Re: [libav-devel] [PATCH] dsputil_mmx: fix incorrect assembly code

2012-07-25 Thread Ronald S. Bultje
Hi, On Mon, Jul 23, 2012 at 5:30 PM, Derek Buitenhuis derek.buitenh...@gmail.com wrote: From: Yang Wang yang.y.w...@intel.com 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,

[libav-devel] [PATCH] dsputil_mmx: fix incorrect assembly code

2012-07-23 Thread Derek Buitenhuis
From: Yang Wang yang.y.w...@intel.com 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 +

Re: [libav-devel] [PATCH] dsputil_mmx: fix incorrect assembly code

2012-07-23 Thread Diego Biurrun
On Mon, Jul 23, 2012 at 07:09:06PM -0400, Derek Buitenhuis wrote: From: Yang Wang yang.y.w...@intel.com 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 the ffmpeg using Intel C++

Re: [libav-devel] [PATCH] dsputil_mmx: fix incorrect assembly code

2012-07-23 Thread Derek Buitenhuis
On 23/07/2012 7:32 PM, Diego Biurrun wrote: Please don't undo the prettyprinting I applied to that file. I don't see a way to keep it pretty short of re-indenting the whole block, which should be a 2nd commit. - Derek ___ libav-devel mailing list

Re: [libav-devel] [PATCH] dsputil_mmx: fix incorrect assembly code

2012-07-23 Thread Diego Biurrun
On Mon, Jul 23, 2012 at 08:13:20PM -0400, Derek Buitenhuis wrote: On 23/07/2012 7:32 PM, Diego Biurrun wrote: Please don't undo the prettyprinting I applied to that file. I don't see a way to keep it pretty short of re-indenting the whole block, which should be a 2nd commit. Have a look at

[libav-devel] [PATCH] dsputil_mmx: fix incorrect assembly code

2012-07-23 Thread Derek Buitenhuis
From: Yang Wang yang.y.w...@intel.com 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 +

Re: [libav-devel] [PATCH] dsputil_mmx: fix incorrect assembly code

2012-07-23 Thread Derek Buitenhuis
On 23/07/2012 8:30 PM, Derek Buitenhuis wrote: How to reproduce: This error is exposed when we build the ffmpeg using Intel C++ Compiler, IPO+PGO optimization. Crashed when decoding an MJPEG video. Forgot to remove the first occurrence. Woops. Amended locally, pending further review.