Bug 26352 - ICE
Summary: ICE
Status: RESOLVED INVALID
Alias: None
Product: gcc
Classification: Unclassified
Component: c++ (show other bugs)
Version: 4.0.2
: P3 normal
Target Milestone: ---
Assignee: Not yet assigned to anyone
URL:
Keywords: memory-hog
Depends on:
Blocks:
 
Reported: 2006-02-18 15:47 UTC by Ivan Godard
Modified: 2007-09-24 17:05 UTC (History)
2 users (show)

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

Attachments
compiler output (697 bytes, text/plain)
2006-02-18 15:48 UTC, Ivan Godard 		Details
source code (compressed) (137.54 KB, application/x-gzip)
2006-02-18 15:48 UTC, Ivan Godard 		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 Ivan Godard 2006-02-18 15:47:43 UTC
Comment 1 Ivan Godard 2006-02-18 15:48:08 UTC 
Created attachment 10873 [details]
compiler output
Comment 2 Ivan Godard 2006-02-18 15:48:30 UTC 
Created attachment 10874 [details]
source code (compressed)
Comment 3 Andrew Pinski 2006-02-18 16:07:45 UTC 
g++: Internal error: Killed (program cc1plus)

means memory was over used.
Comment 4 Andrew Pinski 2006-02-18 16:22:03 UTC 
I cannot reproduce this at all in 4.0.2, 4.0.0, or 4.0.3.
Comment 5 Wolfgang Bangerth 2006-03-08 04:55:31 UTC 
I see you used precompiled headers. Does the same ICE happen if you
don't do that?

W.
Comment 6 Ivan Godard 2006-03-08 06:16:18 UTC 
We don't use pre-compiled headers ourselves. Any precompiles are third-party or standard library, and actually I was unaware that they were used. Can you tell me the files that the header contained? Then I can identify where we got them and might be able to get un-precompiled versions.

Ivan
Comment 7 Wolfgang Bangerth 2006-03-08 14:57:37 UTC 
(In reply to comment #6)
> We don't use pre-compiled headers ourselves. Any precompiles are third-party or
> standard library, and actually I was unaware that they were used. Can you tell
> me the files that the header contained? Then I can identify where we got them
> and might be able to get un-precompiled versions.

All I can say is that in your log file I see the command line

/mnt/export/home/ivan/gcc/bin/../libexec/gcc/i686-pc-linux-gnu/4.0.2/cc1plus -fpreprocessed powersetTest.ii -quiet -dumpbase powersetTest.cc -mtune=pentiumpro -auxbase powersetTest -g -O0 -version -o powersetTest.s

Since I don't have your sources (and don't want to), I can't help you search
further.

W.
Comment 8 Ivan Godard 2006-03-08 19:29:27 UTC 
Well I was really surprised to see this -fpreprocessed option in the -v output, so I did a little digging. It seems that this option is set by the driver whenever -save-temps is set. It does not come from my command line. Try it: compile something -v, and then compile the same thing -v -save-temps and the first -v will not have -fpreprocessed and the second one will.

The bug report instructions say to run with -save-temps so I do. You might ask internally why the driver sets these when the user asks for -save-temps. I think the generated -fpreprocessed is irrelevant to the bug, because it will be set on every  compile that produces a .ii file for you and I've put in a few of those :-)

Ivan
Comment 9 Paolo Carlini 2007-09-24 17:05:09 UTC 
In any case, the pre-processed code doesn't compile anymore with 4.2 and mainline.