[PATCH] fhandler_proc.cc(format_proc_cpuinfo): fix bad bits in last change

[PATCH] fhandler_proc.cc(format_proc_cpuinfo): fix bad bits in last change

[Brian Inglis]

[-- Attachment #1: Type: text/plain, Size: 522 bytes --]


move Linux 5.16 Gobble Gobble flags to 5.17 Superb Owl correct positions:
0x00000007:1 Intel Advanced Matrix eXtensions:
		 EAX:22 amx_bf16 Brain Float 16 dot product
		 EAX:24 amx_tile Tile matrix multiply
		 EAX:25 amx_int8 Int 8 byte dot product
0x00000007:0 Intel Advanced Matrix eXtensions:
		 EDX:22 amx_bf16 Brain Float 16 dot product
		 EDX:24 amx_tile Tile matrix multiply
		 EDX:25 amx_int8 Int 8 byte dot product
---
 winsup/cygwin/fhandler_proc.cc | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-fhandler_proc.cc-format_proc_cpuinfo-fix-bad-bits-in-last-change.patch --]
[-- Type: text/x-patch; name="0001-fhandler_proc.cc-format_proc_cpuinfo-fix-bad-bits-in-last-change.patch", Size: 1599 bytes --]

diff --git a/winsup/cygwin/fhandler_proc.cc b/winsup/cygwin/fhandler_proc.cc
index d51f171a3a41..0b01e68f8e50 100644
--- a/winsup/cygwin/fhandler_proc.cc
+++ b/winsup/cygwin/fhandler_proc.cc
@@ -1456,9 +1456,6 @@ format_proc_cpuinfo (void *, char *&destbuf)
 
 	  ftcprint (features1,  4, "avx_vnni");	    /* vex enc NN vec */
 	  ftcprint (features1,  5, "avx512_bf16");  /* vec bfloat16 short */
-	  ftcprint (features1, 22, "amx_bf16");	    /* Advanced Matrix eXtensions Brain Float 16 dot product */
-	  ftcprint (features1, 24, "amx_tile");	    /* Advanced Matrix eXtensions Tile matrix multiply */
-	  ftcprint (features1, 25, "amx_int8");	    /* Advanced Matrix eXtensions Int 8 byte dot product */
 	}
 
       /* AMD cpuid 0x80000008 ebx */
@@ -1576,7 +1573,10 @@ format_proc_cpuinfo (void *, char *&destbuf)
           ftcprint (features1, 16, "tsxldtrk");		   /* TSX Susp Ld Addr Track */
           ftcprint (features1, 18, "pconfig");		   /* platform config */
           ftcprint (features1, 19, "arch_lbr");		   /* last branch records */
+	  ftcprint (features1, 22, "amx_bf16");	    /* Advanced Matrix eXtensions Brain Float 16 dot product */
           ftcprint (features1, 23, "avx512_fp16");	   /* avx512 fp16 */
+	  ftcprint (features1, 24, "amx_tile");	    /* Advanced Matrix eXtensions Tile matrix multiply */
+	  ftcprint (features1, 25, "amx_int8");	    /* Advanced Matrix eXtensions Int 8 byte dot product */
           ftcprint (features1, 28, "flush_l1d");	   /* flush l1d cache */
           ftcprint (features1, 29, "arch_capabilities");   /* arch cap MSR */
         }

Re: [PATCH] fhandler_proc.cc(format_proc_cpuinfo): fix bad bits in last change

[Corinna Vinschen]

On Feb 15 14:54, Brian Inglis wrote:
> 
> move Linux 5.16 Gobble Gobble flags to 5.17 Superb Owl correct positions:
> 0x00000007:1 Intel Advanced Matrix eXtensions:
> 		 EAX:22 amx_bf16 Brain Float 16 dot product
> 		 EAX:24 amx_tile Tile matrix multiply
> 		 EAX:25 amx_int8 Int 8 byte dot product
> 0x00000007:0 Intel Advanced Matrix eXtensions:
> 		 EDX:22 amx_bf16 Brain Float 16 dot product
> 		 EDX:24 amx_tile Tile matrix multiply
> 		 EDX:25 amx_int8 Int 8 byte dot product
> ---
>  winsup/cygwin/fhandler_proc.cc | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 

Pushed.


Thanks,
Corinna