OK, so one the one hand, we have Rik's latest patch:
>From: Rik van Riel <riel@conectiva.com.br>
>Date: Wed, 26 Apr 2000 10:36:10 -0300 (BRST)
>Subject: [patch] 2.3.99-pre6-3 VM fixed
>
>Hi,
>
>The attached patch should fix most of the VM performance problems
>2.3 was having. It does the following things [ ... ]
and on the other hand, we have this simple patch from Juan:
>From: "Juan J. Quintela" <quintela@fi.udc.es>
>Date: 26 Apr 2000 16:01:24 +0200
>Subject: Re: System hangs when reading or writing many or large files
>
>Hi
>
>>>> EIP; c0138799 <__free_pages_ok+49/324> <=====
>ralf> Trace; c021a8c8 <tvecs+2e40/14ef8>
>ralf> Trace; c021ab55 <tvecs+30cd/14ef8>
>ralf> Trace; c012dc2b <shrink_mmap+457/5c8>
>ralf> Trace; c013823b <do_try_to_free_pages+33/98>
>ralf> Trace; c0138333 <kswapd+93/1b0>
>ralf> Trace; c010907b <kernel_thread+23/30>
>
> I get here similar Oops. They happened after a BUG in
> page_alloc.c. Is that your case also? I am very interested in
> knowing the answer. Are you using a pre6-x, x<6? In the case
> that your Oops are very easily reproducible, could you try this
> small patch and tell me if it solves the problem. It solves the
> problem here, (yes I know that it is not the proper solution,
> but I am interested in knowing if it solves your problems also).
>
>Thanks, Juan.
>
>diff -u -urN --exclude=CVS --exclude=*~ --exclude=.#* 2.3.99-pre6-5/mm/filemap
>.c 2.3.99-pre6-boot/mm/filemap.c
>- --- 2.3.99-pre6-5/mm/filemap.c Tue Apr 4 01:18:56 2000
>+++ 2.3.99-pre6-boot/mm/filemap.c Sun Apr 23 17:48:42 2000
>@@ -467,8 +467,9 @@
> struct page *alias;
> unsigned long flags;
>
>+ LockPage(page);
> flags = page->flags & ~((1 << PG_uptodate) | (1 << PG_error) | (1 << PG
>_dirty));
>- - page->flags = flags | (1 << PG_locked) | (1 << PG_referenced);
>+ page->flags = flags | (1 << PG_referenced);
> get_page(page);
> page->index = offset;
> add_page_to_inode_queue(mapping, page);
I have not tried either one, and I'd like to see if anyone has
opinions on the relationship between these two patches. The memory
problems I have with 2.3.99pre5 are fairly crippling, but I'd like to
use the smallest possible patch. If I don't hear from anyone, I
suppose I'll just try them both out.
--p
-
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/
This archive was generated by hypermail 2b29 : Sun Apr 30 2000 - 21:00:11 EST