2019-10-16 03:37:36

by Dan Williams

[permalink] [raw]
Subject: [PATCH] libata/ahci: Fix PCS quirk application

Commit c312ef176399 "libata/ahci: Drop PCS quirk for Denverton and
beyond" got the polarity wrong on the check for which board-ids should
have the quirk applied. The board type board_ahci_pcs7 is defined at the
end of the list such that "pcs7" boards can be special cased in the
future if they need the quirk. All prior Intel board ids "<
board_ahci_pcs7" should proceed with applying the quirk.

Reported-by: Andreas Friedrich <[email protected]>
Reported-by: Stephen Douthit <[email protected]>
Fixes: c312ef176399 ("libata/ahci: Drop PCS quirk for Denverton and beyond")
Cc: <[email protected]>
Signed-off-by: Dan Williams <[email protected]>
---
drivers/ata/ahci.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
index dd92faf197d5..05c2b32dcc4d 100644
--- a/drivers/ata/ahci.c
+++ b/drivers/ata/ahci.c
@@ -1600,7 +1600,9 @@ static void ahci_intel_pcs_quirk(struct pci_dev *pdev, struct ahci_host_priv *hp
*/
if (!id || id->vendor != PCI_VENDOR_ID_INTEL)
return;
- if (((enum board_ids) id->driver_data) < board_ahci_pcs7)
+
+ /* Skip applying the quirk on Denverton and beyond */
+ if (((enum board_ids) id->driver_data) >= board_ahci_pcs7)
return;

/*


2019-10-16 03:38:24

by Jens Axboe

[permalink] [raw]
Subject: Re: [PATCH] libata/ahci: Fix PCS quirk application

On 10/15/19 1:54 PM, Dan Williams wrote:
> Commit c312ef176399 "libata/ahci: Drop PCS quirk for Denverton and
> beyond" got the polarity wrong on the check for which board-ids should
> have the quirk applied. The board type board_ahci_pcs7 is defined at the
> end of the list such that "pcs7" boards can be special cased in the
> future if they need the quirk. All prior Intel board ids "<
> board_ahci_pcs7" should proceed with applying the quirk.

Thanks Dan (and others), applied.

--
Jens Axboe

2019-10-16 03:56:58

by Stephen Douthit

[permalink] [raw]
Subject: Re: [PATCH] libata/ahci: Fix PCS quirk application

On 10/15/19 3:54 PM, Dan Williams wrote:
> Commit c312ef176399 "libata/ahci: Drop PCS quirk for Denverton and
> beyond" got the polarity wrong on the check for which board-ids should
> have the quirk applied. The board type board_ahci_pcs7 is defined at the
> end of the list such that "pcs7" boards can be special cased in the
> future if they need the quirk. All prior Intel board ids "<
> board_ahci_pcs7" should proceed with applying the quirk.
>
> Reported-by: Andreas Friedrich <[email protected]>
> Reported-by: Stephen Douthit <[email protected]>
> Fixes: c312ef176399 ("libata/ahci: Drop PCS quirk for Denverton and beyond")
> Cc: <[email protected]>
> Signed-off-by: Dan Williams <[email protected]>
> ---
> drivers/ata/ahci.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
> index dd92faf197d5..05c2b32dcc4d 100644
> --- a/drivers/ata/ahci.c
> +++ b/drivers/ata/ahci.c
> @@ -1600,7 +1600,9 @@ static void ahci_intel_pcs_quirk(struct pci_dev *pdev, struct ahci_host_priv *hp
> */
> if (!id || id->vendor != PCI_VENDOR_ID_INTEL)
> return;

Unless I'm missing something this will short-circuit if there are any
older Intel controllers not explicitly listed with a PCI_VDEVICE entry
in ahci_pci_tbl. Those will match on:

/* Generic, PCI class code for AHCI */
{ PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID,
PCI_CLASS_STORAGE_SATA_AHCI, 0xffffff, board_ahci },

and id->vendor will be PCI_ANY_ID so we bail before applying the quirk.
because....

static inline const struct pci_device_id *
pci_match_one_device(const struct pci_device_id *id, const struct pci_dev *dev)
{
if ((id->vendor == PCI_ANY_ID || id->vendor == dev->vendor) &&
(id->device == PCI_ANY_ID || id->device == dev->device) &&
(id->subvendor == PCI_ANY_ID || id->subvendor == dev->subsystem_vendor) &&
(id->subdevice == PCI_ANY_ID || id->subdevice == dev->subsystem_device) &&
!((id->class ^ dev->class) & id->class_mask))
return id;
return NULL;
}

const struct pci_device_id *pci_match_id(const struct pci_device_id *ids,
struct pci_dev *dev)
{
if (ids) {
while (ids->vendor || ids->subvendor || ids->class_mask) {
if (pci_match_one_device(ids, dev))
return ids;
ids++;
}
}
return NULL;
}

So we should check pdev->vendor instead.

-Steve

> - if (((enum board_ids) id->driver_data) < board_ahci_pcs7)
> +
> + /* Skip applying the quirk on Denverton and beyond */
> + if (((enum board_ids) id->driver_data) >= board_ahci_pcs7)
> return;
>
> /*
>

2019-10-16 07:10:45

by Dan Williams

[permalink] [raw]
Subject: Re: [PATCH] libata/ahci: Fix PCS quirk application

On Tue, Oct 15, 2019 at 2:02 PM Stephen Douthit
<[email protected]> wrote:
>
> On 10/15/19 3:54 PM, Dan Williams wrote:
> > Commit c312ef176399 "libata/ahci: Drop PCS quirk for Denverton and
> > beyond" got the polarity wrong on the check for which board-ids should
> > have the quirk applied. The board type board_ahci_pcs7 is defined at the
> > end of the list such that "pcs7" boards can be special cased in the
> > future if they need the quirk. All prior Intel board ids "<
> > board_ahci_pcs7" should proceed with applying the quirk.
> >
> > Reported-by: Andreas Friedrich <[email protected]>
> > Reported-by: Stephen Douthit <[email protected]>
> > Fixes: c312ef176399 ("libata/ahci: Drop PCS quirk for Denverton and beyond")
> > Cc: <[email protected]>
> > Signed-off-by: Dan Williams <[email protected]>
> > ---
> > drivers/ata/ahci.c | 4 +++-
> > 1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
> > index dd92faf197d5..05c2b32dcc4d 100644
> > --- a/drivers/ata/ahci.c
> > +++ b/drivers/ata/ahci.c
> > @@ -1600,7 +1600,9 @@ static void ahci_intel_pcs_quirk(struct pci_dev *pdev, struct ahci_host_priv *hp
> > */
> > if (!id || id->vendor != PCI_VENDOR_ID_INTEL)
> > return;
>
> Unless I'm missing something this will short-circuit if there are any
> older Intel controllers not explicitly listed with a PCI_VDEVICE entry
> in ahci_pci_tbl. Those will match on:
>
> /* Generic, PCI class code for AHCI */
> { PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID,
> PCI_CLASS_STORAGE_SATA_AHCI, 0xffffff, board_ahci },

You're not missing anything, but I think we should stick with explicit
mapping as only newer controllers tend to match on class id rather
than pci-id, and there's no way to know if that class-id match is for
PCS_6 or PCS_7. Hopefully newer controllers are tested with Linux and
the BIOS fixed prior to the breakage leaking into the wild.