On středa 24. ledna 2018 19:17:35 CET, Geert Uytterhoeven wrote:
Using a hog indeed marks it busy, as the hog is a user.
Why do you use a hog in the first place? Shouldn't the SPI driver configure
the GPIO, according to cs-gpios?
That doesn't happen with spi-orion.c. If I remove my gpio-hog for the CS
GPIO pin from the DT, then my device doesn't appear to be selected. Also, I
can run gpio-hammer targetting my CS pin just fine -- and that seems
dangerous.
I don't see any "claim" of a GPIO neither in spi.c nor spi-orion.c. I did
some very hacky grepping [1], and it seems that other drivers probably
handle this on their own. The only other match is spi-lantiq-ssc.c which
doesn't however touch the CS GPIO.
Shouldn't this be really handled by the SPI core? (I don't think I'm
volunteering with patches as I couldn't really test these.)
Should I submit a patch which adapts spi-orion.c to essentially follow how
e.g spi-imx.c allocates a GPIO pin for CS?
Cheers,
Jan
[1] `git grep --files-with-match cs_gpio drivers/spi/*.c | xargs -n1 git
grep -E --files-without-match 'gpio_direction_output|gpio_request' | cat`
--
To unsubscribe from this list: send the line "unsubscribe linux-spi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
