PR/57107 CVS commit: [netbsd-10] src/sys/arch/m68k/m68k

["Martin Husemann" <martin%netbsd.org@localhost>]

The following reply was made to PR port-mac68k/57107; it has been noted by GNATS.

From: "Martin Husemann" <martin%netbsd.org@localhost>
To: gnats-bugs%gnats.NetBSD.org@localhost
Cc: 
Subject: PR/57107 CVS commit: [netbsd-10] src/sys/arch/m68k/m68k
Date: Sat, 7 Jan 2023 14:56:44 +0000

 Module Name:	src
 Committed By:	martin
 Date:		Sat Jan  7 14:56:44 UTC 2023
 
 Modified Files:
 	src/sys/arch/m68k/m68k [netbsd-10]: bus_dma.c
 
 Log Message:
 Pull up following revision(s) (requested by tsutsui in ticket #39):
 
 	sys/arch/m68k/m68k/bus_dma.c: revision 1.39
 
 Fix out of bounds invalidate (and writeback) in bus_dmamap_sync(9) ops.
 
 Detected by the POOL_REDZONE check in sys/kern/subr_pool.c that
 has been activated if options DIAGNOSTIC is enabled on post netbsd-9.
 
 The extra invalidate on DMASYNC_PREREAD op discards redzone pattern
 data allocated right after an mbuf cluster without proper writeback
 to memory so that it triggers false redzone assertions on freeing mbufs.
 
 This bug was my botch in rev 1.25 committed 15 years ago. (sigh)
 
 Fixes PR/57107 (kernel panic on -current when configuring network
 with sn(4) on mac68k), as actually the bus_dma(9) op changes
 in the past days were introduced for mac68k sn(4) improvements
 by using the MI SONIC (src/sys/dev/ic/dp83932.c) driver.
 
 https://mail-index.netbsd.org/port-mac68k/2007/06/01/0001.html
 
 
 To generate a diff of this commit:
 cvs rdiff -u -r1.38 -r1.38.4.1 src/sys/arch/m68k/m68k/bus_dma.c
 
 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.