> I think long term (2.5) the right way is to replace all the iput in the
> slow fail paths with a iput_not_mounted, that will avoid both the iput
> clobbering and the MS_ACTIVE tracking. The differentiation should be
Egads... Andrea, think for a bloody minute. What's the point of adding
a new helper that would share a lot of code with the iput() _and_ would
bring additional calling rules? We get
* more code in inode.c
* more code duplications
* a lot of opportunities to fsck up in fs code
* redundant invalidate_inodes() calls in fs/super.c existing just
to catch these fsckups.
* some filesystems getting out with only iput(), some needing new
helper.
* cut'n'paste programming getting one more source of bugs to
introduce.
And it's not even the case when filesystems could use that distinction in
any sane way...
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/