siginfo cleanups for 4.16-rc1
Subsystems (2)
Linus,
Please pull the siginfo-linus branch from the git tree:
git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git siginfo-linus
HEAD: c0f45555b82450a56c0da11866163b764aeb1a21 signal/memory-failure: Use force_sig_mceerr and send_sig_mceerr
Long ago when 2.4 was just a testing release copy_siginfo_to_user was made to copy individual fields to userspace, possibly for efficiency and to ensure initialized values were not copied to userspace. Unfortunately the design was complex, it's assumptions unstated, and humans are failible and so while work much of the time that design failed to ensure unitialized memory is not copied to userspace.
This set of changes is part of a new design to clean up siginfo and simplify things, and hopefully make the siginfo handling robust enough that a simple inspection of the code can be made to ensure we don't copy any unitializied fields to userspace.
The design is to unifiy struct siginfo and struct compat_siginfo into a single definition that is shared between all architectures so that anyone adding to the set of information shared with struct siginfo can see the whole picture. Hopefully ensuring all future si_code assignments are arch independent.
The design is to unify copy_siginfo_to_user32 and copy_siginfo_from_user32 so that those function are complete and cope with all of the different cases documented in signinfo_layout. I don't think there was a single implementation of either of those functions that was complete and correct before my changes unified them.
The design is to introduce a series of helpers including force_siginfo_fault that take the values that are needed in struct siginfo and build the siginfo structure for their callers. Ensuring struct siginfo is built correctly.
The remaining work for 4.17 (unless someone thinks it is post -rc1 material) is to push usage of those helpers down into the architectures so that architecture specific code will not need to deal with the fiddly work of intializing struct siginfo, and then when struct siginfo is guaranteed to be fully initialized change copy siginfo_to_user into a simple wrapper around copy_to_user.
Further there is work in progress on the issues that have been documented requires arch specific knowledge to sort out.
The changes below fix or at least document all of the issues that have been found with siginfo generation. Then proceed to unify struct siginfo the 32 bit helpers that copy siginfo to and from userspace, and generally clean up anything that is not arch specific with regards to siginfo generation.
It is a lot but with the unification you can of siginfo you can already see the code reduction in the kernel.
Eric
Al Viro (3):
signal/mips: switch mips to generic siginfo
signal: kill __ARCH_SI_UID_T
signal: unify compat_siginfo_t
Eric W. Biederman (42):
signal: Simplify and fix kdb_send_sig
signal/sh: Ensure si_signo is initialized in do_divide_error signal/openrisc: Fix do_unaligned_access to send the proper signal signal/parisc: Document a conflict with SI_USER with SIGFPE signal/metag: Document a conflict with SI_USER with SIGFPE signal/powerpc: Document conflicts with SI_USER and SIGFPE and SIGTRAP signal/arm64: Document conflicts with SI_USER and SIGFPE,SIGTRAP,SIGBUS signal/arm: Document conflicts with SI_USER and SIGFPE
signal: Reduce copy_siginfo to just a memcpy
signal: Introduce clear_siginfo
signal: Ensure generic siginfos the kernel sends have all bits initialized mn10300/misalignment: Use SIGSEGV SEGV_MAPERR to report a failed user copy x86/mm/pkeys: Fix fill_sig_info_pkey
signal: Document all of the signals that use the _sigfault union member
signal: Document the strange si_codes used by ptrace event stops
signal: Document glibc's si_code of SI_ASYNCNL
signal: Ensure no siginfo union member increases the size of struct siginfo
signal: Clear si_sys_private before copying siginfo to userspace
signal: Remove _sys_private and _overrun_incr from struct compat_siginfo ia64/signal: switch to generic struct siginfo
signal/ia64: switch the last arch-specific copy_siginfo_to_user() to generic version
signal: Remove unnecessary ifdefs now that there is only one struct siginfo
signal: Move addr_lsb into the _sigfault union for clarity signal/powerpc: Remove redefinition of NSIGTRAP on powerpc signal/ia64: Move the ia64 specific si_codes to asm-generic/siginfo.h signal/frv: Move the frv specific si_codes to asm-generic/siginfo.h signal/tile: Move the tile specific si_codes to asm-generic/siginfo.h signal/blackfin: Move the blackfin specific si_codes to asm-generic/siginfo.h signal/blackfin: Remove pointless UID16_SIGINFO_COMPAT_NEEDED
