Messages in this thread | | | Date | Sat, 08 May 1999 15:23:05 -0400 | From | Manfred Spraul <> | Subject | Re: Oops assist... [New idea] |
| |
Alexander Viro wrote: > "Fast" reboot is fast only because Windows doesn't do fsck. BTW, > writing to disk after oops may be dangerous or not, but dirty reboot > definitely *is* dangerous. So the whole idea looks pretty bogus.
1) I abandoned my idea because I forgot LOADLIN, and because it would be i386 only. 2) "Fast" reboot means that we could modify sys_reboot(): instead of forcing a cold boot, we could reload lilo. This had nothing to do with oops - these where 2 independant ideas. Sorry, this was just a wild idea, I should have thought about it a bit longer.
But I do think that as soon as add more semaphores to the filesystem and if we implement journaling for ext[2,3], then we must also modify the oops-logging implementation. a) restart immediately (at least a compile option) transaction tracked systems have no problems with crashes, but they can't handle unreliable execution. b) store the oops report without touching VFS.
Regards, Manfred
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |