port-evbarm/51591: NetBSD 7.0.1/evbarm & USB FS errors

[mail%benjaminlorenz.de@localhost]

>Number:         51591
>Category:       port-evbarm
>Synopsis:       NetBSD 7.0.1/evbarm & USB FS errors
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    port-evbarm-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Nov 01 08:45:00 +0000 2016
>Originator:     Benjamin Lorenz
>Release:        NetBSD 7.0.1
>Organization:
	
>Environment:
	
	
System: NetBSD rpi.pocketservices.de 7.0.1 NetBSD 7.0.1 (RPI) #1: Sat Jun 4 10:37:02 CEST 2016 root%rpi.pocketservices.de@localhost:/usr/obj/sys/arch/evbarm/compile/RPI evbarm
Architecture: earm
Machine: evbarm
>Description:
Compiled wedge support in, using usb stick:

umass0 at uhub1 port 3 configuration 1 interface 0
umass0: Intenso Intenso Micro Line, rev 2.00/1.00, addr 5
umass0: using SCSI over Bulk-Only
scsibus0 at umass0: 2 targets, 1 lun per target
sd0 at scsibus0 target 0 lun 0: <Intenso, Micro Line, 1.00> disk removable
sd0: fabricating a geometry
sd0: 30535 MB, 30535 cyl, 64 head, 32 sec, 512 bytes/sect x 62535680 sectors
sd0: fabricating a geometry
sd0: GPT GUID: 0b604279-6ee1-4d8b-bb77-10a9c63d196a
dk0 at sd0: 438D231D-68DF-450C-B9D7-68F27B839EA
dk0: 61863856 blocks at 409640, type: ffs
boot device: ld0

When copying a large file to the stick, after around 1 GB the following exception is raised, and the RPI reboots:

start = 11788, len = 1, fs = /mnt/stick
offset=3086 3086
cg 117
panic: ffs_alloccg: map corrupted
cpu0: Begin traceback...
0x97d479f4: netbsd:db_panic+0xc
0x97d47a24: netbsd:vpanic+0x134
0x97d47a3c: netbsd:snprintf
0x97d47a8c: netbsd:ffs_mapsearch+0x240
0x97d47ad4: netbsd:ffs_alloccgblk.isra.3+0xb0
0x97d47b2c: netbsd:ffs_alloccg+0x4e4
0x97d47b6c: netbsd:ffs_hashalloc+0x34
0x97d47bc4: netbsd:ffs_alloc+0x1c0
0x97d47cf4: netbsd:ffs_balloc+0x1bf0
0x97d47d54: netbsd:ufs_gop_alloc+0x104
0x97d47e24: netbsd:ffs_write+0x5a8
0x97d47e4c: netbsd:VOP_WRITE+0x38
0x97d47e7c: netbsd:vn_write+0xd4
0x97d47ed4: netbsd:dofilewrite+0x98
0x97d47f04: netbsd:sys_write+0x6c
0x97d47f74: netbsd:syscall+0x8c
0x97d47fac: netbsd:swi_handler+0x94
cpu0: End traceback...

dump to dev 92,1 not possible

rpi% cp -p ~/F2016102* .
Write failed: Broken pipe
[machine was rebooting]
rpi% ls -l /mnt/stick/backup/workspace 
total 1168560
-rw-r--r--  1 bjjl  wheel  1196294144 Oct 30 09:58 F20161022T230000Z.zip
rpi% ls -lh /mnt/stick/backup/workspace
total 1.1G
-rw-r--r--  1 bjjl  wheel  1.1G Oct 30 09:58 F20161022T230000Z.zip

Maybe related to a bug which was fixed earlier on amd64? Id 50324, which 
showed this stack trace (using virtio drives on a cloud server):

ldstart() at netbsd:ldstart+0x6b 
ldstrategy() at netbsd:ldstrategy+0x101 
bdev_strategy() at netbsd:bdev_strategy+0x5b 
spec_strategy() at netbsd:spec_strategy+0x2c 
VOP_STRATEGY() at netbsd:VOP_STRATEGY+0x33 
genfs_do_io() at netbsd:genfs_do_io+0x195 
genfs_gop_write() at netbsd:genfs_gop_write+0x52 
genfs_do_putpages() at netbsd:genfs_do_putpages+0x7ca 
VOP_PUTPAGES() at netbsd:VOP_PUTPAGES+0x3a 
ffs_write() at netbsd:ffs_write+0x354 
VOP_WRITE() at netbsd:VOP_WRITE+0x37 
vn_write() at netbsd:vn_write+0xec 
dofilewrite() at netbsd:dofilewrite+0x97 
sys_write() at netbsd:sys_write+0x5f 
syscall() at netbsd:syscall+0x9a 

>How-To-Repeat:
	
>Fix:
	

>Unformatted: