kern/38886: hanging processes with make -j 2

[dieter.NetBSD%pandora.be@localhost]

>Number:         38886
>Category:       kern
>Synopsis:       hanging processes with make -j 2
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Jun 07 20:00:00 +0000 2008
>Originator:     dieter roelants
>Release:        NetBSD 4.99.64
>Organization:
>Environment:
System: NetBSD simult.amelgem.be 4.99.64 NetBSD 4.99.64 (SIMULt) #4: Sat Jun 7 
18:19:43 CEST 2008 
dieter%simult.amelgem.be@localhost:/build/obj.i386.current/sys/arch/i386/compile/SIMULt
 i386
Architecture: i386
Machine: i386
>Description:
        Since about a week, when I upgraded from 4.99.58 to 4.99.64,
        I get stalled builds when invoking build.sh with -j 2 (or
        in pkgsrc with MAKEJOBS set). This is still the case with
        a kernel from todays src. The hanging processes look like
        this:

 UID   PID  PPID   CPU PRI NI  VSZ   RSS WCHAN    STAT TTY       TIME COMMAND
1980 21716     1 36864 127  0  748  3624 tstile   D    ttyE1- 0:00.37 
/build/obj.i386.current/tooldir.NetBSD-4.99.64-i386/bin/nbmake _THISDIR_ 
install 
1980 29215     1 36864 103  0  756   956 biowait  D    ttyE1- 0:00.01 
/build/obj.i386.current/tooldir.NetBSD-4.99.64-i386/bin/i386--netbsdelf-install 
-U -M /build/obj.i386.current/destdir.i386/METALOG -D 
/build/obj.i386.current/destdir.i386 -h sha1 -N /usr/src/etc -l h -r 
/build/obj.i386.current/destdir.i386/usr/share/man/man3/event.3 
/build/obj.i386.current/destdir.i386/usr/share/man/man3/evbuffer_add_buffer.3 
(i386--netbsdelf-)

trace: pid 21716 lid 1 at 0xced4f84c
sleepq_block(0,0,c0524107,c056c59c,d48b1f08,cf6acbec,ced4f8ac,cee7dd84,67,40) 
at netbsd:sleepq_block+0x102
turnstile_block(0,1,d48b1f08,c056c59c,d48e1cb0,d48a5e50,ced4f8dc,c2d6e118,d48a5e50,0)
 at netbsd:turnstile_block+0x1e1
rw_vector_enter(d48b1f08,1,ced4f95c,c034da6d,d48b1f08,d48b1e68,ced4f97c,c026ba01,d48b1f08,2)
 at netbsd:rw_vector_enter+0x281
vlockmgr(d48b1f08,2,0,d48b1e68,d48b1e68,2,ced4f9ac,c0351490,ced4f99c,cc8aadc0) 
at netbsd:vlockmgr+0xee
ffs_lock(ced4f99c,cc8aadc0,ce357bc8,0,c034079a,cf6ac000,c0492740,d48b1e68,2,2) 
at netbsd:ffs_lock+0x41
VOP_LOCK(d48b1e68,2,0,cf3cf100,cf3cf100,d48b1e68,ced4f9fc,c03408b7,d48b1e68,2) 
at netbsd:VOP_LOCK+0x60
vn_lock(d48b1e68,2,ced4f9ec,c2d6e118,c05cbf94,c0492140,cc8aadc0,ced4fc38,ce137540,ced4fc38)
 at netbsd:vn_lock+0x78
cache_lookup(d48a5db0,ced4fc24,ced4fc38,0,0,c05cbf94,0,c02e5144,0,0) at 
netbsd:cache_lookup+0x207
ufs_lookup(ced4fae8,d4478adc,ced4fafc,c0345a6d,d4478ac8,c0492000,d48a5db0,ced4fc24,ced4fc38,d02dfc37)
 at netbsd:ufs_lookup+0xc9
VOP_LOOKUP(d48a5db0,ced4fc24,ced4fc38,c0351490,ced4fb2c,cf6ac000,0,18,20,0) at 
netbsd:VOP_LOOKUP+0x6c
lookup(ced4fc10,20002,400,ced4fc2c,c05a9860,cf6ac00c,8,c0299b29,ced4fc2c,1) at 
netbsd:lookup+0x1fc
namei(ced4fc10,cee7baa4,ced4fc2c,c02d1f02,ce137540,cd6f7d00,ffffffff,bb75b420,0,0)
 at netbsd:namei+0x11d
do_sys_stat(bb75b420,40,ced4fc80,c02e4c05,8,c02d4af1,0,d3464cc0,d3464cc0,cef7b440)
 at netbsd:do_sys_stat+0x37
sys___stat30(cf6ac000,ced4fd00,ced4fd28,bb75b420,bfbfdf38,0,0,10,80010,2) at 
netbsd:sys___stat30+0x2c
syscall(ced4fd48,b3,ab,1f,1f,bb9c6a20,bb75b420,bfbfdfa8,0,e) at 
netbsd:syscall+0xa0

