[Top] [All Lists]

Re: xfs + rmap?

To: Adam McKenna <adam-dated-1013032432.ea4245@xxxxxxxxxxxx>
Subject: Re: xfs + rmap?
From: Steve Lord <lord@xxxxxxx>
Date: 01 Feb 2002 16:05:36 -0600
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <20020201215351.GF23997@xxxxxxxxxxxx>
References: <20020201215351.GF23997@xxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
On Fri, 2002-02-01 at 15:53, Adam McKenna wrote:
> I'm trying to merge Rik van Riel's rmap-12a patch with current CVS, I'm
> having a couple of problems..  I was wondering if someone could give me a
> clue how to work these out (or tell me if they can't be)
> There is a huge hunk of mm/vmscan.c that looks to be totally incompatible,
> the .rej is about 18 kilobytes.
> --Adam

Well, this is the only change we have in the 2.4.17 vmscan.c

--- linux-2.4.17/mm/vmscan.c    Fri Dec 21 11:42:05 2001
+++ xfs-linux.2.4/linux/mm/vmscan.c     Wed Jan 30 15:49:30 2002
@@ -391,7 +391,7 @@
-               if (PageDirty(page) && is_page_cache_freeable(page) && 
page->mapping) {
+               if ((PageDirty(page) || DelallocPage(page)) && 
is_page_cache_freeable(page) && page->mapping) {
                         * It is not critical here to write it only if
                         * the page is unmapped beause any direct writer

This is around a call to writepage. There is probably something similar
in the rmap version of the kernel. So look for where it deals with
dirty pages by calling writepage on them and check for
DelallocPage in the same spot. 



Steve Lord                                      voice: +1-651-683-3511
Principal Engineer, Filesystem Software         email: lord@xxxxxxx

<Prev in Thread] Current Thread [Next in Thread>