Re: [PATCH v1] spi: fix client driver can't register success when use GPIO as CS

[Andy Shevchenko <andriy.shevchenko@linux.intel.com>]

On Thu, May 06, 2021 at 02:30:01PM +0200, Linus Walleij wrote:
> On Thu, May 6, 2021 at 1:45 PM Xin Hao <bier@b-x3vxmd6m-2058.local> wrote:
> > From: Xin Hao <xhao@linux.alibaba.com>
> >
> > When i was testing the TPM2 device, I found that the driver
> > always failed to register which used SPI bus and GPIO as CS
> > signal, i found that the reason for the error was that CS could
> > not be set correctly, so there fixed it.
> >
> > Fixes: 766c6b63aa044e ("spi: fix client driver breakages when using
> > GPIO descriptors")
> > Signed-off-by: Xin Hao <xhao@linux.alibaba.com>
> 
> (...)
> >                                 /* polarity handled by gpiolib */
> >                                 gpiod_set_value_cansleep(spi->cs_gpiod,
> > -                                                        enable1);
> > +                                                        !enable);
> 
> We have been over this code a lot of times, can you
> help us to investigate the root cause here and check
> how the interrupts are provided on this platform.
> 
> TPM2 makes me think that this is an Intel platform
> and maybe ACPI of some kind so you need to run
> it by Andy, who is working on some related fixes.

The above is exactly what my quirk [1] for ACPI does in case the controller
gets GPIOs from the ACPI.

[1]: https://gitlab.com/andy-shev/next/-/commit/5ccbdbb4787d871722f361d77c5f3cb806811c48

-- 
With Best Regards,
Andy Shevchenko