3.18 SDIO driver re-attach issue

3.18 SDIO driver re-attach issue

[Ben Dooks]

We are having an issue with 3.18 on an OMAP3517 and a TI WL1271
Wifi SDIO card attached to an hsmmc port. The module is powered
from the main 3.3/1.8V supply rails so cannot be powered down.

The wifi driver is built as a module, and when the module is
unloaded and then reloaded the driver attach fails. The issue
is sdio_bus_probe() calls the sdio_set_block_size(, func, 0)
which fails with -110 (ETIMEDOUT) on the second load.

Is it possible we need to either ignore the error, or try and
re-set the function when either the driver un-binds / if the
first sdio_set_block_size(, func, 0) call happens?

Currently we ignore the error and the wifi driver seems to be
working with the device. If there is a better fix then we would
like to know so it can be fixed properly up-stream.

-- 
Ben Dooks				http://www.codethink.co.uk/
Senior Engineer				Codethink - Providing Genius

Re: 3.18 SDIO driver re-attach issue

[Ulf Hansson]

On 10 June 2015 at 18:30, Ben Dooks <ben.dooks@codethink.co.uk> wrote:
> We are having an issue with 3.18 on an OMAP3517 and a TI WL1271
> Wifi SDIO card attached to an hsmmc port. The module is powered
> from the main 3.3/1.8V supply rails so cannot be powered down.
>
> The wifi driver is built as a module, and when the module is
> unloaded and then reloaded the driver attach fails. The issue
> is sdio_bus_probe() calls the sdio_set_block_size(, func, 0)
> which fails with -110 (ETIMEDOUT) on the second load.
>
> Is it possible we need to either ignore the error, or try and
> re-set the function when either the driver un-binds / if the
> first sdio_set_block_size(, func, 0) call happens?
>
> Currently we ignore the error and the wifi driver seems to be
> working with the device. If there is a better fix then we would
> like to know so it can be fixed properly up-stream.

I don't think it should fail. So instead of applying a workaround
(ignoring/re-setting/re-trying), I would rather try to find the reason
to *why* the error occurs.

Just a simple test you could do is to *not* call sdio_set_block_size()
the second time in ->probe(), and see if that will give you other
issues.

Kind regards
Uffe