Re: [RFC/RFT PATCH 2/2] gpio: omap: ensure that runtime pm will disable unused gpio banks

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 4/29/15 7:33 AM, Tony Lindgren wrote:
* Grygorii.Strashko@xxxxxxxxxx <grygorii.strashko@xxxxxxxxxx> [150417 10:33]:
Hi Tony,
On 04/16/2015 07:29 PM, Tony Lindgren wrote:
* Tony Lindgren <tony@xxxxxxxxxxx> [150319 16:08]:
* Tony Lindgren <tony@xxxxxxxxxxx> [150307 00:08]:
* grygorii.strashko@xxxxxxxxxx <grygorii.strashko@xxxxxxxxxx> [150306 11:27]:
From: Grygorii Strashko <grygorii.strashko@xxxxxxxxxx>

Now there are two points related to Runtime PM usage:
1) bank state doesn't need to be checked in places where
Rintime PM is used, bacause Runtime PM maintains its
own usage counter:
        if (!BANK_USED(bank))
                 pm_runtime_get_sync(bank->dev);
so, it's safe to drop such checks.

2) There is a call of pm_runtime_get_sync() in omap_gpio_irq_type(),
but no corresponding put. As result, GPIO devices could be
powered on forever if at least one GPIO was used as IRQ.
Hence, allow powering off GPIO banks by adding missed
pm_runtime_put(bank->dev) at the end of omap_gpio_irq_type().

Nice to see this happening, but I think before merging this we need
to test to be sure that the pm_runtime calls actually match.. I'm
not convinced right now.. We may still have uninitialized entry
points similar to 3d009c8c61f9 ("gpio: omap: Fix bad device
access with setup_irq()").

OK so I finally got around testing this along with your bank
removal set. Looks like this patch causes a regression at least
with n900 keyboard LEDs with off-idle. The LED won't come back
on after restore from off-idle. Anyways, now we have something
reproducable :) So I'll try to debug it further at some point,
might be few days before I get to it.

Sorry for the delay, finally got around to this one :)

Here's what I came up with, only lightly tested so far. Note that
we need to keep the PM runtime per bank, not per GPIO. Will repost
a proper patch after some more testing.

I've had a opposite idea actually ;) - use Runtime PM on per-gpio basis as
it maintains all needed counters inside and we can be sure that
PM runtime callbacks will be called once per bank.

Sorry looks like I missed this email earlier, the mail got buried
in all the threads in my inbox.

Anyways, we still need per bank pm runtime for a while until we
can make it per GPIO.

Also, I've thought about moving the code from omap_enable_gpio_module()
into Runtime PM callbacks.
So, final goal - get rid of BANK_USED & LINE_USED.

Were you able to identify broken calls sequence?

No, but it breaks things for omap3 off idle. And then I noticed all
the duplicate code as discussed in the newer thread.

Also, Pay attention pls, that omap2_gpio_prepare_for/resume_after_idle()
will be most probably a NOP now.

Looks like we can't do that change yet, it breaks omap3 off idle.

The issue is across OMAPs. Those stupid debounce clocks won't let the
GPIO block to idle. They are called optional clocks but actually they
aren't.

Regards,
Santosh
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [Linux Arm (vger)]     [ARM Kernel]     [ARM MSM]     [Linux Tegra]     [Linux WPAN Networking]     [Linux Wireless Networking]     [Maemo Users]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite Trails]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux