Re: [Kernel-discuss] Re: SD problems (on h5450)

From: Radek Podgorny <radek_at_podgorny.cz>
Date: Mon, 7 Nov 2005 15:04:19 +0100 (CET)

> On Sat, Nov 05, 2005 at 11:03:06PM +0100, Radek Podgorny wrote:
>> Can't there be a problem in:
>>
>> samcop_sdi_card_power: MMC subsystem requesting bogus clock rate 0.
>> Hardcoding
>> prescaler to 255
>>
>> ?
>
> I thought the same thing. But the mmc system does not request a clock
> rate when it is detecting the card. It doesn't determine a clock rate
> until it has found a card and figured out how fast the card can go.

Yup, the card is being detected just fine AFAIK. There only seems to be a
problem with mounting it as root (cannot read it? whatever?)

>> How is it even possible this message is printed several times during the
>> boot?
>> Shouldn't it be printed only once?
>
> This I don't have an answer to. I would expect the mmc system to only
> require one power on for the card, another for the slot itself and then
> a final call to set the clock rate for the card. But it prints out three
> times at least. And since it doesn't print this when the clock rate is
> properly supplied, it means that this function is called 4 times. A
> little excessive. But this is purely mmc mechanics. The samcop driver is
> just doing what the mmc system wants it to do.

Well, can't this be a bug in mmc? It the mmc code vanilla or are there any
hh specific modifications in your tree?

Is there a way to enable something like "kernel call tree" so I can see
what happens?

>> My observations show the boot process stops for several seconds after
>> the
>>
>> NET: Registered protocol family 1
>>
>> and before
>>
>> samcop_sdi_card_power: MMC subsystem requesting bogus clock rate 0.
>> Hardcoding
>> prescaler to 255
>> Kernel panic - not syncing: VFS: Unable to mount root fs on
>> unknown-block(254,2)
>>
>> Looks like samcop code is waiting for something. Any ideas what should I
>> do to
>> get more info on what's going on?
>
> I see something similar when I load the modules. Matt Reimer
> hypothesised that it was the sdi timeout waiting for a large card to
> respond. Which makes a lot of sense. But I have been unable to detect
> the timeout condition when this happens.

Sounds reasonable, we can investigate this later.

> You can get more information if you turn on mmc debugging (configuration
> option before kernel compilation). But you may need to instrument some
> code in drivers/mmc/samcop_sdi.c to get down to the section you are
> interested in.
>

The debugging is already enabled, looks like I'll have to add more debug
info to the code. The problem is I don't know what info may be valuable.
Can you help me with what can be useful for you, please?

Could you post your bootup log (in the root-from-sd config) so I can see
what it's supposed to look like? Thanks...

Radek P.

> E
>
> --
> Erik Hovland
> mail: erik_at_hovland.org
> web: http://hovland.org/
> PGP/GPG public key available on request
> _______________________________________________
> H5400-port mailing list
> H5400-port_at_handhelds.org
> https://www.handhelds.org/mailman/listinfo/h5400-port
>
Received on Mon Nov 07 2005 - 09:10:37 EST

This archive was generated by hypermail 2.2.0 : Mon Nov 07 2005 - 09:10:46 EST