Re: [PATCH 2/3] ata: libata: allow toggling fua parameter at runtime

[Damien Le Moal <damien.lemoal@opensource.wdc.com>]

On 10/21/22 15:50, Damien Le Moal wrote:
> On 10/21/22 15:21, Hannes Reinecke wrote:
>> On 10/21/22 07:38, Damien Le Moal wrote:
>>> From: "Maciej S. Szmigiero" <maciej.szmigiero@oracle.com>
>>>
>>> Currently, the libata.fua parameter isn't runtime-writable, so a
>>> system restart is required in order to toggle it.
>>> This unnecessarily complicates testing how drives behave with FUA on and
>>> off.
>>>
>>> Let's make this parameter R/W instead, like many others in the kernel.
>>>
>>> Example usage:
>>> Disable the parameter:
>>> echo 0 >/sys/module/libata/parameters/fua
>>>
>>> Revalidate disk cache settings:
>>> F=/sys/class/scsi_disk/0\:0\:0\:0/cache_type; echo `cat $F` >$F
>>>
>>> [Damien]
>>> Enabling fua support by setting libata.fua to 1 will have no effect if
>>> the libata module is loaded with libata.force=[ID]nofua, which disables
>>> fua support for the ata device(s) identified with ID or all ata devices
>>> if no ID is specified.
>>>
>>> Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
>>> Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
>>> ---
>>>   drivers/ata/libata-core.c | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c
>>> index 6008f7ed1c42..1bb9616b10d9 100644
>>> --- a/drivers/ata/libata-core.c
>>> +++ b/drivers/ata/libata-core.c
>>> @@ -128,7 +128,7 @@ module_param(atapi_passthru16, int, 0444);
>>>   MODULE_PARM_DESC(atapi_passthru16, "Enable ATA_16 passthru for ATAPI devices (0=off, 1=on [default])");
>>>   
>>>   int libata_fua = 0;
>>> -module_param_named(fua, libata_fua, int, 0444);
>>> +module_param_named(fua, libata_fua, int, 0644);
>>>   MODULE_PARM_DESC(fua, "FUA support (0=off [default], 1=on)");
>>>   
>>>   static int ata_ignore_hpa;
>> Hmm. I guess you'll need to revalidate the drive when changing that; but 
>> this can be done in a later patch.
> 
> Well, this is not sysfs, we cannot do this automatically easily...
> And thinking about it now that you mention it, going from fua=1 to fua=0
> can actually cause problems. The reverse not, since scsi side would still
> see fua=0 until revalidation.
> 
> So... Unless we find a way to link the param write to reavlidation, we
> should actually not allow this.
> Maciej ? Thoughts ?

I looked at this a little more. We could define the operations (struct
kernel_param_ops) manually together with the fua parameter declaration,
but that would be really ugly...

Given that we are switching to fua=1 by default, do you still need a
dynamic argument ? I am now thinking that this patch should be dropped.

> 
>>
>> Reviewed-by: Hannes Reinecke <hare@suse.de>
>>
>> Cheers,
>>
>> Hannes
> 

-- 
Damien Le Moal
Western Digital Research