Re: [lvm-devel] master - multipathd: fix fd leak when iscsi device logs in

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, 2020-07-13 at 12:13 +0200, Zdenek Kabelac wrote:
> Dne 13. 07. 20 v 12:08 Martin Wilck napsal(a):
> > 
> > Ben has already added support for dm_hold_control_dev() in
> > libmultipath
> > (e24d8b1 ("libmutipath: don't close fd on dm_lib_release")). But
> > this
> > doesn't protect us from calling _open_control() simultaneously in
> > separate code paths, as Lixiaokeng has pointed out.
> > 
> > I don't see how it would, as dm_hold_control_dev() also just
> > changes
> > a static variable.
> 
> Here I'm not familiar with  multipath codebase (more a question for
> Ben 
> likely), but is 'dm_hold...()' being used in non-threaded way - i.e.
> in 
> dmeventd these functions are called  before/after threads are
> finished - so 
> there would be no way 2 threads may call this function in parallel ?

Yes, it's called via pthread_once().

@Lixiaokeng, that would suggest a simpler solution to your fd leak
problem than what I posted before. You'd just need to apply Ben's patch
"libmutipath: don't close fd on dm_lib_release" submitted March
24th.

Contrary to what I said before, this patch is NOT merged yet. The commit id e24d8b1 was from my "upstream-queue" branch only 
(https://github.com/openSUSE/multipath-tools/tree/upstream-queue)

Could you try with that patch (and without mine sent previously)?

Martin


--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/dm-devel




[Index of Archives]     [DM Crypt]     [Fedora Desktop]     [ATA RAID]     [Fedora Marketing]     [Fedora Packaging]     [Fedora SELinux]     [Yosemite Discussion]     [KDE Users]     [Fedora Docs]

  Powered by Linux