Re: [SSI-devel] cfs_writepages(): Is it a bug?
Brought to you by:
brucewalker,
rogertsang
From: Roger T. <rog...@gm...> - 2006-07-04 15:36:51
|
I believe I fixed your problem with CFS over dm with my latest CVS checkin. Roger On 6/21/06, Vladimir Razgulin <one...@gm...> wrote: > All, > Here's what I've found in the cfs code, and believe might be an issue: > > 1. cfs_writepages() function uses test_and_set_bit() and clear_bit() > to manipulate BDI_write_congested bit in &bdi->state, and calls > wake_up_all() to wake up cfs_write_on_congestion(). > > 2. cfs_write_on_congestion() uses a different way to access > BDI_write_congested bit - it calls bdi_write_congested() function. > > 3. bdi_write_congested() checks bdi->congested_fn, and call the > corresponding function if it's not NULL. In that case it just ignore > bdi->state field that was set by cfs_writepages() > > 4. bdi->congested_fn is not NULL, if cfs sits on toop of an ext3 fs > residing on a device-mapped device (it points to dm_any_congested() in > that case) > > The same problem seems to exist for NFS as well. > > Is cfs able to work with dm-ed devices? > > Thanks > Vladimir > > All the advantages of Linux Managed Hosting--Without the Cost and Risk! > Fully trained technicians. The highest number of Red Hat certifications in > the hosting industry. Fanatical Support. Click to learn more > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=107521&bid=248729&dat=121642 > _______________________________________________ > ssic-linux-devel mailing list > ssi...@li... > https://lists.sourceforge.net/lists/listinfo/ssic-linux-devel > |