[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20250902053608.GA11396@lst.de>
Date: Tue, 2 Sep 2025 07:36:08 +0200
From: Christoph Hellwig <hch@....de>
To: Keith Busch <kbusch@...nel.org>
Cc: Christoph Hellwig <hch@....de>, Christoph Hellwig <hch@...radead.org>,
Keith Busch <kbusch@...a.com>, linux-block@...r.kernel.org,
linux-nvme@...ts.infradead.org, axboe@...nel.dk,
iommu@...ts.linux.dev, willy@...radead.org, netdev@...r.kernel.org,
linux-mm@...ck.org
Subject: Re: [PATCHv3 1/2] block: accumulate segment page gaps per bio
On Fri, Aug 29, 2025 at 07:47:40PM -0600, Keith Busch wrote:
> On Wed, Aug 27, 2025 at 09:37:09AM +0200, Christoph Hellwig wrote:
> > On Tue, Aug 26, 2025 at 04:33:15PM -0600, Keith Busch wrote:
> > > virt boundary check. It's looking like replace bvec's "page + offset"
> > > with phys addrs, yeah?!
> >
> > Basically everything should be using physical address. The page + offset
> > is just a weird and inefficient way to represent that and we really
> > need to get rid of it.
>
> I was plowing ahead with converting to phys addrs only to discover
> skb_frag_t overlays a bvec with tightly coupled expectations on its
> layout. I'm not comfortable right now messing with that type. I think it
> may need to be decoupled to proceed on this path. :(
willy got really angry at this for all the right reasons, and we
still need it fixed. Can we just revert the unreviewed crap the
network folks did here to move the kernel forward?
Powered by blists - more mailing lists