On Mon, 20 Jan 2003, Dave Airlie wrote: > > The single PMAZ-A limitation of the driver still applies, but it should > > be fairly easy to relax in the next step. > > can someone try it on a 5000/200 if they have one also .. it's been a long > while since that code was written ... if it works on a PMAZ it'll probably > work on a 5000/200 anyways.. The "discrete" PMAZ-A is identical to the "integrated" one, so if one works, the other does too. Changes that affect the PMAZ-A are as follows: 1. Use bus (physical) addresses everywhere, except when doing PIO copying between host memory and the board's SRAM. 2. An additional write barrier is added before triggering board's DMA transfer from the board's SRAM (a nop for R3k). So they are minimal enough for a single run-time check to suffice. Anyway, testing a current kernel on the /200 would be a good thing. Maciej -- + Maciej W. Rozycki, Technical University of Gdansk, Poland + +--------------------------------------------------------------+ + e-mail: macro@ds2.pg.gda.pl, PGP key available +
