Moger <babu.moger@oracle.com> [sparc] Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-13 05:35:52 +08:00
EXC_COMMON_BEGIN(soft_nmi_common)
mr r10,r1
ld r1,PACAEMERGSP(r13)
subi r1,r1,INT_FRAME_SIZE
INT_COMMON 0x900, PACA_EXGEN, 0, 1, 1, 0, 0
bl save_nvgprs
addi r3,r1,STACK_FRAME_OVERHEAD
bl soft_nmi_interrupt
powerpc/64s: implement arch-specific hardlockup watchdog Implement an arch-speicfic watchdog rather than use the perf-based hardlockup detector. The new watchdog takes the soft-NMI directly, rather than going through perf. Perf interrupts are to be made maskable in future, so that would prevent the perf detector from working in those regions. Additionally, implement a SMP based detector where all CPUs watch one another by pinging a shared cpumask. This is because powerpc Book3S does not have a true periodic local NMI, but some platforms do implement a true NMI IPI. If a CPU is stuck with interrupts hard disabled, the soft-NMI watchdog does not work, but the SMP watchdog will. Even on platforms without a true NMI IPI to get a good trace from the stuck CPU, other CPUs will notice the lockup sufficiently to report it and panic. [npiggin@gmail.com: honor watchdog disable at boot/hotplug] Link: http://lkml.kernel.org/r/20170621001346.5bb337c9@roar.ozlabs.ibm.com [npiggin@gmail.com: fix false positive warning at CPU unplug] Link: http://lkml.kernel.org/r/20170630080740.20766-1-npiggin@gmail.com [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20170616065715.18390-6-npiggin@gmail.com Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Reviewed-by: Don Zickus <dzickus@redhat.com> Tested-by: Babu Moger <babu.moger@oracle.com> [sparc] Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-13 05:35:52 +08:00
b ret_from_except
#else /* CONFIG_PPC_WATCHDOG */
powerpc/64s: implement arch-specific hardlockup watchdog Implement an arch-speicfic watchdog rather than use the perf-based hardlockup detector. The new watchdog takes the soft-NMI directly, rather than going through perf. Perf interrupts are to be made maskable in future, so that would prevent the perf detector from working in those regions. Additionally, implement a SMP based detector where all CPUs watch one another by pinging a shared cpumask. This is because powerpc Book3S does not have a true periodic local NMI, but some platforms do implement a true NMI IPI. If a CPU is stuck with interrupts hard disabled, the soft-NMI watchdog does not work, but the SMP watchdog will. Even on platforms without a true NMI IPI to get a good trace from the stuck CPU, other CPUs will notice the lockup sufficiently to report it and panic. [npiggin@gmail.com: honor watchdog disable at boot/hotplug] Link: http://lkml.kernel.org/r/20170621001346.5bb337c9@roar.ozlabs.ibm.com [npiggin@gmail.com: fix false positive warning at CPU unplug] Link: http://lkml.kernel.org/r/20170630080740.20766-1-npiggin@gmail.com [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20170616065715.18390-6-npiggin@gmail.com Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Reviewed-by: Don Zickus <dzickus@redhat.com> Tested-by: Babu Moger <babu.moger@oracle.com> [sparc] Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-13 05:35:52 +08:00
#define MASKED_DEC_HANDLER_LABEL 2f /* normal return */
#define MASKED_DEC_HANDLER(_H)
#endif /* CONFIG_PPC_WATCHDOG */
/*
* An interrupt came in while soft-disabled. We set paca->irq_happened, then:
* - If it was a decrementer interrupt, we bump the dec to max and and return.
* - If it was a doorbell we return immediately since doorbells are edge
* triggered and won't automatically refire.
* - If it was a HMI we return immediately since we handled it in realmode
* and it won't refire.
powerpc/64s: Fix may_hard_irq_enable() for PMI soft masking The soft IRQ masking code has to hard-disable interrupts in cases where the exception is not cleared by the masked handler. External interrupts used this approach for soft masking. Now recently PMU interrupts do the same thing. The soft IRQ masking code additionally allowed for interrupt handlers to hard-enable interrupts after soft-disabling them. The idea is to allow PMU interrupts through to profile interrupt handlers. So when interrupts are being replayed when there is a pending interrupt that requires hard-disabling, there is a test to prevent those handlers from hard-enabling them if there is a pending external interrupt. may_hard_irq_enable() handles this. After f442d00480 ("powerpc/64s: Add support to mask perf interrupts and replay them"), may_hard_irq_enable() could prematurely enable MSR[EE] when a PMU exception exists, which would result in the interrupt firing again while masked, and MSR[EE] being disabled again. I haven't seen that this could cause a serious problem, but it's more consistent to handle these soft-masked interrupts in the same way. So introduce a define for all types of interrupts that require MSR[EE] masking in their soft-disable handlers, and use that in may_hard_irq_enable(). Fixes: f442d004806e ("powerpc/64s: Add support to mask perf interrupts and replay them") Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Reviewed-by: Madhavan Srinivasan <maddy@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-02-03 15:17:50 +08:00
* - Else it is one of PACA_IRQ_MUST_HARD_MASK, so hard disable and return.
* This is called with r10 containing the value to OR to the paca field.
*/
.macro MASKED_INTERRUPT hsrr
.if \hsrr
masked_Hinterrupt:
.else
masked_interrupt:
.endif
std r11,PACA_EXGEN+EX_R11(r13)
lbz r11,PACAIRQHAPPENED(r13)
or r11,r11,r10
stb r11,PACAIRQHAPPENED(r13)
cmpwi r10,PACA_IRQ_DEC
bne 1f
lis r10,0x7fff
ori r10,r10,0xffff
mtspr SPRN_DEC,r10
b MASKED_DEC_HANDLER_LABEL
1: andi. r10,r10,PACA_IRQ_MUST_HARD_MASK
beq 2f
.if \hsrr
mfspr r10,SPRN_HSRR1
xori r10,r10,MSR_EE /* clear MSR_EE */
mtspr SPRN_HSRR1,r10
.else
mfspr r10,SPRN_SRR1
xori r10,r10,MSR_EE /* clear MSR_EE */
mtspr SPRN_SRR1,r10
.endif
ori r11,r11,PACA_IRQ_HARD_DIS
stb r11,PACAIRQHAPPENED(r13)
2: /* done */
mtcrf 0x80,r9
std r1,PACAR1(r13)
ld r9,PACA_EXGEN+EX_R9(r13)
ld r10,PACA_EXGEN+EX_R10(r13)
ld r11,PACA_EXGEN+EX_R11(r13)
/* returns to kernel where r13 must be set up, so don't restore it */
.if \hsrr
HRFI_TO_KERNEL
.else
RFI_TO_KERNEL
.endif
b .
