public inbox for
help / color / mirror / Atom feed
* [glibc] x86: Prevent SIGSEGV in memcmp-sse2 when data is concurrently modified [BZ #29863]
@ 2022-12-15 17:10 Noah Goldstein
  0 siblings, 0 replies; only message in thread
From: Noah Goldstein @ 2022-12-15 17:10 UTC (permalink / raw)
  To: glibc-cvs;h=b712be52645282c706a5faa038242504feb06db5

commit b712be52645282c706a5faa038242504feb06db5
Author: Noah Goldstein <>
Date:   Wed Dec 14 10:52:10 2022 -0800

    x86: Prevent SIGSEGV in memcmp-sse2 when data is concurrently modified [BZ #29863]
    In the case of INCORRECT usage of `memcmp(a, b, N)` where `a` and `b`
    are concurrently modified as `memcmp` runs, there can be a SIGSEGV
    in `L(ret_nonzero_vec_end_0)` because the sequential logic
    assumes that `(rdx - 32 + rax)` is a positive 32-bit integer.
    To be clear, this change does not mean the usage of `memcmp` is
    supported.  The program behaviour is undefined (UB) in the
    presence of data races, and `memcmp` is incorrect when the values
    of `a` and/or `b` are modified concurrently (data race). This UB
    may manifest itself as a SIGSEGV. That being said, if we can
    allow the idiomatic use cases, like those in yottadb with
    opportunistic concurrency control (OCC), to execute without a
    SIGSEGV, at no cost to regular use cases, then we can aim to
    minimize harm to those existing users.
    The fix replaces a 32-bit `addl %edx, %eax` with the 64-bit variant
    `addq %rdx, %rax`. The 1-extra byte of code size from using the
    64-bit instruction doesn't contribute to overall code size as the
    next target is aligned and has multiple bytes of `nop` padding
    before it. As well all the logic between the add and `ret` still
    fits in the same fetch block, so the cost of this change is
    basically zero.
    The relevant sequential logic can be seen in the following
         * rsi = a
         * rdi = b
         * rdx = len - 32
        /* cmp a[0:15] and b[0:15]. Since length is known to be [17, 32]
        in this case, this check is also assumed to cover a[0:(31 - len)]
        and b[0:(31 - len)].  */
        movups  (%rsi), %xmm0
        movups  (%rdi), %xmm1
        PCMPEQ  %xmm0, %xmm1
        pmovmskb %xmm1, %eax
        subl    %ecx, %eax
        jnz L(END_NEQ)
        /* cmp a[len-16:len-1] and b[len-16:len-1].  */
        movups  16(%rsi, %rdx), %xmm0
        movups  16(%rdi, %rdx), %xmm1
        PCMPEQ  %xmm0, %xmm1
        pmovmskb %xmm1, %eax
        subl    %ecx, %eax
        jnz L(END_NEQ2)
        /* Position first mismatch.  */
        bsfl    %eax, %eax
        /* The sequential version is able to assume this value is a
        positive 32-bit value because the first check included bytes in
        range a[0:(31 - len)] and b[0:(31 - len)] so `eax` must be
        greater than `31 - len` so the minimum value of `edx` + `eax` is
        `(len - 32) + (32 - len) >= 0`. In the concurrent case, however,
        `a` or `b` could have been changed so a mismatch in `eax` less or
        equal than `(31 - len)` is possible (the new low bound is `(16 -
        len)`. This can result in a negative 32-bit signed integer, which
        when zero extended to 64-bits is a random large value this out
        out of bounds. */
        addl %edx, %eax
        /* Crash here because 32-bit negative number in `eax` zero
        extends to out of bounds 64-bit offset.  */
        movzbl  16(%rdi, %rax), %ecx
        movzbl  16(%rsi, %rax), %eax
    This fix is quite simple, just make the `addl %edx, %eax` 64 bit (i.e
    `addq %rdx, %rax`). This prevents the 32-bit zero extension
    and since `eax` is still a low bound of `16 - len` the `rdx + rax`
    is bound by `(len - 32) - (16 - len) >= -16`. Since we have a
    fixed offset of `16` in the memory access this must be in bounds.

 sysdeps/x86_64/multiarch/memcmp-sse2.S | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/sysdeps/x86_64/multiarch/memcmp-sse2.S b/sysdeps/x86_64/multiarch/memcmp-sse2.S
index afd450d020..51bc9344f0 100644
--- a/sysdeps/x86_64/multiarch/memcmp-sse2.S
+++ b/sysdeps/x86_64/multiarch/memcmp-sse2.S
@@ -308,7 +308,17 @@ L(ret_nonzero_vec_end_0):
 	setg	%dl
 	leal	-1(%rdx, %rdx), %eax
 #  else
-	addl	%edx, %eax
+	/* Use `addq` instead of `addl` here so that even if `rax` + `rdx`
+       is negative value of the sum will be usable as a 64-bit offset
+       (negative 32-bit numbers zero-extend to a large and often
+       out-of-bounds 64-bit offsets).  Note that `rax` + `rdx` >= 0 is
+       an invariant when `memcmp` is used correctly, but if the input
+       strings `rsi`/`rdi` are concurrently modified as the function
+       runs (there is a Data-Race) it is possible for `rax` + `rdx` to
+       be negative.  Given that there is virtually no extra to cost
+       using `addq` instead of `addl` we may as well protect the
+       data-race case.  */
+	addq	%rdx, %rax
 	movzbl	(VEC_SIZE * -1 + SIZE_OFFSET)(%rsi, %rax), %ecx
 	movzbl	(VEC_SIZE * -1 + SIZE_OFFSET)(%rdi, %rax), %eax
 	subl	%ecx, %eax

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2022-12-15 17:10 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-12-15 17:10 [glibc] x86: Prevent SIGSEGV in memcmp-sse2 when data is concurrently modified [BZ #29863] Noah Goldstein

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).