We know that the called function (ff_chroma_inter_body_mmxext)
doesn't touch the redzone, and thus will be kept intact - thus,
this doesn't fix any bug per se.
However, valgrind's memcheck tool intentionally assumes that the
redzone is clobbered on every function call and function return
(see a long comment in valgrind/memcheck/mc_main.c). This avoids
false positives in that tool, at the cost of an extra stack pointer
adjustment.
The other alternative would be a valgrind suppression for this issue,
but that's an extra burden for everybody that wants to run libavcodec
within valgrind.
Signed-off-by: Martin Storsjö <martin@martin.st>
; int8_t *tc0)
;-----------------------------------------------------------------------------
cglobal deblock_h_chroma_8, 5,7
-%if UNIX64
- %define buf0 [rsp-24]
- %define buf1 [rsp-16]
-%elif WIN64
+%if ARCH_X86_64
+ ; This could use the red zone on 64 bit unix to avoid the stack pointer
+ ; readjustment, but valgrind assumes the red zone is clobbered on
+ ; function calls and returns.
sub rsp, 16
%define buf0 [rsp]
%define buf1 [rsp+8]
movq m0, buf0
movq m3, buf1
TRANSPOSE8x4B_STORE PASS8ROWS(t5, r0, r1, t6)
-%if WIN64
+%if ARCH_X86_64
add rsp, 16
%endif
RET