Re: [PATCH v2] scsi: core: Drop Scsi_Host.default_lock

Re: [PATCH v2] scsi: core: Drop Scsi_Host.default_lock

[Damien Le Moal]

On 2026/08/06 4:39, John Garry wrote:
> From: John Garry <john.garry@linux.dev>
> 
> Back in the 2.6.xx days, it was possible to for low-level drivers to set

s/to for/for

> the shost lock pointer. That is why there is a default lock and a pointer
> to the shost lock.
> 
> However, support for this has long been removed, so drop
> Scsi_Host.default_lock and make Scsi_Host.host_lock as the actual lock.
> 
> Getting the address of embedded host_lock structure just requires adding
> a fixed offset value to the shost pointer. However, getting the value of
> the host_lock pointer requires loading from a fixed offset to the shost
> pointer. The latter should be very slightly slower, which is relevant as
> this lock is used a lot throughout the core code and drivers.
> 
> Signed-off-by: John Garry <john.g.garry@oracle.com>
For the ata bits,

Acked-by: Damien Le Moal <dlemoal@kernel.org>

-- 
Damien Le Moal
Western Digital Research

Re: [PATCH v2] scsi: core: Drop Scsi_Host.default_lock

[Bart Van Assche]

On 8/6/26 4:39 AM, John Garry wrote:
 > Getting the address of embedded host_lock structure just requires adding
 > a fixed offset value to the shost pointer. However, getting the value of
 > the host_lock pointer requires loading from a fixed offset to the shost
 > pointer. The latter should be very slightly slower, which is relevant as
 > this lock is used a lot throughout the core code and drivers.

Hmm ... SCSI drivers for which performance matters shouldn't use the 
SCSI host lock in their hot path.

 > @Bart, Can you kindly build test this? s390 and sh were causing some 
build
 > issues for me. Thanks

This patch passes my build tests, but an update for the references to
default_lock in the Documentation/ directory is missing.

Otherwise this patch looks good to me. So with or without the
Documentation/ update, feel free to add:

Reviewed-by: Bart Van Assche <bvanassche@acm.org>

Re: [PATCH v2] scsi: core: Drop Scsi_Host.default_lock

[John Garry]

> 
> On 8/6/26 4:39 AM, John Garry wrote:
>    > Getting the address of embedded host_lock structure just requires adding
>    > a fixed offset value to the shost pointer. However, getting the value of
>    > the host_lock pointer requires loading from a fixed offset to the shost
>    > pointer. The latter should be very slightly slower, which is relevant as
>    > this lock is used a lot throughout the core code and drivers.
> 
> Hmm ... SCSI drivers for which performance matters shouldn't use the
> SCSI host lock in their hot path.

Any driver which uses DEF_SCSI_QCMD uses this lock in the fastpath. 
However I don't think that many are high performance HBAs. Maybe 
usb/storage/ or pmcraid could be consider higher performing.

Then there is stuff like xen-scsifront.c which uses this lock in the cmd 
done handler. It's used a lot in lpfc and also in code which looks to be 
fastpath...but that drivers is very complex.

> 
>    > @Bart, Can you kindly build test this? s390 and sh were causing some
> build
>    > issues for me. Thanks
> 
> This patch passes my build tests, but an update for the references to
> default_lock in the Documentation/ directory is missing.

I'll fix it

> 
> Otherwise this patch looks good to me. So with or without the
> Documentation/ update, feel free to add:
> 
> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> 

thanks