Subject: Re: de0/pciide0 performance conflict?
To: Manuel Bouyer <bouyer@antioche.lip6.fr>
From: Allen Briggs <briggs@ninthwonder.com>
List: port-i386
Date: 10/04/1999 22:30:04
> Setting wdcdebug_pciide_mask to 0x10 will show it also, for the pciide
> part at last.

Actually, it doesn't appear to.  I've added some code, including some
code to drop the PCI latency down from 0x40 to 0x20 (you said that 32
should be OK?).  This gives somewhat better performance on the ftp test:
around 90KB/s.  Still pretty bad.  It's possible that the BIOS is not
really setting the VIA up properly...  Here's dmesg with
wdcdebug_pciide_mask set to 0x10:

apollo_setup_chip: old APO_IDECONF=0xba09e20a, APO_CTLMISC=0xc00368, APO_DATATIM=0x20a8a8a8, APO_UDMA=0x60030303
apollo_setup_chip: OLD latency timer 0x4000
apollo_setup_chip: OLD Interrupt/gnt/latency=0x0, 0x0
pciide_dma_table_setup: table at 0xf7d82000 len 136, phy 0x430000
pciide_dma_table_setup: phy addr of table 0x430000
wd0(pciide0:0:0): using PIO mode 4, Ultra-DMA mode 2 (using DMA data transfers)
apollo_setup_chip: Latency timer 0x2000
apollo_setup_chip: Interrupt/gnt/latency=0x0, 0x0
apollo_setup_chip: APO_DATATIM=0x20000000, APO_UDMA=0xe0000000
pciide: command/status register=2800007

If it is the BIOS being lame, I expect that it's also the source of my
keyboard/mouse problems.

-allen