From: Kai-Heng Feng <kai.heng.feng@canonical.com>
To: Mario.Limonciello@dell.com
Cc: Keith Busch <kbusch@kernel.org>,
hch@lst.de, axboe@fb.com, sagi@grimberg.me, rafael@kernel.org,
linux-pm@vger.kernel.org, rafael.j.wysocki@intel.com,
linux-kernel@vger.kernel.org, linux-nvme@lists.infradead.org,
keith.busch@intel.com
Subject: Re: [PATCH] nvme-pci: Use non-operational power state instead of D3 on Suspend-to-Idle
Date: Fri, 10 May 2019 14:05:42 +0800 [thread overview]
Message-ID: <BC5EB1D0-8718-48B3-ACAB-F7E5571D821D@canonical.com> (raw)
In-Reply-To: <495d76c66aec41a8bfbbf527820f8eb9@AUSX13MPC101.AMER.DELL.COM>
at 06:19, <Mario.Limonciello@dell.com> <Mario.Limonciello@dell.com> wrote:
>> -----Original Message-----
>> From: Keith Busch <kbusch@kernel.org>
>> Sent: Thursday, May 9, 2019 4:54 PM
>> To: Limonciello, Mario
>> Cc: kai.heng.feng@canonical.com; hch@lst.de; axboe@fb.com;
>> sagi@grimberg.me; rafael@kernel.org; linux-pm@vger.kernel.org;
>> rafael.j.wysocki@intel.com; linux-kernel@vger.kernel.org; linux-
>> nvme@lists.infradead.org; keith.busch@intel.com
>> Subject: Re: [PATCH] nvme-pci: Use non-operational power state instead
>> of D3 on
>> Suspend-to-Idle
>>
>>
>> [EXTERNAL EMAIL]
>>
>> On Thu, May 09, 2019 at 09:37:58PM +0000, Mario.Limonciello@dell.com
>> wrote:
>>>> +int nvme_set_power(struct nvme_ctrl *ctrl, unsigned npss)
>>>> +{
>>>> + int ret;
>>>> +
>>>> + mutex_lock(&ctrl->scan_lock);
>>>> + nvme_start_freeze(ctrl);
>>>> + nvme_wait_freeze(ctrl);
>>>> + ret = nvme_set_features(ctrl, NVME_FEAT_POWER_MGMT, npss, NULL, 0,
>>>> + NULL);
>>>> + nvme_unfreeze(ctrl);
>>>> + mutex_unlock(&ctrl->scan_lock);
>>>> +
>>>> + return ret;
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(nvme_set_power);
>>>
>>> I believe without memory barriers at the end disks with HMB this will
>>> still kernel panic (Such as Toshiba BG3).
>>
>> Well, the mutex has an implied memory barrier, but your HMB explanation
>> doesn't make much sense to me anyway. The "mb()" in this thread's original
>> patch is a CPU memory barrier, and the CPU had better not be accessing
>> HMB memory. Is there something else going on here?
>
> Kai Heng will need to speak up a bit in his time zone as he has this disk
> on hand,
> but what I recall from our discussion was that DMA operation MemRd after
> resume was the source of the hang.
Yes, that’ what I was told by the NVMe vendor, so all I know is to impose a
memory barrier.
If mb() shouldn’t be used here, what’s the correct variant to use in this
context?
>
>>> This still allows D3 which we found at least failed to go into deepest
>>> state and
>> blocked
>>> platform s0ix for the following SSDs (maybe others):
>>> Hynix PC601
>>> LiteOn CL1
>>
>> We usually write features to spec first, then quirk non-compliant
>> devices after.
>
> NVME spec doesn't talk about a relationship between SetFeatures w/
> NVME_FEAT_POWER_MGMGT and D3 support, nor order of events.
>
> This is why we opened a dialog with storage vendors, including
> contrasting the behavior
> of Microsoft Windows inbox NVME driver and Intel's Windows RST driver.
>
> Those two I mention that come to mind immediately because they were most
> recently
> tested to fail. Our discussion with storage vendors overwhelmingly
> requested
> that we don't use D3 under S2I because their current firmware
> architecture won't
> support it.
>
> For example one vendor told us with current implementation that receiving
> D3hot
> after NVME shutdown will prevent being able to enter L1.2. D3hot entry
> was supported
> by an IRQ handler that isn't serviced in NVME shutdown state.
>
> Another vendor told us that with current implementation it's impossible
> to transition
> to PS4 (at least via APST) while L1.2 D3hot is active.
I tested the patch from Keith and it has two issues just as simply skipping
nvme_dev_disable():
1) It consumes more power in S2I
2) System freeze after resume
Also I don’t think it’s a spec. It’s a guide to let OEM/ODM pick and
assemble Modern Standby compliant machines, so what Windows NVMe driver
really does is still opaque to us.
Kai-Heng
next prev parent reply other threads:[~2019-05-10 6:05 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-05-08 18:59 [PATCH] nvme-pci: Use non-operational power state instead of D3 on Suspend-to-Idle Kai-Heng Feng
2019-05-08 19:15 ` Chaitanya Kulkarni
2019-05-08 19:16 ` Keith Busch
2019-05-08 19:30 ` Kai-Heng Feng
2019-05-08 19:38 ` Mario.Limonciello
2019-05-08 19:51 ` Christoph Hellwig
2019-05-08 20:28 ` Mario.Limonciello
2019-05-09 6:12 ` Christoph Hellwig
2019-05-09 6:48 ` Kai-Heng Feng
2019-05-09 6:52 ` Christoph Hellwig
2019-05-09 9:19 ` Rafael J. Wysocki
2019-05-09 9:25 ` Christoph Hellwig
2019-05-09 20:48 ` Rafael J. Wysocki
2019-05-09 9:07 ` Rafael J. Wysocki
2019-05-09 9:42 ` Kai-Heng Feng
2019-05-09 9:56 ` Christoph Hellwig
2019-05-09 10:28 ` Kai-Heng Feng
2019-05-09 10:31 ` Christoph Hellwig
2019-05-09 11:59 ` Kai-Heng Feng
2019-05-09 18:57 ` Mario.Limonciello
2019-05-09 19:28 ` Keith Busch
2019-05-09 20:54 ` Rafael J. Wysocki
2019-05-09 21:16 ` Keith Busch
2019-05-09 21:39 ` Rafael J. Wysocki
2019-05-09 21:37 ` Mario.Limonciello
2019-05-09 21:54 ` Keith Busch
2019-05-09 22:19 ` Mario.Limonciello
2019-05-10 6:05 ` Kai-Heng Feng [this message]
2019-05-10 8:23 ` Rafael J. Wysocki
2019-05-10 13:52 ` Keith Busch
2019-05-10 15:15 ` Kai Heng Feng
2019-05-10 15:36 ` Keith Busch
2019-05-10 14:02 ` Keith Busch
2019-05-10 15:18 ` Kai Heng Feng
2019-05-10 15:49 ` hch
2019-05-10 5:30 ` Christoph Hellwig
2019-05-10 13:51 ` Keith Busch
2019-05-09 16:20 ` Keith Busch
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=BC5EB1D0-8718-48B3-ACAB-F7E5571D821D@canonical.com \
--to=kai.heng.feng@canonical.com \
--cc=Mario.Limonciello@dell.com \
--cc=axboe@fb.com \
--cc=hch@lst.de \
--cc=kbusch@kernel.org \
--cc=keith.busch@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
--cc=linux-pm@vger.kernel.org \
--cc=rafael.j.wysocki@intel.com \
--cc=rafael@kernel.org \
--cc=sagi@grimberg.me \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).