[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <Zn1_WtRUW-XeZVoL@casper.infradead.org>
Date: Thu, 27 Jun 2024 16:03:54 +0100
From: Matthew Wilcox <willy@...radead.org>
To: Jakub Kicinski <kuba@...nel.org>
Cc: Paolo Abeni <pabeni@...hat.com>, Sagi Grimberg <sagi@...mberg.me>,
netdev@...r.kernel.org, Eric Dumazet <edumazet@...gle.com>,
David Howells <dhowells@...hat.com>
Subject: Re: [PATCH v2] net: allow skb_datagram_iter to be called from any
context
On Tue, Jun 25, 2024 at 07:10:28AM -0700, Jakub Kicinski wrote:
> On Tue, 25 Jun 2024 15:27:41 +0200 Paolo Abeni wrote:
> > On Sun, 2024-06-23 at 11:12 +0300, Sagi Grimberg wrote:
> > > We only use the mapping in a single context, so kmap_local is sufficient
> > > and cheaper. Make sure to use skb_frag_foreach_page as skb frags may
> > > contain highmem compound pages and we need to map page by page.
> > >
> > > Signed-off-by: Sagi Grimberg <sagi@...mberg.me>
> >
> > V1 is already applied to net-next, you need to either send a revert
> > first or share an incremental patch (that would be a fix, and will need
> > a fixes tag).
> >
> > On next revision, please include the target tree in the subj prefix.
>
> I think the bug exists in net (it just requires an arch with HIGHMEM
> to be hit). So send the fix based on net/master and we'll deal with
> the net-next conflict? Or you can send a revert for net-next at the
> same time, but I think the fix should target net.
It does not require an arch with highmem. I was quite clear about this
in my earlier email.
Powered by blists - more mailing lists