[PATCH 2/2] NVMe: Simplify device resume on io queue failure
keith.busch at intel.com
Fri Oct 2 07:17:30 PDT 2015
On Fri, 2 Oct 2015, Christoph Hellwig wrote:
> On Thu, Oct 01, 2015 at 05:14:11PM -0600, Keith Busch wrote:
>> this patch removes the unnecessary indirection, and places all work
>> tasks onto the global work queue.
> Can you explain why nvme_workq is removed here? I don't really
> understand why it's realted, but that might be because I don't
> even know why it exists in the first place. Maybe this should be a
> separate patch with a separate changelog entry?
That's the part I thought may need to be separated. I noticed using the
single threaded work queue for recovery was a bad idea when confirming
the failure injection on my drives to test the rest of the patch.
So yes, the relationship is a bit dubious. I ran out of time to split
before posting yesterday. I'll resend 2/2 as two patches.
> Otherwise looks good,
> Reviewed-by: Christoph Hellwig <hch at lst.de>
More information about the Linux-nvme