signal: Unify and correct copy_siginfo_from_user32
signal: Remove the code to clear siginfo before calling copy_siginfo_from_user32
signal: Unify and correct copy_siginfo_to_user32
ptrace: Use copy_siginfo in setsiginfo and getsiginfo
signal/arm64: Better isolate the COMPAT_TASK portion of ptrace_hbptriggered
signal: Don't use structure initializers for struct siginfo
signal: Replace memset(info,...) with clear_siginfo for clarity
signal: Add send_sig_fault and force_sig_fault
signal: Helpers for faults with specialized siginfo layouts signal/powerpc: Remove unnecessary signal_code parameter of do_send_trap signal/ptrace: Add force_sig_ptrace_errno_trap and use it where needed mm/memory_failure: Remove unused trapno from memory_failure signal/memory-failure: Use force_sig_mceerr and send_sig_mceerr
arch/arc/kernel/traps.c | 14 +- arch/arm/include/uapi/asm/siginfo.h | 13 + arch/arm/kernel/ptrace.c | 8 +- arch/arm/vfp/vfpmodule.c | 2 +- arch/arm64/include/asm/compat.h | 64 ---- arch/arm64/include/uapi/asm/siginfo.h | 21 ++ arch/arm64/kernel/debug-monitors.c | 13 +- arch/arm64/kernel/fpsimd.c | 2 +- arch/arm64/kernel/ptrace.c | 42 +-- arch/arm64/kernel/signal32.c | 80 ----- arch/arm64/mm/fault.c | 114 +++---- arch/blackfin/include/uapi/asm/siginfo.h | 34 -- arch/frv/include/uapi/asm/Kbuild | 1 + arch/frv/include/uapi/asm/siginfo.h | 13 - arch/ia64/include/uapi/asm/siginfo.h | 96 ------ arch/ia64/kernel/signal.c | 52 --- arch/m68k/mm/fault.c | 3 +- arch/metag/include/uapi/asm/siginfo.h | 7 + arch/metag/kernel/traps.c | 2 +- arch/mips/include/asm/compat.h | 73 ----- arch/mips/include/uapi/asm/siginfo.h | 86 +---- arch/mips/kernel/signal32.c | 67 ---- arch/mips/kernel/traps.c | 29 +- arch/mn10300/mm/misalignment.c | 2 +- arch/openrisc/kernel/traps.c | 10 +- arch/parisc/include/asm/compat.h | 64 ---- arch/parisc/include/uapi/asm/siginfo.h | 7 + arch/parisc/kernel/pdt.c | 2 +- arch/parisc/kernel/signal32.c | 106 ------- arch/parisc/kernel/signal32.h | 3 - arch/parisc/kernel/traps.c | 2 +- arch/powerpc/include/asm/compat.h | 65 ---- arch/powerpc/include/asm/debug.h | 2 +- arch/powerpc/include/uapi/asm/siginfo.h | 16 +- arch/powerpc/kernel/process.c | 13 +- arch/powerpc/kernel/signal_32.c | 66 ---- arch/powerpc/kernel/traps.c | 22 +- .../powerpc/platforms/powernv/opal-memory-errors.c | 2 +- arch/s390/include/asm/compat.h | 73 ----- arch/s390/kernel/compat_signal.c | 100 ------ arch/sh/kernel/traps_32.c | 3 +- arch/sparc/include/asm/compat.h | 59 ---- arch/sparc/kernel/signal32.c | 69 ---- arch/tile/include/asm/compat.h | 62 ---- arch/tile/include/uapi/asm/siginfo.h | 8 - arch/tile/kernel/compat_signal.c | 73 ----- arch/tile/kernel/single_step.c | 24 +- arch/tile/kernel/traps.c | 4 +- arch/tile/kernel/unaligned.c | 46 +-- arch/um/kernel/trap.c | 2 +- arch/x86/include/asm/compat.h | 86 +---- arch/x86/include/asm/fpu/signal.h | 6 - arch/x86/kernel/cpu/mcheck/mce.c | 6 +- arch/x86/kernel/signal_compat.c | 123 +------- arch/x86/mm/fault.c | 7 +- arch/xtensa/kernel/ptrace.c | 8 +- drivers/acpi/apei/ghes.c | 2 +- drivers/base/memory.c | 2 +- drivers/ras/cec.c | 2 +- drivers/usb/core/devio.c | 4 +- fs/fcntl.c | 1 + include/linux/compat.h | 100 +++++- include/linux/mm.h | 4 +- include/linux/sched/signal.h | 28 ++ include/linux/signal.h | 15 +- include/uapi/asm-generic/siginfo.h | 109 +++++-- ipc/mqueue.c | 1 + kernel/debug/kdb/kdb_main.c | 10 +- kernel/debug/kdb/kdb_private.h | 2 +- kernel/ptrace.c | 5 +- kernel/seccomp.c | 2 +- kernel/signal.c | 350 +++++++++++++++++++-- kernel/time/posix-timers.c | 2 +- mm/hwpoison-inject.c | 2 +- mm/madvise.c | 2 +- mm/memory-failure.c | 48 ++- 76 files changed, 858 insertions(+), 1810 deletions(-)
Quoted 47 lines · [email protected] (Eric W. Biederman) writes:
Linus,
Please pull the siginfo-linus branch from the git tree:
git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git siginfo-linus HEAD: c0f45555b82450a56c0da11866163b764aeb1a21 signal/memory-failure: Use force_sig_mceerr and send_sig_mceerrLong ago when 2.4 was just a testing release
copy_siginfo_to_userwas made to copy individual fields to userspace, possibly for efficiency and to ensure initialized values were not copied to userspace. Unfortunately the design was complex, it's assumptions unstated, and humans are failible and so while work much of the time that design failed to ensure unitialized memory is not copied to userspace.This set of changes is part of a new design to clean up siginfo and simplify things, and hopefully make the siginfo handling robust enough that a simple inspection of the code can be made to ensure we don't copy any unitializied fields to userspace.
