Problem about the new sdhci driver

[Alexander Motin]

Henry Hu wrote:
> After doing more experiments, I found the cause:
> hw.pci.do_power_resume=3
> If I set it to 0 and load the driver, it works well with and without the card.
> If I set it to 3 and load the driver, then
> 1. If the card is present, then it would detect the card and the disk
> shows up. If I remove the card and insert it again, it cannot detect
> it again. And if I unload the driver now and reload it with the card,
> it cannot detect the card and it cannot detect card insertion /
> removal.
> 2. If the card is not present, then it cannot detect the card, and if
> I remove the card and insert it again, nothing happens.
> The 0x01fa0000 state is normal: I found that the detection of
> write-protection is independent from card detection. If I pull the
> card out completely, the present register is 0x01f20000. If I pull the
> card out half-way, the present register is 0x01fa000, the
> write-protection bit is 1 and the card-present bit is 0...
> 
> So the problem is about power states, not the driver, and no patch is
> needed. Since I rarely use the card reader when the laptop is using
> battery, I decided to remove it from sysctl .conf ....
> Thanks for the help. Good driver!

And still it would be good to try the patch. Some revisions of this 
controller known behave wrong if being reseted without card inserted. 
This patch enables special quirk to avoid that situation. I think power 
management can be not a reason of problem, but just a trigger.

> On Sun, Feb 1, 2009 at 6:08 PM, Alexander Motin <mav at freebsd.org> wrote:
>> Henry Hu wrote:
>>> I'm using FreeBSD 7-STABLE, with sdhci driver
>>> sdhci.20081029.tgz
>>> mmc.20081029.tgz
>>>
>>> I've found that if I load sdhci driver with card inserted, everything
>>> works fine.
>>> But if I load sdhci driver first, and then insert card, the card is
>>> not detected.Additionally, there is no interrupt.
>>> After I've load sdhci driver without card, the controller seems to be
>>> in a problematic state. The register dump outputed when the driver
>>> loaded showed that Present is 0x01f20000 which is correct since the
>>> card is not inserted then. But cards is not detected.
>>> Even if I unload the driver and reload with card inserted, the card
>>> still cannot be detected, and there's no interrupt. But the Present
>>> register changed to 0x01fa0000 which is strange.
>>>
>>> So I guess there's something wrong with the driver when the card is not
>>> present.
>>>
>>> sdhci0: <RICOH R5C822 SD> mem 0xf0401000-0xf04010ff irq 18 at device 9.1
>>> on pci5
>>> sdhci0-slot0: 33MHz 4bits 3.3V DMA
>>> sdhci0-slot0: ============== REGISTER DUMP ==============
>>> sdhci0-slot0: Sys addr: 0x00000000 | Version:  0x00000200
>>> sdhci0-slot0: Blk size: 0x00000000 | Blk cnt:  0x00000000
>>> sdhci0-slot0: Argument: 0x00000000 | Trn mode: 0x00000000
>>> sdhci0-slot0: Present:  0x01f20000 | Host ctl: 0x00000000
>>> sdhci0-slot0: Power:    0x00000000 | Blk gap:  0x00000000
>>> sdhci0-slot0: Wake-up:  0x00000000 | Clock:    0x00000000
>>> sdhci0-slot0: Timeout:  0x0000000c | Int stat: 0x00000000
>>> sdhci0-slot0: Int enab: 0x01ff00fb | Sig enab: 0x01ff00fb
>>> sdhci0-slot0: AC12 err: 0x00000000 | Slot int: 0x00000000
>>> sdhci0-slot0: Caps:     0x01c021a1 | Max curr: 0x00000040
>>> sdhci0-slot0: ===========================================
>> Try this patch please:
>> --- sdhci.c.prev        2009-01-29 00:53:41.000000000 +0200
>> +++ sdhci.c     2009-02-01 11:58:42.000000000 +0200
>> @@ -82,7 +82,8 @@ static const struct sdhci_device {
>>        u_int           quirks;
>>  } sdhci_devices[] = {
>>        { 0x08221180,   0xffff, "RICOH R5C822 SD",
>> -           SDHCI_QUIRK_FORCE_DMA },
>> +           SDHCI_QUIRK_FORCE_DMA |
>> +           SDHCI_QUIRK_NO_CARD_NO_RESET },
>>        { 0x8034104c,   0xffff, "TI XX21/XX11 SD",
>>            SDHCI_QUIRK_FORCE_DMA },
>>        { 0x05501524,   0xffff, "ENE CB712 SD",
>>
>>
>> Could everybody having this controller send me pciconf and Version register
>> from their controller and report me is this bug present on their system? I
>> would like to collect some statistics.
>>
>> --
>> Alexander Motin
>> _______________________________________________
>> freebsd-mobile at freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-mobile
>> To unsubscribe, send any mail to "freebsd-mobile-unsubscribe at freebsd.org"
>>


-- 
Alexander Motin