pkg/40622: Build of lang/mono loops on NetBSD/macppc 4.0

[he%NetBSD.org@localhost]

>Number:         40622
>Category:       pkg
>Synopsis:       Build of lang/mono loops on NetBSD/macppc 4.0
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    pkg-manager
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Feb 12 10:50:00 +0000 2009
>Originator:     Havard Eidnes
>Release:        NetBSD 4.0
>Organization:
        I Try...
>Environment:
System: NetBSD malus.urc.uninett.no 4.0 NetBSD 4.0 (GENERIC) #0: Sun Dec 16 
00:27:58 PST 2007  
builds@wb30:/home/builds/ab/netbsd-4-0-RELEASE/macppc/200712160005Z-obj/home/builds/ab/netbsd-4-0-RELEASE/src/sys/arch/macppc/compile/GENERIC
 macppc
Architecture: powerpc
Machine: macppc
>Description:
        It appears that during the build of lang/mono, a mono
        executable is used at some stage in the build process.

        On NetBSD/macppc 4.0, this mono invocation loops, consuming
        100% system CPU, as can be seen here:

load averages:  1.01,  1.02,  1.00                 up 84 days, 23:46   11:35:53
91 processes:  1 runnable, 88 sleeping, 1 stopped, 1 on processor
CPU states:  0.0% user,  0.0% nice,  100% system,  0.0% interrupt,  0.0% idle
Memory: 241M Act, 123M Inact, 4740K Wired, 52M Exec, 291M File, 1632K Free
Swap: 1300M Total, 14M Used, 1286M Free

  PID USERNAME PRI NICE   SIZE   RES STATE      TIME   WCPU    CPU COMMAND
13413 root      97   20    35M 3688K RUN       38.1H 99.02% 99.02% mono

        and in this output from "systat vm":

    5 users    Load  1.01  1.02  1.00                  Thu Feb 12 11:36:07

Proc:r  d  s  w     Csw    Trp    Sys   Int   Sof    Flt      PAGING   SWAPPING
     2    13         42 833838          109    45 833801      in  out   in  out
                                                        ops
 100.0% Sy   0.0% Us   0.0% Ni   0.0% In   0.0% Id    pages
|    |    |    |    |    |    |    |    |    |    |
==================================================                        forks
                                                                          fkppw
           memory totals (in kB)             149 Interrupts               fksvm
          real  virtual     free             100 cpu0 clock               pwait
Active  247132   261364     1320              11 cpu0 soft clock          relck
All     499252   513484  1318284              16 cpu0 soft net            rlkok
                                                 openpic irq 29           noram
Namei         Sys-cache     Proc-cache           openpic irq 39           ndcpy
    Calls     hits    %     hits     %           openpic irq 40           fltcp
        6        6  100                       11 openpic irq 41           zfod
                                              11 gem0 interrupts          cow
Disks:   md0   cd0   wd0                                               64 fmin
 seeks                                                                 85 ftarg
 xfers                                                                    itarg
 bytes                                                               1186 wired
 %busy                                                                    pdfre

        This was experienced during a bulk pkgsrc build, so of course
        the build stalls at this point.

        The process which loops in this manner is running:

13413 ttyp2 RN   2290:10.59 
/usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono --config 
/usr/pkgsrc/lang/mono/work/mono-2.0.1/runtime/etc/mono/config 
./class/lib/monolite/mcs.exe /out:build/deps/basic-profile-check.exe 
build/deps/basic-profile-check.cs

        according to the output from "ps axww".

