Subject: kern/24455: current hangs when running mv
To: None <gnats-bugs@gnats.netbsd.org>
From: None <milos@openbsd.cz>
List: netbsd-bugs
Date: 02/17/2004 13:40:05
>Number:         24455
>Category:       kern
>Synopsis:       current hangs when running mv
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Feb 17 12:40:00 UTC 2004
>Closed-Date:
>Last-Modified:
>Originator:     Milos Urbanek
>Release:        NetBSD 1.6ZJ
>Organization:
>Environment:
System: NetBSD oakland 1.6ZJ NetBSD 1.6ZJ (GENERIC) #0: Thu Feb 12 12:35:00 MET 2004  milos@oakland:/export/crash/usr/src/current_base/sys/arch/i386/compile/obj/GENERIC

Architecture: i386
Machine: i386
>Description:
	Current panics when running command "mv file1 file2". file1 is
	located on the smbfs filesystem, file2 is on the ffs partition.

	The only usefull information I can provide is a manually rewritten
	ddb backtrace (current 1.6ZJ):

	kernel: page fault trap, code = 0
	Stopped in pid 567.1 (mv) at netbsd: bgetvp +0x16: movl 0x3c(%ecx),%eax
	bgetvp(0, c0fa8a9c, 0, 0, 2) at netbsd: bgetvp + 0x16
	getblk(0, f36530, 0, 2000, 0) at netbsd: getblk + 0x1fa
	bread(0, f36530, 0, 2000, ffffffff) at netbsd: bread +0x37
	ffs_nodealloccg(c75962c0, f8, 1d86b2, 0, 81ed) at netbsd: ffs_nodealloccg + 0xcb
	ffs_hashalloc(c75962c0, f8, 1d86b2, 0, 81ed) at netbsd: ffs_hashalloc +0x23
	ffs_valloc(c7607be4, c7593c4c, c7607c1c, c0eb1a00, c05a4640) at netbsd: ffs_valloc + 0xbc
	VOP_VALLOC(c7593c4c, 81ed, c0fb1880, c7607c30, c05a3e40) at netbsd: VOP_VALLOC + 0x34
	ufs_makeinode(81ed, c7593c4c, c7607ec4, c7607ed8, 0) at netbsd: ufs_makeinode + 0x48
	ufs_create(c7607d494, 40308259, c7607dcc, c03912c8, c05a3d00) at netbsd: ufs_create +0x26
	VOP_CREATE(c7593c4c, c7607ec4, c7607ed8, c7607df4, c7593c4c) at netbsd: VOP_CREATE +0x34
	vn_open(c7390c64, c7607f64, c6707f5c, 0, c7607fa4) at netbsd: sys_open +0xbc
	syscall_plain(c7607fa8, 1f,1f,1f, 1f) at netbsd: syscall_plain +0x7e


	I initialy found the problem when running current 1.6ZE, for
	current 1.6ZE the backtrace is somehow different:

	kernel: page fault trap, code = 0
	Stopped in pid 1557.1 (mv) at netbsd: bgetvp +0x16: movl 0x3c(%ecx),%eax
	bgetvp(0, c5dfc930, 0, f368e0, f368e0) at netbsd: bgetvp + 0x16
	getblk(0, f368e0, 0, 2000, 0) at netbsd: getblk + 0x1df
	bread(0, f368e0, 0, 2000, ffffffff) at netbsd: bread +0x37
	ffs_update(d413ed84, d418e4fc, 0,0, c0563940) at netbsd: ffs_update +0x2af
	VOP_UPDATE(d4291230, 0, 0, 0, 0,) at netbsd: VOP_UPDATE +0x34
	ufs_inactive(d413edf4, 40307d4e, d413ee0c, c037c689, c056358) at netbsd:ufs_inactive + 0xc7
	VOP_INACTIVE(d4291230, d418e4fc, de13ef1c, c037b0e2, 12) at netbsd: VOP_INACTIVE +0x28
	vrele(d4291230, d413ee88, c0e9ab80, 2, d3c0f1ec) at netbsd: vrele +0x92
	rename_files(bfbffc4b, bfbffc7a, d418e4fc, 0, 0) at netbsd: rename_files +0x1de
	sys_rename(d417o448, d413ef64, d413ef5c, 0, d413efa4) at netbsd: sys_rename + 0x1b
	syscall_plain(d413efa8, 1f, 1f, 1f, 1f) at netbsd: syscall_plain +0x7e
	
	
>How-To-Repeat:
	mount smbfs share from somewhere. try to move some file to your
	local fs.
>Fix:
	no idea
>Release-Note:
>Audit-Trail:
>Unformatted: