Re: msync(2) bug(?), returns AOP_WRITEPAGE_ACTIVATE to userland

[Erez Zadok]

In message <Pine.LNX.4.64.0710222101420.23513@xxxxxxxxxxxxxxxxxxxxxx>, Hugh Dickins writes:
> On Mon, 15 Oct 2007, Pekka Enberg wrote:
> > 
> > I wonder whether _not setting_ BDI_CAP_NO_WRITEBACK implies that
> > ->writepage() will never return AOP_WRITEPAGE_ACTIVATE for
> > !wbc->for_reclaim case which would explain why we haven't hit this bug
> > before. Hugh, Andrew?
> 
> Only ramdisk and shmem have been returning AOP_WRITEPAGE_ACTIVATE.
> Both of those set BDI_CAP_NO_WRITEBACK.  ramdisk never returned it
> if !wbc->for_reclaim.  I contend that shmem shouldn't either: it's
> a special code to get the LRU rotation right, not useful elsewhere.
> Though Documentation/filesystems/vfs.txt does imply wider use.

Yes, based on vfs.txt I figured unionfs should return
AOP_WRITEPAGE_ACTIVATE.  But, now that unionfs has ->writepages which won't
even call the lower writepage if BDI_CAP_NO_WRITEBACK is on, then perhaps I
no longer need unionfs_writepage to bother checking for
AOP_WRITEPAGE_ACTIVATE, or even return it up?

But, a future file system _could_ return AOP_WRITEPAGE_ACTIVATE w/o setting
BDI_CAP_NO_WRITEBACK, right?  In that case, unionfs will still need to
handle AOP_WRITEPAGE_ACTIVATE in ->writepage, right?

> I think this is where people use the phrase "go figure" ;)
> 
> Hugh

Erez.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/