http://bugzilla.novell.com/show_bug.cgi?id=573478 http://bugzilla.novell.com/show_bug.cgi?id=573478#c0 Summary: Floating point register state corruption after handling SIGSEGV Classification: openSUSE Product: openSUSE 11.0 Version: Final Platform: x86-64 OS/Version: SLES 10 Status: NEW Severity: Critical Priority: P5 - None Component: Kernel AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: henrik.osterdahl@oracle.com QAContact: qa@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.9.1.2) Gecko/20090729 Firefox/3.5.2 (.NET CLR 3.5.30729) Earlier I found a problem on Red Hat Enterprise Linux and Oracle Enterprise Linux with SSE2 register state being corrupted on context switch. It turns out that it was a general Linux kernel problem. The Red Hat bugzilla entry is here: https://bugzilla.redhat.com/show_bug.cgi?id=547893 In the RH bugzilla entry are three attachments. 1. Reproducer that consistently fails on all x86_64 kernels I've tried it on. 2. Patch with fix for RHEL4u8 kernel. 3. Patch with fix for RHEL5u4 kernel. My run data for the xmmtest4.c reproducer: SLES 9: hosterda@sthx6404:~/src$ cat /etc/SuSE-release SUSE LINUX Enterprise Server 9 (x86_64) VERSION = 9 PATCHLEVEL = 3 hosterda@sthx6404:~/src$ ./xmmtest4 mapped memory: 0x2a9556c000 +++ creating threads !!! incorrect xmm0 value after trap: 0.000000 !!! incorrect xmm0 value after trap: 0.000000 !!! incorrect xmm0 value after trap: 0.000000 --- joined threads uname result: SYSNAME=Linux NODENAME=sthx6404 RELEASE=2.6.5-7.244-smp VERSION=#1 SMP Mon Dec 12 18:32:25 UTC 2005 MACHINE=x86_64 SLES 10: hosterda@sthx6442:~/src$ cat /etc/SuSE-release SUSE Linux Enterprise Server 10 (x86_64) VERSION = 10 PATCHLEVEL = 1 hosterda@sthx6442:~/src$ ./xmmtest4 mapped memory: 0x2ac4a240b000 +++ creating threads --- joined threads (...repeating message...) +++ creating threads --- joined threads +++ creating threads !!! incorrect xmm0 value after trap: 0.000000 --- joined threads uname result: SYSNAME=Linux NODENAME=sthx6442 RELEASE=2.6.16.46-0.12-smp VERSION=#1 SMP Thu May 17 14:00:09 UTC 2007 MACHINE=x86_64 SLES 11: hosterda@sthsles03:~/src$ cat /etc/SuSE-release SUSE Linux Enterprise Server 11 (x86_64) VERSION = 11 PATCHLEVEL = 0 hosterda@sthsles03:~/src$ ./xmmtest4 mapped memory: 0x7f270d761000 +++ creating threads --- joined threads (...repeating message...) +++ creating threads --- joined threads +++ creating threads !!! incorrect xmm0 value after trap: 0.000000 --- joined threads uname result: SYSNAME=Linux NODENAME=sthsles03 RELEASE=2.6.27.19-5-default VERSION=#1 SMP 2009-02-28 04:40:21 +0100 MACHINE=x86_64 As far as I understand, it is probably a simple matter to patch the SuSE Linux kernel with the same or similar patch. Reproducible: Always Steps to Reproduce: 1. Compile repro: "gcc lpthread xmmtest4.c -o xmmtest4" 2. Run repro on SLES 9/10/11 on x86_64. Actual Results: Repro terminates (after some amount of time) with error message as described. It's clearly timing-dependent. Expected Results: Repro should never terminate. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.