[hatari-devel] Hatari asserts that should be fixed for the release

npomarede at corp.free.fr npomarede at corp.free.fr
Mon Apr 19 00:19:04 CEST 2010


On Mon, 19 Apr 2010, Eero Tamminen wrote:

> Hi,
>
> On Sunday 18 April 2010, npomarede at corp.free.fr wrote:
>>>>> 1) Run the EPSS STE music demo.
>>>>
>>>> Hello,
>>>>
>>>> what is the pouet link for this one ?
>>>
>>> http://www.pouet.net/prod.php?which=30912
>>>
>>> (it's in compat list)
>>
>> OK, I thought it was another one.
>>
>> I didn't see an assert, but I got some graphics corruption in the lower
>> border scrolling (happens if you start the demo and do nothing, no fast
>> forward). This happened nearly at the beggining or when the bottom
>> scroller goes in mirror mode (2-3 mins)
>>
>> But in fact, this was caused by the dma mixing code and applying david's
>> latest patch (which I just committed) fixed the problem in my case.
>
> Hm.  When I tested Mudflap, it gave some warnings about arrays it cannot
> track because their sizes aren't declared (in headers).  Was this for one of
> them?

I don't really know, I haven't looked at all the potential problems the 
previous version could cause, but seeing the result I had it was clearly 
overwritting some other variables/array of Hatari's memory.

>> In case assert is still there, can you tell after how many time it happens
>> (couting at the start of the demo, after depacking) ?
>
> I've been able to reproduce it each time (4/4) just by waiting enough.
>
> The issue happens once the demo has loaded and you leave it running for
> maybe 5 minutes. It's after the bottom scroller goes to mirror mode.
>
> Unlike the mono issue, this happens without fast-forward.
>
> I'm compiling code with -O3 and have borders & sound (44100Hz) enabled,
> in case that matters.
>

OK, I can reproduce the problem (my version was unfortunately compiled 
with assert turned off ... :( ).
I will look in more details during the week.

Nicolas




More information about the hatari-devel mailing list