Mailing List Archive: Hard 6 Meg limit on shared pages

I've been doing some stress testing on how much memory a domain can
share (after a HD and a NIC front end).
It seems that no matter what I set ORDER_GRANT_FRAMES (in xen/include/
asm/grant_table.h) and NR_GRANT_FRAMES
(in linux-2.6-xen/include/xen/gnttab.h) to, attempting to share 6
megabytes or more from any one domain causes really terrible
errors which very quickly corrupt the virtual hard disk (I have tried
allocating up to 32 pages for the grant table and the problems persist).

A sample of the error messages generated by the kernel:
EXT2-fs error (device hda1): ext2_get_inode: unable to read inode
block - inode=346073, block=688192
end_request: I/O error, dev hda1, sector 5737616
end_request: I/O error, dev hda1, sector 5505536
end_request: I/O error, dev hda1, sector 0
EXT2-fs error (device hda1): ext2_get_inode: unable to read inode
block - inode=346073, block=688192
end_request: I/O error, dev hda1, sector 5737616
end_request: I/O error, dev hda1, sector 5505536
end_request: I/O error, dev hda1, sector 0
EXT2-fs error (device hda1): ext2_get_inode: unable to read inode
block - inode=346073, block=688192

(sorry I can't get the first error message generated, it scrolls by
too quickly and my terminals buffer fills up)

I've tried this on xen-3.0-testing.hg (3.0.2-3), and xen-unstable
trees with the same results.

I have not managed to find any reason why this should be happening.
Is this a known issue?
Is there any additional information I can provide that would help
track this down?

Thanks
-aaron

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com
> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of
> James Pendergrass
> Sent: 03 November 2006 17:19
> To: xen-devel@lists.xensource.com
> Subject: [Xen-devel] Hard 6 Meg limit on shared pages
>
> I've been doing some stress testing on how much memory a domain can
> share (after a HD and a NIC front end).
> It seems that no matter what I set ORDER_GRANT_FRAMES (in
> xen/include/
> asm/grant_table.h) and NR_GRANT_FRAMES
> (in linux-2.6-xen/include/xen/gnttab.h) to, attempting to share 6
> megabytes or more from any one domain causes really terrible
> errors which very quickly corrupt the virtual hard disk (I
> have tried
> allocating up to 32 pages for the grant table and the
> problems persist).
>
> A sample of the error messages generated by the kernel:
> EXT2-fs error (device hda1): ext2_get_inode: unable to read inode
> block - inode=346073, block=688192
> end_request: I/O error, dev hda1, sector 5737616
> end_request: I/O error, dev hda1, sector 5505536
> end_request: I/O error, dev hda1, sector 0
> EXT2-fs error (device hda1): ext2_get_inode: unable to read inode
> block - inode=346073, block=688192
> end_request: I/O error, dev hda1, sector 5737616
> end_request: I/O error, dev hda1, sector 5505536
> end_request: I/O error, dev hda1, sector 0
> EXT2-fs error (device hda1): ext2_get_inode: unable to read inode
> block - inode=346073, block=688192
>
> (sorry I can't get the first error message generated, it scrolls by
> too quickly and my terminals buffer fills up)
>
> I've tried this on xen-3.0-testing.hg (3.0.2-3), and xen-unstable
> trees with the same results.
>
> I have not managed to find any reason why this should be happening.
> Is this a known issue?
> Is there any additional information I can provide that would help
> track this down?
>
> Thanks
> -aaron
>
I don't know, but I think it would help if you tried compiling Xen with
"debug=y", to allow any debug code to catch some inconsistency that
occurs...

--
Mats

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel