kernel/arch/unicore32/include, branch linux-5.1.y

x86/mpx, mm/core: Fix recursive munmap() corruption

2019-05-25T16:16:40Z

commit 5a28fc94c9143db766d1ba5480cae82d856ad080 upstream. This is a bit of a mess, to put it mildly. But, it's a bug that only seems to have showed up in 4.20 but wasn't noticed until now, because nobody uses MPX. MPX has the arch_unmap() hook inside of munmap() because MPX uses bounds tables that protect other areas of memory. When memory is unmapped, there is also a need to unmap the MPX bounds tables. Barring this, unused bounds tables can eat 80% of the address space. But, the recursive do_munmap() that gets called vi arch_unmap() wreaks havoc with __do_munmap()'s state. It can result in freeing populated page tables, accessing bogus VMA state, double-freed VMAs and more. See the "long story" further below for the gory details. To fix this, call arch_unmap() before __do_unmap() has a chance to do anything meaningful. Also, remove the 'vma' argument and force the MPX code to do its own, independent VMA lookup. == UML / unicore32 impact == Remove unused 'vma' argument to arch_unmap(). No functional change. I compile tested this on UML but not unicore32. == powerpc impact == powerpc uses arch_unmap() well to watch for munmap() on the VDSO and zeroes out 'current->mm->context.vdso_base'. Moving arch_unmap() makes this happen earlier in __do_munmap(). But, 'vdso_base' seems to only be used in perf and in the signal delivery that happens near the return to userspace. I can not find any likely impact to powerpc, other than the zeroing happening a little earlier. powerpc does not use the 'vma' argument and is unaffected by its removal. I compile-tested a 64-bit powerpc defconfig. == x86 impact == For the common success case this is functionally identical to what was there before. For the munmap() failure case, it's possible that some MPX tables will be zapped for memory that continues to be in use. But, this is an extraordinarily unlikely scenario and the harm would be that MPX provides no protection since the bounds table got reset (zeroed). I can't imagine anyone doing this: ptr = mmap(); // use ptr ret = munmap(ptr); if (ret) // oh, there was an error, I'll // keep using ptr. Because if you're doing munmap(), you are *done* with the memory. There's probably no good data in there _anyway_. This passes the original reproducer from Richard Biener as well as the existing mpx selftests/. The long story: munmap() has a couple of pieces: 1. Find the affected VMA(s) 2. Split the start/end one(s) if neceesary 3. Pull the VMAs out of the rbtree 4. Actually zap the memory via unmap_region(), including freeing page tables (or queueing them to be freed). 5. Fix up some of the accounting (like fput()) and actually free the VMA itself. This specific ordering was actually introduced by: dd2283f2605e ("mm: mmap: zap pages with read mmap_sem in munmap") during the 4.20 merge window. The previous __do_munmap() code was actually safe because the only thing after arch_unmap() was remove_vma_list(). arch_unmap() could not see 'vma' in the rbtree because it was detached, so it is not even capable of doing operations unsafe for remove_vma_list()'s use of 'vma'. Richard Biener reported a test that shows this in dmesg: [1216548.787498] BUG: Bad rss-counter state mm:0000000017ce560b idx:1 val:551 [1216548.787500] BUG: non-zero pgtables_bytes on freeing mm: 24576 What triggered this was the recursive do_munmap() called via arch_unmap(). It was freeing page tables that has not been properly zapped. But, the problem was bigger than this. For one, arch_unmap() can free VMAs. But, the calling __do_munmap() has variables that *point* to VMAs and obviously can't handle them just getting freed while the pointer is still in use. I tried a couple of things here. First, I tried to fix the page table freeing problem in isolation, but I then found the VMA issue. I also tried having the MPX code return a flag if it modified the rbtree which would force __do_munmap() to re-walk to restart. That spiralled out of control in complexity pretty fast. Just moving arch_unmap() and accepting that the bonkers failure case might eat some bounds tables seems like the simplest viable fix. This was also reported in the following kernel bugzilla entry: https://bugzilla.kernel.org/show_bug.cgi?id=203123 There are some reports that this commit triggered this bug: dd2283f2605 ("mm: mmap: zap pages with read mmap_sem in munmap") While that commit certainly made the issues easier to hit, I believe the fundamental issue has been with us as long as MPX itself, thus the Fixes: tag below is for one of the original MPX commits. [ mingo: Minor edits to the changelog and the patch. ] Reported-by: Richard Biener Reported-by: H.J. Lu Signed-off-by: Dave Hansen Reviewed-by Thomas Gleixner Reviewed-by: Yang Shi Acked-by: Michael Ellerman Cc: Andrew Morton Cc: Andy Lutomirski Cc: Anton Ivanov Cc: Benjamin Herrenschmidt Cc: Borislav Petkov Cc: Guan Xuetao Cc: H. Peter Anvin Cc: Jeff Dike Cc: Linus Torvalds Cc: Michal Hocko Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Richard Weinberger Cc: Rik van Riel Cc: Vlastimil Babka Cc: linux-arch@vger.kernel.org Cc: linux-mm@kvack.org Cc: linux-um@lists.infradead.org Cc: linuxppc-dev@lists.ozlabs.org Cc: stable@vger.kernel.org Fixes: dd2283f2605e ("mm: mmap: zap pages with read mmap_sem in munmap") Link: http://lkml.kernel.org/r/20190419194747.5E1AD6DC@viggo.jf.intel.com Signed-off-by: Ingo Molnar Signed-off-by: Greg Kroah-Hartman

