Re: [patch 1/5] MMC OMAP driver

[Anderson Briglia <anderson.briglia@indt.org.br>]

Tony Lindgren wrote:
> * Russell King <rmk+lkml@arm.linux.org.uk> [060201 04:44]:
>>>+static inline int is_broken_card(struct mmc_card *card)
>>>+{
>>>+	int i;
>>>+	struct mmc_cid *c = &card->cid;
>>>+	static const struct broken_card_cid {
>>>+		unsigned int manfid;
>>>+		char prod_name[8];
>>>+		unsigned char hwrev;
>>>+		unsigned char fwrev;
>>>+	} broken_cards[] = {
>>>+		{ 0x00150000, "\x30\x30\x30\x30\x30\x30\x15\x00", 0x06, 0x03 },
>>>+	};
>>>+
>>>+	for (i = 0; i < sizeof(broken_cards)/sizeof(broken_cards[0]); i++) {
>>>+		const struct broken_card_cid *b = broken_cards + i;
>>>+
>>>+		if (b->manfid != c->manfid)
>>>+			continue;
>>>+		if (memcmp(b->prod_name, c->prod_name, sizeof(b->prod_name)) != 0)
>>>+			continue;
>>>+		if (b->hwrev != c->hwrev || b->fwrev != c->fwrev)
>>>+			continue;
>>>+		return 1;
>>>+	}
>>>+	return 0;
>>>+}
>>
>>I've already mentioned this to the OMAP folk... What problem is this
>>trying to work around?  If it's a card problem, it's at the wrong
>>level.  If it's a problem with the host not waiting the mandatory
>>80 cycles before starting a command, that could be the upper layers
>>or a host problem.
>>
>>Either way, the right place to fix this is _not_ in the request
>>function but in the set_ios function.  The request function does
>>not know if the card has just been powered up.
> 
> 
> Anderson, can you pull out the broken card check from omap.c, and put
> it into a separate patch? Let's fix the omap.c issues first, and have
> that integrated. Then we can start working on the additional patches
> and test them one at a time.
> 

Ok. It's already done. I'll wait the omap clock framework fix to post another patch for
omap.c, ok?

Regards,

Anderson Briglia
INdT - Manaus - Brazil