This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug c++/81389] New: _mm_cmpestri segfault on -O0


https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81389

            Bug ID: 81389
           Summary: _mm_cmpestri segfault on -O0
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rockeet at gmail dot com
  Target Milestone: ---

With option -O0, intrinsic _mm_cmpestri generate a mmload and a _mm_cmpestri,
in which mmload raise a CPU hardware exception.

The reason is: mmload require memory align to 16, but _mm_cmpestri directly on
memory does not require memory align.

It is OK with option -O3, because the compiler does not generate mmload, it
generate an mm_cmpestri directly on memory.

I had noticed this bug in gcc-4.7, serveral years later, the bug is still
there(gcc-6.3)...

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]