Re: [PATCH] NVMe: Added another device ID with stripe quirk

Re: [PATCH] NVMe: Added another device ID with stripe quirk

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 04:12:31PM -0600, David Wayne Fugate wrote:
> Adds a fourth Intel controller which has the "stripe" quirk.

NVMe has stadardized a way to communicate this information through
the Namespace Optimal IO Boundary (NOIOB) field in the Identify
Namespace structure, and Keith and Amber at Intel helped to define
this, so please actually implement it in your controllers.

[PATCH] NVMe: Added another device ID with stripe quirk

[Christoph Hellwig]

On Thu, Jul 06, 2017@04:12:31PM -0600, David Wayne Fugate wrote:
> Adds a fourth Intel controller which has the "stripe" quirk.

NVMe has stadardized a way to communicate this information through
the Namespace Optimal IO Boundary (NOIOB) field in the Identify
Namespace structure, and Keith and Amber at Intel helped to define
this, so please actually implement it in your controllers.

Re: [PATCH] NVMe: Added another device ID with stripe quirk

[Keith Busch]

On Fri, Jul 07, 2017 at 12:17:54AM +0200, Christoph Hellwig wrote:
> On Thu, Jul 06, 2017 at 04:12:31PM -0600, David Wayne Fugate wrote:
> > Adds a fourth Intel controller which has the "stripe" quirk.
> 
> NVMe has stadardized a way to communicate this information through
> the Namespace Optimal IO Boundary (NOIOB) field in the Identify
> Namespace structure, and Keith and Amber at Intel helped to define
> this, so please actually implement it in your controllers.

That's true for all Intel controllers going forward, but this
is actually an older controller that pre-dates NOIOB. It's the exact
same as the 8086:0A54 model, but a particular vendor decided their
rebranded device needs to be made special with a different DID.

We all agree that's a terrible way to go about this for mutliple reasons,
but we can't go back in time to tell the decision makers of this folly. So
I think we need to let this last one go through with the quirk.

Acked-by: Keith Busch <keith.busch@intel.com>

[PATCH] NVMe: Added another device ID with stripe quirk

[Keith Busch]

On Fri, Jul 07, 2017@12:17:54AM +0200, Christoph Hellwig wrote:
> On Thu, Jul 06, 2017@04:12:31PM -0600, David Wayne Fugate wrote:
> > Adds a fourth Intel controller which has the "stripe" quirk.
> 
> NVMe has stadardized a way to communicate this information through
> the Namespace Optimal IO Boundary (NOIOB) field in the Identify
> Namespace structure, and Keith and Amber at Intel helped to define
> this, so please actually implement it in your controllers.

That's true for all Intel controllers going forward, but this
is actually an older controller that pre-dates NOIOB. It's the exact
same as the 8086:0A54 model, but a particular vendor decided their
rebranded device needs to be made special with a different DID.

We all agree that's a terrible way to go about this for mutliple reasons,
but we can't go back in time to tell the decision makers of this folly. So
I think we need to let this last one go through with the quirk.

Acked-by: Keith Busch <keith.busch at intel.com>

Re: [PATCH] NVMe: Added another device ID with stripe quirk

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 06:39:56PM -0400, Keith Busch wrote:
> That's true for all Intel controllers going forward, but this
> is actually an older controller that pre-dates NOIOB. It's the exact
> same as the 8086:0A54 model, but a particular vendor decided their
> rebranded device needs to be made special with a different DID.

Meh..

> 
> We all agree that's a terrible way to go about this for mutliple reasons,
> but we can't go back in time to tell the decision makers of this folly. So
> I think we need to let this last one go through with the quirk.
> 
> Acked-by: Keith Busch <keith.busch@intel.com>

Reluctantly-accepted-by: Christoph Hellwig <hch@lst.de>

[PATCH] NVMe: Added another device ID with stripe quirk

[Christoph Hellwig]

On Thu, Jul 06, 2017@06:39:56PM -0400, Keith Busch wrote:
> That's true for all Intel controllers going forward, but this
> is actually an older controller that pre-dates NOIOB. It's the exact
> same as the 8086:0A54 model, but a particular vendor decided their
> rebranded device needs to be made special with a different DID.

Meh..

> 
> We all agree that's a terrible way to go about this for mutliple reasons,
> but we can't go back in time to tell the decision makers of this folly. So
> I think we need to let this last one go through with the quirk.
> 
> Acked-by: Keith Busch <keith.busch at intel.com>

Reluctantly-accepted-by: Christoph Hellwig <hch at lst.de>

Re: [PATCH] NVMe: Added another device ID with stripe quirk

[Sagi Grimberg]

>> That's true for all Intel controllers going forward, but this
>> is actually an older controller that pre-dates NOIOB. It's the exact
>> same as the 8086:0A54 model, but a particular vendor decided their
>> rebranded device needs to be made special with a different DID.
> 
> Meh..
> 
>>
>> We all agree that's a terrible way to go about this for mutliple reasons,
>> but we can't go back in time to tell the decision makers of this folly. So
>> I think we need to let this last one go through with the quirk.
>>
>> Acked-by: Keith Busch <keith.busch@intel.com>
> 
> Reluctantly-accepted-by: Christoph Hellwig <hch@lst.de>
> 

David, can you resend the patch? For some reason I
can't locate it in my mailbox...

[PATCH] NVMe: Added another device ID with stripe quirk

[Sagi Grimberg]

>> That's true for all Intel controllers going forward, but this
>> is actually an older controller that pre-dates NOIOB. It's the exact
>> same as the 8086:0A54 model, but a particular vendor decided their
>> rebranded device needs to be made special with a different DID.
> 
> Meh..
> 
>>
>> We all agree that's a terrible way to go about this for mutliple reasons,
>> but we can't go back in time to tell the decision makers of this folly. So
>> I think we need to let this last one go through with the quirk.
>>
>> Acked-by: Keith Busch <keith.busch at intel.com>
> 
> Reluctantly-accepted-by: Christoph Hellwig <hch at lst.de>
> 

David, can you resend the patch? For some reason I
can't locate it in my mailbox...

[PATCH] NVMe: Added another device ID with stripe quirk

[David Wayne Fugate]

Adds a fourth Intel controller which has the "stripe" quirk.

Signed-off-by: David Wayne Fugate <david.fugate at intel.com>
---
 drivers/nvme/host/pci.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 48d3ed3..9101b3e 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -2462,6 +2462,9 @@ static const struct pci_device_id nvme_id_table[] = {
 	{ PCI_VDEVICE(INTEL, 0x0a54),
 		.driver_data = NVME_QUIRK_STRIPE_SIZE |
 				NVME_QUIRK_DEALLOCATE_ZEROES, },
+	{ PCI_VDEVICE(INTEL, 0x0a55),
+		.driver_data = NVME_QUIRK_STRIPE_SIZE |
+				NVME_QUIRK_DEALLOCATE_ZEROES, },
 	{ PCI_VDEVICE(INTEL, 0xf1a5),	/* Intel 600P/P3100 */
 		.driver_data = NVME_QUIRK_NO_DEEPEST_PS },
 	{ PCI_VDEVICE(INTEL, 0x5845),	/* Qemu emulated controller */
-- 
2.7.4

[PATCH] NVMe: Added another device ID with stripe quirk

[Sagi Grimberg]

queued to nvme-4.13