KVM: export and iif KVM is supported

2019-03-28T16:27:42Z

I do not see any consistency about headers_install of and . According to my analysis of Linux 5.1-rc1, there are 3 groups: [1] Both and are exported alpha, arm, hexagon, mips, powerpc, s390, sparc, x86 [2] is exported, but is not arc, arm64, c6x, h8300, ia64, m68k, microblaze, nios2, openrisc, parisc, sh, unicore32, xtensa [3] Neither nor is exported csky, nds32, riscv This does not match to the actual KVM support. At least, [2] is half-baked. Nor do arch maintainers look like they care about this. For example, commit 0add53713b1c ("microblaze: Add missing kvm_para.h to Kbuild") exported to user-space in order to fix an in-kernel build error. We have two ways to make this consistent: [A] export both and for all architectures, irrespective of the KVM support [B] Match the header export of and to the KVM support My first attempt was [A] because the code looks cleaner, but Paolo suggested [B]. So, this commit goes with [B]. For most architectures, was moved to the kernel-space. I changed include/uapi/linux/Kbuild so that it checks generated asm/kvm_para.h as well as check-in ones. After this commit, there will be two groups: [1] Both and are exported arm, arm64, mips, powerpc, s390, x86 [2] Neither nor is exported alpha, arc, c6x, csky, h8300, hexagon, ia64, m68k, microblaze, nds32, nios2, openrisc, parisc, riscv, sh, sparc, unicore32, xtensa Signed-off-by: Masahiro Yamada Acked-by: Cornelia Huck Signed-off-by: Paolo Bonzini

kbuild: force all architectures except um to include mandatory-y

2019-03-17T03:56:32Z

Currently, every arch/*/include/uapi/asm/Kbuild explicitly includes the common Kbuild.asm file. Factor out the duplicated include directives to scripts/Makefile.asm-generic so that no architecture would opt out of the mandatory-y mechanism. um is not forced to include mandatory-y since it is a very exceptional case which does not support UAPI. Signed-off-by: Masahiro Yamada

Merge tag 'y2038-syscall-abi' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/playground into timers/2038

2019-02-27T20:45:27Z

Pull additional syscall ABI cleanup for y2038 from Arnd Bergmann: This is a follow-up to the y2038 syscall patches already merged in the tip tree. As the final 32-bit RISC-V syscall ABI is still being decided on, this is the last chance to make a few corrections to leave out interfaces based on 32-bit time_t along with the old off_t and rlimit types. The series achieves this in a few steps: - A couple of bug fixes for minor regressions I introduced in the original series - A couple of older patches from Yury Norov that I had never merged in the past, these fix up the openat/open_by_handle_at and getrlimit/setrlimit syscalls to disallow the old versions of off_t and rlimit. - Hiding the deprecated system calls behind an #ifdef in include/uapi/asm-generic/unistd.h - Change arch/riscv to drop all these ABIs. Originally, the plan was to also leave these out on C-Sky, but that now has a glibc port that uses the older interfaces, so we need to leave them in place.

unicore32: Fix __ARCH_WANT_STAT64 definition

2019-02-19T20:27:53Z

The __ARCH_WANT_STAT64 macro must be defined before including asm-generic/unistd.h. I got this right for everything except unicore32. Fixes: bf4b6a7d371e ("y2038: Remove stat64 family from default syscall set") Signed-off-by: Arnd Bergmann

asm-generic: Make time32 syscall numbers optional

2019-02-19T20:27:32Z

We don't want new architectures to even provide the old 32-bit time_t based system calls any more, or define the syscall number macros. Add a new __ARCH_WANT_TIME32_SYSCALLS macro that gets enabled for all existing 32-bit architectures using the generic system call table, so we don't change any current behavior. Since this symbol is evaluated in user space as well, we cannot use a Kconfig CONFIG_* macro but have to define it in uapi/asm/unistd.h. On 64-bit architectures, the same system call numbers mostly refer to the system calls we want to keep, as they already pass 64-bit time_t. As new architectures no longer provide these, we need new exceptions in checksyscalls.sh. Signed-off-by: Arnd Bergmann

