Re: [PATCH v2 2/2] ata: ahci: Protect users from setting policies their drives don't support

[Damien Le Moal]

On 4/5/22 04:39, Limonciello, Mario wrote:
> [AMD Official Use Only]
> 
> 
> 
>> -----Original Message-----
>> From: Damien Le Moal <damien.lemoal@xxxxxxxxxxxxxxxxxx>
>> Sent: Sunday, April 3, 2022 20:11
>> To: Limonciello, Mario <Mario.Limonciello@xxxxxxx>
>> Cc: open list:LIBATA SUBSYSTEM (Serial and Parallel ATA drivers) <linux-
>> ide@xxxxxxxxxxxxxxx>; open list <linux-kernel@xxxxxxxxxxxxxxx>;
>> hdegoede@xxxxxxxxxx
>> Subject: Re: [PATCH v2 2/2] ata: ahci: Protect users from setting policies their
>> drives don't support
>>
>> On 3/3/22 12:49, Mario Limonciello wrote:
>>> As the default low power policy applies to more chipsets and drives, it's
>>> important to make sure that drives actually support the policy that a user
>>> selected in their kernel configuration.
>>>
>>> If the drive doesn't support slumber, don't let the default policies
>>> dependent upon slumber (`min_power` or `min_power_with_partial`) affect
>> the
>>> disk.
>>>
>>> Signed-off-by: Mario Limonciello <mario.limonciello@xxxxxxx>
>>
>> Mario,
>>
>> Can you resend a rebased version of this, on top of libata for-5.19 branch ?
> 
> 
> OK.
> 
>>
>>> ---
>>> Changes from v1->v2:
>>> * Move deeper into codepaths
>>> * Reset to MED_POWER rather than ignore
>>>   drivers/ata/libata-sata.c | 11 +++++++++++
>>>   1 file changed, 11 insertions(+)
>>>
>>> diff --git a/drivers/ata/libata-sata.c b/drivers/ata/libata-sata.c
>>> index 071158c0c44c..0dc03888c62b 100644
>>> --- a/drivers/ata/libata-sata.c
>>> +++ b/drivers/ata/libata-sata.c
>>> @@ -13,6 +13,7 @@
>>>   #include <scsi/scsi_device.h>
>>>   #include <linux/libata.h>
>>>
>>> +#include "ahci.h"
>>>   #include "libata.h"
>>>   #include "libata-transport.h"
>>>
>>> @@ -368,10 +369,20 @@ int sata_link_scr_lpm(struct ata_link *link, enum
>> ata_lpm_policy policy,
>>>   		      bool spm_wakeup)
>>>   {
>>>   	struct ata_eh_context *ehc = &link->eh_context;
>>> +	struct ata_port *ap = link->ap;
>>> +	struct ahci_host_priv *hpriv;
>>>   	bool woken_up = false;
>>>   	u32 scontrol;
>>>   	int rc;
>>>
>>> +	hpriv = ap->host->private_data;
>>> +	if (policy >= ATA_LPM_MIN_POWER_WITH_PARTIAL &&
>>> +	  !(hpriv->cap & HOST_CAP_SSC)) {
>>> +		dev_warn(ap->host->dev,
>>> +			"This drive doesn't support slumber; restting policy to
>> MED_POWER\n");
>>
>> Typo here: s/restting/resetting. Also, s/doesn't/does not.
>>
>>> +		policy = ATA_LPM_MED_POWER;
>>
>> Here, shouldn't we use the default policy defined by
>> CONFIG_SATA_LPM_POLICY ?
> 
> If they set it too aggressively we still don't want to honor it if the drive
> can't do slumber I would expect.

True. But if the default is set to a higher performance mode, we should
not fall back to the med-power mode.

We should either (1) fallback to the closest higher performance policy
supported, or (2) not change the current policy at all. no ?

See what ahci_update_initial_lpm_policy() does to check the possible
"initial" (the default ?) policy.



> 
>>
>>> +	}
>>> +
>>>   	rc = sata_scr_read(link, SCR_CONTROL, &scontrol);
>>>   	if (rc)
>>>   		return rc;
>>
>>
>> --
>> Damien Le Moal
>> Western Digital Research


-- 
Damien Le Moal
Western Digital Research