From: DRC <information@libjpeg-turbo.org>
Date: Fri, 28 Sep 2018 20:46:35 +0000 (-0500)
Subject: SIMD: Fix c000001d exception on Win 7 w/o SP1
X-Git-Tag: 2.0.1~9
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=d5f281b734425fc1d930ff2c3f8441aad731343e;p=libjpeg-turbo

SIMD: Fix c000001d exception on Win 7 w/o SP1

Apparently Windows 7 without SP1 has O/S support for XSAVE but not for
YMM registers, and this exposed a bug in our usage of xgetbv.  The test
instruction will set ZF only if none of the bits match between the two
operarands, so in effect, we were enabling AVX2 instructions if the O/S
supported XSAVE and the CPU supported AVX2 but the O/S only supported
XMM registers.  This bug was not exposed on, for instance, Windows XP or
RHEL 5 because those O/S's do not support XSAVE.

Fixes #288
---

diff --git a/ChangeLog.md b/ChangeLog.md
index 3d99801..47c730b 100644
--- a/ChangeLog.md
+++ b/ChangeLog.md
@@ -14,6 +14,9 @@ functions in the MIPS DSPr2 SIMD extensions are now disabled at compile time
 if the soft float ABI is enabled.  Those functions use instructions that are
 incompatible with the soft float ABI.
 
+3. Fixed an issue in the SIMD feature detection code that caused libjpeg-turbo
+to crash on Windows 7 if Service Pack 1 was not installed.
+
 
 2.0.0
 =====
diff --git a/simd/i386/jsimdcpu.asm b/simd/i386/jsimdcpu.asm
index 50a0d51..faddd38 100644
--- a/simd/i386/jsimdcpu.asm
+++ b/simd/i386/jsimdcpu.asm
@@ -94,9 +94,10 @@ EXTN(jpeg_simd_cpu_support):
 
     xor         ecx, ecx
     xgetbv
-    test        eax, 6                  ; O/S does not manage XMM/YMM state
+    and         eax, 6
+    cmp         eax, 6                  ; O/S does not manage XMM/YMM state
                                         ; using XSAVE
-    jz          short .no_avx2
+    jnz         short .no_avx2
 
     or          edi, JSIMD_AVX2
 .no_avx2:
diff --git a/simd/x86_64/jsimdcpu.asm b/simd/x86_64/jsimdcpu.asm
index 42979be..38e1a7b 100644
--- a/simd/x86_64/jsimdcpu.asm
+++ b/simd/x86_64/jsimdcpu.asm
@@ -60,9 +60,10 @@ EXTN(jpeg_simd_cpu_support):
 
     xor         rcx, rcx
     xgetbv
-    test        rax, 6                  ; O/S does not manage XMM/YMM state
+    and         rax, 6
+    cmp         rax, 6                  ; O/S does not manage XMM/YMM state
                                         ; using XSAVE
-    jz          short .return
+    jnz         short .return
 
     or          rdi, JSIMD_AVX2