NetBSD-Bugs archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

PR/57359 CVS commit: src/sys/dev/pci



The following reply was made to PR kern/57359; it has been noted by GNATS.

From: "Izumi Tsutsui" <tsutsui%netbsd.org@localhost>
To: gnats-bugs%gnats.NetBSD.org@localhost
Cc: 
Subject: PR/57359 CVS commit: src/sys/dev/pci
Date: Tue, 6 Jan 2026 15:21:38 +0000

 Module Name:	src
 Committed By:	tsutsui
 Date:		Tue Jan  6 15:21:38 UTC 2026
 
 Modified Files:
 	src/sys/dev/pci: ehci_pci.c
 
 Log Message:
 ehci_pci: Apply AMD SB600/SB700 quirk after enabling the device
 
 With src/sys/dev/pci/pci_map.c rev. 1.40 we now disable I/O or memory
 decoding before probing BARs and restore the command register afterward.
 
 However, after that change, on some SB600-based systems (at least
 my ASRock M3A-UCC) applying the quirk before enabling the device
 made the first EHCI MMIO accesses unreliable, as mentioned in
 PR kern/57359.
 
 Moving the quirk handling (writing the vendor specific SBx00
 workaround register) to after the device is enabled resolves
 issue even with pci_map.c rev. 1.40, and this should not affect
 systems other than SB600 and certain early SB700 revisions.
 
 Should be pullup to netbsd-10 and netbsd-11.
 
 
 To generate a diff of this commit:
 cvs rdiff -u -r1.78 -r1.79 src/sys/dev/pci/ehci_pci.c
 
 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.
 


Home | Main Index | Thread Index | Old Index