The design is to unifiy struct siginfo and struct
compat_siginfointo a single definition that is shared between all architectures so that anyone adding to the set of information shared with struct siginfo can see the whole picture. Hopefully ensuring all futuresi_codeassignments are arch independent.The design is to unify
copy_siginfo_to_user32andcopy_siginfo_from_user32so that those function are complete and cope with all of the different cases documented insigninfo_layout. I don't think there was a single implementation of either of those functions that was complete and correct before my changes unified them.The design is to introduce a series of helpers including
force_siginfo_faultthat take the values that are needed in struct siginfo and build the siginfo structure for their callers. Ensuring struct siginfo is built correctly.The remaining work for 4.17 (unless someone thinks it is post -rc1 material) is to push usage of those helpers down into the architectures so that architecture specific code will not need to deal with the fiddly work of intializing struct siginfo, and then when struct siginfo is guaranteed to be fully initialized change copy
siginfo_to_userinto a simple wrapper aroundcopy_to_user.Further there is work in progress on the issues that have been documented requires arch specific knowledge to sort out.
Quoted 8 lines
The changes below fix or at least document all of the issues that have been found with siginfo generation. Then proceed to unify struct siginfo the 32 bit helpers that copy siginfo to and from userspace, and generally clean up anything that is not arch specific with regards to siginfo generation.
It is a lot but with the unification you can of siginfo you can already see the code reduction in the kernel.
I should add there is a trivial conflict with powerpc. The powerpc branch in linux-next adds a new use of the memory_failure function, while this code removes the unused trapno argument from thememory_failure function. Conflict resolution just remove
the middle argument to memory_failure from the powerpc branch.
Quoted 126 lines
Al Viro (3):
signal/mips: switch mips to generic siginfo
signal: kill__ARCH_SI_UID_T
signal: unifycompat_siginfo_tEric W. Biederman (42):
signal: Simplify and fixkdb_send_sig
signal/sh: Ensuresi_signois initialized indo_divide_errorsignal/openrisc: Fixdo_unaligned_accessto send the proper signal signal/parisc: Document a conflict withSI_USERwith SIGFPE signal/metag: Document a conflict withSI_USERwith SIGFPE signal/powerpc: Document conflicts withSI_USERand SIGFPE and SIGTRAP signal/arm64: Document conflicts withSI_USERand SIGFPE,SIGTRAP,SIGBUS signal/arm: Document conflicts withSI_USERand SIGFPE
signal: Reducecopy_siginfoto just a memcpy
signal: Introduceclear_siginfo
signal: Ensure generic siginfos the kernel sends have all bits initialized mn10300/misalignment: Use SIGSEGVSEGV_MAPERRto report a failed user copy x86/mm/pkeys: Fixfill_sig_info_pkey
signal: Document all of the signals that use the_sigfaultunion member
signal: Document the strangesi_codesused by ptrace event stops
signal: Document glibc'ssi_codeofSI_ASYNCNL
signal: Ensure no siginfo union member increases the size of struct siginfo
signal: Clearsi_sys_privatebefore copying siginfo to userspace
signal: Remove_sys_privateand_overrun_incrfrom structcompat_siginfoia64/signal: switch to generic struct siginfo
signal/ia64: switch the last arch-specificcopy_siginfo_to_user()to generic version
signal: Remove unnecessary ifdefs now that there is only one struct siginfo
