[Bug libc/25966] New: Incorrect access of __x86_shared_non_temporal_threshold for x32

[Bug libc/25966] New: Incorrect access of __x86_shared_non_temporal_threshold for x32

[hjl.tools at gmail dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=25966

            Bug ID: 25966
           Summary: Incorrect access of
                    __x86_shared_non_temporal_threshold for x32
           Product: glibc
           Version: 2.32
            Status: NEW
          Severity: normal
          Priority: P2
         Component: libc
          Assignee: unassigned at sourceware dot org
          Reporter: hjl.tools at gmail dot com
                CC: drepper.fsp at gmail dot com
  Target Milestone: ---
            Target: x86_64

There are

sysdeps/x86/cacheinfo.c:long int __x86_shared_non_temporal_threshold
attribute_hidden;
sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:   cmpq 
__x86_shared_non_temporal_threshold(%rip), %rdx
sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:   cmpq 
__x86_shared_non_temporal_threshold(%rip), %rdx
sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:   cmpq 
__x86_shared_non_temporal_threshold(%rip), %rdx

But long int is 4 bytes for x32.  The upper 32 bits have random values.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug libc/25966] Incorrect access of __x86_shared_non_temporal_threshold for x32

[hjl.tools at gmail dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=25966

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> ---
I will check in this patch:

diff --git a/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S
b/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S
index c763b7d871..74953245aa 100644
--- a/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S
+++ b/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S
@@ -244,7 +244,7 @@ L(return):
   ret

 L(movsb):
-  cmpq  __x86_shared_non_temporal_threshold(%rip), %rdx
+  cmp   __x86_shared_non_temporal_threshold(%rip), %RDX_LP
   jae   L(more_8x_vec)
   cmpq  %rsi, %rdi
   jb 1f
@@ -402,7 +402,7 @@ L(more_8x_vec):
   addq  %r8, %rdx
 #if (defined USE_MULTIARCH || VEC_SIZE == 16) && IS_IN (libc)
   /* Check non-temporal store threshold.  */
-  cmpq  __x86_shared_non_temporal_threshold(%rip), %rdx
+  cmp   __x86_shared_non_temporal_threshold(%rip), %RDX_LP
   ja L(large_forward)
 #endif
 L(loop_4x_vec_forward):
@@ -454,7 +454,7 @@ L(more_8x_vec_backward):
   subq  %r8, %rdx
 #if (defined USE_MULTIARCH || VEC_SIZE == 16) && IS_IN (libc)
   /* Check non-temporal store threshold.  */
-  cmpq  __x86_shared_non_temporal_threshold(%rip), %rdx
+  cmp   __x86_shared_non_temporal_threshold(%rip), %RDX_LP
   ja L(large_backward)
 #endif
 L(loop_4x_vec_backward):

after verifying there are no binary changes for x86-64 and no regressions
on x32.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug libc/25966] Incorrect access of __x86_shared_non_temporal_threshold for x32

[hjl.tools at gmail dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=25966

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |2.32
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #2 from H.J. Lu <hjl.tools at gmail dot com> ---
Fixed for 2.32 and on 2.30/2.31 branches.

-- 
You are receiving this mail because:
You are on the CC list for the bug.