[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAGXJAmx4bB15iT1OwNyTmEDYMLvDhCVUXQBhKCftAeBt932uUw@mail.gmail.com>
Date: Tue, 7 Jan 2025 13:51:41 -0800
From: John Ousterhout <ouster@...stanford.edu>
To: Eric Dumazet <edumazet@...gle.com>
Cc: Jakub Kicinski <kuba@...nel.org>, netdev@...r.kernel.org, pabeni@...hat.com,
horms@...nel.org
Subject: Re: [PATCH net-next v5 06/12] net: homa: create homa_peer.h and homa_peer.c
On Tue, Jan 7, 2025 at 1:02 PM Eric Dumazet <edumazet@...gle.com> wrote:
>
> On Tue, Jan 7, 2025 at 9:54 PM John Ousterhout <ouster@...stanford.edu> wrote:
> >
> > I have removed the cast now.
> >
> > -John-
> >
> >
> > On Tue, Jan 7, 2025 at 6:15 AM Jakub Kicinski <kuba@...nel.org> wrote:
> > >
> > > On Mon, 6 Jan 2025 10:12:12 -0800 John Ousterhout wrote:
> > > > +void homa_dst_refresh(struct homa_peertab *peertab, struct homa_peer *peer,
> > > > + struct homa_sock *hsk)
> > > > +{
> > > > + struct dst_entry *dst;
> > > > +
> > > > + spin_lock_bh(&peertab->write_lock);
> > > > + dst = homa_peer_get_dst(peer, &hsk->inet);
> > > > + if (!IS_ERR(dst)) {
> > > > + struct homa_dead_dst *dead = (struct homa_dead_dst *)
> > > > + kmalloc(sizeof(*dead), GFP_KERNEL);
> > >
>
> While you are at it, I suggest you test your patch with LOCKDEP enabled,
> and CONFIG_DEBUG_ATOMIC_SLEEP=y
>
> Using GFP_KERNEL while BH are blocked is not good.
I will follow up on all of this. Thanks for the pointers, and sorry
that there is so much I don't know.
-John-
Powered by blists - more mailing lists