On Sat, Mar 26, 2011 at 1:24 PM, Thomas Gleixner [off-list ref] wrote:

On Sun, 27 Mar 2011, Varun Wadekar wrote:

quoted

Thomas, then how do you think we should handle restoring of gpio states
across suspend-resume cycles?

That code is unused. Period. No caller, nothing nada. So what does it
handle?

Tegra suspend support didn't make it into 2.6.39, but should get
merged in 2.6.40, and will call tegra_gpio_suspend/resume.

quoted

quoted

-
-   for (i = INT_GPIO_BASE; i < (INT_GPIO_BASE + TEGRA_NR_GPIOS); i++) {
-           struct irq_desc *desc = irq_to_desc(i);
-           if (!desc || (desc->status & IRQ_WAKEUP))
-                   continue;
-           enable_irq(i);
-   }

And this part is totally unacceptable and should have never been
merged. Further it is in the way of cleanups to the core code and as
there is no user I'm not willing to even think about what it does and
why it is there.

FYI, the core code deals with interrupt suspending/resuming
already. So if there is a problem with that which does not cover your
specific problem, then you better talk to me before hacking up such
private workarounds and expecting that I tolerate them in unused code.

Yes, the existing code wrong, and unnecessary.  It was copied from
mach-tegra/irq.c, which I later fixed, but I missed this one.  I'll
take this patch for 2.6.39-rc1, which will prevent merge conflicts
between your tree and the tegra tree in 2.6.40.