Re: [PATCH v2] Input: joystick - Use ktime for measuring timing

At Wed, 17 Sep 2014 21:23:37 +0200,
Andreas Mohr wrote:

Hi,

On Wed, Sep 10, 2014 at 05:57:17PM +0200, Takashi Iwai wrote:

quoted

The current codes in gameport and analog joystick drivers for the time
accounting have a long-standing problem when the system is running
with CPU freq; since the timing is measured via TSC or sample counter,
the calculation isn't reliable.

Thank you very much for having followed through on this!
(somehow you never seem to disappoint me :)

While working on testing this and doing various gameport/soundcard modifications,
I'm afraid I have seen the following checkpatch.pl (of v3.16) warnings:

WARNING: Missing a blank line after declarations
#164: FILE: drivers/input/joystick/analog.c:192:
+               unsigned int x;
+               GET_TIME(x);

WARNING: line over 80 characters
#224: FILE: drivers/input/joystick/analog.c:301:
+                               port->axes[j] = (delta(start, time[i]) << ANALOG_FUZZ_BITS) / port->loop;

Both are cosmetic and I won't be bothered if it were sent to my tree,
for example ;)  It depends on the maintainer, so if Dmitry wants these
to be fixed, I'm willing to resubmit, of course.

BTW the commit as-is will not be compatible with v3.16
since there's no ktime_get_ns() there yet
and (to add insult to injury) even the #include header files have changed, too.

That's a bit of a pity since I just had intended to say
that it's a very good idea
to release a quick(&dirty) initial timing hotfix for gameport handling
*prior* to possibly doing any subsequent less-related gameport cleanup commits,
since the quick initial timing hotfix would be very prominent -stable material,
except it... ain't so (--> life sucks :-).

Hey, why not using 3.17 kernel?
If the fix is *seriously* demanded, one can backport easily with the
equivalent function and the header change, and send to stable kernel,
too.  The important point is to track the upstream commit ID (once if
it's really merged).


Takashi