[PATCH] x86/testsuite: Refine AVX10.2 rounding testcases

[Haochen Jiang]

Hi all,

As discussed in AVX10.2 media instructions, using exact byte code
is definitely not a good idea. Change that for AVX10.2 rounding
testcases.

Ok for trunk?

Thx,
Haochen

---
Using hard byte code is not a good idea in dump file. Add a label
for intel syntax test check to avoid that.

gas/ChangeLog:

	* testsuite/gas/i386/avx10_2-rounding-intel.d: Use label for
	test split.
	* testsuite/gas/i386/avx10_2-rounding.s: Add label to avoid
	hard coding in dump file.
---
 gas/testsuite/gas/i386/avx10_2-rounding-intel.d | 4 ++--
 gas/testsuite/gas/i386/avx10_2-rounding.s       | 1 +
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/gas/testsuite/gas/i386/avx10_2-rounding-intel.d b/gas/testsuite/gas/i386/avx10_2-rounding-intel.d
index 09535f6a1f7..f40c296d823 100644
--- a/gas/testsuite/gas/i386/avx10_2-rounding-intel.d
+++ b/gas/testsuite/gas/i386/avx10_2-rounding-intel.d
@@ -6,9 +6,9 @@
 
 Disassembly of section \.text:
 
-0+ <_start>:
 #...
-\s*a83:\s*62 f1 d1 18 c2 ec 7b\s+vcmppd k5,ymm5,ymm4\{sae\},0x7b
+[a-f0-9]+ <_intel>:
+\s*[a-f0-9]+:\s*62 f1 d1 18 c2 ec 7b\s+vcmppd k5,ymm5,ymm4\{sae\},0x7b
 \s*[a-f0-9]+:\s*62 f1 d1 1f c2 ec 7b\s+vcmppd k5\{k7\},ymm5,ymm4\{sae\},0x7b
 \s*[a-f0-9]+:\s*62 f2 f9 18 42 f5\s+vgetexppd ymm6,ymm5\{sae\}
 \s*[a-f0-9]+:\s*62 f2 f9 1f 42 f5\s+vgetexppd ymm6\{k7\},ymm5\{sae\}
diff --git a/gas/testsuite/gas/i386/avx10_2-rounding.s b/gas/testsuite/gas/i386/avx10_2-rounding.s
index cb0a6b9a900..048e7150635 100644
--- a/gas/testsuite/gas/i386/avx10_2-rounding.s
+++ b/gas/testsuite/gas/i386/avx10_2-rounding.s
@@ -175,6 +175,7 @@ _start:
 	vcvt\n\()2ph	{rz-sae}, %ymm5, %ymm6{%k7}{z}
 	.endr
 
+_intel:
 	.intel_syntax noprefix
 	.irp m, pd, ph, ps
 	vcmp\m	k5, ymm5, ymm4{sae}, 123
-- 
2.31.1