[Openchrome-users] CN700 stability

[Hindrik Hettema]

Christiaan van Dijk wrote:
> Hi,
> 
> [...]

> Main problem remaining is the handling of concurrent DMA transfers. I 
> first blamed the Openchrome driver (for which my apologies :-) ) but the 
> real problem is caused by the chipset/initial settings from the BIOS. I 
> recently reported the problem to Jetway and got a reply saying they 
> would look into it (reply within some days =-O ).
> Right now the system is running with only one DMA device enabled (VGA) 
> and is stable as far I have tested.
> 
> [...]
> I do not know if the VIA CN700 boards have the same problem with 
> concurrent DMA, for the VT8235 there were DMA problems for a very long 
> time, a BIOS patch solved  (??)  the problem.
> 

Hi there,

Same problem here - after reading your emails I did some experimenting 
in this way and found the following:

     + Disabling only DRI for X11 did not solve the problem - although
       it reduced the chance of hangup's. Number of freezes estimated
       once a day on normal desktop use.
     + Disabling only DMA for my drives did not solve the problem,
       although it greatly reduced the number of freezes (I noticed many
       times the system would hang while accessing the harddisk) - on
       use of the system (openoffice, internet) once every 3 days.
     + Disabling ethernet optimizations in the kernel (MMIO) did not
       solve the problem - this to test whether it was related to
       VIA-RHINE problems experienced with the EPIA M10000.
     + Changing between kernel versions (2.6.18 vs 2.6.19) had no effect
       (did not expect it though, but one has to try).

I am currently thinking of disabling both X11 and harddisk DMA, are 
there any extra DMA channels I should disable to get the system stable?

Hindrik

PS: I own an upgraded Hush system, I would have to lift the cover to 
check the motherboard. If you have any updates, please keep the list posted.



>[...]