Hi,
the attached patch is strictly identical to this x264 one:
http://git.videolan.org/?p=x264/x264-sandbox.git;a=commit;h=a7a909fc7b82df85ff369adcbc048927ebec2289
except for the edited paths.
The goal is to catch that last batch of issues in external asm:
http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=fc7e02f0ff345d5331b7c78f2400668d2c79a8b0;hp=4ccd7cb45b9aa46d94c29dbd1c065b652bda2319
It is currently in x264 'sandbox', ie not in a released version.
--
Christophe
From 1ade9345fbb392c0612ff0053523857abd5da563 Mon Sep 17 00:00:00 2001
From: Christophe Gisquet <christophe.gisq...@gmail.com>
Date: Tue, 3 Feb 2015 20:40:41 +0100
Subject: [PATCH] x86inc: also warn on sse2 instruction
When using movq/movh/movd in e.g. SSE code, 974f2e78 may not warn
about it.
---
libavutil/x86/x86inc.asm | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/libavutil/x86/x86inc.asm b/libavutil/x86/x86inc.asm
index 33b22cd..b0ef2c6 100644
--- a/libavutil/x86/x86inc.asm
+++ b/libavutil/x86/x86inc.asm
@@ -1074,6 +1074,11 @@ INIT_XMM
%if notcpuflag(%2)
%error use of ``%1'' %2 instruction in cpuname function: current_function
%endif
+ %ifidn %2, mmx
+ %if __sizeofreg > 8 && notcpuflag(sse2)
+ %error use of ``%1'' sse2 instruction in cpuname function: current_function
+ %endif
+ %endif
%endif
%endif
@@ -1209,7 +1214,7 @@ AVX_INSTR minsd, sse2, 1, 0, 1
AVX_INSTR minss, sse, 1, 0, 1
AVX_INSTR movapd, sse2
AVX_INSTR movaps, sse
-AVX_INSTR movd
+AVX_INSTR movd, mmx
AVX_INSTR movddup, sse3
AVX_INSTR movdqa, sse2
AVX_INSTR movdqu, sse2
@@ -1225,7 +1230,7 @@ AVX_INSTR movntdq, sse2
AVX_INSTR movntdqa, sse4
AVX_INSTR movntpd, sse2
AVX_INSTR movntps, sse
-AVX_INSTR movq
+AVX_INSTR movq, mmx
AVX_INSTR movsd, sse2, 1, 0, 0
AVX_INSTR movshdup, sse3
AVX_INSTR movsldup, sse3
--
1.9.2.msysgit.0
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