trace: pid 29215 lid 1 at 0xceafba1c
sleepq_block(0,0,c05271f8,c056b950,d48b1e68,0,c05c1abc,c30614ec,c306157c,d6dbf964)
 at netbsd:sleepq_block+0x102
cv_wait(c306157c,d48b1e68,c0492800,d48b1e68,c30614ec,d48b1e68,ceafbaac,c033ea89,c30614ec,c30614ec)
 at netbsd:cv_wait+0xcd
biowait(c30614ec,c30614ec,ce137540,1000000,c30614ec,d48af200,ceafbadc,c035236c,c30614ec,1)
 at netbsd:biowait+0x42
bwrite(c30614ec,1,6000,0,ceafbca0,c30614ec,ceafbafc,c0491fc0,c30614ec,14) at 
netbsd:bwrite+0x109
VOP_BWRITE(c30614ec,5400,0,ceafbb0c,ceafbb08,1,ceafbb1c,c03429e2,d48b1e68,c30614ec)
 at netbsd:VOP_BWRITE+0x6c
ufs_dirremove(d48b1e68,d6dbf964,840c,0,0,cf6acbe0,ceafbb5c,c034e878,cd6e2c38,ce363a18)
 at netbsd:ufs_dirremove+0x16c
ufs_rename(ceafbbcc,c056ada8,ceafbbec,c0351426,ceafbbdc,8418,c0492500,d48b1e68,d6dbe7f8,ceafbca0)
 at netbsd:ufs_rename+0xd1e
VOP_RENAME(d48b1e68,d6dbe7f8,ceafbca0,d48b1e68,0,ceafbc58,0,cdf195b0,0,1) at 
netbsd:VOP_RENAME+0x7c
do_sys_rename(bfbfca4c,bfbffb6b,0,1,c056a16c,cf6acbe0,ceafbd3c,c03a7c70,cf6acbe0,ceafbd00)
 at netbsd:do_sys_rename+0x3c5
sys___posix_rename(cf6acbe0,ceafbd00,ceafbd28,bfbfca4c,bfbffb6b,0,0,2,0,1) at 
netbsd:sys___posix_rename+0x29
syscall(ceafbd48,b3,ab,1f,1f,bfbffb6b,bfbfca4c,bfbfce58,0,0) at 
netbsd:syscall+0xa0

        Also, FWIW, the processes don't seem related (in the sense
        that one is the parent of the other), because their CWDs
        are lib/libcurses and lib/libevent.

        /build is /dev/wd0p type ffs (noatime, local)
        Please let me know what more information I need to provide...
>How-To-Repeat:
        In my case;
        cd /usr/src; ./build.sh -j2 -O /build/obj.i386.current -U distribution
>Fix:
        ?