Messages in this thread | | | Date | Sun, 21 Mar 2004 15:21:51 -0800 (PST) | From | Linus Torvalds <> | Subject | Re: Linux 2.6.5-rc1 |
| |
On Sun, 21 Mar 2004, Peter Osterlund wrote: > > For some reason I don't understand, this makes the UDF filesystem lock > up when I write a bunch of mp3 files to a CDRW using the packet > writing patch. Both "cp" and pdflush get stuck in __down. Reverting > the semaphore changes as in the patch below makes the problem go away, > but it's probably not the right solution to re-introduce lock_kernel() > calls.
Looks correct. It looks like memory pressure while doign write() on the semaphore will try to writeback dirty inodes and data, and the kernel lock allowed that fine, but using the inode semaphore will just deadlock, since the write() already holds the semaphore.
I think the prealloc stuff could probably be protected by the inode spinlock instead, but for now your backout patch seems to be the same thing to do.
Linus - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |