login
Header Space

 
 

Re: [PATCH 3/3] Consolidate send_sigqueue and send_group_sigqueue

Previous thread: [PATCH] fs/inode.c: use hlist_for_each_entry() by Matthias Kaehlcke on Thursday, February 28, 2008 - 6:11 pm. (1 message)

Next thread: Re: + markers-dont-risk-null-deref-in-marker-checkpatch-fixes.patch added to -mm tree by Jesper Juhl on Thursday, February 28, 2008 - 6:36 pm. (1 message)
To: Roland McGrath <roland@...>
Cc: Oleg Nesterov <oleg@...>, Pavel Emelyanov <xemul@...>, Andrew Morton <akpm@...>, Linux Kernel Mailing List <linux-kernel@...>
Date: Thursday, February 28, 2008 - 6:32 pm

Right. I was working on a fix for that for a different reason. What we
really want is a check for such a timer on a transition from SIGIGN to
another state.  This would allow us to drop the ugly auto rearm code
for ignored timers completely. I had some working proof of concept
code, but got distracted. Need to find it again and dust if off.

Thanks,

	tglx
--
To: Thomas Gleixner <tglx@...>
Cc: Oleg Nesterov <oleg@...>, Pavel Emelyanov <xemul@...>, Andrew Morton <akpm@...>, Linux Kernel Mailing List <linux-kernel@...>
Date: Thursday, February 28, 2008 - 6:41 pm

That is certainly what came first to mind.  (Note that it's not strictly
SIG_IGN alone, but sigaction_says_ignore.)  We'll have to contemplate a lot
of corners to be sure we get it right.  It's not real clear to me how you
go about finding all the affected timers sanely from sigaction.


Thanks,
Roland
--
To: Roland McGrath <roland@...>
Cc: Oleg Nesterov <oleg@...>, Pavel Emelyanov <xemul@...>, Andrew Morton <akpm@...>, Linux Kernel Mailing List <linux-kernel@...>
Date: Friday, February 29, 2008 - 4:04 am

My idea was to dequeue the timer signal and put it into some separate
"check me on unignore" queue. That's the easy part. The nasty part is
to rearm the timer without running into locking madness.

Thanks,

	tglx
--
Previous thread: [PATCH] fs/inode.c: use hlist_for_each_entry() by Matthias Kaehlcke on Thursday, February 28, 2008 - 6:11 pm. (1 message)

Next thread: Re: + markers-dont-risk-null-deref-in-marker-checkpatch-fixes.patch added to -mm tree by Jesper Juhl on Thursday, February 28, 2008 - 6:36 pm. (1 message)
speck-geostationary