x86/purgatory: do not use __builtin_memcpy and __builtin_memset
PATCH v2 1/2 x86/purgatory: do not use __builtin_memcpy and __builtin_memset
Implementing memcpy and memset in terms of __builtin_memcpy and __builtin_memset is problematic.
GCC at -O2 will replace calls to the builtins with calls to memcpy and memset (but will generate an inline implementation at -Os). Clang will replace the builtins with these calls regardless of optimization level.
$ llvm-objdump -dr arch/x86/purgatory/string.o | tail
0000000000000339 memcpy:
339: 48 b8 00 00 00 00 00 00 00 00 movabsq $0, %rax
000000000000033b: R_X86_64_64 memcpy
343: ff e0 jmpq *%rax0000000000000345 memset:
345: 48 b8 00 00 00 00 00 00 00 00 movabsq $0, %rax
0000000000000347: R_X86_64_64 memset
34f: ff e0Such code results in infinite recursion at runtime. This is observed when doing kexec.
Instead, reuse an implementation from arch/x86/boot/compressed/string.c if we define warn as a symbol.
---
Changes v1 -> v2:
* Add Fixes tag.
* Move this patch to first in the series.
arch/x86/purgatory/Makefile | 3 +++ arch/x86/purgatory/purgatory.c | 6 ++++++ arch/x86/purgatory/string.c | 23 ----------------------- 3 files changed, 9 insertions(+), 23 deletions(-) delete mode 100644 arch/x86/purgatory/string.c
arch/x86/purgatory/Makefile +3 -0
@@ -6,6 +6,9 @@ purgatory-y := purgatory.o stack.o setup-x86_$(BITS).o sha256.o entry64.o string targets += $(purgatory-y) PURGATORY_OBJS = $(addprefix $(obj)/,$(purgatory-y)) +$(obj)/string.o: $(srctree)/arch/x86/boot/compressed/string.c FORCE + $(call if_changed_rule,cc_o_c) + $(obj)/sha256.o: $(srctree)/lib/sha256.c FORCE $(call if_changed_rule,cc_o_c)
arch/x86/purgatory/purgatory.c +6 -0
@@ -68,3 +68,9 @@ void purgatory(void) }copy_backup_region (); } + +/* +* Defined in order to reuse memcpy() and memset() from +* arch/x86/boot/compressed/string.c +*/ +void warn (const char *msg) {}
arch/x86/purgatory/string.c deleted +0 -23
@@ -1,23 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0-only -/* -* Simple string functions. -* -* Copyright (C) 2014 Red Hat Inc. -* -* Author: -* Vivek Goyal <[email protected]> -*/ - -#include <linux/types.h> - -#include "../boot/string.c" - -void *memcpy (void *dst,const void *src,size_t len) -{ -return __builtin_memcpy (dst, src, len); -} - -void *memset (void *dst,int c,size_t len) -{ -return __builtin_memset (dst, c, len); -}
Signature · 1 line
2.22.0.657.g960e92d24f-goog
PATCH v2 2/2 x86/purgatory: use CFLAGS_REMOVE rather than reset KBUILD_CFLAGS
KBUILD_CFLAGS is very carefully built up in the top level Makefile, particularly when cross compiling or using different build tools. Resetting KBUILD_CFLAGS via := assignment is an antipattern.
The comment above the reset mentions that -pg is problematic. Other Makefiles like arch/x86/xen/vdso/Makefile use CFLAGS_REMOVE_file.o = -pg when CONFIG_FUNCTION_TRACER is set. Prefer that pattern to wiping out all of the important KBUILD_CFLAGS then manually having to re-add them.
---
Rather than manually add -mno-sse, -mno-mmx, -mno-sse2, prefer to filter -pg flags.
