Re: [MiNT] Xaaes 0.998/0.999 and CTCPI/Radeon

["Helmut Karlowski" <helmut.karlowski@ish.de>]

Am 03.10.2010, 15:13 Uhr, schrieb Didier MEQUIGNON  
<didiermequignon@free.fr>:

> Hi,
>
> Sorry for my poor english.

No problem - we're so good at it we understand everything ;-)

> This week I tried to search why MiNT1.16.3/Xaaes 0.998 (2006 release)  
> AES "freeze" on the Radeon/CTPCI when you are inside the system/log  
> window and you move the slider to the left. CTRL-ALT-DEL continue to  
> works, and I see always periodic calls to vq_key_s.

As I see it this is called on every mouse-move for example (don't know  
what it does though)

> When this problem arrives, there are a VDI call inside another call, and

I guess they are not inside one another but rather that two threads call  
VDI independently. If this is true it's a serious issue, that has to be  
fixed.

> This function was always vq_key_s or vex_motv, now with the Xaaes CVS  
> sources I think understand a little bit, because vex_movt is just called  
> inside k_main.c. Maybe for some reason, k_exit is called.

Yes, during mouse-initialization and at exit.

> This tests was with NVDI, I got white screen with fVDI/radeon.sys during

What is in the boot-log in the fvdi-case?

> I tried XAAESDEB.KM and now all works fine with TOS VDI, NVDI, or
> fVDI/radeon.sys. Very strange, I not understand why the GENERATE_DIAG  
> code added fix this problem.

Maybe the debug-version has another timing due to the additional  
debug-code, and the case of two threads calling VDI simultaneously does  
not happen by chance.

> Note that the VDI works fine with without fVDI / NVDI with MyAES, XAAES  
> with kernel module, MagiC and TOS.

You mean the built-in VDI?

--
Helmut Karlowski