On Fri, Jan 15, 2010 at 02:41:57PM +0100, Jiri Moskovcak wrote: > On 01/15/2010 01:10 PM, Neal Becker wrote: > >This is the second time in a row, generating backtrace seems to run forever. > >Something wrong? > >abrt-1.0.3-1.fc12.x86_64 > > > > What is it you're trying to process? If it's some bigger thing like > Firefox or Openoffice it might take a really long time. > > Jirka What kernel are you using? I committed a change to how the core pipe stuff works in the kernel for security reasons, which requires processes to have a non-zero core limit. It hit in 2.6.32 IIRC and probably just hit fedora. That could be the cause as well. Someone from canonical said they wanted to rewrite some of do_coredump to allow zero core limits to dump to pipes while still preventing recursive cores, but I've not heard from them in quite some time.
If thats the problem, the solution should be pretty straightforward, if you set the core limit for init to 1, (or simply set the session core limit value to 1 as a test case for user processes, that should work. Neil > begin:vcard > fn:Jiri Moskovcak > n:Moskovcak;Jiri > email;internet:jmosk...@redhat.com > x-mozilla-html:FALSE > version:2.1 > end:vcard > > -- > devel mailing list > devel@lists.fedoraproject.org > https://admin.fedoraproject.org/mailman/listinfo/devel -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
