From: Marcin Bukat <marcin.bukat_at_gmail.com>
Date: Sat, 1 Oct 2011 14:19:56 +0200

2011/10/1 Marcin Bukat <marcin.bukat_at_gmail.com>:
> 2011/10/1 Torne Wuff <torne+rockboxdev_at_wolfpuppy.org.uk>:
>> On 1 October 2011 10:32, Marcin Bukat <marcin.bukat_at_gmail.com> wrote:
>>> Hello folks,
>>> http://www.rockbox.org/tracker/task/12302 adds backtrace log to the
>>> panic screen on ARMs. This is simple port of
>>> http://www.mcternan.me.uk/ArmStackUnwinding/. The approach to take
>>> this backtrace is not perfect and has limitations BUT it doesn't need
>>> any special symbols or something to unwind stack frames. I am aware
>>> that the code itself would need some polish (unneeded typedefs, dos
>>> line ends etc.) but first I am interested if we want something like
>>> this at all. I played a bit with this inserting asm snippet causing
>>> data abort in various places and produced backtrace looked reasonable
>>> - still more tests needed.
>>
>> Thanks for doing this! I'll experiment with it later, but for now I
>> was wondering if you've measured the binsize impact of it? I expect
>> it's small since that's the point of the original library, but I'm
>> curious exactly what it comes to on Rockbox.
>>
>> --
>> Torne Wuff
>> torne_at_wolfpuppy.org.uk
>>
>
> 5596 bytes  (rk27xx build)
>

If the binsize is at premium we can: 1) compile this with -Os (I think
we compile with -O by default on arm) 2) compile this as thumb
Received on 2011-10-01