asm-generic: Drop getrlimit and setrlimit syscalls from default list

2019-02-19T09:10:06Z

The newer prlimit64 syscall provides all the functionality of getrlimit and setrlimit syscalls and adds the pid of target process, so future architectures won't need to include getrlimit and setrlimit. Therefore drop getrlimit and setrlimit syscalls from the generic syscall list unless __ARCH_WANT_SET_GET_RLIMIT is defined by the architecture's unistd.h prior to including asm-generic/unistd.h, and adjust all architectures using the generic syscall list to define it so that no in-tree architectures are affected. Cc: Arnd Bergmann Cc: linux-arch@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org Cc: linux-hexagon@vger.kernel.org Cc: uclinux-h8-devel@lists.sourceforge.jp Signed-off-by: Yury Norov Acked-by: Arnd Bergmann Acked-by: Mark Salter [c6x] Acked-by: James Hogan [metag] Acked-by: Ley Foon Tan [nios2] Acked-by: Stafford Horne [openrisc] Acked-by: Will Deacon [arm64] Acked-by: Vineet Gupta #arch/arc bits Signed-off-by: Yury Norov Signed-off-by: Arnd Bergmann

arch: unexport asm/shmparam.h for all architectures

2019-02-01T23:46:22Z

Most architectures do not export shmparam.h to user-space. $ find arch -name shmparam.h | sort arch/alpha/include/asm/shmparam.h arch/arc/include/asm/shmparam.h arch/arm64/include/asm/shmparam.h arch/arm/include/asm/shmparam.h arch/csky/include/asm/shmparam.h arch/ia64/include/asm/shmparam.h arch/mips/include/asm/shmparam.h arch/nds32/include/asm/shmparam.h arch/nios2/include/asm/shmparam.h arch/parisc/include/asm/shmparam.h arch/powerpc/include/asm/shmparam.h arch/s390/include/asm/shmparam.h arch/sh/include/asm/shmparam.h arch/sparc/include/asm/shmparam.h arch/x86/include/asm/shmparam.h arch/xtensa/include/asm/shmparam.h Strangely, some users of the asm-generic wrapper export shmparam.h $ git grep 'generic-y += shmparam.h' arch/c6x/include/uapi/asm/Kbuild:generic-y += shmparam.h arch/h8300/include/uapi/asm/Kbuild:generic-y += shmparam.h arch/hexagon/include/uapi/asm/Kbuild:generic-y += shmparam.h arch/m68k/include/uapi/asm/Kbuild:generic-y += shmparam.h arch/microblaze/include/uapi/asm/Kbuild:generic-y += shmparam.h arch/openrisc/include/uapi/asm/Kbuild:generic-y += shmparam.h arch/riscv/include/asm/Kbuild:generic-y += shmparam.h arch/unicore32/include/uapi/asm/Kbuild:generic-y += shmparam.h The newly added riscv correctly creates the asm-generic wrapper in the kernel space, but the others (c6x, h8300, hexagon, m68k, microblaze, openrisc, unicore32) create the one in the uapi directory. Digging into the git history, now I guess fcc8487d477a ("uapi: export all headers under uapi directories") was the misconversion. Prior to that commit, no architecture exported to shmparam.h As its commit description said, that commit exported shmparam.h for c6x, h8300, hexagon, m68k, openrisc, unicore32. 83f0124ad81e ("microblaze: remove asm-generic wrapper headers") accidentally exported shmparam.h for microblaze. This commit unexports shmparam.h for those architectures. There is no more reason to export include/uapi/asm-generic/shmparam.h, so it has been moved to include/asm-generic/shmparam.h Link: http://lkml.kernel.org/r/1546904307-11124-1-git-send-email-yamada.masahiro@socionext.com Signed-off-by: Masahiro Yamada Acked-by: Stafford Horne Cc: Geert Uytterhoeven Cc: Michal Simek Cc: Yoshinori Sato Cc: Richard Kuo Cc: Guan Xuetao Cc: Nicolas Dichtel Cc: Arnd Bergmann Cc: Aurelien Jacquiot Cc: Greentime Hu Cc: Guo Ren Cc: Palmer Dabbelt Cc: Stefan Kristiansson Cc: Mark Salter Cc: Albert Ou Cc: Jonas Bonn Cc: Vincent Chen Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

arch: restore generic-y += shmparam.h for some architectures

2019-01-07T02:16:11Z

For some reasons, I accidentally got rid of "generic-y += shmparam.h" from some architectures. Restore them to fix building c6x, h8300, hexagon, m68k, microblaze, openrisc, and unicore32. Fixes: d6e4b3e326d8 ("arch: remove redundant UAPI generic-y defines") Signed-off-by: Masahiro Yamada Signed-off-by: Linus Torvalds

arch: remove redundant UAPI generic-y defines

2019-01-06T01:22:15Z

Now that Kbuild automatically creates asm-generic wrappers for missing mandatory headers, it is redundant to list the same headers in generic-y and mandatory-y. Suggested-by: Sam Ravnborg Signed-off-by: Masahiro Yamada Acked-by: Sam Ravnborg