3 systems in a Openmosix cluster all exhibit the same behavior. All randomly segfault during emerges. Settings in make.conf were altered but did not seem to effect. Each would randomly segfault at different stages while doing various emerges. Rebooting each with a non OpenMosix Kernel would solve the segfaulting problem. Kernel version used was 2.4.20-openmosix-r2. System 1 dual 350 with smp kernel releas 1.4_rc2 System 2 single 650 1.4_rc2 System 3 Singel 1G Celron 1.4_rc3 The Dual system did seem to exhibit the segfaults more readily than the others.
This problem is known to the openmosix developers and has been fixed in their cvs. As soon as a new version will be released this problem will be fixed.
I also have the same problem, thus emerging mozilla also segfault, don't know if it's related, but I'll give it a try and then post back.
I am now using openmosix cvs, and I still get the same problem emerging mozilla. The segfault happens on the same place everytime, but emerging it with a vanilla 2.4.20 kernel works out fine.
Just for making clear, that it's openmosix: Stop openmosix by executing /etc/init.d/openmosix stop and try to emerge then, if it's openmosix, it should not segfault.
Yep, it did not segfault, perhaps it's an issue with Mozilla's ebuild that should strip down make -j?
Please try with openmosix-sources-2.4.20-r3 and report if the problem still exists.
I have confirmed it still behaves the same way with -r3. Other processes randomly segfault as well; but for whatever reason, it seems like emerges exploit this bug a lot more reliably that anything else. What was the ChangeLog entry for -r3 anyway? I didn't see anything new in the ChangLog in my portage tree. Cheers, Jordan. - - - 8-< *snip* - - - carbonara root # uname -a Linux carbonara.cnt.com 2.4.20-openmosix-r3 #1 SMP Wed Apr 16 10:16:48 CDT 2003 i686 Pentium III (Coppermine) GenuineIntel GNU/Linux carbonara root # emerge sync Segmentation fault carbonara root # /etc/init.d/openmosix stop * Stopping openMosix... * openMosix: All remote processes were expelled and no further remote processes accep [ ok ] carbonara root # emerge sync >>> starting rsync with rsync://trogdor/gentoo-portage... This is TROGDOR! receiving file list ... 40764 files to consider rsync[4144] (receiver) heap statistics: arena: 5902840 (bytes from sbrk) ordblks: 606 (chunks not in use) smblks: 2 hblks: 1 (chunks from mmap) hblkhd: 258048 (bytes from mmap) usmblks: 0 fsmblks: 40 uordblks: 5139232 (bytes used) fordblks: 763608 (bytes free) keepcost: 112920 (bytes in releasable chunk) Number of files: 40764 Number of files transferred: 0 Total file size: 33477788 bytes Total transferred file size: 0 bytes Literal data: 0 bytes Matched data: 0 bytes File list size: 965234 Total bytes written: 153 Total bytes read: 965288 wrote 153 bytes read 965288 bytes 39405.76 bytes/sec total size is 33477788 speedup is 34.68 >>> Updating Portage cache... ...done! carbonara root #
The problem is that you are using different machines with different capabilities (as seen in cpuinfo: mmx, sse etc...) When a proccess is migrated from a "fast" machine to a "slow" one, than it can happen that code compiled with e.g. sse2 optimizations is run on a processor without e.g. sse2. It segfaults then. This issue is known to the openmosix-developers and will be fixed in the future.
This is happening to me too. I can't even run emerge. It segfaults almost immediately. Cant even get the list of commands. Stop openmosix or run emerge with mosrun -L and it runs correctly. I'm re-emerging gcc+glibc+python to see if it solves it.
Forgot to specify kernel version. I'm running openmosix-sources-2.4.22-r2
