I discover the following:
LVM 1.0.7 patch:
modifies file fs/buffer.c:
add function int fsync_dev_lockfs(kdev_t dev)
the function use macro DQUOT_SYNC(dev);
XFS 1.2 patch:
modifies file fs/buffer.c:
modifies function int fsync_super(struct super_block *sb)
from DQUOT_SYNC(dev) --> to DQUOT_SYNC_SB(sb);
modifies function int fsync_dev(kdev_t dev)
from DQUOT_SYNC(dev) --> DQUOT_SYNC_DEV(dev);
unmodified function int fsync_dev_lockfs(kdev_t dev) (becouse is not
know to xfs patch)
still use DQUOT_SYNC(dev);
I've tried to change file fs/buffer.c:
from DQUOT_SYNC(dev) --> to DQUOT_SYNC_DEV(dev);
I used *_DEV becouse it is called on kdev_t variable (but really I do
not known nothing about these function)
The compile was successfull, but is there anybody that can test it work?
I'll wait for comments...
Bye Roberto
> Hi Nathan
>
> Thank you for the respone! .... :)
>
> Did you mean that I've to change the LVM patch?
> it sound a big trouble to me (beginner)... but I'll try.... : P
>
> I've also found the new 2.4.20 patch. (Mar. 18)
> is that the release stable patch file?
>
> Have a nice day ^_^
>
> Rocky
>
> ----- Original Message -----
> From: "Nathan Scott" <nathans@xxxxxxx>
> To: "Rocky Lee" <rocky_lee@xxxxxxxxxxxxxx>
> Cc: <linux-xfs@xxxxxxxxxxx>
> Sent: Thursday, March 27, 2003 2:18 PM
> Subject: Re: kernel patch fail with LVM
>>
>> We have Jan Kara's 32 bit quota patches in our tree, because
>> we make use of some quota extensions they provide. The LVM
>> patch most likely is just for a stock-standard 2.4.20, which
>> does not have these patches.
>>
>> So, Jan's patches contain a change which removes DQUOT_SYNC
>> and instead splits the functionality into a DQUOT_SYNC_DEV
>> and a DQUOT_SYNC_SB - you'll need to change the code (looks
>> like fsync_dev_lockfs needs this change) to make use of one
>> of these new macros.
>>
>> cheers.
>>
>> --
>> Nathan
>>
>>
buffer-orig-2419.c.gz
Description: Binary data
buffer-xfs12-lvm107.c.gz
Description: Binary data
buffer-xfs12-lvm107-fixed.c.gz
Description: Binary data
|