Re: [GIT PULL] notification tree - try 37!

[Andreas Gruenbacher <agruen@suse.de>]

On Monday 23 August 2010 18:46:13 Eric Paris wrote:
> Spent a bit of the weekend trying to figure out what you were doing and
> couldn't reproduce it or find it in the code because I had already fixed
> it (albeit for slightly different reasons).  The patch in question was:
> 
> http://marc.info/?l=linux-kernel&m=128214903125780&w=2
> 
> the vfsmount_test_mask was always initialized but since no vfsmount
> marks were found it was never cleared.  This left the code thinking that
> the given (inode) mark was interested in the event.

That seems to explain it.  I can no longer reproduce the bug with your current 
for-linus tree (3ba67c8a) using my previous method.

Sorry this bug was difficult for you to track down; I was not even aware of 
this fix until today.

> I think you could reproduce it differently
> 
> inotifywait -m -e open /mnt/tmp
> inotifywait -m -e close /mnt/tmp
> 
> and you would get both event types for both watches.

No, the open listener gets:
	d/ OPEN,ISDIR

and the close listener gets:
	d/ CLOSE_NOWRITE,CLOSE,ISDIR

I tried to reproduce the previously failing case with inotifywait. I can see 
that inotifywait receives five inotify events at the syscall level, but it 
will not report events with mask == 0.

Thanks,
Andreas