[libav-commits] dsputil_mmx: fix incorrect assembly code

Yang Wang git at libav.org
Wed Jul 25 20:59:41 CEST 2012


Module: libav
Branch: master
Commit: 845e92fd6abf749a960354becdc5a9bc498f1f96

Author:    Yang Wang <yang.y.wang at intel.com>
Committer: Derek Buitenhuis <derek.buitenhuis at gmail.com>
Date:      Tue Jul 24 00:51:10 2012 +0200

dsputil_mmx: fix incorrect assembly code

In ff_put_pixels_clamped_mmx(), there are two assembly code blocks.
In the first block (in the unrolled loop), the instructions
"movq 8%3, %%mm1 \n\t", and so forth, have problems.

>From above instruction, it is clear what the programmer wants: a load from
p + 8. But this assembly code doesn’t guarantee that. It only works if the
compiler puts p in a register to produce an instruction like this:
"movq 8(%edi), %mm1". During compiler optimization, it is possible that the
compiler will be able to constant propagate into p. Suppose p = &x[10000].
Then operand 3 can become 10000(%edi), where %edi holds &x. And the instruction
becomes "movq 810000(%edx)". That is, it will stride by 810000 instead of 8.

This will cause a segmentation fault.

This error was fixed in the second block of the assembly code, but not in
the unrolled loop.

How to reproduce:
    This error is exposed when we build using Intel C++ Compiler, with
    IPO+PGO optimization enabled. Crashed when decoding an MJPEG video.

Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>

---

 libavcodec/x86/dsputil_mmx.c |   18 +++++++++---------
 1 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 5eb4a24..522a565 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -245,14 +245,14 @@ void ff_put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels,
     pix = pixels;
     /* unrolled loop */
     __asm__ volatile (
-        "movq        %3, %%mm0          \n\t"
-        "movq       8%3, %%mm1          \n\t"
-        "movq      16%3, %%mm2          \n\t"
-        "movq      24%3, %%mm3          \n\t"
-        "movq      32%3, %%mm4          \n\t"
-        "movq      40%3, %%mm5          \n\t"
-        "movq      48%3, %%mm6          \n\t"
-        "movq      56%3, %%mm7          \n\t"
+        "movq      (%3), %%mm0          \n\t"
+        "movq     8(%3), %%mm1          \n\t"
+        "movq    16(%3), %%mm2          \n\t"
+        "movq    24(%3), %%mm3          \n\t"
+        "movq    32(%3), %%mm4          \n\t"
+        "movq    40(%3), %%mm5          \n\t"
+        "movq    48(%3), %%mm6          \n\t"
+        "movq    56(%3), %%mm7          \n\t"
         "packuswb %%mm1, %%mm0          \n\t"
         "packuswb %%mm3, %%mm2          \n\t"
         "packuswb %%mm5, %%mm4          \n\t"
@@ -262,7 +262,7 @@ void ff_put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels,
         "movq     %%mm4, (%0, %1, 2)    \n\t"
         "movq     %%mm6, (%0, %2)       \n\t"
         :: "r"(pix), "r"((x86_reg)line_size), "r"((x86_reg)line_size * 3),
-           "m"(*p)
+           "r"(p)
         : "memory");
     pix += line_size * 4;
     p   += 32;



More information about the libav-commits mailing list