        A small session with gdb may point to where it's working:

malus# gdb /usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono
GNU gdb 6.5
Copyright (C) 2006 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "powerpc--netbsd"...
(gdb) attach 13413
Attaching to program: /usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono, 
process 13413
0x018c8d0c in mono_jit_info_table_find (domain=0x1a94ea0, 
    addr=0xefd2a0c0 "L\203") at domain.c:406
406             } while (chunk_pos < table->num_chunks);
(gdb) where
#0  0x018c8d0c in mono_jit_info_table_find (domain=0x1a94ea0, 
    addr=0xefd2a0c0 "L\203") at domain.c:406
#1  0x018c8cb8 in mono_jit_info_table_find (domain=<value optimized out>, 
    addr=0xefd2a0c0 "L\203") at domain.c:425
#2  0x018c8cb8 in mono_jit_info_table_find (domain=<value optimized out>, 
    addr=0xefd2a0c0 "L\203") at domain.c:425
Previous frame inner to this frame (corrupt stack?)
(gdb) det
Detaching from program: /usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono, 
process 13413
(gdb) attach 13413
Attaching to program: /usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono, 
process 13413
0x018c8d0c in mono_jit_info_table_find (domain=0x1a94ea0, 
    addr=0xefd2a0c0 "L\203") at domain.c:406
406             } while (chunk_pos < table->num_chunks);
(gdb) where
#0  0x018c8d0c in mono_jit_info_table_find (domain=0x1a94ea0, 
    addr=0xefd2a0c0 "L\203") at domain.c:406
#1  0x018c8cb8 in mono_jit_info_table_find (domain=<value optimized out>, 
    addr=0xefd2a0c0 "L\203") at domain.c:425
#2  0x018c8cb8 in mono_jit_info_table_find (domain=<value optimized out>, 
    addr=0xefd2a0c0 "L\203") at domain.c:425
Previous frame inner to this frame (corrupt stack?)
(gdb) det
Detaching from program: /usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono, 
process 13413
(gdb) attach 13413
Attaching to program: /usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono, 
process 13413
0x018c8d0c in mono_jit_info_table_find (domain=0x1a94ea0, 
    addr=0xefd2a0c0 "L\203") at domain.c:406
406             } while (chunk_pos < table->num_chunks);
(gdb) where
#0  0x018c8d0c in mono_jit_info_table_find (domain=0x1a94ea0, 
    addr=0xefd2a0c0 "L\203") at domain.c:406
#1  0x018c8cb8 in mono_jit_info_table_find (domain=<value optimized out>, 
    addr=0xefd2a0c0 "L\203") at domain.c:425
#2  0x018c8cb8 in mono_jit_info_table_find (domain=<value optimized out>, 
    addr=0xefd2a0c0 "L\203") at domain.c:425
Previous frame inner to this frame (corrupt stack?)
(gdb) 
(gdb) i locals
chunk = (MonoJitInfoTableChunk *) 0x1aa2000
table = (MonoJitInfoTable *) 0x1a66280
ji = (MonoJitInfo *) 0x1a94ea0
chunk_pos = 1
pos = 0
hp = (MonoThreadHazardPointers *) 0x0
__PRETTY_FUNCTION__ = "ble_remove\000\000jit_info_tabl"
(gdb) 
(gdb) list
401                                     goto not_found;
402                     }
403     
404                     ++chunk_pos;
405                     pos = 0;
406             } while (chunk_pos < table->num_chunks);
407     
408      not_found:
409             mono_hazard_pointer_clear (hp, JIT_INFO_TABLE_HAZARD_INDEX);
410             mono_hazard_pointer_clear (hp, JIT_INFO_HAZARD_INDEX);
(gdb) p table->num_chunks
$1 = 1
(gdb) p *table
$2 = {domain = 0x1a94ea0, num_chunks = 1, chunks = 0x1a66288}
(gdb) p *chunk
$3 = {refcount = 1, num_elements = 0, last_code_end = 0x0, data = {
    0x0 <repeats 64 times>}}
(gdb) 

        At this point I tried a "next", which appears to have been a
        bad idea: gdb appeared to hang, and when I killed gdb, it
        pulled mono with itself:

malus# kill %2
Detaching from program: /usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono, 
process 13413
stat 2 flag 0
Quitting: ptrace: Device busy.
sorry, pid 13413 was killed: orphaned traced process
[2]    Done                          gdb 
/usr/pkgsrc/lang/mono/work/mono-2.0.1/mono/mini/mono
malus# 

        Hopefully this will give someone at least some leads as to
        what the problem is.

>How-To-Repeat:
        Try to build lang/mono on NetBSD/macppc 4.0.

>Fix:
        Sorry, don't know.