Re: [PATCH v1 2/2] mmc: core: Let BKOPS timeout readable/writable via sysfs

[Bean Huo]

On Tue, 2021-08-24 at 16:33 +0200, Ulf Hansson wrote:
> On Wed, 18 Aug 2021 at 00:42, Bean Huo <huobean@xxxxxxxxx> wrote:
> > From: Bean Huo <beanhuo@xxxxxxxxxx>
> > 
> > For special cases, the application in the userspace wants to change
> > BKOPS operation timeout value, also, wants eMMC back to
> > R1_STATE_TRAN
> > after BKOPS timeouts. A fixed BKOPS timeout value(120s) is no
> > longer
> > feasible, therefore, it is better to let the user controls its
> > timeout
> > value.
> 
> I am not fond of exporting tweakable timeout values through sysfs.
> Primarily because it's ABI and it becomes difficult to change.
> 
> Can you perhaps explain in more detail when you want to have a
> different timeout?
> 
> Perhaps we can do something similar as we currently do for
> mmc_santize(), where we allow userspace to pass the cmd-timeout?
> 
> Kind regards
> Uffe
> 
> > Signed-off-by: Bean Huo <beanhuo@xxxxxxxxxx>
> > 

Hi Ullf,
Thanks for your review. 

Actually, taking the first patch "MMC: core: Issue HPI in case the
BKOPS timed out" is good enough in the most of cases. And the current
120s timeout for the manual BKOPS is also enough, we don't need to
change this timeout value through sysfs.

Kind regards,
Bean