From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Red Hat/1.0.6-1.4.1 Firefox/1.0.6 Description of problem: rpmbuild --rebuild glibc-2.3.4-2.12.src.rpm (also glibc-2.3.4-2.9.src.rpm) fails on both i386 and x86_64 machines (--target i686 also fails on i386 machine). These only fail if either kernel-2.6.9-11.EL or kernel-2.6.9-16.EL (or the smp kernels of the same version are installed). If kernel-2.6.9-5.0.5.EL or kernel-2.6.9-5.0.3.EL or kernel-2.6.9-5.EL are installed, the rebuild completes normally. I have verified this issue on i386 single processor (pIII and p4), x86_64 (AMD Athlon 64), i386 dual processor (p4 xeon) The log file is large, so I am going to gzip it. Version-Release number of selected component (if applicable): glibc-2.3.4-2.12.src.rpm (and glibc-2.3.4-2.9.src.rpm) How reproducible: Always Steps to Reproduce: rpmbuild --rebuild glibc-2.3.4-2.12.src.rpm OR rpmbuild --rebuild glibc-2.3.4-2.9.src.rpm Actual Results: See the attached log file ... if you notice on line 9456 is the first error. make[2]: *** [/usr/src/redhat/BUILD/glibc-2.3-20050704T1133/build-i486-linuxnptl/nptl/tst-join4.out] Error 1 Failed to kill test process: No child processes The build process continues (and there are several releated errors) until the end of the log. At this point the actual build process is hung (line 9598 of the log). Pressing ctrl-c (or killing the defunct process) produces more log lines (starting with 9599 to 9785) and then all the processes fo rthe build exit. If kernel-2.6.9-5.0.5.EL or kernel-2.6.9-5.0.3.EL or kernel-2.6.9-5.EL are installed, the rebuild completes normally. Expected Results: Normal build. Additional info:
Created attachment 117956 [details] gzipped build log
The build log contains "2>&1" information ... and this build was from an i386 install, single processor machine, RHEL4-U2 beta. Kernel was 2.6.9-16.EL.i686 same results with building glibc-2.3.4-2.9.src.rpm using kernel-2.6.9-11.EL.i686
The same bug makes the check_dns nagios plugin (and perhaps some other plugins) segfault most of the time, and is still there with kernel 2.6.9-22.EL. This seems to be related to bug #141896 which was fixed in 2.6.10. The same fix applies here.
Thibaut, I appreciate the update, and the pointer to 141896. This case was a bit confusing at first until I realized that it wasn't related to rpmbuild, but rather the NPTL test within it. Thanks, Dave
(i.e., the NPTL test within the glibc package...)
*** Bug 168775 has been marked as a duplicate of this bug. ***
Confirmation that the problem still exists with kernel-2.6.9-22.EL-i686 and glibc-2.3.4-2.13 (single processor Pentium III). Is the patch referenced in bug #141896 (see comment #4 above) suitable for use with this kernel, or is that patch specific to v2.6.10?
That patch is 2.6.10-specific and would need slight adjustments for RHEL4, but everything from it needs to be applied except for this part: case TASK_TRACED: - flag = 1; if (!my_ptrace_child(p)) continue; /*FALLTHROUGH*/ That particular "flag = 1;" should be left as is.
Test kernels containing a fix for this bug may be found in: http://people.redhat.com/~jbaron/rhel4
*** Bug 153323 has been marked as a duplicate of this bug. ***
*** Bug 180353 has been marked as a duplicate of this bug. ***
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHSA-2006-0132.html