Benjamin Herrenschmidt <[EMAIL PROTECTED]> writes: > Thanks. Now, I'd like if you could try to get more precise values because > those I sent you are really too high. > Let me first say that >> In function dma_bits_to_command(), comment out the MDMA 2 case gave me Multiword DMA 1 instead of 2, but didn't have any other effect.
> Can you try changing only one of them, and then try decreasing them until > you find the "breaking" point ? wait_for_ready 5000 / IDE_WAKEUP_DELAY_MS 5000 is too short. So the truth should lie somewhere in the middle. I had to notice that even with wait_for_ready 5000 / IDE_WAKEUP_DELAY_MS 10000, which is fine at bootup, the HDD won't spin up at all (which locks the machine), whereas the CD-ROM does. I have purged pmud for now. > Also, to get rid of the interrupt, can you try changing > pmac_ide_do_setfeature() this way: (hand-made diff) I'll give it a try, but as it takes almost an hour to compile a kernel here, it may take some time. Andre
