Re: [linux-lvm] [PATCH] fix oops when snapshots get full

[Andreas Dilger <adilger@turbolabs.com>]

On Oct 11, 2001  17:03 -0400, Chris Mason wrote:
> > 3) The alternative would be to check lv->lv_block_exception inside
> >    lvm_snapshot_remap_block() instead of in the callers (returning "1"
> >    if it is NULL, and we don't want to do remap).  This would avoid any
> >    problems in the future if someone else calls lvm_snapshot_remap_block()
> >    without checking lv_block_exception first.  Untested patch below which
> >    should be equivalent to your previous patch.
> 
> The only thing I don't like about this is that it makes it hides
> the fact that lvm_snapshot_COW and a few other calls depend on 
> block_exception being valid.  The code certainly looks right though,
> just a matter of style.

Well, lvm_snapshot_COW() is only called from one place, AFAICS, which
is in __remap_snapshot() after we call lvm_snapshot_remap_block(), so
we should be OK (could add a comment about this, and also the fact that
you need to hold the lv_lock to call this function in the first place.

Cheers, Andreas
--
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert