On Fri, Oct 04, 2019 at 04:33:02PM +0200, Christian Brauner wrote:
On Fri, Oct 04, 2019 at 03:27:48PM +0100, Al Viro wrote:
On Fri, Oct 04, 2019 at 04:05:03PM +0200, Christian Brauner wrote:
From: Will Deacon will@kernel.org
Closing /dev/pts/ptmx removes the corresponding pty under /dev/pts/ without synchronizing against concurrent path walkers. This can lead to 'dcache_readdir()' tripping over a 'struct dentry' with a NULL 'd_inode' field:
FWIW, vfs.git#fixes (or #next.dcache) ought to deal with that one.
Is it feasible to backport your changes? Or do we want to merge the one here first and backport?
I'm not sure. The whole pile is backportable, all right (and the first commit alone should take care of devpts problem). However, there's a performance regression on some loads; it *is* possible to get the thing reasonably lockless without fucking it up (as the original conversion had been). Still not in the series, since cifs (ab)use of dcache_readdir() needs to be clarified to figure out the right way to do it. Asked CIFS folks, got no reaction whatsoever, will ask again...
Al, mostly back after flu, digging through the piles of mail