Bug 474148
| Summary: | Segfault when no "-mno-tls-direct-seg-refs" for Fortran/OpenMP test case | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 5 | Reporter: | Michal Nowak <mnowak> |
| Component: | kernel-xen | Assignee: | Xen Maintainance List <xen-maint> |
| Status: | CLOSED DUPLICATE | QA Contact: | Martin Jenner <mjenner> |
| Severity: | medium | Docs Contact: | |
| Priority: | medium | ||
| Version: | 5.3 | CC: | clalance, ohudlick |
| Target Milestone: | rc | Keywords: | Reopened |
| Target Release: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | kernel-xen-2.6.18-126.el5 | Doc Type: | Bug Fix |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2009-07-08 12:32:56 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
Description of problem: From the `man gcc': -mno-tls-direct-seg-refs Controls whether TLS variables may be accessed with offsets from the TLS segment register (%gs for 32-bit, %fs for 64-bit), or whether the thread base pointer must be added. Whether or not this is legal depends on the operating system, and whether it maps the segment to cover the entire TLS area. For systems that use GNU libc, the default is on. Consider this test case: program foo implicit none common /bobcom/ bob(2) !$omp threadprivate (/bobcom/) integer i real*8 bob do i=1,2 write(*,*) i bob(i)=0.0d0 enddo end program Compile it: gfortran -o reproducer -fopenmp reproducer.f90 or gfortran43 -o reproducer -fopenmp reproducer.f90 And run: newman@dhcp-lab-198 445666-OpenMP-segv $ ./reproducer 1 Segmentation fault Plus output from the kernel ring buffer: Dec 2 15:31:11 dhcp-lab-198 kernel: 4gb seg fixup, process reproducer (pid 6069), cs:ip 73:00a0758d Version-Release number of selected component (if applicable): kernel-xen-2.6.18-124.el5 How reproducible: always Expected results: Run slowly but without Segfault. Additional info: newman@dhcp-lab-198 445666-OpenMP-segv $ gfortran -o reproducer -fopenmp reproducer.f90 -mno-tls-direct-seg-refs newman@dhcp-lab-198 445666-OpenMP-segv $ ./reproducer 1 2 From GDB: Run -> Segfault -> bt -> i r ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ gdb) bt #0 0x08048659 in MAIN__ () #1 0x080486b7 in main () gdb) i r eax 0x1 1 ecx 0x0 0 edx 0x0 0 ebx 0xbfd19cf0 -1076781840 esp 0xbfd19cd0 0xbfd19cd0 ebp 0xbfd19e08 0xbfd19e08 esi 0xa13cc0 10566848 edi 0x0 0 eip 0x8048659 0x8048659 <MAIN__+133> eflags 0x210282 [ SF IF RF ID ] cs 0x73 115 ss 0x7b 123 ds 0x7b 123 es 0x7b 123 fs 0x0 0 gs 0x33 51 (gdb) disas $pc-32 $pc+32 Dump of assembler code from 0x8048639 to 0x8048679: 0x08048639 <MAIN__+101>: and $0x8,%al 0x0804863b <MAIN__+103>: add $0x0,%al 0x0804863d <MAIN__+105>: add %al,(%eax) 0x0804863f <MAIN__+107>: lea -0x8(%ebp),%eax 0x08048642 <MAIN__+110>: mov %eax,0x4(%esp) 0x08048646 <MAIN__+114>: mov %ebx,(%esp) 0x08048649 <MAIN__+117>: call 0x80484a8 <_gfortran_transfer_integer@plt> 0x0804864e <MAIN__+122>: mov %ebx,(%esp) 0x08048651 <MAIN__+125>: call 0x8048488 <_gfortran_st_write_done@plt> 0x08048656 <MAIN__+130>: mov -0x8(%ebp),%eax 0x08048659 <MAIN__+133>: movl $0x0,%gs:-0x18(,%eax,8) 0x08048665 <MAIN__+145>: movl $0x0,%gs:-0x14(,%eax,8) 0x08048671 <MAIN__+157>: lea 0x1(%eax),%edx 0x08048674 <MAIN__+160>: mov %edx,-0x8(%ebp) 0x08048677 <MAIN__+163>: cmp $0x2,%eax End of assembler dump.