rules: avoid traversing parent of every single device?

rules: avoid traversing parent of every single device?

[Alan Jenkins]

Hi,

I've been looking at what is responsible for all the path lookup 
activity in coldplug.  On my debian stable system, it looks like every 
device gets its parent looked up in sysfs.  I think this is due to 
SUBSYSTEMS matches.

I see the udev default rules are different, but it looks like they still 
test for SUBSYSTEMS on every single device:

SUBSYSTEMS="scsi", ATTRS{type}="1|8", GROUP="tape"
SUBSYSTEMS="scsi", ATTRS{type}="4|5", GROUP="cdrom"
SUBSYSTEMS="scsi", ATTRS{type}="0", GROUP="disk"


Should we add SUBSYSTEM="scsi_generic" to these three rules?

Thanks
Alan

Re: rules: avoid traversing parent of every single device?

[Kay Sievers]

On Fri, May 22, 2009 at 16:15, Alan Jenkins <alan-jenkins@tuffmail.co.uk> wrote:
> I've been looking at what is responsible for all the path lookup activity in
> coldplug.  On my debian stable system, it looks like every device gets its
> parent looked up in sysfs.  I think this is due to SUBSYSTEMS matches.
>
> I see the udev default rules are different, but it looks like they still
> test for SUBSYSTEMS on every single device:
>
> SUBSYSTEMS="scsi", ATTRS{type}="1|8", GROUP="tape"
> SUBSYSTEMS="scsi", ATTRS{type}="4|5", GROUP="cdrom"
> SUBSYSTEMS="scsi", ATTRS{type}="0", GROUP="disk"
>
>
> Should we add SUBSYSTEM="scsi_generic" to these three rules?

Sounds good. I think the "tape" will need a match for the actual
scsi-tape devices too? The real "cdrom" and "disk" devices seems
covered by explicit rules.

Thanks,
Kay

Re: rules: avoid traversing parent of every single device?

[Alan Jenkins]

Kay Sievers wrote:
> On Fri, May 22, 2009 at 16:15, Alan Jenkins <alan-jenkins@tuffmail.co.uk> wrote:
>   
>> I've been looking at what is responsible for all the path lookup activity in
>> coldplug.  On my debian stable system, it looks like every device gets its
>> parent looked up in sysfs.  I think this is due to SUBSYSTEMS matches.
>>
>> I see the udev default rules are different, but it looks like they still
>> test for SUBSYSTEMS on every single device:
>>
>> SUBSYSTEMS="scsi", ATTRS{type}="1|8", GROUP="tape"
>> SUBSYSTEMS="scsi", ATTRS{type}="4|5", GROUP="cdrom"
>> SUBSYSTEMS="scsi", ATTRS{type}="0", GROUP="disk"
>>
>>
>> Should we add SUBSYSTEM="scsi_generic" to these three rules?
>>     
>
> Sounds good. I think the "tape" will need a match for the actual
> scsi-tape devices too? The real "cdrom" and "disk" devices seems
> covered by explicit rules.
>   

Yes, good catch.

Re: rules: avoid traversing parent of every single device?

[Kay Sievers]

On Fri, May 22, 2009 at 16:33, Alan Jenkins <alan-jenkins@tuffmail.co.uk> wrote:
>>> Should we add SUBSYSTEM="scsi_generic" to these three rules?
>>
>> Sounds good. I think the "tape" will need a match for the actual
>> scsi-tape devices too? The real "cdrom" and "disk" devices seems
>> covered by explicit rules.
>
> Yes, good catch.

SUBSYSTEM="scsi_generic|scsi_tape", SUBSYSTEMS="scsi", ... should be fine.

Should I add all that, or do you have a patch already?

Kay

Re: rules: avoid traversing parent of every single device?

[Alan Jenkins]

Kay Sievers wrote:
> On Fri, May 22, 2009 at 16:33, Alan Jenkins <alan-jenkins@tuffmail.co.uk> wrote:
>   
>>>> Should we add SUBSYSTEM="scsi_generic" to these three rules?
>>>>         
>>> Sounds good. I think the "tape" will need a match for the actual
>>> scsi-tape devices too? The real "cdrom" and "disk" devices seems
>>> covered by explicit rules.
>>>       
>> Yes, good catch.
>>     
>
> SUBSYSTEM="scsi_generic|scsi_tape", SUBSYSTEMS="scsi", ... should be fine.
>
> Should I add all that, or do you have a patch already?
>   

You can go ahead with that.

Re: rules: avoid traversing parent of every single device?

[Kay Sievers]

On Fri, May 22, 2009 at 16:42, Alan Jenkins <alan-jenkins@tuffmail.co.uk> wrote:
>> Should I add all that, or do you have a patch already?
>
> You can go ahead with that.

Done.

Thanks,
Kay