[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190905161134.GA22363@lst.de>
Date: Thu, 5 Sep 2019 18:11:34 +0200
From: Christoph Hellwig <hch@....de>
To: Keith Busch <kbusch@...nel.org>
Cc: Christoph Hellwig <hch@....de>, linux-nvme@...ts.infradead.org,
Sagi Grimberg <sagi@...mberg.me>,
Logan Gunthorpe <logang@...tatee.com>,
Hannes Reinecke <hare@...e.com>,
"Martin K . Petersen" <martin.petersen@...cle.com>,
linux-kernel@...r.kernel.org, Jens Axboe <axboe@...com>
Subject: Re: [PATCH] nvme: Restore device naming sanity
On Wed, Sep 04, 2019 at 11:31:59AM -0600, Keith Busch wrote:
> The namespace names must be unique for the lifetime of the subsystem.
> This was accomplished by using their parent subsystems' instances which
> was independent of the controllers connected to that subsystem.
>
> The consequence of that naming scheme meant that name prefixes given to
> namespaces may match a controller from an unrelated subsystem. This has
> understandbly invited confusion when examining device nodes.
>
> Ensure the namespace's subsystem instance never clashes with a
> controller instance of another subsystem by transferring the instance
> ownership to parent subsystem from the first controller discovered in
> that subsystem.
Sanitity sounds a little exaggerated. The nvme naming isn't really
that different except that the block devices uses number where say
scsi uses letters. So maybe tone down that claim a bit, but otherwise
the patch looks fine.
Powered by blists - more mailing lists