Re: [FFmpeg-devel] [FFmpeg-cvslog] tests/checkasm: Add partial support for wasm

[Martin Storsjö]

On Wed, 4 Dec 2024, Zhao Zhili wrote:



On Dec 4, 2024, at 17:56, Martin Storsjö <mar...@martin.st> wrote:

On Wed, 4 Dec 2024, Zhao Zhili wrote:

ffmpeg | branch: master | Zhao Zhili <zhiliz...@tencent.com> | Wed Nov 20 
20:45:35 2024 +0800| [ea3d21c349db5bbfee7840857de81e4b2fc757fa] | committer: Zhao 
Zhili

tests/checkasm: Add partial support for wasm

WASI mssing signal and siglongjmp support. This patch workaround
build error and add simd128 flag. Please note that many tests use
large array on stack, so you need to increase the stack size when
build checkasm, e.g., --extra-ldflags='-Wl,-z,stack-size=10485760'

Signed-off-by: Zhao Zhili <zhiliz...@tencent.com>

http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=ea3d21c349db5bbfee7840857de81e4b2fc757fa
---

tests/checkasm/checkasm.c |  8 ++++++++
tests/checkasm/checkasm.h | 15 +++++++++++++--
2 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/tests/checkasm/checkasm.c b/tests/checkasm/checkasm.c
index f30c53cec2..544e3ee9ba 100644
--- a/tests/checkasm/checkasm.c
+++ b/tests/checkasm/checkasm.c
@@ -363,6 +363,8 @@ static const struct {
#elif ARCH_LOONGARCH
   { "LSX",      "lsx",      AV_CPU_FLAG_LSX },
   { "LASX",     "lasx",     AV_CPU_FLAG_LASX },
+#elif ARCH_WASM
+    { "SIMD128",    "simd128",  AV_CPU_FLAG_SIMD128 },
#endif
   { NULL }
};
@@ -770,6 +772,8 @@ static LONG NTAPI signal_handler(EXCEPTION_POINTERS *e) {
}
#endif
#else
+
+#ifndef _WASI_EMULATED_SIGNAL
static void signal_handler(int s);

static const struct sigaction signal_handler_act = {
@@ -784,6 +788,8 @@ static void signal_handler(int s) {
       checkasm_load_context(s);
   }
}
+#endif // _WASI_EMULATED_SIGNAL
+
#endif

/* Compares a string with a wildcard pattern. */
@@ -933,10 +939,12 @@ int main(int argc, char *argv[])
   AddVectoredExceptionHandler(0, signal_handler);
#endif
#else
+#ifndef _WASI_EMULATED_SIGNAL
   sigaction(SIGBUS,  &signal_handler_act, NULL);
   sigaction(SIGFPE,  &signal_handler_act, NULL);
   sigaction(SIGILL,  &signal_handler_act, NULL);
   sigaction(SIGSEGV, &signal_handler_act, NULL);
+#endif // _WASI_EMULATED_SIGNAL
#endif
#if HAVE_PRCTL && defined(PR_SET_UNALIGN)
   prctl(PR_SET_UNALIGN, PR_UNALIGN_SIGBUS);
diff --git a/tests/checkasm/checkasm.h b/tests/checkasm/checkasm.h
index 0ba5c3040d..a1bd19c54a 100644
--- a/tests/checkasm/checkasm.h
+++ b/tests/checkasm/checkasm.h
@@ -66,12 +66,23 @@ typedef struct { CONTEXT c; int status; } checkasm_context;
#define checkasm_save_context() 0
#define checkasm_load_context() do {} while (0)
#endif
-#else
+#else  // _WIN32
+
+#ifdef _WASI_EMULATED_SIGNAL

Why the nested #else + #ifdef, why not #elif defined()? I think that would keep 
the logic slightly less complex.

Should be fixed by
https://ffmpeg.org/pipermail/ffmpeg-devel/2024-December/337041.html

By the way, how to make nested conditional compile readable? The code use 
nested #if already:

#ifdef _WIN32
#include <windows.h>
#if ARCH_X86_32
#include <setjmp.h>

I remember this patch use nested #else + #ifdef is for easy to comment. 
Should we use indent like in VLC?

I don't mind adding indentation to the ifdefs, and adding comments can 
help readability as well.

// Martin
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".