Help request - ASoC recursion issue

[Lee Jones]

On 23/07/12 19:12, Will Deacon wrote:
> On Mon, Jul 23, 2012 at 03:56:02PM +0100, Lee Jones wrote:
>> On 23/07/12 15:50, Lee Jones wrote:
>>>> I was wondering if I may bother you for some help. I've been having
>>>> serious issues with testing the new mop500 sound system you have in your
>>>> ASoC for-next branch. I've fixed a few issues and will be submitting
>>>> patches shortly. The most serious issue I came across was with
>>>> recursion.
>>>
>>> It appears you can ignore this. I just tried another piece of hardware
>>> and this problem doesn't arise, leading me to believe I have some kind
>>> of eMMC corruption error going on with one of my boards.
>>
>> FYI, just so you don't think I'm mad: http://paste.ubuntu.com/1106535/
>
> I still maintain that you're getting memory corruption as a result of your
> linear mapping getting screwed up:
>
>    Truncating RAM at 18000000-3fffffff to -2c3fffff (vmalloc region overlap).
>
> looks wrong to me. Your command line also has some non-standard mem= options
> that I don't understand:

Yes, I am inclined to agree with you. I just thought it was 
uncharacteristic of a memory corruption error, as can reliably reproduce 
the bootlog above. What's even more strange is that the issue only 
arises on one of my two _identical_ development boards.

>    vmalloc=300M mem=128M at 0 mali.mali_mem=64M at 128M hwmem=168M at 192M mem=22M at 360M mem_issw=1M at 383M mem=640M at 384M
>
> I'd guess you need highmem to map that lot, but you don't appear to have that
> enabled and you end up with a rather strange amount of reported memory to go
> with your lowmem mapping:
>
>    Memory: 474156k/474156k available, 11220k reserved, 0K highmem
>    Virtual kernel memory layout:
>        vector  : 0xffff0000 - 0xffff1000   (   4 kB)
>        fixmap  : 0xfff00000 - 0xfffe0000   ( 896 kB)
>        vmalloc : 0xec800000 - 0xff000000   ( 296 MB)
>        lowmem  : 0xc0000000 - 0xec400000   ( 708 MB)

Enabling HIGHMEM is actually on my TODO list. We have it enabled in our 
release kernel, it's just on in Mainline yet. I'll bring it forward and 
make it the next think I do.

> Now, I don't know a thing about Snowball, so maybe this is all normal for
> that board but if you're seeing memory corruption I'd get to the bottom of
> this before trying to debug instances of it cropping up in random driver
> code :)

I don't have any allocated time to debug issues which only arise on 
Snowball. I'm just waiting on another piece of hardware, then I can 
start testing on the u9540. This issue will be the first thing I test for.

Thanks for your time chaps.

-- 
Lee Jones
Linaro ST-Ericsson Landing Team Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog