On 1/8/26 1:15 PM, Richard Henderson wrote:
On 1/9/26 07:09, Pierrick Bouvier wrote:
On 1/7/26 9:29 PM, Richard Henderson wrote:
Since x86_64 always has SSE2, we can remove the fallback
that was present for i686.
Signed-off-by: Richard Henderson <[email protected]>
---
host/include/x86_64/host/bufferiszero.c.inc | 5 -----
1 file changed, 5 deletions(-)
diff --git a/host/include/x86_64/host/bufferiszero.c.inc
b/host/include/x86_64/host/
bufferiszero.c.inc
index 74ae98580f..7e9d896a8d 100644
--- a/host/include/x86_64/host/bufferiszero.c.inc
+++ b/host/include/x86_64/host/bufferiszero.c.inc
@@ -3,7 +3,6 @@
* buffer_is_zero acceleration, x86 version.
*/
-#if defined(CONFIG_AVX2_OPT) || defined(__SSE2__)
#include <immintrin.h>
/* Helper for preventing the compiler from reassociating
@@ -119,7 +118,3 @@ static unsigned best_accel(void)
#endif
return info & CPUINFO_SSE2 ? 1 : 0;
}
-
-#else
-# include "host/include/generic/host/bufferiszero.c.inc"
-#endif
The only other user for this file is now
host/include/aarch64/host/bufferiszero.c.inc.
Code could be directly moved there instead, so host/include/generic/host/
bufferiszero.c.inc can be removed.
It can be done in another commit though.
No. Everyone who *doesn't* have such a file uses generic.
r~
Oops, missed util/bufferiszero.c indeed.
