Re: [Emutos-devel] Blue demo regression with EmuTOS v0.8.7

[Roger B. <rfb...@ym...>]

On 24 Jun 2014 at 10:31, Vincent Rivière wrote:

> >> On keskiviikko 01 elokuu 2012, Thomas Huth wrote:
> >>> Now that's an ugly problem ... seems like the the demo calls Ikbdws()
> >>> while the stack A7 points to a very, very low address in RAM (0x22 when
> >>> I've got it right). With EmuTOS 0.8.6, there was still enough space to
> >>> survive this XBIOS call, but with recent EmuTOS versions, Ikbdws() used
> >>> a little bit of more stack space, so it crashed very badly.
> 
> I wonder how the stack could go down to 0x22.
> 
> I wonder:
> - if this is because EmuTOS uses too much stack
> - if some EmuTOS bug (null pointer somewhere?) causes the stack to go so low
> - if the Blue Demo should be categorized as badly written software
> 
Another possibility (arguably the same as #3) is that the Blue Demo reads into 
a memory address that is free under some version(s) of TOS but not EmuTOS, 
thereby overwriting some internal TOS variable and leading to the strange stack 
address.

Roger