RE: [PATCH] nvme-pci: Save PCI state before putting drive into deepest state

From: Mario.Limonciello
Date: Wed Sep 18 2019 - 18:00:57 EST


> -----Original Message-----
> From: Rafael J. Wysocki <rjw@xxxxxxxxxxxxx>
> Sent: Wednesday, September 18, 2019 4:57 PM
> To: Limonciello, Mario
> Cc: kbusch@xxxxxxxxxx; axboe@xxxxxx; hch@xxxxxx; sagi@xxxxxxxxxxx; linux-
> nvme@xxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Hong, Ryan; Wang,
> Crag; sjg@xxxxxxxxxx; Dominguez, Jared; linux-pci@xxxxxxxxxxxxxxx; linux-
> pm@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH] nvme-pci: Save PCI state before putting drive into deepest
> state
>
>
> [EXTERNAL EMAIL]
>
> On Wednesday, September 18, 2019 11:43:28 PM CEST
> Mario.Limonciello@xxxxxxxx wrote:
> > > -----Original Message-----
> > > From: Rafael J. Wysocki <rjw@xxxxxxxxxxxxx>
> > > Sent: Wednesday, September 18, 2019 4:31 PM
> > > To: Limonciello, Mario
> > > Cc: Keith Busch; Jens Axboe; Christoph Hellwig; Sagi Grimberg; linux-
> > > nvme@xxxxxxxxxxxxxxxxxxx; LKML; Hong, Ryan; Wang, Crag; sjg@xxxxxxxxxx;
> > > Dominguez, Jared; Linux PCI; Linux PM
> > > Subject: Re: [PATCH] nvme-pci: Save PCI state before putting drive into
> deepest
> > > state
> > >
> > >
> > > [EXTERNAL EMAIL]
> > >
> > > On Thursday, September 12, 2019 1:42:33 AM CEST Mario Limonciello wrote:
> > > > The action of saving the PCI state will cause numerous PCI configuration
> > > > space reads which depending upon the vendor implementation may cause
> > > > the drive to exit the deepest NVMe state.
> > > >
> > > > In these cases ASPM will typically resolve the PCIe link state and APST
> > > > may resolve the NVMe power state. However it has also been observed
> > > > that this register access after quiesced will cause PC10 failure
> > > > on some device combinations.
> > > >
> > > > To resolve this, move the PCI state saving to before SetFeatures has been
> > > > called. This has been proven to resolve the issue across a 5000 sample
> > > > test on previously failing disk/system combinations.
> > >
> > > This sounds reasonable to me, but it would be nice to CC that to linux-pm
> > > and/or linux-pci too.
> > >
> > > > Signed-off-by: Mario Limonciello <mario.limonciello@xxxxxxxx>
> > > > ---
> > > > drivers/nvme/host/pci.c | 13 +++++++------
> > > > 1 file changed, 7 insertions(+), 6 deletions(-)
> > > >
> > > > diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
> > > > index 732d5b6..9b3fed4 100644
> > > > --- a/drivers/nvme/host/pci.c
> > > > +++ b/drivers/nvme/host/pci.c
> > > > @@ -2894,6 +2894,13 @@ static int nvme_suspend(struct device *dev)
> > > > if (ret < 0)
> > > > goto unfreeze;
> > > >
> > > > + /*
> > > > + * A saved state prevents pci pm from generically controlling the
> > > > + * device's power. If we're using protocol specific settings, we don't
> > > > + * want pci interfering.
> > > > + */
> > > > + pci_save_state(pdev);
> > > > +
> > > > ret = nvme_set_power_state(ctrl, ctrl->npss);
> > > > if (ret < 0)
> > > > goto unfreeze;
> > > > @@ -2908,12 +2915,6 @@ static int nvme_suspend(struct device *dev)
> > >
> > > This is the case in which the PCI layer is expected to put the device into
> > > D3, so you need
> > >
> > > pdev->state_saved = 0;
> > >
> > > at this point, because you have saved the config space already.
> > >
> > > > ret = 0;
> > > > goto unfreeze;
> > >
> > > And here you don't need to jump to "unfreeze" any more.
> > >
> > > > }
> > > > - /*
> > > > - * A saved state prevents pci pm from generically controlling the
> > > > - * device's power. If we're using protocol specific settings, we don't
> > > > - * want pci interfering.
> > > > - */
> > > > - pci_save_state(pdev);
> > > > unfreeze:
> > > > nvme_unfreeze(ctrl);
> > > > return ret;
> > > >
> > >
> > >
> > >
> >
> > Thanks, I actually followed up with something along that line in a v2 sent out
> > today. My apology you weren't in CC, but here is a weblink to it.
> > http://lists.infradead.org/pipermail/linux-nvme/2019-September/027251.html
> >
>
> I don't think that pci_load_saved_state() will work, because it sets
> state_saved at the end again (if all goes well). You simply only need to
> clear state_saved here.

Explicitly calling it with NULL as the saved state to restore seemed to have that effect
of clearing state (there is an explicit check in there if it's NULL to just return 0).