linux-kernel - Re: [PATCH v2 2/8] IB/core: Replace semaphore sm

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20161025154822.GA27240@obsidianresearch.com>
Date:   Tue, 25 Oct 2016 09:48:22 -0600
From:   Jason Gunthorpe <jgunthorpe@...idianresearch.com>
To:     Binoy Jayan <binoy.jayan@...aro.org>
Cc:     Jack Wang <xjtuwjp@...il.com>, Doug Ledford <dledford@...hat.com>,
        Sean Hefty <sean.hefty@...el.com>,
        Hal Rosenstock <hal.rosenstock@...il.com>,
        Arnd Bergmann <arnd@...db.de>,
        "linux-rdma@...r.kernel.org" <linux-rdma@...r.kernel.org>,
        Linux kernel mailing list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2 2/8] IB/core: Replace semaphore sm_sem with completion

On Tue, Oct 25, 2016 at 08:38:21PM +0530, Binoy Jayan wrote:
> On 25 October 2016 at 18:13, Jack Wang <xjtuwjp@...il.com> wrote:
> > Hi Binoy,
> >
> > snip
> >>
> >>         port->ib_dev   = device;
> >>         port->port_num = port_num;
> >> -       sema_init(&port->sm_sem, 1);
> >> +       init_completion(&port->sm_comp);
> >> +       complete(&port->sm_comp);
> >
> > Why complete here?
> >
> >>         mutex_init(&port->file_mutex);
> >>         INIT_LIST_HEAD(&port->file_list);
> >>
> > KR,
> > Jinpu
> 
> 
> Hi Jack,
> 
> ib_umad_sm_open() calls wait_for_completion_interruptible() which
> comes before ib_umad_sm_close() that calls complete(). In the
> initial open() there will not be anybody to signal the completion,
> so the complete is called to mark the initial state.  I am not sure
> if this is the right way to do it, though.

Using a completion to model exclusive ownership seems convoluted to
me - is that a thing now? What about an atomic?

open:

while (true) {
   wait_event_interruptible(priv->queue,test_bit(CLAIMED_BIT, &priv->flags));
   if (!test_and_set_bit(CLAIMED_BIT, &priv->flags))
        break;
}

close():

   clear_bit(CLAIMED_BIT, &priv->flags)
   wake_up(&priv->queue);

??

Jason