[PATCH 1/2] ata: ahci: Add new board low_power_no_debounce_delay

[PATCH 1/2] ata: ahci: Add new board low_power_no_debounce_delay

[Paul Menzel]

Some adapters d onot require the default 200 ms debounce delay in
`sata_link_resume()`. So, create the new board
`board_ahci_low_power_no_debounce_delay` with the link flag
`ATA_LFLAG_NO_DEBOUNCE_DELAY`.

Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Mario Limonciello <mario.limonciello@amd.com>
---
 drivers/ata/ahci.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
index 84456c05e845..0fc09b86a559 100644
--- a/drivers/ata/ahci.c
+++ b/drivers/ata/ahci.c
@@ -51,6 +51,7 @@ enum board_ids {
 	board_ahci,
 	board_ahci_ign_iferr,
 	board_ahci_low_power,
+	board_ahci_low_power_no_debounce_delay,
 	board_ahci_no_debounce_delay,
 	board_ahci_nomsi,
 	board_ahci_noncq,
@@ -142,6 +143,14 @@ static const struct ata_port_info ahci_port_info[] = {
 		.udma_mask	= ATA_UDMA6,
 		.port_ops	= &ahci_ops,
 	},
+	[board_ahci_low_power_no_debounce_delay] = {
+		AHCI_HFLAGS	(AHCI_HFLAG_USE_LPM_POLICY),
+		.flags		= AHCI_FLAG_COMMON,
+		.link_flags	= ATA_LFLAG_NO_DEBOUNCE_DELAY,
+		.pio_mask	= ATA_PIO4,
+		.udma_mask	= ATA_UDMA6,
+		.port_ops	= &ahci_ops,
+	},
 	[board_ahci_no_debounce_delay] = {
 		.flags		= AHCI_FLAG_COMMON,
 		.link_flags	= ATA_LFLAG_NO_DEBOUNCE_DELAY,
-- 
2.30.2

[PATCH 2/2] [RFC] ata: ahci: Skip debounce delay for AMD FCH SATA Controller

[Paul Menzel]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1584 bytes --]

AMD devices with the FCH SATA Controller 0x1022:0x7901 do not need the
default debounce delay of 200 ms.

    07:00.2 SATA controller [0106]: Advanced Micro Devices, Inc. [AMD] FCH SATA Controller [AHCI mode] [1022:7901] (rev 51)

So skip it, by mapping it to the board with no debounce delay.

Tested on the MSI MS-7A37/B350M MORTAR (MS-7A37).

To-do: Add test details and results.

Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Mario Limonciello <mario.limonciello@amd.com>
---
I am travelling so could not test this exact patch just yet, but I ran
something similar for several weeks already. It’d be great, if the
desktop and AMD folks could also give this a try.

 drivers/ata/ahci.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
index 0fc09b86a559..44b79fe43d13 100644
--- a/drivers/ata/ahci.c
+++ b/drivers/ata/ahci.c
@@ -456,7 +456,7 @@ static const struct pci_device_id ahci_pci_tbl[] = {
 	{ PCI_VDEVICE(AMD, 0x7800), board_ahci }, /* AMD Hudson-2 */
 	{ PCI_VDEVICE(AMD, 0x7801), board_ahci_no_debounce_delay }, /* AMD Hudson-2 (AHCI mode) */
 	{ PCI_VDEVICE(AMD, 0x7900), board_ahci }, /* AMD CZ */
-	{ PCI_VDEVICE(AMD, 0x7901), board_ahci_low_power }, /* AMD Green Sardine */
+	{ PCI_VDEVICE(AMD, 0x7901), board_ahci_low_power_no_debounce_delay }, /* AMD Green Sardine */
 	/* AMD is using RAID class only for ahci controllers */
 	{ PCI_VENDOR_ID_AMD, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID,
 	  PCI_CLASS_STORAGE_RAID << 8, 0xffffff, board_ahci },
-- 
2.30.2

Re: [PATCH 1/2] ata: ahci: Add new board low_power_no_debounce_delay

[Sergei Shtylyov]

On 3/3/22 1:04 PM, Paul Menzel wrote:

> Some adapters d onot require the default 200 ms debounce delay in

   Do not. Perhaps could be fixed while applying...

> `sata_link_resume()`. So, create the new board
> `board_ahci_low_power_no_debounce_delay` with the link flag
> `ATA_LFLAG_NO_DEBOUNCE_DELAY`.
> 
> Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
> Cc: Hans de Goede <hdegoede@redhat.com>
> Cc: Mario Limonciello <mario.limonciello@amd.com>
[...]

MBR, Sergey

Re: [PATCH 2/2] [RFC] ata: ahci: Skip debounce delay for AMD FCH SATA Controller

[Damien Le Moal]

On 2022/03/03 12:04, Paul Menzel wrote:
> AMD devices with the FCH SATA Controller 0x1022:0x7901 do not need the
> default debounce delay of 200 ms.
> 
>     07:00.2 SATA controller [0106]: Advanced Micro Devices, Inc. [AMD] FCH SATA Controller [AHCI mode] [1022:7901] (rev 51)
> 
> So skip it, by mapping it to the board with no debounce delay.
> 
> Tested on the MSI MS-7A37/B350M MORTAR (MS-7A37).
> 
> To-do: Add test details and results.

Please squash this patch together with patch 1. Since you are adding a new board
entry definition, it is better to have a user for it in the same patch (this
avoids reverts to leave unused code behind).

> 
> Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
> Cc: Hans de Goede <hdegoede@redhat.com>
> Cc: Mario Limonciello <mario.limonciello@amd.com>
> ---
> I am travelling so could not test this exact patch just yet, but I ran
> something similar for several weeks already. It’d be great, if the
> desktop and AMD folks could also give this a try.
> 
>  drivers/ata/ahci.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
> index 0fc09b86a559..44b79fe43d13 100644
> --- a/drivers/ata/ahci.c
> +++ b/drivers/ata/ahci.c
> @@ -456,7 +456,7 @@ static const struct pci_device_id ahci_pci_tbl[] = {
>  	{ PCI_VDEVICE(AMD, 0x7800), board_ahci }, /* AMD Hudson-2 */
>  	{ PCI_VDEVICE(AMD, 0x7801), board_ahci_no_debounce_delay }, /* AMD Hudson-2 (AHCI mode) */
>  	{ PCI_VDEVICE(AMD, 0x7900), board_ahci }, /* AMD CZ */
> -	{ PCI_VDEVICE(AMD, 0x7901), board_ahci_low_power }, /* AMD Green Sardine */
> +	{ PCI_VDEVICE(AMD, 0x7901), board_ahci_low_power_no_debounce_delay }, /* AMD Green Sardine */

Really long name, but I cannot think of anything better...

>  	/* AMD is using RAID class only for ahci controllers */
>  	{ PCI_VENDOR_ID_AMD, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID,
>  	  PCI_CLASS_STORAGE_RAID << 8, 0xffffff, board_ahci },


-- 
Damien Le Moal
Western Digital Research

Re: [PATCH 2/2] [RFC] ata: ahci: Skip debounce delay for AMD FCH SATA Controller

[Mario Limonciello]

On 3/3/22 06:23, Damien Le Moal wrote:
> On 2022/03/03 12:04, Paul Menzel wrote:
>> AMD devices with the FCH SATA Controller 0x1022:0x7901 do not need the
>> default debounce delay of 200 ms.
>>
>>      07:00.2 SATA controller [0106]: Advanced Micro Devices, Inc. [AMD] FCH SATA Controller [AHCI mode] [1022:7901] (rev 51)
>>
>> So skip it, by mapping it to the board with no debounce delay.
>>
>> Tested on the MSI MS-7A37/B350M MORTAR (MS-7A37).
>>
>> To-do: Add test details and results.
> 
> Please squash this patch together with patch 1. Since you are adding a new board
> entry definition, it is better to have a user for it in the same patch (this
> avoids reverts to leave unused code behind). >
>>
>> Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
>> Cc: Hans de Goede <hdegoede@redhat.com>
>> Cc: Mario Limonciello <mario.limonciello@amd.com>
>> ---
>> I am travelling so could not test this exact patch just yet, but I ran
>> something similar for several weeks already. It’d be great, if the
>> desktop and AMD folks could also give this a try.

As we are trying to drop the low power definition for 5.18, maybe can 
this wait until 5.19 so we can see if that sticks so this doesn't get 
caught up in possible reverts?

