From mboxrd@z Thu Jan 1 00:00:00 1970 From: axboe@kernel.dk (Jens Axboe) Date: Thu, 17 May 2018 09:33:34 -0600 Subject: [PATCH] nvme: handle completions outside of the queue lock In-Reply-To: <20180517153007.GB19042@infradead.org> References: <20180517034750.GA22063@localhost.localdomain> <20180517071651.GC30079@infradead.org> <20180517142159.GC22063@localhost.localdomain> <39c39949-1aea-4892-1d80-5ea58d1ced24@kernel.dk> <20180517153007.GB19042@infradead.org> Message-ID: <6df24c77-bdc8-4265-30ea-8140eedc1fbd@kernel.dk> On 5/17/18 9:30 AM, Christoph Hellwig wrote: > On Thu, May 17, 2018@08:22:01AM -0600, Jens Axboe wrote: >> Just apply it on top, doesn't really warrant a rebase imho. But yeah, >> we should probably cool the jets for a day next time :-) > > Honestly, between the poll issue and things like the volatile > annotations that should have been a separate prep patch I'd rather > back out and restart the thing slowly. We should still have plenty > of time to get it done for this merge window. The poll issue is almost moot imho, unless you have a ton of other IO on the same hardware queue while you poll. It'll work fine, the testing I ran didn't find anything. Agree the volatile should have been a prep, but again not really a major thing. -- Jens Axboe