Re: MiNT and Milan

[Konrad Kokoszkiewicz <draco@piwo.bl.pg.gda.pl>]

> There is only one thing I know about, and I already mailed it to  this
> list asking to include it. It is concerning the cache-handling of Mint:
> in cpu.spp there is a check for cpu=68040 because then caches are handled
> different. This needs to be changed to cpu>=68040 and everything works
> fine, even copyback-cache is running.

Yes, this is applied in pl 7.

> concerning the serial stuff which started this discussion thread:
> Mint definitely accesses the hw directly. try grep <address of
> controlregister A of scc> to the mint-source tree. There are three
> files you should find then two of them are: bios.c and xbios.c I don`t
> remember the third one. I reported this bug a while ago and asked to
> include a solution I worked out for this.
> The reason why it does not crash on hades is that hades-escc is on the
> same adresses as tt-scc.

xbios.c accesses the hardware in Rsconf() apparently in order to
workaround some bug in TOS. That's the only place where xbios.c does this
(i.e. it _was_ the only place... before I inserted there the Sync-error
patch for Videl, works fine, by the way). 

bios.c also does that, right (i am looking to the sources right now).
Besides, as of version 1.14.6 the kernel also accesses MFP when
Ssystem(TIOCMGET,...) is called. I just moved this part from MiNT libs
(ioctl.c) to avoid switching to supervisor there. I realize that's rather
a quick hack.

Konrad M.Kokoszkiewicz
mail:draco@bl.pg.gda.pl
http://www.orient.uw.edu.pl/~conradus/

** Quem Iuppiter vult perdere, dementat prius.
*******************************************************
** Kogo Jowisz chce zgubic, temu wpierw rozum odbiera.