>>
>>   drivers/ata/ahci.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
>> index 0fc09b86a559..44b79fe43d13 100644
>> --- a/drivers/ata/ahci.c
>> +++ b/drivers/ata/ahci.c
>> @@ -456,7 +456,7 @@ static const struct pci_device_id ahci_pci_tbl[] = {
>>   	{ PCI_VDEVICE(AMD, 0x7800), board_ahci }, /* AMD Hudson-2 */
>>   	{ PCI_VDEVICE(AMD, 0x7801), board_ahci_no_debounce_delay }, /* AMD Hudson-2 (AHCI mode) */
>>   	{ PCI_VDEVICE(AMD, 0x7900), board_ahci }, /* AMD CZ */
>> -	{ PCI_VDEVICE(AMD, 0x7901), board_ahci_low_power }, /* AMD Green Sardine */
>> +	{ PCI_VDEVICE(AMD, 0x7901), board_ahci_low_power_no_debounce_delay }, /* AMD Green Sardine */
> 
> Really long name, but I cannot think of anything better...
> 
>>   	/* AMD is using RAID class only for ahci controllers */
>>   	{ PCI_VENDOR_ID_AMD, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID,
>>   	  PCI_CLASS_STORAGE_RAID << 8, 0xffffff, board_ahci },
> 
> 

Re: [PATCH 2/2] [RFC] ata: ahci: Skip debounce delay for AMD FCH SATA Controller

[Damien Le Moal]

On 2022/03/03 15:19, Mario Limonciello wrote:
> On 3/3/22 06:23, Damien Le Moal wrote:
>> On 2022/03/03 12:04, Paul Menzel wrote:
>>> AMD devices with the FCH SATA Controller 0x1022:0x7901 do not need the
>>> default debounce delay of 200 ms.
>>>
>>>      07:00.2 SATA controller [0106]: Advanced Micro Devices, Inc. [AMD] FCH SATA Controller [AHCI mode] [1022:7901] (rev 51)
>>>
>>> So skip it, by mapping it to the board with no debounce delay.
>>>
>>> Tested on the MSI MS-7A37/B350M MORTAR (MS-7A37).
>>>
>>> To-do: Add test details and results.
>>
>> Please squash this patch together with patch 1. Since you are adding a new board
>> entry definition, it is better to have a user for it in the same patch (this
>> avoids reverts to leave unused code behind). >
>>>
>>> Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
>>> Cc: Hans de Goede <hdegoede@redhat.com>
>>> Cc: Mario Limonciello <mario.limonciello@amd.com>
>>> ---
>>> I am travelling so could not test this exact patch just yet, but I ran
>>> something similar for several weeks already. It’d be great, if the
>>> desktop and AMD folks could also give this a try.
> 
> As we are trying to drop the low power definition for 5.18, maybe can 
> this wait until 5.19 so we can see if that sticks so this doesn't get 
> caught up in possible reverts?

The "no debounce delay" addition is a valid change regardless of the "low_power
is default" change. So I do not think this is an issue. But you will need to
rebase your patch to include this change :)

Going forward, I would also like to have the "no debounce delay" as the default
too, after the low power default change settles and hopefully sticks.

> 
>>>
>>>   drivers/ata/ahci.c | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
>>> index 0fc09b86a559..44b79fe43d13 100644
>>> --- a/drivers/ata/ahci.c
>>> +++ b/drivers/ata/ahci.c
>>> @@ -456,7 +456,7 @@ static const struct pci_device_id ahci_pci_tbl[] = {
>>>   	{ PCI_VDEVICE(AMD, 0x7800), board_ahci }, /* AMD Hudson-2 */
>>>   	{ PCI_VDEVICE(AMD, 0x7801), board_ahci_no_debounce_delay }, /* AMD Hudson-2 (AHCI mode) */
>>>   	{ PCI_VDEVICE(AMD, 0x7900), board_ahci }, /* AMD CZ */
>>> -	{ PCI_VDEVICE(AMD, 0x7901), board_ahci_low_power }, /* AMD Green Sardine */
>>> +	{ PCI_VDEVICE(AMD, 0x7901), board_ahci_low_power_no_debounce_delay }, /* AMD Green Sardine */
>>
>> Really long name, but I cannot think of anything better...
>>
>>>   	/* AMD is using RAID class only for ahci controllers */
>>>   	{ PCI_VENDOR_ID_AMD, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID,
>>>   	  PCI_CLASS_STORAGE_RAID << 8, 0xffffff, board_ahci },
>>
>>
> 


-- 
Damien Le Moal
Western Digital Research