Hello Sebastian,
On Tue, 19 Mar 2019 18:27:37 +0100
Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx> wrote:
> Since commit 39ee32ce48675 ("mmc: sdhci-omap: drop ->get_ro()
> implementation") my
> |Model: TI AM5728 IDK
> |Board: AM572x IDK REV 1.3B
>
> does not boot from MMC anymore:
Thanks for the feedback. It's really weird. Before my change, the
sdhci-omap driver was setting:
host->mmc_host_ops.get_ro = mmc_gpio_get_ro;
But thanks to another change I did, the core is now doing:
else if (host->ops->get_ro)
is_readonly = host->ops->get_ro(host);
+ else if (mmc_can_gpio_ro(host->mmc))
+ is_readonly = mmc_gpio_get_ro(host->mmc);
else
is_readonly = !(sdhci_readl(host, SDHCI_PRESENT_STATE)
& SDHCI_WRITE_PROTECT);
So basically, if there's no host->ops->get_ro, it calls
mmc_gpio_get_ro(), which should do the exact same thing.
The only difference that I can see is that it only calls
mmc_gpio_get_ro() if mmc_can_gpio_ro() returns true. But looking at the
code of mmc_can_gpio_ro(), I don't see how it can be borked, and
mmc_gpio_get_ro() be working.
So there is something odd going on here. Do you have the time to add
some traces in sdhci_check_ro(), to see what is the return value of
mmc_can_gpio_ro(), whether we call mmc_gpio_get_ro(), and what does it
return ?
This would help debugging his issue. Thanks a lot!
Best regards,
Thomas
--
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
