Re: kern/55540 (aceride Fails To Downgrade Transfer Mode)

[Martin Husemann <martin%duskware.de@localhost>]

The following reply was made to PR kern/55540; it has been noted by GNATS.

From: Martin Husemann <martin%duskware.de@localhost>
To: Kyra Sable <kyra.sable%mail.ru@localhost>
Cc: gnats-bugs%netbsd.org@localhost
Subject: Re: kern/55540 (aceride Fails To Downgrade Transfer Mode)
Date: Sun, 9 Aug 2020 07:43:02 +0200

 On Sat, Aug 08, 2020 at 11:20:14PM -0400, Kyra Sable wrote:
 > Scarcely "not relevant any more" given NetBSD's intended target audience.
 
 Thanks for hunting this done, but I think you misunderstood the commit log
 message. We certainly still care for this devices still.
 
 I have never seen this exact version of the error / downgrade path, and
 I am typing this on a machine that has one:
 
 aceride0 at pci3 dev 13 function 0: Acer Labs M5229 UDMA IDE Controller (rev. 0xc4)
 aceride0: bus-master DMA support present
 aceride0: using PIO transfers above 137GB as workaround for 48bit DMA access bug, expect reduced performance
 aceride0: primary channel configured to native-PCI mode
 aceride0: using ivec 1f98 for native-PCI interrupt
 [..]
 cd0 at atapibus0 drive 0: <JLMS XJ-HD166S, , D3S4> cdrom removable
 cd0: drive supports PIO mode 4, DMA mode 2, Ultra-DMA mode 2 (Ultra/33)
 cd0(aceride0:1:0): using PIO mode 4, Ultra-DMA mode 2 (Ultra/33) (using DMA)
 
 and the cd works fine.
 
 BUT: I have a ATA<->SD card converter on that ATA bus that I use to boot the
 machine (all "disks" on SATA that firmware does not know about), and I have
 to disable UDMA on that, otherwise it will not be able write to the "disk".
 
 (but then failure mode looks quite different to what you reported).
 
 I always assumed this is a bug in the firmware of the SD card converter.
 
 I will put another (real) ATA disk in this machine and test, and also check
 if I have more machines with aceride.
 
 This might be a driver bug, or a quirk in combination with certain ATA disks
 and aceride. We need to dig deeper and avoid the original HBA error, or
 recognize the issue and adjust capabilities, so we end up with a working
 setup w/o the error -> degration path.
 
 Martin
 
 P.S.: Jaromir - could this be similar to the cmdide issue we fixed recently?