Bug 46147 - g++ segfault compiling gnat 0.8.8 on mips-linux n32 and -fvisibility-inlines-hidden -fPIC -march=loongson2f
Summary: g++ segfault compiling gnat 0.8.8 on mips-linux n32 and -fvisibility-inlines-...
Status: UNCONFIRMED
Alias: None
Product: gcc
Classification: Unclassified
Component: middle-end (show other bugs)
Version: 4.5.1
: P3 normal
Target Milestone: ---
Assignee: Not yet assigned to anyone
URL:
Keywords: ice-on-valid-code, visibility
Depends on:
Blocks: visibility
  Show dependency tree / graph
 
Reported: 2010-10-23 13:36 UTC by Lluís Batlle i Rossell
Modified: 2021-09-17 00:10 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:
Known to work:
Known to fail:
Last reconfirmed:

Attachments
The preprocessed source causing the trouble. (351.42 KB, application/octet-stream)
2010-10-23 13:36 UTC, Lluís Batlle i Rossell 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Lluís Batlle i Rossell 2010-10-23 13:36:48 UTC 
Created attachment 22132 [details]
The preprocessed source causing the trouble.

At the middle of building gnat 0.8.8 (with boost 1.44.0) g++ dies with an internal error related to a segmentation fault.

I'm using gcc 4.5.1 with glibc 2.12.1 in a mips64-linux running n32 userland code, and binutils from a snapshot of the end of august (for the loongson2f fix).

The gcc is built, as a special flag, with --with-arch=loongson2f.

I attach the preprocessed file that triggers the error. Compile the file with the same flags the gnash build system decided:
g++ -g -O2 -fvisibility-inlines-hidden -fPIC -c
Removing only "-fPIC", the problem goes away. Using "-O0" instead, also makes the problem go away. Removing only "-fvisibility-inlines-hidden" also makes the problem go away.

The system has 1GiB of RAM, and monitoring with 'top', I see g++ never goes beyond 300MiB.
Comment 1 Paolo Carlini 2012-10-05 13:58:59 UTC 
CC-ing mips maintainers.