linux-kernel - Re: [PATCH 14/15] nvme: enable FDP support

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <Zz0V6qv3oyKV_Kzp@kbusch-mbp.dhcp.thefacebook.com>
Date: Tue, 19 Nov 2024 15:49:14 -0700
From: Keith Busch <kbusch@...nel.org>
To: Christoph Hellwig <hch@....de>
Cc: Jens Axboe <axboe@...nel.dk>, Christian Brauner <brauner@...nel.org>,
	Sagi Grimberg <sagi@...mberg.me>,
	Kanchan Joshi <joshi.k@...sung.com>, Hui Qi <hui81.qi@...sung.com>,
	Nitesh Shetty <nj.shetty@...sung.com>, Jan Kara <jack@...e.cz>,
	Pavel Begunkov <asml.silence@...il.com>,
	linux-block@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-nvme@...ts.infradead.org, linux-fsdevel@...r.kernel.org,
	io-uring@...r.kernel.org
Subject: Re: [PATCH 14/15] nvme: enable FDP support

On Tue, Nov 19, 2024 at 07:24:27PM +0100, Christoph Hellwig wrote:
> On Tue, Nov 19, 2024 at 11:17:36AM -0700, Keith Busch wrote:
> > > +	if (le32_to_cpu(configs[result.fdpcidx].nrg) > 1) {
> > > +		dev_warn(ns->ctrl->device, "FDP NRG > 1 not supported\n");
> > 
> > Why not support multiple reclaim groups?
> 
> Can you come up with a sane API for that? 

Haven't really thought about it. If it's there, it's probably useful for
RU's that are not "Persistently Isolated". But let's not worry about it
now, we can just say you don't get to use write streams for these.

> And can you find devices in
> the wild that actually support it?

I haven't come across any, no.

But more about the return codes for pretty much all the errors here.
They'll prevent the namespace from being visible, but I think you just
want to set the limits to disable write streams instead. Otherwise it'd
be a regression since namespaces configured this way are currently
usable.