Re: 2.6.0-test2-bk3 phantom I/O errors

[Andrew Morton <akpm@osdl.org>]

Sander van Malssen <svm@kozmix.org> wrote:
>
> If I try this on 2.6.0-test2-mm1 with the dump_stack() added (ext3, no
>  initrd) I get the following:
> 
> 
>  Buffer I/O error on device hda1, logical block 25361
>  Call Trace:
>   [<c0150f02>] buffer_io_error+0x42/0x50

OK, looks like the new readahead stuff confused the error reporting.

Does this make the error messages go away?


diff -puN mm/readahead.c~a mm/readahead.c
--- 25/mm/readahead.c~a	2003-07-30 11:58:07.000000000 -0700
+++ 25-akpm/mm/readahead.c	2003-07-30 11:58:20.000000000 -0700
@@ -96,7 +96,7 @@ static int read_pages(struct address_spa
 	struct pagevec lru_pvec;
 	int ret = 0;
 
-	current->flags |= PF_READAHEAD;
+//	current->flags |= PF_READAHEAD;
 
 	if (mapping->a_ops->readpages) {
 		ret = mapping->a_ops->readpages(filp, mapping, pages, nr_pages);

_



Tell me how hard this is to hit.  Does it only happen when there is a large
amount of IO happening?  What is the system doing at the time?