Re: DEC PWS500au (MiataGL) EIDE Issues
Jason Thorpe <[email protected]>
| Newsgroups | gmane.os.netbsd.ports.alpha |
|---|---|
| Message-ID | <[email protected]> |
I’ll poke around … -- thorpej Sent from my iPhone. > On May 16, 2021, at 5:36 AM, Rhialto <[email protected]> wrote: > > On Sat 15 May 2021 at 21:27:33 -0400, Alexander Jacocks wrote: >> All, >> >> I?m trying to install NetBSD 9.1 on my Personal Workstation 500au >> (MiataGL), and I?m running into EIDE issues. Whenever I try to access any >> attached EIDE disk (or SATA disk, with a SATA/IDE bridge) I get the >> following: >> >> wd0c: device timeout reading fsbn 9 (wd0 bn 0; cn 0 tn 0 sn 0), xfer 38, >> retry 0 >> cypide0:0:0: geometry timed out >> >> This happens across a bunch of disks, with several different IDE cables, >> both 40 and 80 conductor. >> >> Does anyone have any ideas? > > This sounds very similar to a known issue with DMA not being able to > cross some boundaries (64 K?) in the Miata's chipset. > > A simple solution is to turn off DMA for the built-in IDE. > At some point I plugged in a separate PCI IDE card to get around it in a > different way. There was this in my kernel config for NetBSD 1.6: > > # ST506/ESDI/IDE devices > wd* at pciide0 channel ? flags 0xffc # Use PIO mode 4, no DMA > wd* at pciide? channel ? # Allow DMA on IDE cards > > or this from 2.0: > > # ATA (IDE) bus support > # atabus0 at ata0 > atabus0 at cmdide0 channel 0 > atabus1 at cmdide0 channel 1 > atabus* at ata? > > # IDE drives > wd* at atabus0 drive ? flags 0xffc # Use PIO mode 4, no DMA > wd* at atabus1 drive ? flags 0xffc # Use PIO mode 4, no DMA > wd* at atabus? drive ? # Allow DMA on IDE cards > > That's the latest version I have on my Alpha though (and currently it is > unused in storage). > > I do notice your message mentions "cypide0" and not "cmdide0" so maybe > it's a different issue. > > -Olaf. > -- > ___ Q: "What's an anagram of Banach-Tarski?" -- Olaf "Rhialto" Seibert > \X/ A: "Banach-Tarski Banach-Tarski." -- rhialto at falu dot nl