MASKED_DEC_HANDLER(\hsrr\())
.endm
TRAMP_REAL_BEGIN(stf_barrier_fallback)
std r9,PACA_EXRFI+EX_R9(r13)
std r10,PACA_EXRFI+EX_R10(r13)
sync
ld r9,PACA_EXRFI+EX_R9(r13)
ld r10,PACA_EXRFI+EX_R10(r13)
ori 31,31,0
.rept 14
b 1f
1:
.endr
blr
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
TRAMP_REAL_BEGIN(rfi_flush_fallback)
SET_SCRATCH0(r13);
GET_PACA(r13);
powerpc/64s: Make rfi_flush_fallback a little more robust Because rfi_flush_fallback runs immediately before the return to userspace it currently runs with the user r1 (stack pointer). This means if we oops in there we will report a bad kernel stack pointer in the exception entry path, eg: Bad kernel stack pointer 7ffff7150e40 at c0000000000023b4 Oops: Bad kernel stack pointer, sig: 6 [#1] LE SMP NR_CPUS=32 NUMA PowerNV Modules linked in: CPU: 0 PID: 1246 Comm: klogd Not tainted 4.18.0-rc2-gcc-7.3.1-00175-g0443f8a69ba3 #7 NIP: c0000000000023b4 LR: 0000000010053e00 CTR: 0000000000000040 REGS: c0000000fffe7d40 TRAP: 4100 Not tainted (4.18.0-rc2-gcc-7.3.1-00175-g0443f8a69ba3) MSR: 9000000002803031 <SF,HV,VEC,VSX,FP,ME,IR,DR,LE> CR: 44000442 XER: 20000000 CFAR: c00000000000bac8 IRQMASK: c0000000f1e66a80 GPR00: 0000000002000000 00007ffff7150e40 00007fff93a99900 0000000000000020 ... NIP [c0000000000023b4] rfi_flush_fallback+0x34/0x80 LR [0000000010053e00] 0x10053e00 Although the NIP tells us where we were, and the TRAP number tells us what happened, it would still be nicer if we could report the actual exception rather than barfing about the stack pointer. We an do that fairly simply by loading the kernel stack pointer on entry and restoring the user value before returning. That way we see a regular oops such as: Unrecoverable exception 4100 at c00000000000239c Oops: Unrecoverable exception, sig: 6 [#1] LE SMP NR_CPUS=32 NUMA PowerNV Modules linked in: CPU: 0 PID: 1251 Comm: klogd Not tainted 4.18.0-rc3-gcc-7.3.1-00097-g4ebfcac65acd-dirty #40 NIP: c00000000000239c LR: 0000000010053e00 CTR: 0000000000000040 REGS: c0000000f1e17bb0 TRAP: 4100 Not tainted (4.18.0-rc3-gcc-7.3.1-00097-g4ebfcac65acd-dirty) MSR: 9000000002803031 <SF,HV,VEC,VSX,FP,ME,IR,DR,LE> CR: 44000442 XER: 20000000 CFAR: c00000000000bac8 IRQMASK: 0 ... NIP [c00000000000239c] rfi_flush_fallback+0x3c/0x80 LR [0000000010053e00] 0x10053e00 Call Trace: [c0000000f1e17e30] [c00000000000b9e4] system_call+0x5c/0x70 (unreliable) Note this shouldn't make the kernel stack pointer vulnerable to a meltdown attack, because it should be flushed from the cache before we return to userspace. The user r1 value will be in the cache, because we load it in the return path, but that is harmless. Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
2018-07-26 20:42:44 +08:00
std r1,PACA_EXRFI+EX_R12(r13)
ld r1,PACAKSAVE(r13)
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
std r9,PACA_EXRFI+EX_R9(r13)
std r10,PACA_EXRFI+EX_R10(r13)
std r11,PACA_EXRFI+EX_R11(r13)
mfctr r9
ld r10,PACA_RFI_FLUSH_FALLBACK_AREA(r13)
ld r11,PACA_L1D_FLUSH_SIZE(r13)
srdi r11,r11,(7 + 3) /* 128 byte lines, unrolled 8x */
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
mtctr r11
DCBT_BOOK3S_STOP_ALL_STREAM_IDS(r11) /* Stop prefetch streams */
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
/* order ld/st prior to dcbt stop all streams with flushing */
sync
/*
* The load adresses are at staggered offsets within cachelines,
* which suits some pipelines better (on others it should not
* hurt).
*/
1:
ld r11,(0x80 + 8)*0(r10)
ld r11,(0x80 + 8)*1(r10)
ld r11,(0x80 + 8)*2(r10)
ld r11,(0x80 + 8)*3(r10)
ld r11,(0x80 + 8)*4(r10)
ld r11,(0x80 + 8)*5(r10)
ld r11,(0x80 + 8)*6(r10)
ld r11,(0x80 + 8)*7(r10)
addi r10,r10,0x80*8
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
bdnz 1b
mtctr r9
ld r9,PACA_EXRFI+EX_R9(r13)
ld r10,PACA_EXRFI+EX_R10(r13)
ld r11,PACA_EXRFI+EX_R11(r13)
powerpc/64s: Make rfi_flush_fallback a little more robust Because rfi_flush_fallback runs immediately before the return to userspace it currently runs with the user r1 (stack pointer). This means if we oops in there we will report a bad kernel stack pointer in the exception entry path, eg: Bad kernel stack pointer 7ffff7150e40 at c0000000000023b4 Oops: Bad kernel stack pointer, sig: 6 [#1] LE SMP NR_CPUS=32 NUMA PowerNV Modules linked in: CPU: 0 PID: 1246 Comm: klogd Not tainted 4.18.0-rc2-gcc-7.3.1-00175-g0443f8a69ba3 #7 NIP: c0000000000023b4 LR: 0000000010053e00 CTR: 0000000000000040 REGS: c0000000fffe7d40 TRAP: 4100 Not tainted (4.18.0-rc2-gcc-7.3.1-00175-g0443f8a69ba3) MSR: 9000000002803031 <SF,HV,VEC,VSX,FP,ME,IR,DR,LE> CR: 44000442 XER: 20000000 CFAR: c00000000000bac8 IRQMASK: c0000000f1e66a80 GPR00: 0000000002000000 00007ffff7150e40 00007fff93a99900 0000000000000020 ... NIP [c0000000000023b4] rfi_flush_fallback+0x34/0x80 LR [0000000010053e00] 0x10053e00 Although the NIP tells us where we were, and the TRAP number tells us what happened, it would still be nicer if we could report the actual exception rather than barfing about the stack pointer. We an do that fairly simply by loading the kernel stack pointer on entry and restoring the user value before returning. That way we see a regular oops such as: Unrecoverable exception 4100 at c00000000000239c Oops: Unrecoverable exception, sig: 6 [#1] LE SMP NR_CPUS=32 NUMA PowerNV Modules linked in: CPU: 0 PID: 1251 Comm: klogd Not tainted 4.18.0-rc3-gcc-7.3.1-00097-g4ebfcac65acd-dirty #40 NIP: c00000000000239c LR: 0000000010053e00 CTR: 0000000000000040 REGS: c0000000f1e17bb0 TRAP: 4100 Not tainted (4.18.0-rc3-gcc-7.3.1-00097-g4ebfcac65acd-dirty) MSR: 9000000002803031 <SF,HV,VEC,VSX,FP,ME,IR,DR,LE> CR: 44000442 XER: 20000000 CFAR: c00000000000bac8 IRQMASK: 0 ... NIP [c00000000000239c] rfi_flush_fallback+0x3c/0x80 LR [0000000010053e00] 0x10053e00 Call Trace: [c0000000f1e17e30] [c00000000000b9e4] system_call+0x5c/0x70 (unreliable) Note this shouldn't make the kernel stack pointer vulnerable to a meltdown attack, because it should be flushed from the cache before we return to userspace. The user r1 value will be in the cache, because we load it in the return path, but that is harmless. Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
2018-07-26 20:42:44 +08:00
ld r1,PACA_EXRFI+EX_R12(r13)
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
GET_SCRATCH0(r13);
rfid
TRAMP_REAL_BEGIN(hrfi_flush_fallback)
SET_SCRATCH0(r13);
GET_PACA(r13);
powerpc/64s: Make rfi_flush_fallback a little more robust Because rfi_flush_fallback runs immediately before the return to userspace it currently runs with the user r1 (stack pointer). This means if we oops in there we will report a bad kernel stack pointer in the exception entry path, eg: Bad kernel stack pointer 7ffff7150e40 at c0000000000023b4 Oops: Bad kernel stack pointer, sig: 6 [#1] LE SMP NR_CPUS=32 NUMA PowerNV Modules linked in: CPU: 0 PID: 1246 Comm: klogd Not tainted 4.18.0-rc2-gcc-7.3.1-00175-g0443f8a69ba3 #7 NIP: c0000000000023b4 LR: 0000000010053e00 CTR: 0000000000000040 REGS: c0000000fffe7d40 TRAP: 4100 Not tainted (4.18.0-rc2-gcc-7.3.1-00175-g0443f8a69ba3) MSR: 9000000002803031 <SF,HV,VEC,VSX,FP,ME,IR,DR,LE> CR: 44000442 XER: 20000000 CFAR: c00000000000bac8 IRQMASK: c0000000f1e66a80 GPR00: 0000000002000000 00007ffff7150e40 00007fff93a99900 0000000000000020 ... NIP [c0000000000023b4] rfi_flush_fallback+0x34/0x80 LR [0000000010053e00] 0x10053e00 Although the NIP tells us where we were, and the TRAP number tells us what happened, it would still be nicer if we could report the actual exception rather than barfing about the stack pointer. We an do that fairly simply by loading the kernel stack pointer on entry and restoring the user value before returning. That way we see a regular oops such as: Unrecoverable exception 4100 at c00000000000239c Oops: Unrecoverable exception, sig: 6 [#1] LE SMP NR_CPUS=32 NUMA PowerNV Modules linked in: CPU: 0 PID: 1251 Comm: klogd Not tainted 4.18.0-rc3-gcc-7.3.1-00097-g4ebfcac65acd-dirty #40 NIP: c00000000000239c LR: 0000000010053e00 CTR: 0000000000000040 REGS: c0000000f1e17bb0 TRAP: 4100 Not tainted (4.18.0-rc3-gcc-7.3.1-00097-g4ebfcac65acd-dirty) MSR: 9000000002803031 <SF,HV,VEC,VSX,FP,ME,IR,DR,LE> CR: 44000442 XER: 20000000 CFAR: c00000000000bac8 IRQMASK: 0 ... NIP [c00000000000239c] rfi_flush_fallback+0x3c/0x80 LR [0000000010053e00] 0x10053e00 Call Trace: [c0000000f1e17e30] [c00000000000b9e4] system_call+0x5c/0x70 (unreliable) Note this shouldn't make the kernel stack pointer vulnerable to a meltdown attack, because it should be flushed from the cache before we return to userspace. The user r1 value will be in the cache, because we load it in the return path, but that is harmless. Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
2018-07-26 20:42:44 +08:00
std r1,PACA_EXRFI+EX_R12(r13)
ld r1,PACAKSAVE(r13)
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
std r9,PACA_EXRFI+EX_R9(r13)
std r10,PACA_EXRFI+EX_R10(r13)
std r11,PACA_EXRFI+EX_R11(r13)
mfctr r9
ld r10,PACA_RFI_FLUSH_FALLBACK_AREA(r13)
ld r11,PACA_L1D_FLUSH_SIZE(r13)
srdi r11,r11,(7 + 3) /* 128 byte lines, unrolled 8x */
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
mtctr r11
DCBT_BOOK3S_STOP_ALL_STREAM_IDS(r11) /* Stop prefetch streams */
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
/* order ld/st prior to dcbt stop all streams with flushing */
sync
/*
* The load adresses are at staggered offsets within cachelines,
* which suits some pipelines better (on others it should not
* hurt).
*/
1:
ld r11,(0x80 + 8)*0(r10)
ld r11,(0x80 + 8)*1(r10)
ld r11,(0x80 + 8)*2(r10)
ld r11,(0x80 + 8)*3(r10)
ld r11,(0x80 + 8)*4(r10)
ld r11,(0x80 + 8)*5(r10)
ld r11,(0x80 + 8)*6(r10)
ld r11,(0x80 + 8)*7(r10)
addi r10,r10,0x80*8
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
bdnz 1b
mtctr r9
ld r9,PACA_EXRFI+EX_R9(r13)
ld r10,PACA_EXRFI+EX_R10(r13)
ld r11,PACA_EXRFI+EX_R11(r13)
powerpc/64s: Make rfi_flush_fallback a little more robust Because rfi_flush_fallback runs immediately before the return to userspace it currently runs with the user r1 (stack pointer). This means if we oops in there we will report a bad kernel stack pointer in the exception entry path, eg: Bad kernel stack pointer 7ffff7150e40 at c0000000000023b4 Oops: Bad kernel stack pointer, sig: 6 [#1] LE SMP NR_CPUS=32 NUMA PowerNV Modules linked in: CPU: 0 PID: 1246 Comm: klogd Not tainted 4.18.0-rc2-gcc-7.3.1-00175-g0443f8a69ba3 #7 NIP: c0000000000023b4 LR: 0000000010053e00 CTR: 0000000000000040 REGS: c0000000fffe7d40 TRAP: 4100 Not tainted (4.18.0-rc2-gcc-7.3.1-00175-g0443f8a69ba3) MSR: 9000000002803031 <SF,HV,VEC,VSX,FP,ME,IR,DR,LE> CR: 44000442 XER: 20000000 CFAR: c00000000000bac8 IRQMASK: c0000000f1e66a80 GPR00: 0000000002000000 00007ffff7150e40 00007fff93a99900 0000000000000020 ... NIP [c0000000000023b4] rfi_flush_fallback+0x34/0x80 LR [0000000010053e00] 0x10053e00 Although the NIP tells us where we were, and the TRAP number tells us what happened, it would still be nicer if we could report the actual exception rather than barfing about the stack pointer. We an do that fairly simply by loading the kernel stack pointer on entry and restoring the user value before returning. That way we see a regular oops such as: Unrecoverable exception 4100 at c00000000000239c Oops: Unrecoverable exception, sig: 6 [#1] LE SMP NR_CPUS=32 NUMA PowerNV Modules linked in: CPU: 0 PID: 1251 Comm: klogd Not tainted 4.18.0-rc3-gcc-7.3.1-00097-g4ebfcac65acd-dirty #40 NIP: c00000000000239c LR: 0000000010053e00 CTR: 0000000000000040 REGS: c0000000f1e17bb0 TRAP: 4100 Not tainted (4.18.0-rc3-gcc-7.3.1-00097-g4ebfcac65acd-dirty) MSR: 9000000002803031 <SF,HV,VEC,VSX,FP,ME,IR,DR,LE> CR: 44000442 XER: 20000000 CFAR: c00000000000bac8 IRQMASK: 0 ... NIP [c00000000000239c] rfi_flush_fallback+0x3c/0x80 LR [0000000010053e00] 0x10053e00 Call Trace: [c0000000f1e17e30] [c00000000000b9e4] system_call+0x5c/0x70 (unreliable) Note this shouldn't make the kernel stack pointer vulnerable to a meltdown attack, because it should be flushed from the cache before we return to userspace. The user r1 value will be in the cache, because we load it in the return path, but that is harmless. Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
2018-07-26 20:42:44 +08:00
ld r1,PACA_EXRFI+EX_R12(r13)
powerpc/64s: Add support for RFI flush of L1-D cache On some CPUs we can prevent the Meltdown vulnerability by flushing the L1-D cache on exit from kernel to user mode, and from hypervisor to guest. This is known to be the case on at least Power7, Power8 and Power9. At this time we do not know the status of the vulnerability on other CPUs such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale CPUs. As more information comes to light we can enable this, or other mechanisms on those CPUs. The vulnerability occurs when the load of an architecturally inaccessible memory region (eg. userspace load of kernel memory) is speculatively executed to the point where its result can influence the address of a subsequent speculatively executed load. In order for that to happen, the first load must hit in the L1, because before the load is sent to the L2 the permission check is performed. Therefore if no kernel addresses hit in the L1 the vulnerability can not occur. We can ensure that is the case by flushing the L1 whenever we return to userspace. Similarly for hypervisor vs guest. In order to flush the L1-D cache on exit, we add a section of nops at each (h)rfi location that returns to a lower privileged context, and patch that with some sequence. Newer firmwares are able to advertise to us that there is a special nop instruction that flushes the L1-D. If we do not see that advertised, we fall back to doing a displacement flush in software. For guest kernels we support migration between some CPU versions, and different CPUs may use different flush instructions. So that we are prepared to migrate to a machine with a different flush instruction activated, we may have to patch more than one flush instruction at boot if the hypervisor tells us to. In the end this patch is mostly the work of Nicholas Piggin and Michael Ellerman. However a cast of thousands contributed to analysis of the issue, earlier versions of the patch, back ports testing etc. Many thanks to all of them. Tested-by: Jon Masters <jcm@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-01-10 00:07:15 +08:00
GET_SCRATCH0(r13);
hrfid
/*
* Real mode exceptions actually use this too, but alternate
* instruction code patches (which end up in the common .text area)
* cannot reach these if they are put there.
*/
USE_FIXED_SECTION(virt_trampolines)
MASKED_INTERRUPT EXC_STD
MASKED_INTERRUPT EXC_HV
#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
TRAMP_REAL_BEGIN(kvmppc_skip_interrupt)
/*
* Here all GPRs are unchanged from when the interrupt happened
* except for r13, which is saved in SPRG_SCRATCH0.
*/
mfspr r13, SPRN_SRR0
addi r13, r13, 4
mtspr SPRN_SRR0, r13
GET_SCRATCH0(r13)
RFI_TO_KERNEL
b .
TRAMP_REAL_BEGIN(kvmppc_skip_Hinterrupt)
/*
* Here all GPRs are unchanged from when the interrupt happened
* except for r13, which is saved in SPRG_SCRATCH0.
*/
mfspr r13, SPRN_HSRR0
addi r13, r13, 4
mtspr SPRN_HSRR0, r13
GET_SCRATCH0(r13)
HRFI_TO_KERNEL
b .
#endif
/*
* Ensure that any handlers that get invoked from the exception prologs
* above are below the first 64KB (0x10000) of the kernel image because
* the prologs assemble the addresses of these handlers using the
* LOAD_HANDLER macro, which uses an ori instruction.
*/
/*** Common interrupt handlers ***/
/*
* Relocation-on interrupts: A subset of the interrupts can be delivered
* with IR=1/DR=1, if AIL==2 and MSR.HV won't be changed by delivering
* it. Addresses are the same as the original interrupt addresses, but
* offset by 0xc000000000004000.
* It's impossible to receive interrupts below 0x300 via this mechanism.
* KVM: None of these traps are from the guest ; anything that escalated
* to HV=1 from HV=0 is delivered via real mode handlers.
*/
/*
* This uses the standard macro, since the original 0x300 vector
* only has extra guff for STAB-based processors -- which never
* come here.
*/
EXC_COMMON_BEGIN(ppc64_runlatch_on_trampoline)
b __ppc64_runlatch_on
USE_FIXED_SECTION(virt_trampolines)
powerpc/book3s64: Fix branching to OOL handlers in relocatable kernel Some of the interrupt vectors on 64-bit POWER server processors are only 32 bytes long (8 instructions), which is not enough for the full first-level interrupt handler. For these we need to branch to an out-of-line (OOL) handler. But when we are running a relocatable kernel, interrupt vectors till __end_interrupts marker are copied down to real address 0x100. So, branching to labels (ie. OOL handlers) outside this section must be handled differently (see LOAD_HANDLER()), considering relocatable kernel, which would need at least 4 instructions. However, branching from interrupt vector means that we corrupt the CFAR (come-from address register) on POWER7 and later processors as mentioned in commit 1707dd16. So, EXCEPTION_PROLOG_0 (6 instructions) that contains the part up to the point where the CFAR is saved in the PACA should be part of the short interrupt vectors before we branch out to OOL handlers. But as mentioned already, there are interrupt vectors on 64-bit POWER server processors that are only 32 bytes long (like vectors 0x4f00, 0x4f20, etc.), which cannot accomodate the above two cases at the same time owing to space constraint. Currently, in these interrupt vectors, we simply branch out to OOL handlers, without using LOAD_HANDLER(), which leaves us vulnerable when running a relocatable kernel (eg. kdump case). While this has been the case for sometime now and kdump is used widely, we were fortunate not to see any problems so far, for three reasons: 1. In almost all cases, production kernel (relocatable) is used for kdump as well, which would mean that crashed kernel's OOL handler would be at the same place where we end up branching to, from short interrupt vector of kdump kernel. 2. Also, OOL handler was unlikely the reason for crash in almost all the kdump scenarios, which meant we had a sane OOL handler from crashed kernel that we branched to. 3. On most 64-bit POWER server processors, page size is large enough that marking interrupt vector code as executable (see commit 429d2e83) leads to marking OOL handler code from crashed kernel, that sits right below interrupt vector code from kdump kernel, as executable as well. Let us fix this by moving the __end_interrupts marker down past OOL handlers to make sure that we also copy OOL handlers to real address 0x100 when running a relocatable kernel. This fix has been tested successfully in kdump scenario, on an LPAR with 4K page size by using different default/production kernel and kdump kernel. Also tested by manually corrupting the OOL handlers in the first kernel and then kdump'ing, and then causing the OOL handlers to fire - mpe. Fixes: c1fb6816fb1b ("powerpc: Add relocation on exception vector handlers") Cc: stable@vger.kernel.org Signed-off-by: Hari Bathini <hbathini@linux.vnet.ibm.com> Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2016-04-15 20:48:02 +08:00
/*
* The __end_interrupts marker must be past the out-of-line (OOL)
* handlers, so that they are copied to real address 0x100 when running
* a relocatable kernel. This ensures they can be reached from the short
* trampoline handlers (like 0x4f00, 0x4f20, etc.) which branch
* directly, without using LOAD_HANDLER().
