Re: Re: triggering do_writepages

Re: Re: triggering do_writepages

[Kapish K]

Hello,
   Thanks for your inputs. I do understand
that wrietpage is triggered via the swap or
bflush paths of syncing or wrting back to
disk, from looking at the code. No issues
there. But, the thing I observed, for my
situation, where I want to insert some stub
in do-writepages path and want to test its
triggering, I do not find it getting
triggered.
The situation being that I want to invoke
some code in do_writepages path for a
particular filesystem ( identified by the
super_block, say ). In that case, when I
issue writes on the filesystem, I do see the
commit/prepare write path activated ASAP as
it it should be. But on a explicit sync on
the filesystem, I don't see writepages for
that filesystem activated. And there lies
the confusion. 
Again, have I missed something obvious out
here?
Thanks, in advance
 

---- On Tue, 19 Nov 2002, Bryan Henderson
(hbryan@us.ibm.com) wrote:

> 
> writepage also runs when the system
synchronizes a file's cache in any of
> numerous ways (fsync() system call, sync()
system call, filesystem driver
> calls f_datasync() kernel service, ...).
> 
> 
> 
> 
> 


________________________________________________
Get your own "800" number
Voicemail, fax, email, and a lot more
http://www.ureach.com/reg/tag

Re: Re: triggering do_writepages

[girish]

> The situation being that I want to invoke
> some code in do_writepages path for a
> particular filesystem ( identified by the
> super_block, say ). In that case, when I
> issue writes on the filesystem, I do see the
> commit/prepare write path activated ASAP as
> it it should be. But on a explicit sync on
> the filesystem, I don't see writepages for
> that filesystem activated.
sys_sync->sync_fdev->sync_inodes>__sync_one->filemap_fdatasync->writepage if
"drty pages are avilable"