trace
nfs_reclaim() at nfs_reclaim+0x150 0xffffffff807f6aa0
vop_reclaim() at vop_reclaim+0x4b 0xffffffff8070b3cb
vgone_vxlocked() at vgone_vxlocked+0x266 0xffffffff806f5cf6
vrecycle() at vrecycle+0x48 0xffffffff806f5f08
nfs_inactive() at nfs_inactive+0xfe 0xffffffff807f694e
vop_inactive() at vop_inactive+0x4b 0xffffffff8070b30b
vrele() at vrele+0x16e 0xffffffff806f6f6e
vn_close() at vn_close+0x2d 0xffffffff807095ad
fdrop() at fdrop+0x20f 0xffffffff806315df
closef() at closef+0x32 0xffffffff80631f22
kern_close() at kern_close+0xef 0xffffffff8063212f
syscall2() at syscall2+0x11e 0xffffffff80bd6a9e
db> 
--- later captures (identical) ---
v2-fix typelie variant (stock mechanism, fixed-v1 kernel):
nfs_reclaim() at nfs_reclaim+0x150
vop_reclaim() at vop_reclaim+0x4b
vgone_vxlocked() at vgone_vxlocked+0x266
vrecycle() at vrecycle+0x48
nfs_inactive() at nfs_inactive+0xfe
vop_inactive() at vop_inactive+0x4b
vrele() at vrele+0x16e
vn_close() at vn_close+0x2d
fdrop() at fdrop+0x20f
closef() at closef+0x32
kern_close() at kern_close+0xef
syscall2() at syscall2+0x11e

--- fix-v2 dirfh rejection-path panic (flaw in the candidate fix, NOT the original bug) ---
panic: lockmgr: LK_RELEASE: no lock held
lockmgr_release() at lockmgr_release+0x11a
vput() at vput+0x11
vop_compat_nremove() at vop_compat_nremove+0xf7
vop_nremove() at vop_nremove+0x97
sys_unlink() at sys_unlink+0x4c
syscall2() at syscall2+0x11e