*/
.align 7
.globl __end_interrupts
__end_interrupts:
DEFINE_FIXED_SYMBOL(__end_interrupts)
#ifdef CONFIG_PPC_970_NAP
/*
* Called by exception entry code if _TLF_NAPPING was set, this clears
* the NAPPING flag, and redirects the exception exit to
* power4_fixup_nap_return.
*/
.globl power4_fixup_nap
EXC_COMMON_BEGIN(power4_fixup_nap)
andc r9,r9,r10
std r9,TI_LOCAL_FLAGS(r11)
LOAD_REG_ADDR(r10, power4_idle_nap_return)
std r10,_NIP(r1)
blr
power4_idle_nap_return:
blr
#endif
CLOSE_FIXED_SECTION(real_vectors);
CLOSE_FIXED_SECTION(real_trampolines);
CLOSE_FIXED_SECTION(virt_vectors);
CLOSE_FIXED_SECTION(virt_trampolines);
USE_TEXT_SECTION()
/* MSR[RI] should be clear because this uses SRR[01] */
enable_machine_check:
mflr r0
bcl 20,31,$+4
0: mflr r3
addi r3,r3,(1f - 0b)
mtspr SPRN_SRR0,r3
mfmsr r3
ori r3,r3,MSR_ME
mtspr SPRN_SRR1,r3
RFI_TO_KERNEL
1: mtlr r0
blr
/* MSR[RI] should be clear because this uses SRR[01] */
disable_machine_check:
mflr r0
bcl 20,31,$+4
0: mflr r3
addi r3,r3,(1f - 0b)
mtspr SPRN_SRR0,r3
mfmsr r3
li r4,MSR_ME
andc r3,r3,r4
mtspr SPRN_SRR1,r3
RFI_TO_KERNEL
1: mtlr r0
blr
/*
* Hash table stuff
*/
.balign IFETCH_ALIGN_BYTES
do_hash_page:
#ifdef CONFIG_PPC_BOOK3S_64
lis r0,(DSISR_BAD_FAULT_64S | DSISR_DABRMATCH | DSISR_KEYFAULT)@h
ori r0,r0,DSISR_BAD_FAULT_64S@l
and. r0,r5,r0 /* weird error? */
bne- handle_page_fault /* if not, try to insert a HPTE */
ld r11, PACA_THREAD_INFO(r13)
powerpc: Allow perf_counters to access user memory at interrupt time This provides a mechanism to allow the perf_counters code to access user memory in a PMU interrupt routine. Such an access can cause various kinds of interrupt: SLB miss, MMU hash table miss, segment table miss, or TLB miss, depending on the processor. This commit only deals with 64-bit classic/server processors, which use an MMU hash table. 32-bit processors are already able to access user memory at interrupt time. Since we don't soft-disable on 32-bit, we avoid the possibility of reentering hash_page or the TLB miss handlers, since they run with interrupts disabled. On 64-bit processors, an SLB miss interrupt on a user address will update the slb_cache and slb_cache_ptr fields in the paca. This is OK except in the case where a PMU interrupt occurs in switch_slb, which also accesses those fields. To prevent this, we hard-disable interrupts in switch_slb. Interrupts are already soft-disabled at this point, and will get hard-enabled when they get soft-enabled later. This also reworks slb_flush_and_rebolt: to avoid hard-disabling twice, and to make sure that it clears the slb_cache_ptr when called from other callers than switch_slb, the existing routine is renamed to __slb_flush_and_rebolt, which is called by switch_slb and the new version of slb_flush_and_rebolt. Similarly, switch_stab (used on POWER3 and RS64 processors) gets a hard_irq_disable() to protect the per-cpu variables used there and in ste_allocate. If a MMU hashtable miss interrupt occurs, normally we would call hash_page to look up the Linux PTE for the address and create a HPTE. However, hash_page is fairly complex and takes some locks, so to avoid the possibility of deadlock, we check the preemption count to see if we are in a (pseudo-)NMI handler, and if so, we don't call hash_page but instead treat it like a bad access that will get reported up through the exception table mechanism. An interrupt whose handler runs even though the interrupt occurred when soft-disabled (such as the PMU interrupt) is considered a pseudo-NMI handler, which should use nmi_enter()/nmi_exit() rather than irq_enter()/irq_exit(). Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2009-08-17 13:17:54 +08:00
lwz r0,TI_PREEMPT(r11) /* If we're in an "NMI" */
andis. r0,r0,NMI_MASK@h /* (i.e. an irq when soft-disabled) */
bne 77f /* then don't call hash_page now */
/*
* r3 contains the trap number
* r4 contains the faulting address
* r5 contains dsisr
* r6 msr
*
powerpc: Rework lazy-interrupt handling The current implementation of lazy interrupts handling has some issues that this tries to address. We don't do the various workarounds we need to do when re-enabling interrupts in some cases such as when returning from an interrupt and thus we may still lose or get delayed decrementer or doorbell interrupts. The current scheme also makes it much harder to handle the external "edge" interrupts provided by some BookE processors when using the EPR facility (External Proxy) and the Freescale Hypervisor. Additionally, we tend to keep interrupts hard disabled in a number of cases, such as decrementer interrupts, external interrupts, or when a masked decrementer interrupt is pending. This is sub-optimal. This is an attempt at fixing it all in one go by reworking the way we do the lazy interrupt disabling from the ground up. The base idea is to replace the "hard_enabled" field with a "irq_happened" field in which we store a bit mask of what interrupt occurred while soft-disabled. When re-enabling, either via arch_local_irq_restore() or when returning from an interrupt, we can now decide what to do by testing bits in that field. We then implement replaying of the missed interrupts either by re-using the existing exception frame (in exception exit case) or via the creation of a new one from an assembly trampoline (in the arch_local_irq_enable case). This removes the need to play with the decrementer to try to create fake interrupts, among others. In addition, this adds a few refinements: - We no longer hard disable decrementer interrupts that occur while soft-disabled. We now simply bump the decrementer back to max (on BookS) or leave it stopped (on BookE) and continue with hard interrupts enabled, which means that we'll potentially get better sample quality from performance monitor interrupts. - Timer, decrementer and doorbell interrupts now hard-enable shortly after removing the source of the interrupt, which means they no longer run entirely hard disabled. Again, this will improve perf sample quality. - On Book3E 64-bit, we now make the performance monitor interrupt act as an NMI like Book3S (the necessary C code for that to work appear to already be present in the FSL perf code, notably calling nmi_enter instead of irq_enter). (This also fixes a bug where BookE perfmon interrupts could clobber r14 ... oops) - We could make "masked" decrementer interrupts act as NMIs when doing timer-based perf sampling to improve the sample quality. Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org> --- v2: - Add hard-enable to decrementer, timer and doorbells - Fix CR clobber in masked irq handling on BookE - Make embedded perf interrupt act as an NMI - Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want to retrigger an interrupt without preventing hard-enable v3: - Fix or vs. ori bug on Book3E - Fix enabling of interrupts for some exceptions on Book3E v4: - Fix resend of doorbells on return from interrupt on Book3E v5: - Rebased on top of my latest series, which involves some significant rework of some aspects of the patch. v6: - 32-bit compile fix - more compile fixes with various .config combos - factor out the asm code to soft-disable interrupts - remove the C wrapper around preempt_schedule_irq v7: - Fix a bug with hard irq state tracking on native power7
