On Mon, Apr 20, 2015 at 01:08:16PM -0600, Andreas Dilger wrote: > On Apr 20, 2015, at 12:12 PM, Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote: > > > > From: Al Viro <viro@xxxxxxxxxxxxxxxxxx> > > > > Signed-off-by: Al Viro <viro@xxxxxxxxxxxxxxxxxx> > > Al, the patch itself looks good, thanks. > > However, if this is applied at the start of the series it could > allow tests to easily cause a stack overflow during a bisection (I > don't think users would see a kernel in the middle of the series). > > Could this be converted over to checking nd->link_count along with > the [02/24] patch until closer to the end of the series when the > recursion has been removed? Er... You do realize that struct nameidata is opaque for anything outside of fs/namei.c and has been that way for a while now? Sure, we can export a helper that would return that and rip it out in the end of the series, but... > It isn't fatal if that doesn't happen, since this whole series should > land at one time and the chance of testing Lustre symlinks right > in the middle of the series is low, just something I thought when > reviewing the patch. -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
