Description of problem: Tracking a `syscall-reset' testcase bug for i386-debugger-on-x86_64-kernel. Version-Release number of selected component (if applicable): kernel-2.6.24.3-34.fc8.x86_64 (biarch: -m32) Working for native variants: kernel-2.6.24.3-34.fc8.i686 (-m32) kernel-2.6.24.3-34.fc8.x86_64 (-m64) How reproducible: Always. Steps to Reproduce: 1. wget -O syscall-reset.c 'http://sources.redhat.com/cgi-bin/cvsweb.cgi/~checkout~/tests/ptrace-tests/tests/syscall-reset.c?cvsroot=systemtap' 2. gcc -o syscall-reset syscall-reset.c -Wall -ggdb2 -D_GNU_SOURCE -m32 3. ./syscall-reset; echo $? Actual results: 1 Expected results: 0 Additional info: non-biarch (native arch size) runs work fine. Rawhide kernel-2.6.25-0.121.rc5.git4.fc9.x86_64 - non-utrace one, based on 2.6.25-rc5-git4 - fails both natively (-m64) and biarch (-m32).
No upstream kernel did this right for 32-bit ptrace callers on x86_64 in the past. We probably won't worry about it for any pre-2.6.25 kernel in Fedora. So this bug should be for rawhide/f9 if those kernels have a problem. F8 update kernels will probably be 2.6.25 eventually after F9 kernels are stable.
Waiting till a new rawhide utrace lands - rawhide ptrace is broken (Comment 0 kernel-2.6.25-0.121.rc5.git4.fc9.x86_64).