On Fri, Aug 5, 2022 at 4:04 PM Dmitry Rokosov [off-list ref] wrote:

On Thu, Aug 04, 2022 at 08:30:12PM +0200, Andy Shevchenko wrote:

quoted

quoted

quoted

quoted

quoted

+               dev_err(dev, "cannot %s register %u from debugfs (%d)\n",
+                       readval ? "read" : "write", reg, err);

You may consider taking [1] as a precursor here and use str_read_write().

[1]: https://lore.kernel.org/linux-i2c/20220703154232.55549-1-andriy.shevchenko@linux.intel.com/ (local)

Oh, really... Thank you for suggestion!

I have taken it closer, and it's really helpful and nice, but looks like
it's not merged to linux-next.
Please advise how I can use it in the driver. Should I provide
"Depends-On:" tag as I did for my HZ units patchset?

No, just take that patch into your series.

Do you mean include your patch to this reply-thread through the
message-id linking?

No, just take it as a part of your series. Ah, I wrote almost the same
thing above...

The idea is you rebase your tree interactively and put a breakpoint to
the beginning of your series, then you take a link and run `b4 am -s
...` (-s is important) followed by `git am ...` (b4 will show you the
command you need to run). Then you continue your rebasing. Now, when
you send a new version of the series, take one more patch into it by
changing depth from 3 (the number of patches in your series) to 4 (+
my patch).

Generally speaking the HZ series is something different. It's a series
which can't be simply taken because it might touch the different
subsystem(s). Luckily for you the "different subsystem(s)" is the same
subsystem you are taking these patches with. Hence it might not be a
problem to attach it as well into a chain.

Speaking of lib/ code almost any maintainer can take it via their
trees. So taking a _single_ patch against lib/ is fine in most cases.

-- 
With Best Regards,
Andy Shevchenko