[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MiNT] Problem with mint 1.16 installation

Hi! :)

Odd Skancke wrote:

 Hello,

fre, 10,.06.2005 kl. 14.34 -0700, skrev Standa Opichal:

 Agreed it is a hack. Agreed this is not a good idea. Agreed this is not
a thing the AES should do by design. I have been saying that all along,
right?


ok. Let's make the approach description clear then...


 Now, what choice do I have? Altho the AES is not the place to do this,
XaAES is a kernel module, and have access to the right API's so it can
make sure CPU caches are turned off before calling v_opnwk() then
restored immediately after. This is the best solution given the
situation. And since Ssystem() is a well documented system call, this
will work on any systems/architectures. The ideal situation would be for
such functionality to be available when the driver runs.
So the real solution you ET whatever problem is to fix/wrap the HW driver to do what is necessary. If there is no API for the particular thing (a VDI card video driver in this case) that the closest documented API level is to be fixed IMHO (the VDI, right?). 


 This is a problem most Hades owners (and all versions of the Nova VDI,
mach64, et6000 and Rage) suffers from, not just me. I did patch the et6k
driver to get rid of this problem, and it worked fine. However, patching
the video-driver like that will make it work only on 060 and compatible.
And what is wrong with having it only 060 compatible? In my eyes this is definitely not an excuse to hack AES in such a way....
That is just like the older NOVA HW driver on my friends TT doesn't have the VDI scrninfo() implemented. Would I patch HighWire to not to use this function or would I rather go and implement the VDI function for the single TT I have problem with? 

 I do not understand what you are saying here? Please explain?
What would you do if a particular VDI driver you use (IIRC Matrix TT card) doesn't implement vq_scrninfo() method? 
  1) hack every application that is using that to not to
  2) create AUTO TSR application that would hook on this and imlement
    the function?

I think that the only way to go is 2).
If you think I added this for my own amusement,
No, I would never even think about that. I just do not like hacks in new apps because of some other SW imperfections. TSR seems to be the right solution for me in this case. And so far I cannot think of any better. 

Best Regards

STanda