[PATCH] scsi: Delete scsi_use_blk_mq

[PATCH] scsi: Delete scsi_use_blk_mq

[John Garry]

Module param scsi_use_blk_mq has not been referenced for some time, so
zap it.

Signed-off-by: John Garry <john.garry@huawei.com>

diff --git a/drivers/scsi/scsi.c b/drivers/scsi/scsi.c
index 930e4803d888..4b9fdfab77d9 100644
--- a/drivers/scsi/scsi.c
+++ b/drivers/scsi/scsi.c
@@ -764,10 +764,6 @@ MODULE_LICENSE("GPL");
 module_param(scsi_logging_level, int, S_IRUGO|S_IWUSR);
 MODULE_PARM_DESC(scsi_logging_level, "a bit mask of logging levels");
 
-/* This should go away in the future, it doesn't do anything anymore */
-bool scsi_use_blk_mq = true;
-module_param_named(use_blk_mq, scsi_use_blk_mq, bool, S_IWUSR | S_IRUGO);
-
 static int __init init_scsi(void)
 {
 	int error;
diff --git a/drivers/scsi/scsi_priv.h b/drivers/scsi/scsi_priv.h
index 3bff9f7aa684..25b0aaaf5ae8 100644
--- a/drivers/scsi/scsi_priv.h
+++ b/drivers/scsi/scsi_priv.h
@@ -29,7 +29,6 @@ extern int scsi_init_hosts(void);
 extern void scsi_exit_hosts(void);
 
 /* scsi.c */
-extern bool scsi_use_blk_mq;
 int scsi_init_sense_cache(struct Scsi_Host *shost);
 void scsi_init_command(struct scsi_device *dev, struct scsi_cmnd *cmd);
 #ifdef CONFIG_SCSI_LOGGING
-- 
2.17.1

Re: [PATCH] scsi: Delete scsi_use_blk_mq

[Bart Van Assche]

On 2/10/20 9:33 AM, John Garry wrote:
> -module_param_named(use_blk_mq, scsi_use_blk_mq, bool, S_IWUSR | S_IRUGO);

Will this change cause trouble to shell scripts that set or read this 
parameter (/sys/module/scsi_mod/parameters/use_blk_mq)? What will the 
impact be on systems where scsi_mod.use_blk_mq=Y is passed by GRUB to 
the kernel at boot time, e.g. because it has been set in the 
GRUB_CMDLINE_LINUX variable in /etc/default/grub?

Thanks,

Bart.

Re: [PATCH] scsi: Delete scsi_use_blk_mq

[John Garry]

On 10/02/2020 22:37, Bart Van Assche wrote:
> On 2/10/20 9:33 AM, John Garry wrote:
>> -module_param_named(use_blk_mq, scsi_use_blk_mq, bool, S_IWUSR | 
>> S_IRUGO);
> 

Hi Bart,

> Will this change cause trouble to shell scripts that set or read this 
> parameter (/sys/module/scsi_mod/parameters/use_blk_mq)? 

The entry in Documentation/admin-guide/kernel-parameters.txt is gone for 
2 years now.

And it is not an archaic module param, it was introduced 6 years ago. As 
such, I'd say that if a shell script was setup to access this parameter, 
then it would prob also pre-check if it exists and gracefully accept 
that it may not.

I will also note that there is still scsi_sysfs.c:show_use_blk_mq(), 
which would stay.

What will the
> impact be on systems where scsi_mod.use_blk_mq=Y is passed by GRUB to 
> the kernel at boot time, e.g. because it has been set in the 
> GRUB_CMDLINE_LINUX variable in /etc/default/grub?

The kernel should any params that does not recognize.

> 

Having said all that, I don't feel too strongly about deleting this - 
it's only some tidy-up.

Thanks,
John

Re: [PATCH] scsi: Delete scsi_use_blk_mq

[John Garry]

On 11/02/2020 11:50, John Garry wrote:
> On 10/02/2020 22:37, Bart Van Assche wrote:
>> On 2/10/20 9:33 AM, John Garry wrote:
>>> -module_param_named(use_blk_mq, scsi_use_blk_mq, bool, S_IWUSR | 
>>> S_IRUGO);
>>
> 
> Hi Bart,
> 
>> Will this change cause trouble to shell scripts that set or read this 
>> parameter (/sys/module/scsi_mod/parameters/use_blk_mq)? 
> 
> The entry in Documentation/admin-guide/kernel-parameters.txt is gone for 
> 2 years now.
> 
> And it is not an archaic module param, it was introduced 6 years ago. As 
> such, I'd say that if a shell script was setup to access this parameter, 
> then it would prob also pre-check if it exists and gracefully accept 
> that it may not.
> 
> I will also note that there is still scsi_sysfs.c:show_use_blk_mq(), 
> which would stay.
> 
> What will the
>> impact be on systems where scsi_mod.use_blk_mq=Y is passed by GRUB to 
>> the kernel at boot time, e.g. because it has been set in the 
>> GRUB_CMDLINE_LINUX variable in /etc/default/grub?
> 
> The kernel should any params that does not recognize.

	          ^ ignore

> 
>>
> 
> Having said all that, I don't feel too strongly about deleting this - 
> it's only some tidy-up.
> 
> Thanks,
> John
> 
> .

Re: [PATCH] scsi: Delete scsi_use_blk_mq

[Ewan D. Milne]

On Tue, 2020-02-11 at 11:50 +0000, John Garry wrote:
> On 10/02/2020 22:37, Bart Van Assche wrote:
> > On 2/10/20 9:33 AM, John Garry wrote:
> > > -module_param_named(use_blk_mq, scsi_use_blk_mq, bool, S_IWUSR | 
> > > S_IRUGO);
> 
> Hi Bart,
> 
> > Will this change cause trouble to shell scripts that set or read this 
> > parameter (/sys/module/scsi_mod/parameters/use_blk_mq)? 
> 
> The entry in Documentation/admin-guide/kernel-parameters.txt is gone for 
> 2 years now.
> 
> And it is not an archaic module param, it was introduced 6 years ago. As 
> such, I'd say that if a shell script was setup to access this parameter, 
> then it would prob also pre-check if it exists and gracefully accept 
> that it may not.
> 
> I will also note that there is still scsi_sysfs.c:show_use_blk_mq(), 
> which would stay.
> 
> What will the
> > impact be on systems where scsi_mod.use_blk_mq=Y is passed by GRUB to 
> > the kernel at boot time, e.g. because it has been set in the 
> > GRUB_CMDLINE_LINUX variable in /etc/default/grub?
> 
> The kernel should any params that does not recognize.
> 
> 
> Having said all that, I don't feel too strongly about deleting this - 
> it's only some tidy-up.
> 
> Thanks,
> John
> 

I think we should remove it.  It is not good to have a kernel parameter
that people used to be able to set to "N" that no longer does that.

-Ewan

Re: [PATCH] scsi: Delete scsi_use_blk_mq

[Martin K. Petersen]

John,

> Module param scsi_use_blk_mq has not been referenced for some time, so
> zap it.

Applied to 5.7/scsi-queue, thanks!

-- 
Martin K. Petersen	Oracle Linux Engineering