2012-03-06 15:27:59 +08:00
* at return r3 = 0 for success, 1 for page fault, negative for error
*/
bl __hash_page /* build HPTE if possible */
cmpdi r3,0 /* see if __hash_page succeeded */
powerpc: Rework lazy-interrupt handling The current implementation of lazy interrupts handling has some issues that this tries to address. We don't do the various workarounds we need to do when re-enabling interrupts in some cases such as when returning from an interrupt and thus we may still lose or get delayed decrementer or doorbell interrupts. The current scheme also makes it much harder to handle the external "edge" interrupts provided by some BookE processors when using the EPR facility (External Proxy) and the Freescale Hypervisor. Additionally, we tend to keep interrupts hard disabled in a number of cases, such as decrementer interrupts, external interrupts, or when a masked decrementer interrupt is pending. This is sub-optimal. This is an attempt at fixing it all in one go by reworking the way we do the lazy interrupt disabling from the ground up. The base idea is to replace the "hard_enabled" field with a "irq_happened" field in which we store a bit mask of what interrupt occurred while soft-disabled. When re-enabling, either via arch_local_irq_restore() or when returning from an interrupt, we can now decide what to do by testing bits in that field. We then implement replaying of the missed interrupts either by re-using the existing exception frame (in exception exit case) or via the creation of a new one from an assembly trampoline (in the arch_local_irq_enable case). This removes the need to play with the decrementer to try to create fake interrupts, among others. In addition, this adds a few refinements: - We no longer hard disable decrementer interrupts that occur while soft-disabled. We now simply bump the decrementer back to max (on BookS) or leave it stopped (on BookE) and continue with hard interrupts enabled, which means that we'll potentially get better sample quality from performance monitor interrupts. - Timer, decrementer and doorbell interrupts now hard-enable shortly after removing the source of the interrupt, which means they no longer run entirely hard disabled. Again, this will improve perf sample quality. - On Book3E 64-bit, we now make the performance monitor interrupt act as an NMI like Book3S (the necessary C code for that to work appear to already be present in the FSL perf code, notably calling nmi_enter instead of irq_enter). (This also fixes a bug where BookE perfmon interrupts could clobber r14 ... oops) - We could make "masked" decrementer interrupts act as NMIs when doing timer-based perf sampling to improve the sample quality. Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org> --- v2: - Add hard-enable to decrementer, timer and doorbells - Fix CR clobber in masked irq handling on BookE - Make embedded perf interrupt act as an NMI - Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want to retrigger an interrupt without preventing hard-enable v3: - Fix or vs. ori bug on Book3E - Fix enabling of interrupts for some exceptions on Book3E v4: - Fix resend of doorbells on return from interrupt on Book3E v5: - Rebased on top of my latest series, which involves some significant rework of some aspects of the patch. v6: - 32-bit compile fix - more compile fixes with various .config combos - factor out the asm code to soft-disable interrupts - remove the C wrapper around preempt_schedule_irq v7: - Fix a bug with hard irq state tracking on native power7
2012-03-06 15:27:59 +08:00
/* Success */
beq fast_exc_return_irq /* Return from exception on success */
powerpc: Rework lazy-interrupt handling The current implementation of lazy interrupts handling has some issues that this tries to address. We don't do the various workarounds we need to do when re-enabling interrupts in some cases such as when returning from an interrupt and thus we may still lose or get delayed decrementer or doorbell interrupts. The current scheme also makes it much harder to handle the external "edge" interrupts provided by some BookE processors when using the EPR facility (External Proxy) and the Freescale Hypervisor. Additionally, we tend to keep interrupts hard disabled in a number of cases, such as decrementer interrupts, external interrupts, or when a masked decrementer interrupt is pending. This is sub-optimal. This is an attempt at fixing it all in one go by reworking the way we do the lazy interrupt disabling from the ground up. The base idea is to replace the "hard_enabled" field with a "irq_happened" field in which we store a bit mask of what interrupt occurred while soft-disabled. When re-enabling, either via arch_local_irq_restore() or when returning from an interrupt, we can now decide what to do by testing bits in that field. We then implement replaying of the missed interrupts either by re-using the existing exception frame (in exception exit case) or via the creation of a new one from an assembly trampoline (in the arch_local_irq_enable case). This removes the need to play with the decrementer to try to create fake interrupts, among others. In addition, this adds a few refinements: - We no longer hard disable decrementer interrupts that occur while soft-disabled. We now simply bump the decrementer back to max (on BookS) or leave it stopped (on BookE) and continue with hard interrupts enabled, which means that we'll potentially get better sample quality from performance monitor interrupts. - Timer, decrementer and doorbell interrupts now hard-enable shortly after removing the source of the interrupt, which means they no longer run entirely hard disabled. Again, this will improve perf sample quality. - On Book3E 64-bit, we now make the performance monitor interrupt act as an NMI like Book3S (the necessary C code for that to work appear to already be present in the FSL perf code, notably calling nmi_enter instead of irq_enter). (This also fixes a bug where BookE perfmon interrupts could clobber r14 ... oops) - We could make "masked" decrementer interrupts act as NMIs when doing timer-based perf sampling to improve the sample quality. Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org> --- v2: - Add hard-enable to decrementer, timer and doorbells - Fix CR clobber in masked irq handling on BookE - Make embedded perf interrupt act as an NMI - Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want to retrigger an interrupt without preventing hard-enable v3: - Fix or vs. ori bug on Book3E - Fix enabling of interrupts for some exceptions on Book3E v4: - Fix resend of doorbells on return from interrupt on Book3E v5: - Rebased on top of my latest series, which involves some significant rework of some aspects of the patch. v6: - 32-bit compile fix - more compile fixes with various .config combos - factor out the asm code to soft-disable interrupts - remove the C wrapper around preempt_schedule_irq v7: - Fix a bug with hard irq state tracking on native power7
