In my environment I have large, sequential I/O requests (e.g. 4 MB)
split into much smaller requests (e.g. 8 KB) and passed to a kernel
module, which in turn submits them to an NVMe device. This splitting
of requests seems to hurt performance badly. Because of architectural
reasons I cannot avoid
On 08/10/2017 11:15 PM, Ritesh Harjani wrote:
> Hi Jens,
>
> Could you please take a look at below patch and the issue it is trying
> to solve. Please let us know your thoughts on the below problem and the
> patch.
It looks reasonable to me. I'll queue it up for 4.14.
--
Jens Axboe
On 08/10/2017 01:55 PM, Paolo Valente wrote:
>
>> Il giorno 04 ago 2017, alle ore 07:35, Paolo Valente
>> ha scritto:
>>
>> Hi,
>> these two patches improve throughput-boosting logic in two
>> aspects. The first patch refactors the parts of the device-idling
>> logic,
On Fri, Aug 11, 2017 at 03:49:29PM +0200, Benjamin Block wrote:
> On Fri, Aug 11, 2017 at 11:14:15AM +0200, Christoph Hellwig wrote:
> > But patch 1 still creates an additional copy of the sense data for
> > all bsg users.
> >
>
> Huh? What additional copy? There is one reply-buffer and that is
On Fri, 2017-08-11 at 01:11 -0700, Christoph Hellwig wrote:
> [+ Martin and linux-scsi]
>
> Given that we need this big pile and a few bfq fixes to avoid
> major regressesions I'm tempted to revert the default to scsi-mq
> for 4.14, but bring it back a little later for 4.15.
>
> What do you
On 08/11/2017 02:00 AM, Christoph Hellwig wrote:
> A few more small fixes - the fc/lpfc update is the biggest by far.
>
> The following changes since commit 7dd1ab163c17e11473a65b11f7e748db30618ebb:
>
> nvme: validate admin queue before unquiesce (2017-07-26 17:41:41 +0200)
>
> are available
On Fri, Aug 11, 2017 at 11:14:15AM +0200, Christoph Hellwig wrote:
> But patch 1 still creates an additional copy of the sense data for
> all bsg users.
>
Huh? What additional copy? There is one reply-buffer and that is copied
into the user-buffer should it contain valid data. Just like in your
On Mon, Jun 19, 2017 at 11:18 PM, Tomas Winkler wrote:
> That's correct, I guess someone didn't read the spec till the end when
> adding rpmb block device.
> though also looks like that the software guys where drinking up in the
> bar while jdec committee has met.
:D
>> +/*
But patch 1 still creates an additional copy of the sense data for
all bsg users.
Can you test the patch below which implements my suggestion? Your
other patches should still apply fine on top modulo minor context
changes.
---
>From 4cd32ee48e334b62b55bff0d380833b978454040 Mon Sep 17 00:00:00
My point was that we now gurantee that that the sense data is not
a stack pointer an a driver can DMA to it. Now for BSG the sense
data is "just" abused as reply, but the point still stands - we
don't want to pass a possible stack pointer to drivers in a data
buffer because we want to allow DMA
[+ Martin and linux-scsi]
Given that we need this big pile and a few bfq fixes to avoid
major regressesions I'm tempted to revert the default to scsi-mq
for 4.14, but bring it back a little later for 4.15.
What do you think? Maybe for 4.15 we could also do it through the
block tree where all
A few more small fixes - the fc/lpfc update is the biggest by far.
The following changes since commit 7dd1ab163c17e11473a65b11f7e748db30618ebb:
nvme: validate admin queue before unquiesce (2017-07-26 17:41:41 +0200)
are available in the git repository at:
git://git.infradead.org/nvme.git
12 matches
Mail list logo