Does this show the expected answer for your MMC?
 

So, we've started running into a problem where the kernel lies to us regarding which MMC is the internal, or the external one, and we would like to quiz you people (N8x0 users, on Maemo) on the outputs of the following commands (non-destructive):

1. Note down which mmcblkX you think is the internal one, and which you think is the external one.

You can possibly use sfdisk -uM -s to see the sizes of the MMC cards in kilobytes.

2. Run these two commands:

# cat /sys/block/mmcblk0/device/../slot_name
# cat /sys/block/mmcblk1/device/../slot_name

3. Did they correspond to what you thought in 1)?

If yes, vote yes and your tablet type, if no, vote no and your tablet type.

Thanks for your cooperation!

Re: Does the kernel lie to us regarding MMC slot_name?
 

The slot names are defined in /arch/arm/mach-omapX/board-nXXXX-XXX.c in a omap_mmc_platform_data struct. So in theory if i switch them and give this kernel to you, you will have switched slots ;)

So the kernel is not lying - it just tells what it is told to. :)

Re: Does the kernel lie to us regarding MMC slot_name?
 

Yeah, I know - except we have a clean N810 with nokia-provided kernel telling us it's internal is external and such - and we're just trying to find out if it's a freak, or not :) (It matters a great deal when wanting to partition internal or external card, that we can figure out which is which)

Re: Does the kernel lie to us regarding MMC slot_name?
 

So try to detect it with slot_name AND size :) Maybe Nokia has some faulty units which have the internal n810 mmc wired to the other controller slot...

Re: Does the kernel lie to us regarding MMC slot_name?
 

inb4 N810 lusers!

(Really, it's odd... I'm the third N800 voter, and you'd think the N810ers would get back quicker with their h/w keyboards -- are there more N800ers around? or are we just faster?)

This message typed on a 4-row on-screen thumbboard.

Re: Does the kernel lie to us regarding MMC slot_name?
 

CRAP. I misread the question and I voted wrongly, and I can't change my vote.

The thread title is, "Does the kernel lie to us...?" and the poll question is, "Did you get the expected result?" I answered the thread title instead of the poll question.

So my vote SHOULD be, YES, I got the expected result on my N800, not NO, my kernel didn't lie to me.

Re: Does the kernel lie to us regarding MMC slot_name?
 

Either the hackers who can be bothered to type two commands into the terminal all use N800s, or there are more N800 users with way too much time on their hands checking the new posts on these forums.

Re: Does the kernel lie to us regarding MMC slot_name?
 

It would have made more sense to skip the mysterious "what do you expect" part and just provide the commands, the proper results, and the poll then. As it is, you're counting on everyone to know how Linux device & Maemo SD slot naming works and therefore, what to expect. Those who don't grok this will either skip your poll or do it, possibly guessing wrong, and provide bogus feedback.

Re: Does the kernel lie to us regarding MMC slot_name?
 

Output was exactly as expected on my n800 running the latest diablo.

Re: Does the kernel lie to us regarding MMC slot_name?
 

So N800s are clean, and N810s apparently are surprisingly rare in this audience, and not yet conclusive.

I seem to remember a similar bug being confirmed previously -- something like the mmcblk[01] files were assigned one way in the initfs and reversed in the rootfs, which may be related to this...