Hi,
On 25/02/2011 20:31, Arnaud Patard (Rtp) wrote:
Eric Benard<eric@xxxxxxxxxx> writes:
may you please test the attached patch. It may give someone with a
better knowledge of sdhci than me an idea of what is wrong.
I've tested this patch on my efikamx and this patch does solve the
issue. I didn't test on the efika smartbook but I guess it'll be fine
here too. Thanks.
good. Now we have to find can how this code be integrated properly in the
sdhci driver.
Here are the workaround this patch add :
- we can't let enable or disable irq enabled when the card is
present/not present, else the irq triger again which explains why you
get the freeze -> so we must rely on the card presence bit to enable
the right interrupt,
so, we're getting an interrupt storm, right ? can't it be fixed by
setting a different irq type ?
no this seems to be the way the SDHCI works, at least the i.MX51 ref manual says :
- When the CRM bit is cleared, if no card is inserted it is immediately set
again: this can be prevented by clearing the card removal status enable bit in
interrupt status enable register.
- When the CIN bit is cleared, if a card is inserted it is immediately set
again: this can be prevented by clearing the card inserted status enable bit
in interrupt status enable register.
As, unless I'm mistaken, sdhci host actually consider card detect as broken by
default, I think this code in not actually used so it may not be really tested.
- we can't turn the clock off if we want the card detect to work when
the card is removed -> as a quick workaround this patch prevents
sdhci_set_clock from turning off the clocks when the
SDHCI_INT_CARD_INSERT interrupt is enabled.
Also, I had to change the MX51_PAD_GPIO1_0__SD1_CD pad setting as
follows to enable the internal pull up :
_MX51_PAD_GPIO1_0__SD1_CD | MUX_PAD_CTRL(PAD_CTL_PUS_22K_UP |
PAD_CTL_PKE | PAD_CTL_SRE_FAST |
PAD_CTL_DSE_HIGH | PAD_CTL_PUE | PAD_CTL_HYS),
It worked without changing this.
this could mean you have an external pull up resistor on the board which was
not the case on the prototype I was working on which is partially mounted ;-)
Eric
--
To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
