Mailing List Archive: Re: [PATCH 2/2] Fix (improve) deadlock condition on module removal netfilter socket option removal

On Wed, 2007-09-05 at 16:26 -0400, Jon Masters wrote:
> On Tue, 2007-09-04 at 16:30 -0400, Neil Horman wrote:
>
> > 2nd of two patches. This patch enhances modprobe to operate like rmmod
> > in non-blocking mode. It also adds a -w option to allow for explicit blocking
> > operation.
>
> As I suspected, this patch isn't in the tree. I am going to commit it
> now because it makes sense. I'm also going to sort out moving things to
> kernel.org this afternoon while I'm at it - I don't want to confuse
> people with kerneltools.org any more now I've got a kernel.org acc.

1) You don't want to hand the "wait" flag (ie ~O_NONBLOCK) to
sub-rmmods,

2) You need to do something about this code if wait is specified:

if (usecount != 0) {
if (!ignore_inuse)
error("Module %s is in use.\n", name);
goto remove_rest;
}

Cheers,
Rusty.

On Thu, 2007-09-06 at 08:41 +1000, Rusty Russell wrote:
> On Wed, 2007-09-05 at 16:26 -0400, Jon Masters wrote:
> > On Tue, 2007-09-04 at 16:30 -0400, Neil Horman wrote:
> >
> > > 2nd of two patches. This patch enhances modprobe to operate like rmmod
> > > in non-blocking mode. It also adds a -w option to allow for explicit blocking
> > > operation.
> >
> > As I suspected, this patch isn't in the tree. I am going to commit it
> > now because it makes sense. I'm also going to sort out moving things to
> > kernel.org this afternoon while I'm at it - I don't want to confuse
> > people with kerneltools.org any more now I've got a kernel.org acc.
>
> 1) You don't want to hand the "wait" flag (ie ~O_NONBLOCK) to
> sub-rmmods,
>
> 2) You need to do something about this code if wait is specified:
>
> if (usecount != 0) {
> if (!ignore_inuse)
> error("Module %s is in use.\n", name);
> goto remove_rest;
> }

Goodness, I suck. I'll get it fixed properly.

Jon.