Re: port-arm/55737: Apparent bug in evbarm64 DMA code causes filesystem corruption

[Richard Todd <rmtodd%servalan.servalan.com@localhost>]

The following reply was made to PR port-arm/55737; it has been noted by GNATS.

From: Richard Todd <rmtodd%servalan.servalan.com@localhost>
To: Nick Hudson <nick.hudson%gmx.co.uk@localhost>
Cc: matthew green <mrg%eterna.com.au@localhost>,  gnats-bugs%netbsd.org@localhost,
  port-arm-maintainer%netbsd.org@localhost,  gnats-admin%netbsd.org@localhost,
  netbsd-bugs%netbsd.org@localhost
Subject: Re: port-arm/55737: Apparent bug in evbarm64 DMA code causes
 filesystem corruption
Date: Fri, 23 Oct 2020 02:04:29 -0500

 Nick Hudson <nick.hudson%gmx.co.uk@localhost> writes:
 >
 > I think this patch is correct and covers the case that your patch
 > doesn't where there are no ranges in the parent tag.
 >
 > I think the virtio patch is still valid, however.
 
 Okay, here's the (somewhat delayed) results from my testing that I
 promised earlier. 
 
 1) The virtio patch you provided earlier (here 
 http://mail-index.netbsd.org/port-arm/2020/10/19/msg007017.html)
 seems to do the trick -- applying that to a recentish -current tree
 (21a63fef3aab date:        Mon Oct 19 09:07:29 2020 +0000, if it matters)
 resulted in a kernel that could withstand three rounds of
 sysupgrade installing kernel and world without a single panic.  So that
 patch looks good.
 
 2) The patch you just gave for bus_dma.c, applied by itself on top of
 the same -current, does *not* seem to fix the problem; we get what looks
 like the same panic at about the same place.