[Bug pch/60010] New: AArch64: sigsegv in cc1plus using pch without defining TRY_EMPTY_VM_SPACE

[kyle at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60010

Bug ID: 60010
   Summary: AArch64: sigsegv in cc1plus using pch without defining
TRY_EMPTY_VM_SPACE
   Product: gcc
   Version: 4.8.2
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: pch
  Assignee: unassigned at gcc dot gnu.org
  Reporter: kyle at redhat dot com

Created attachment 32002
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32002&action=edit
add TRY_EMPTY_VM_SPACE define for __aarch64__

Building Fedora with randomize_vm_space on, we've noticed segmentation faults
in cc1plus when precompiled headers are used. While I was debugging the PCH
internals, Jakub kindly pointed out that AArch64 would probably need a
TRY_EMPTY_VM_SPACE define to start looking for free memory at, instead of the
default 0. (I suspect that buffer_size needs to be bigger than 32MB now.)

Attached is a patch that survived rebuilding our easiest reproducing test-case
(wxGTK) 30 times without issue.

Thanks, Kyle M.

[Bug pch/60010] AArch64: sigsegv in cc1plus using pch without defining TRY_EMPTY_VM_SPACE

[kyle at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60010

--- Comment #1 from Kyle McMartin  ---
https://bugzilla.redhat.com/show_bug.cgi?id=1058991

Just FYI, this is RHBZ#1058991.

[Bug pch/60010] AArch64: sigsegv in cc1plus using pch without defining TRY_EMPTY_VM_SPACE

[kyle at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60010

--- Comment #3 from Kyle McMartin  ---
Will do. Thanks much, Andrew. (I'll use you ChangeLog from adding it for mips
as a template.)