2012-03-06 15:27:59 +08:00
/* Error */
blt- 13f
/* Reload DAR/DSISR into r4/r5 for the DABR check below */
ld r4,_DAR(r1)
ld r5,_DSISR(r1)
#endif /* CONFIG_PPC_BOOK3S_64 */
/* Here we have a page fault that hash_page can't handle. */
handle_page_fault:
11: andis. r0,r5,DSISR_DABRMATCH@h
bne- handle_dabr_fault
addi r3,r1,STACK_FRAME_OVERHEAD
bl do_page_fault
cmpdi r3,0
powerpc/watchpoint: Restore NV GPRs while returning from exception powerpc hardware triggers watchpoint before executing the instruction. To make trigger-after-execute behavior, kernel emulates the instruction. If the instruction is 'load something into non-volatile register', exception handler should restore emulated register state while returning back, otherwise there will be register state corruption. eg, adding a watchpoint on a list can corrput the list: # cat /proc/kallsyms | grep kthread_create_list c00000000121c8b8 d kthread_create_list Add watchpoint on kthread_create_list->prev: # perf record -e mem:0xc00000000121c8c0 Run some workload such that new kthread gets invoked. eg, I just logged out from console: list_add corruption. next->prev should be prev (c000000001214e00), \ but was c00000000121c8b8. (next=c00000000121c8b8). WARNING: CPU: 59 PID: 309 at lib/list_debug.c:25 __list_add_valid+0xb4/0xc0 CPU: 59 PID: 309 Comm: kworker/59:0 Kdump: loaded Not tainted 5.1.0-rc7+ #69 ... NIP __list_add_valid+0xb4/0xc0 LR __list_add_valid+0xb0/0xc0 Call Trace: __list_add_valid+0xb0/0xc0 (unreliable) __kthread_create_on_node+0xe0/0x260 kthread_create_on_node+0x34/0x50 create_worker+0xe8/0x260 worker_thread+0x444/0x560 kthread+0x160/0x1a0 ret_from_kernel_thread+0x5c/0x70 List corruption happened because it uses 'load into non-volatile register' instruction: Snippet from __kthread_create_on_node: c000000000136be8: addis r29,r2,-19 c000000000136bec: ld r29,31424(r29) if (!__list_add_valid(new, prev, next)) c000000000136bf0: mr r3,r30 c000000000136bf4: mr r5,r28 c000000000136bf8: mr r4,r29 c000000000136bfc: bl c00000000059a2f8 <__list_add_valid+0x8> Register state from WARN_ON(): GPR00: c00000000059a3a0 c000007ff23afb50 c000000001344e00 0000000000000075 GPR04: 0000000000000000 0000000000000000 0000001852af8bc1 0000000000000000 GPR08: 0000000000000001 0000000000000007 0000000000000006 00000000000004aa GPR12: 0000000000000000 c000007ffffeb080 c000000000137038 c000005ff62aaa00 GPR16: 0000000000000000 0000000000000000 c000007fffbe7600 c000007fffbe7370 GPR20: c000007fffbe7320 c000007fffbe7300 c000000001373a00 0000000000000000 GPR24: fffffffffffffef7 c00000000012e320 c000007ff23afcb0 c000000000cb8628 GPR28: c00000000121c8b8 c000000001214e00 c000007fef5b17e8 c000007fef5b17c0 Watchpoint hit at 0xc000000000136bec. addis r29,r2,-19 => r29 = 0xc000000001344e00 + (-19 << 16) => r29 = 0xc000000001214e00 ld r29,31424(r29) => r29 = *(0xc000000001214e00 + 31424) => r29 = *(0xc00000000121c8c0) 0xc00000000121c8c0 is where we placed a watchpoint and thus this instruction was emulated by emulate_step. But because handle_dabr_fault did not restore emulated register state, r29 still contains stale value in above register state. Fixes: 5aae8a5370802 ("powerpc, hw_breakpoints: Implement hw_breakpoints for 64-bit server processors") Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com> Cc: stable@vger.kernel.org # 2.6.36+ Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2019-06-13 11:30:14 +08:00
beq+ ret_from_except_lite
bl save_nvgprs
mr r5,r3
addi r3,r1,STACK_FRAME_OVERHEAD
ld r4,_DAR(r1)
bl bad_page_fault
b ret_from_except
/* We have a data breakpoint exception - handle it */
handle_dabr_fault:
bl save_nvgprs
ld r4,_DAR(r1)
ld r5,_DSISR(r1)
addi r3,r1,STACK_FRAME_OVERHEAD
bl do_break
powerpc/watchpoint: Restore NV GPRs while returning from exception powerpc hardware triggers watchpoint before executing the instruction. To make trigger-after-execute behavior, kernel emulates the instruction. If the instruction is 'load something into non-volatile register', exception handler should restore emulated register state while returning back, otherwise there will be register state corruption. eg, adding a watchpoint on a list can corrput the list: # cat /proc/kallsyms | grep kthread_create_list c00000000121c8b8 d kthread_create_list Add watchpoint on kthread_create_list->prev: # perf record -e mem:0xc00000000121c8c0 Run some workload such that new kthread gets invoked. eg, I just logged out from console: list_add corruption. next->prev should be prev (c000000001214e00), \ but was c00000000121c8b8. (next=c00000000121c8b8). WARNING: CPU: 59 PID: 309 at lib/list_debug.c:25 __list_add_valid+0xb4/0xc0 CPU: 59 PID: 309 Comm: kworker/59:0 Kdump: loaded Not tainted 5.1.0-rc7+ #69 ... NIP __list_add_valid+0xb4/0xc0 LR __list_add_valid+0xb0/0xc0 Call Trace: __list_add_valid+0xb0/0xc0 (unreliable) __kthread_create_on_node+0xe0/0x260 kthread_create_on_node+0x34/0x50 create_worker+0xe8/0x260 worker_thread+0x444/0x560 kthread+0x160/0x1a0 ret_from_kernel_thread+0x5c/0x70 List corruption happened because it uses 'load into non-volatile register' instruction: Snippet from __kthread_create_on_node: c000000000136be8: addis r29,r2,-19 c000000000136bec: ld r29,31424(r29) if (!__list_add_valid(new, prev, next)) c000000000136bf0: mr r3,r30 c000000000136bf4: mr r5,r28 c000000000136bf8: mr r4,r29 c000000000136bfc: bl c00000000059a2f8 <__list_add_valid+0x8> Register state from WARN_ON(): GPR00: c00000000059a3a0 c000007ff23afb50 c000000001344e00 0000000000000075 GPR04: 0000000000000000 0000000000000000 0000001852af8bc1 0000000000000000 GPR08: 0000000000000001 0000000000000007 0000000000000006 00000000000004aa GPR12: 0000000000000000 c000007ffffeb080 c000000000137038 c000005ff62aaa00 GPR16: 0000000000000000 0000000000000000 c000007fffbe7600 c000007fffbe7370 GPR20: c000007fffbe7320 c000007fffbe7300 c000000001373a00 0000000000000000 GPR24: fffffffffffffef7 c00000000012e320 c000007ff23afcb0 c000000000cb8628 GPR28: c00000000121c8b8 c000000001214e00 c000007fef5b17e8 c000007fef5b17c0 Watchpoint hit at 0xc000000000136bec. addis r29,r2,-19 => r29 = 0xc000000001344e00 + (-19 << 16) => r29 = 0xc000000001214e00 ld r29,31424(r29) => r29 = *(0xc000000001214e00 + 31424) => r29 = *(0xc00000000121c8c0) 0xc00000000121c8c0 is where we placed a watchpoint and thus this instruction was emulated by emulate_step. But because handle_dabr_fault did not restore emulated register state, r29 still contains stale value in above register state. Fixes: 5aae8a5370802 ("powerpc, hw_breakpoints: Implement hw_breakpoints for 64-bit server processors") Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com> Cc: stable@vger.kernel.org # 2.6.36+ Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2019-06-13 11:30:14 +08:00
/*
* do_break() may have changed the NV GPRS while handling a breakpoint.
* If so, we need to restore them with their updated values. Don't use
* ret_from_except_lite here.
*/
b ret_from_except
#ifdef CONFIG_PPC_BOOK3S_64
/* We have a page fault that hash_page could handle but HV refused
* the PTE insertion
*/
13: bl save_nvgprs
mr r5,r3
addi r3,r1,STACK_FRAME_OVERHEAD
ld r4,_DAR(r1)
bl low_hash_fault
b ret_from_except
#endif
powerpc: Allow perf_counters to access user memory at interrupt time This provides a mechanism to allow the perf_counters code to access user memory in a PMU interrupt routine. Such an access can cause various kinds of interrupt: SLB miss, MMU hash table miss, segment table miss, or TLB miss, depending on the processor. This commit only deals with 64-bit classic/server processors, which use an MMU hash table. 32-bit processors are already able to access user memory at interrupt time. Since we don't soft-disable on 32-bit, we avoid the possibility of reentering hash_page or the TLB miss handlers, since they run with interrupts disabled. On 64-bit processors, an SLB miss interrupt on a user address will update the slb_cache and slb_cache_ptr fields in the paca. This is OK except in the case where a PMU interrupt occurs in switch_slb, which also accesses those fields. To prevent this, we hard-disable interrupts in switch_slb. Interrupts are already soft-disabled at this point, and will get hard-enabled when they get soft-enabled later. This also reworks slb_flush_and_rebolt: to avoid hard-disabling twice, and to make sure that it clears the slb_cache_ptr when called from other callers than switch_slb, the existing routine is renamed to __slb_flush_and_rebolt, which is called by switch_slb and the new version of slb_flush_and_rebolt. Similarly, switch_stab (used on POWER3 and RS64 processors) gets a hard_irq_disable() to protect the per-cpu variables used there and in ste_allocate. If a MMU hashtable miss interrupt occurs, normally we would call hash_page to look up the Linux PTE for the address and create a HPTE. However, hash_page is fairly complex and takes some locks, so to avoid the possibility of deadlock, we check the preemption count to see if we are in a (pseudo-)NMI handler, and if so, we don't call hash_page but instead treat it like a bad access that will get reported up through the exception table mechanism. An interrupt whose handler runs even though the interrupt occurred when soft-disabled (such as the PMU interrupt) is considered a pseudo-NMI handler, which should use nmi_enter()/nmi_exit() rather than irq_enter()/irq_exit(). Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2009-08-17 13:17:54 +08:00
/*
* We come here as a result of a DSI at a point where we don't want
* to call hash_page, such as when we are accessing memory (possibly
* user memory) inside a PMU interrupt that occurred while interrupts
* were soft-disabled. We want to invoke the exception handler for
* the access, or panic if there isn't a handler.
*/
77: bl save_nvgprs
powerpc: Allow perf_counters to access user memory at interrupt time This provides a mechanism to allow the perf_counters code to access user memory in a PMU interrupt routine. Such an access can cause various kinds of interrupt: SLB miss, MMU hash table miss, segment table miss, or TLB miss, depending on the processor. This commit only deals with 64-bit classic/server processors, which use an MMU hash table. 32-bit processors are already able to access user memory at interrupt time. Since we don't soft-disable on 32-bit, we avoid the possibility of reentering hash_page or the TLB miss handlers, since they run with interrupts disabled. On 64-bit processors, an SLB miss interrupt on a user address will update the slb_cache and slb_cache_ptr fields in the paca. This is OK except in the case where a PMU interrupt occurs in switch_slb, which also accesses those fields. To prevent this, we hard-disable interrupts in switch_slb. Interrupts are already soft-disabled at this point, and will get hard-enabled when they get soft-enabled later. This also reworks slb_flush_and_rebolt: to avoid hard-disabling twice, and to make sure that it clears the slb_cache_ptr when called from other callers than switch_slb, the existing routine is renamed to __slb_flush_and_rebolt, which is called by switch_slb and the new version of slb_flush_and_rebolt. Similarly, switch_stab (used on POWER3 and RS64 processors) gets a hard_irq_disable() to protect the per-cpu variables used there and in ste_allocate. If a MMU hashtable miss interrupt occurs, normally we would call hash_page to look up the Linux PTE for the address and create a HPTE. However, hash_page is fairly complex and takes some locks, so to avoid the possibility of deadlock, we check the preemption count to see if we are in a (pseudo-)NMI handler, and if so, we don't call hash_page but instead treat it like a bad access that will get reported up through the exception table mechanism. An interrupt whose handler runs even though the interrupt occurred when soft-disabled (such as the PMU interrupt) is considered a pseudo-NMI handler, which should use nmi_enter()/nmi_exit() rather than irq_enter()/irq_exit(). Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2009-08-17 13:17:54 +08:00
addi r3,r1,STACK_FRAME_OVERHEAD
li r5,SIGSEGV
bl bad_page_fault
b ret_from_except
/*
* When doorbell is triggered from system reset wakeup, the message is
* not cleared, so it would fire again when EE is enabled.
*
* When coming from local_irq_enable, there may be the same problem if
* we were hard disabled.
*
* Execute msgclr to clear pending exceptions before handling it.
*/
h_doorbell_common_msgclr:
LOAD_REG_IMMEDIATE(r3, PPC_DBELL_MSGTYPE << (63-36))
PPC_MSGCLR(3)
b h_doorbell_common
doorbell_super_common_msgclr:
LOAD_REG_IMMEDIATE(r3, PPC_DBELL_MSGTYPE << (63-36))
PPC_MSGCLRP(3)
b doorbell_super_common
/*
* Called from arch_local_irq_enable when an interrupt needs
* to be resent. r3 contains 0x500, 0x900, 0xa00 or 0xe80 to indicate
* which kind of interrupt. MSR:EE is already off. We generate a
* stackframe like if a real interrupt had happened.
*
* Note: While MSR:EE is off, we need to make sure that _MSR
* in the generated frame has EE set to 1 or the exception
* handler will not properly re-enable them.
*
* Note that we don't specify LR as the NIP (return address) for
* the interrupt because that would unbalance the return branch
* predictor.
*/
_GLOBAL(__replay_interrupt)
/* We are going to jump to the exception common code which
* will retrieve various register values from the PACA which
* we don't give a damn about, so we don't bother storing them.
*/
mfmsr r12
LOAD_REG_ADDR(r11, replay_interrupt_return)
mfcr r9
ori r12,r12,MSR_EE
cmpwi r3,0x900
beq decrementer_common
cmpwi r3,0x500
BEGIN_FTR_SECTION
beq h_virt_irq_common
FTR_SECTION_ELSE
beq hardware_interrupt_common
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_300)
cmpwi r3,0xf00
beq performance_monitor_common
BEGIN_FTR_SECTION
cmpwi r3,0xa00
beq h_doorbell_common_msgclr
cmpwi r3,0xe60
beq hmi_exception_common
FTR_SECTION_ELSE
cmpwi r3,0xa00
beq doorbell_super_common_msgclr
ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE)
replay_interrupt_return:
blr
_ASM_NOKPROBE_SYMBOL(__replay_interrupt)