Hide Forgot
abrt version: 1.1.14 architecture: i686 Attached file: backtrace cmdline: /usr/libexec/gcc/i686-redhat-linux/4.5.1/cc1plus -quiet -D_GNU_SOURCE main.cpp -quiet -dumpbase main.cpp -mtune=generic -march=i686 -auxbase main -o /tmp/cc6UuSWz.s component: gcc crash_function: gt_ggc_mx_pending_template executable: /usr/libexec/gcc/i686-redhat-linux/4.5.1/cc1plus kernel: 2.6.35.11-83.fc14.i686.PAE package: gcc-c++-4.5.1-4.fc14 rating: 4 reason: Process /usr/libexec/gcc/i686-redhat-linux/4.5.1/cc1plus was killed by signal 11 (SIGSEGV) release: Fedora release 14 (Laughlin) time: 1298283871 uid: 500 How to reproduce ----- 1.Compile 2. 3.
Created attachment 479863 [details] File: backtrace
Please provide preprocessed source (usually g++ should prepare a /tmp/cc*.out file for you and tell it where it is, alternatively preprocess it yourself (either with -save-temps or -E instead of -c).
I've managed to reproduce it on an artificial testcase and am tracking this upstream now, nevertheless the preprocessed source would be useful to help figuring out if this actually is a regression from earlier versions of g++ or not.
Created attachment 479920 [details] Source code Its tested source code for this bug.
Can be hardly considered a regression, given that g++ 4.3 and 4.4 instead of segfaulting on it within minutes keep compiling it for terribly long (killed cc1plus after 72 minutes).
Should be fixed in gcc-4.6.0-0.11.fc15 in rawhide.