Hello. Bartlomiej Zolnierkiewicz wrote:
Index: linux-2.6/drivers/ide/pci/cmd64x.c =================================================================== --- linux-2.6.orig/drivers/ide/pci/cmd64x.c +++ linux-2.6/drivers/ide/pci/cmd64x.c
While this was always incorrectly setting PIO4, the PIO4 is "the usual" case and for this driver we need to program PIO explicitly even when using DMA.
Hm, why it's *so* special, i.e. why almost all the other drivers can get away without it (the majority seems to have autotune set *only* if hwif->dma_base is seen as 0 in the init_hwif() method? :-/
The core code doesn't program PIO mode unless told to (->autotune flag == 1) so after the above change PIO mode won't be programmed et all.
I think that we now need to set ->autotune unconditionally in init_hwif_cmd64x().
Don't think we *need* to. Look at the code at the end of init_chipset() method. Those values it writes to ARTTIM/DRWTIM registers already matches PIO4! That's another question what this code is doing there, being both duplicate and misplaced. :-)
For me it looks like a bunch of hacks to get things working on BIOS/firmware-less and non-x86 systems. :)
[ This code pre-dates PIO4 forcing in config_cmd64x_chipset_for_pio(). ]
The "#ifdef __i386__" part looked the most mysterous: why anyone would want to program less *IDE* address setup on x86 than on non-x86? :-O I was going to remove all that crap as well but decided to defer it, mostly because of that part...
Bart
MBR, Sergei - To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