signal: Moveaddr_lsbinto the_sigfaultunion for clarity signal/powerpc: Remove redefinition of NSIGTRAP on powerpc signal/ia64: Move the ia64 specificsi_codestoasm-generic/siginfo.hsignal/frv: Move the frv specificsi_codestoasm-generic/siginfo.hsignal/tile: Move the tile specificsi_codestoasm-generic/siginfo.hsignal/blackfin: Move the blackfin specificsi_codestoasm-generic/siginfo.hsignal/blackfin: Remove pointlessUID16_SIGINFO_COMPAT_NEEDED
signal: Unify and correctcopy_siginfo_from_user32
signal: Remove the code to clear siginfo before callingcopy_siginfo_from_user32
signal: Unify and correctcopy_siginfo_to_user32
ptrace: Usecopy_siginfoin setsiginfo and getsiginfo
signal/arm64: Better isolate theCOMPAT_TASKportion ofptrace_hbptriggered
signal: Don't use structure initializers for struct siginfo
signal: Replacememset(info,...)withclear_siginfofor clarity
signal: Addsend_sig_faultandforce_sig_fault
signal: Helpers for faults with specialized siginfo layouts signal/powerpc: Remove unnecessarysignal_codeparameter ofdo_send_trapsignal/ptrace: Addforce_sig_ptrace_errno_trapand use it where neededmm/memory_failure: Remove unused trapno frommemory_failuresignal/memory-failure: Useforce_sig_mceerrandsend_sig_mceerrarch/arc/kernel/traps.c | 14 +- arch/arm/include/uapi/asm/siginfo.h | 13 + arch/arm/kernel/ptrace.c | 8 +- arch/arm/vfp/vfpmodule.c | 2 +- arch/arm64/include/asm/compat.h | 64 ---- arch/arm64/include/uapi/asm/siginfo.h | 21 ++ arch/arm64/kernel/debug-monitors.c | 13 +- arch/arm64/kernel/fpsimd.c | 2 +- arch/arm64/kernel/ptrace.c | 42 +-- arch/arm64/kernel/signal32.c | 80 ----- arch/arm64/mm/fault.c | 114 +++---- arch/blackfin/include/uapi/asm/siginfo.h | 34 -- arch/frv/include/uapi/asm/Kbuild | 1 + arch/frv/include/uapi/asm/siginfo.h | 13 - arch/ia64/include/uapi/asm/siginfo.h | 96 ------ arch/ia64/kernel/signal.c | 52 --- arch/m68k/mm/fault.c | 3 +- arch/metag/include/uapi/asm/siginfo.h | 7 + arch/metag/kernel/traps.c | 2 +- arch/mips/include/asm/compat.h | 73 ----- arch/mips/include/uapi/asm/siginfo.h | 86 +---- arch/mips/kernel/signal32.c | 67 ---- arch/mips/kernel/traps.c | 29 +- arch/mn10300/mm/misalignment.c | 2 +- arch/openrisc/kernel/traps.c | 10 +- arch/parisc/include/asm/compat.h | 64 ---- arch/parisc/include/uapi/asm/siginfo.h | 7 + arch/parisc/kernel/pdt.c | 2 +- arch/parisc/kernel/signal32.c | 106 ------- arch/parisc/kernel/signal32.h | 3 - arch/parisc/kernel/traps.c | 2 +- arch/powerpc/include/asm/compat.h | 65 ---- arch/powerpc/include/asm/debug.h | 2 +- arch/powerpc/include/uapi/asm/siginfo.h | 16 +- arch/powerpc/kernel/process.c | 13 +- arch/powerpc/kernel/signal_32.c | 66 ---- arch/powerpc/kernel/traps.c | 22 +- .../powerpc/platforms/powernv/opal-memory-errors.c | 2 +- arch/s390/include/asm/compat.h | 73 ----- arch/s390/kernel/compat_signal.c | 100 ------ arch/sh/kernel/traps_32.c | 3 +- arch/sparc/include/asm/compat.h | 59 ---- arch/sparc/kernel/signal32.c | 69 ---- arch/tile/include/asm/compat.h | 62 ---- arch/tile/include/uapi/asm/siginfo.h | 8 - arch/tile/kernel/compat_signal.c | 73 ----- arch/tile/kernel/single_step.c | 24 +- arch/tile/kernel/traps.c | 4 +- arch/tile/kernel/unaligned.c | 46 +-- arch/um/kernel/trap.c | 2 +- arch/x86/include/asm/compat.h | 86 +---- arch/x86/include/asm/fpu/signal.h | 6 - arch/x86/kernel/cpu/mcheck/mce.c | 6 +- arch/x86/kernel/signal_compat.c | 123 +------- arch/x86/mm/fault.c | 7 +- arch/xtensa/kernel/ptrace.c | 8 +- drivers/acpi/apei/ghes.c | 2 +- drivers/base/memory.c | 2 +- drivers/ras/cec.c | 2 +- drivers/usb/core/devio.c | 4 +- fs/fcntl.c | 1 + include/linux/compat.h | 100 +++++- include/linux/mm.h | 4 +- include/linux/sched/signal.h | 28 ++ include/linux/signal.h | 15 +- include/uapi/asm-generic/siginfo.h | 109 +++++-- ipc/mqueue.c | 1 + kernel/debug/kdb/kdb_main.c | 10 +- kernel/debug/kdb/kdb_private.h | 2 +- kernel/ptrace.c | 5 +- kernel/seccomp.c | 2 +- kernel/signal.c | 350 +++++++++++++++++++-- kernel/time/posix-timers.c | 2 +- mm/hwpoison-inject.c | 2 +- mm/madvise.c | 2 +- mm/memory-failure.c | 48 ++- 76 files changed, 858 insertions(+), 1810 deletions(-)
[GIT PULL] siginfo fix for 4.16-rc1
Linus,
Please pull the siginfo-linus branch from the git tree:
git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git siginfo-linus
HEAD: f20239c0d2d720d7deb8a40fd05c31849ddc8293 mm/memory_failure: update powerpc for memory_failure() API change
I had a brain fart when I received Stephen email about a compile failure in linux-next. I thought the powerpc call to memory_failure had not merged yet. Instead it simply failed to compile :(
Linus can you please pull Stephen's fix so that powerpc builds correctly.
Thank you,
Eric
From: Stephen Rothwell <[email protected]>
Date: Fri, 26 Jan 2018 12:05:12 +1100
Subject: [PATCH] mm/memory_failure: update powerpc for memory_failure() API change
After merging the userns tree, today's linux-next build (powerpc allyesconfig) failed like this:
arch/powerpc/kernel/mce.c: In function 'machine_process_ue_event':arch/powerpc/kernel/mce.c:276:5: error: too many arguments to function 'memory_failure'
memory_failure(pfn, SIGBUS, 0);
^~~~~~~~~~~~~~In file included from include/linux/pid_namespace.h:7:0,
from include/linux/ptrace.h:10, from arch/powerpc/kernel/mce.c:27:include/linux/mm.h:2571:12: note: declared here
extern int memory_failure(unsigned long pfn, int flags);
^~~~~~~~~~~~~~
Caused by commit
83b57531c58f ("mm/memory_failure: Remove unused trapno from memory_failure")
--- arch/powerpc/kernel/mce.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/powerpc/kernel/mce.c +1 -1
@@ -273,7 +273,7 @@ static void machine_process_ue_event(struct work_struct *work) pfn = evt->u.ue_error.physical_address >>PAGE_SHIFT ; -memory_failure (pfn,SIGBUS ,0 ); +memory_failure (pfn,0 ); }else pr_warn ("Failed to identify bad address from " "where the uncorrectable error (UE) "
Signature · 1 line
2.14.1
Re: [GIT PULL] siginfo fix for 4.16-rc1
Quoted 6 lines · Eric W. Biederman
I had a brain fart when I received Stephen email about a compile failure in linux-next. I thought the powerpc call to
memory_failurehad not merged yet. Instead it simply failed to compile :(Linus can you please pull Stephen's fix so that powerpc builds correctly.
I fixed this up in the merge already, afaik. Or is there some other failure?
Linus
Re: [GIT PULL] siginfo fix for 4.16-rc1
Quoted 10 lines · Linus Torvalds
Quoted 6 lines · Eric W. Biederman
I had a brain fart when I received Stephen email about a compile failure in linux-next. I thought the powerpc call to
memory_failurehad not merged yet. Instead it simply failed to compile :(Linus can you please pull Stephen's fix so that powerpc builds correctly.
I fixed this up in the merge already, afaik. Or is there some other failure?
My bad. I missed you fixed it in the merge.
I received a kbuild failure against linux-next this morning, and I realized the build failure was against code that was already in 2.15 so I should have patched and fixed that myself.
At which point I assumed the failure was because my siginfo code had been merged into your tree.
Looking again you definitely fixed this in the merge. So I don't know the story of the broken kbuild of linux-next. But I assume it will sort itself out when the dust clears.
Apologies for the noise,
Eric