[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ext4
Subject:    Re: [Ext2-devel] [Fwd: Re: raid5d, page_launder and scheduling latency]
From:       Andrew Morton <andrewm () uow ! edu ! au>
Date:       2001-07-17 5:59:32
[Download RAW message or body]

Andrew Morton wrote:
> 
> Hi, Neil.
> 
> Seems we're still not out of the woods with RAID5 and ext3.  Mike's
> testing shows that the system still gets congested and disk throughput
> falls away as the number of tiobench threads is increased.
> 

Mike,  I just found a bug which could conceivably cause large
stalls within ext3.  There's a fix in CVS HEAD now.  The
relevant chunk is:


diff -u -r1.21 checkpoint.c
--- fs/jbd/checkpoint.c	2001/07/13 14:11:19	1.21
+++ fs/jbd/checkpoint.c	2001/07/17 05:38:42
@@ -136,9 +136,13 @@
 		}
 		
 		if (jh->b_transaction != NULL) {
+			transaction_t *transaction = jh->b_transaction;
+			tid_t tid = transaction->t_tid;
+
 			spin_unlock(&journal_datalist_lock);
+			log_start_commit(journal, transaction);
 			unlock_journal(journal);
-			log_wait_commit(journal, jh->b_transaction->t_tid);
+			log_wait_commit(journal, tid);
 			goto out_return_1;
 		}
 

Doubtful if this is the problem, but...

_______________________________________________
Ext2-devel mailing list
Ext2-devel@lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/ext2-devel

[prev in list] [next in list] [prev in thread] [next in thread]