arch/x86/purgatory/Makefile | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-)
arch/x86/purgatory/Makefile +7 -5
@@ -20,11 +20,13 @@ KCOV_INSTRUMENT := n # Default KBUILD_CFLAGS can have -pg option set when FTRACE is enabled. That # in turn leaves some undefined symbols like __fentry__ in purgatory and not -# sure how to relocate those. Like kexec-tools, use custom flags. - -KBUILD_CFLAGS := -fno-strict-aliasing -Wall -Wstrict-prototypes -fno-zero-initialized-in-bss -fno-builtin -ffreestanding -c -Os -mcmodel=large -KBUILD_CFLAGS += -m$(BITS) -KBUILD_CFLAGS += $(call cc-option,-fno-PIE) +# sure how to relocate those. +ifdef CONFIG_FUNCTION_TRACER +CFLAGS_REMOVE_sha256.o = -pg +CFLAGS_REMOVE_purgatory.o = -pg +CFLAGS_REMOVE_string.o = -pg +CFLAGS_REMOVE_kexec-purgatory.o = -pg +endif $(obj)/purgatory.ro: $(PURGATORY_OBJS) FORCE $(call if_changed,ld)
Signature · 1 line
2.22.0.657.g960e92d24f-goog
Re: [PATCH v2 2/2] x86/purgatory: use CFLAGS_REMOVE rather than reset KBUILD_CFLAGS
Quoted 27 lines · Nick Desaulniers
KBUILD_CFLAGSis very carefully built up in the top levelMakefile, particularly when cross compiling or using different build tools. ResettingKBUILD_CFLAGSvia := assignment is an antipattern.The comment above the reset mentions that -pg is problematic. Other Makefiles like
arch/x86/xen/vdso/MakefileuseCFLAGS_REMOVE_file.o = -pgwhenCONFIG_FUNCTION_TRACERis set. Prefer that pattern to wiping out all of the importantKBUILD_CFLAGSthen manually having to re-add them.Fixes 8fc5b4d4121c ("purgatory: core purgatory functionality")Reported-by Vaibhav Rustagi <[email protected]>Suggested-by Peter Zijlstra <[email protected]>Signed-off-by Nick Desaulniers <[email protected]>---
Rather than manually add -mno-sse, -mno-mmx, -mno-sse2, prefer to filter(quoted diff) +0
-1-pg flags. arch/x86/purgatory/Makefile | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-)arch/x86/purgatory/Makefile +0
-0
arch/x86/purgatory/Makefile KCOV_INSTRUMENT := n · line 26
Quoted 7 lines
arch/x86/purgatory/Makefile +0
-4@@ -20,11 +20,13 @@ KCOV_INSTRUMENT := n # Default KBUILD_CFLAGS can have -pg option set when FTRACE is enabled. That # in turn leaves some undefined symbols like __fentry__ in purgatory and not -# sure how to relocate those. Like kexec-tools, use custom flags. - -KBUILD_CFLAGS := -fno-strict-aliasing -Wall -Wstrict-prototypes -fno-zero-initialized-in-bss -fno-builtin -ffreestanding -c -Os -mcmodel=large -KBUILD_CFLAGS += -m$(BITS)
Is purgatory/kexec supported for CONFIG_X86_32? Should I be keeping -m$(BITS)? arch/x86/purgatory/Makefile mentions setup-x86_$(BITS).o which I assume is broken as there is no arch/x86/purgatory/setup-x86_32.S?
Quoted 14 lines
arch/x86/purgatory/Makefile +7
-2-KBUILD_CFLAGS += $(call cc-option,-fno-PIE) +# sure how to relocate those. +ifdef CONFIG_FUNCTION_TRACER +CFLAGS_REMOVE_sha256.o = -pg +CFLAGS_REMOVE_purgatory.o = -pg +CFLAGS_REMOVE_string.o = -pg +CFLAGS_REMOVE_kexec-purgatory.o = -pg +endif $(obj)/purgatory.ro: $(PURGATORY_OBJS) FORCE $(call if_changed,ld) --2.22.0.657.g960e92d24f-goog
Signature · 2 lines
Thanks, ~Nick Desaulniers
Re: [PATCH v2 2/2] x86/purgatory: use CFLAGS_REMOVE rather than reset KBUILD_CFLAGS
Quoted 43 lines · Nick Desaulniers
Quoted 27 lines · Nick Desaulniers
KBUILD_CFLAGSis very carefully built up in the top levelMakefile, particularly when cross compiling or using different build tools. ResettingKBUILD_CFLAGSvia := assignment is an antipattern.The comment above the reset mentions that -pg is problematic. Other Makefiles like
arch/x86/xen/vdso/MakefileuseCFLAGS_REMOVE_file.o = -pgwhenCONFIG_FUNCTION_TRACERis set. Prefer that pattern to wiping out all of the importantKBUILD_CFLAGSthen manually having to re-add them.Fixes 8fc5b4d4121c ("purgatory: core purgatory functionality")Reported-by Vaibhav Rustagi <[email protected]>Suggested-by Peter Zijlstra <[email protected]>Signed-off-by Nick Desaulniers <[email protected]>---
Rather than manually add -mno-sse, -mno-mmx, -mno-sse2, prefer to filter(quoted diff) +0
-1-pg flags. arch/x86/purgatory/Makefile | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-)arch/x86/purgatory/Makefile +0
-0arch/x86/purgatory/Makefile KCOV_INSTRUMENT := n · line 26
Quoted 7 lines
arch/x86/purgatory/Makefile +0
-4@@ -20,11 +20,13 @@ KCOV_INSTRUMENT := n # Default KBUILD_CFLAGS can have -pg option set when FTRACE is enabled. That # in turn leaves some undefined symbols like __fentry__ in purgatory and not -# sure how to relocate those. Like kexec-tools, use custom flags. - -KBUILD_CFLAGS := -fno-strict-aliasing -Wall -Wstrict-prototypes -fno-zero-initialized-in-bss -fno-builtin -ffreestanding -c -Os -mcmodel=large -KBUILD_CFLAGS += -m$(BITS)Is purgatory/kexec supported for
CONFIG_X86_32? Should I be keeping-m$(BITS)?arch/x86/purgatory/Makefilementionssetup-x86_$(BITS).owhich I assume is broken as there is noarch/x86/purgatory/setup-x86_32.S?
Quoted 9 lines
arch/x86/purgatory/Makefile KCOV_INSTRUMENT := n · line 30
Quoted 9 lines
arch/x86/purgatory/Makefile +7
-1-KBUILD_CFLAGS += $(call cc-option,-fno-PIE) +# sure how to relocate those. +ifdef CONFIG_FUNCTION_TRACER +CFLAGS_REMOVE_sha256.o = -pg +CFLAGS_REMOVE_purgatory.o = -pg +CFLAGS_REMOVE_string.o = -pg +CFLAGS_REMOVE_kexec-purgatory.o = -pg +endif
The changes suggested will cause undefined symbols while loading the new kernel. On doing 'nm purgatory.ro', I found below undefined symbols:
U bmcp
U __stack_chk_fail
Quoted 10 lines
Quoted 5 lines
arch/x86/purgatory/Makefile +0
-1$(obj)/purgatory.ro: $(PURGATORY_OBJS) FORCE $(call if_changed,ld) --2.22.0.657.g960e92d24f-goog
--
Thanks,
~Nick Desaulniers
Re: [PATCH v2 2/2] x86/purgatory: use CFLAGS_REMOVE rather than reset KBUILD_CFLAGS
Quoted 5 lines · Vaibhav Rustagi
The changes suggested will cause undefined symbols while loading the new kernel. On doing 'nm purgatory.ro', I found below undefined symbols:
U bmcp
U__stack_chk_fail
Thanks for the report, a v3: https://lkml.org/lkml/2019/7/23/864
(Tested v3 with defconfig, defconfig+CONFIG_FUNCTION_TRACER, +CONFIG_STACKPROTECTOR_STRONG, and allyesconfig)
Signature · 2 lines
Thanks, ~Nick Desaulniers
A message that is not in the archive (sent off-list or not yet received): [email protected]
Re: [PATCH v2 0/2] Support kexec/kdump for clang built kernel
Joe,
Is it possible to have scripts/get_maintainer.pl always cc [email protected]? I just sent out a series, and it seems the cover letter didn't get sent to LKML. I usually use this shell function to send patches:
```
function kpatch () {
patch=$1
shift
git send-email \
--cc-cmd="./scripts/get_maintainer.pl --norolestats $patch" \
$@ $patch
}
```Invoked via:
``` $ mkdir purgatory $ git format-patch HEAD~2 --cover-letter -o purgatory -v2 $ kpatch purgatory/v2-000* --cc [email protected] --cc [email protected] ```
Maybe I should just add --cc [email protected] to my shell function?
Quoted 26 lines · Nick Desaulniers
1. Reuse the implementation of memcpy and memset instead of relying on
__builtin_memcpyand__builtin_memsetas it causes infinite recursion in Clang (at any opt level) or GCC at -O2.
2. Don't resetKBUILD_CFLAGS, rather filterCONFIG_FUNCTION_TRACERflags viaCFLAGS_REMOVE_<file>.o = -pg.The order of the patches are reversed; in case we ever need to bisect, the memcpy/memset infinite recursion would occur with just patch 2/2 applied.
V2 of: https://lkml.org/lkml/2019/7/17/755
Nick Desaulniers (2):
x86/purgatory: do not use__builtin_memcpyand__builtin_memsetx86/purgatory: useCFLAGS_REMOVErather than resetKBUILD_CFLAGSarch/x86/purgatory/Makefile | 15 ++++++++++----- arch/x86/purgatory/purgatory.c | 6 ++++++ arch/x86/purgatory/string.c | 23 ----------------------- 3 files changed, 16 insertions(+), 28 deletions(-) delete mode 100644 arch/x86/purgatory/string.c--
2.22.0.657.g960e92d24f-goog
Signature · 2 lines
Thanks, ~Nick Desaulniers