Here's a summary of the Linux Test Project run on
the latest sparc git tree. I'm focussing on the blocking
tests first, but it's slow going.
Test systems:
palantir9 SS20, 2x ROSS Hypersparc, 320MB
palantir13 SS10, 2x SuperSparc II, 128MB
Kernel:
2.6.23-mph4 SMP from the sparc-2.6.git tree. Last commit:
4fa4d23fa20de67df919030c1216295664866ad7 Merge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6
LTP version: ltp-full-20070930
Command run: ./runltplite.sh -p -l ltplite-211007.log
Summary:
Total Tests: 759
Total Failures: 16
2 tests block the test run from completing on SS20: mkdir09 and rename14
On SS10 pipe04 also blocked completion.
Blocking tests:
---------------
1) mkdir09
In SMP mode, the process threads end up in state D (Uninterruptible sleep).
With the same kernel in UP mode (with num_cpus=0), the test always
succeeds.
Could someone with a SPARC64 try this test?
Typical ps output in SMP mode:
palantir13:/opt# ps -luopendev
F S UID PID PPID C PRI NI ADDR SZ WCHAN TTY TIME CMD
5 S 1001 468 464 0 80 0 - 2331 - ? 00:00:00 sshd
0 S 1001 469 468 2 80 0 - 960 - pts/0 00:00:06 bash
0 S 1001 564 469 0 80 0 - 405 - pts/0 00:00:00 mkdir09
1 D 1001 565 564 0 80 0 - 405 - pts/0 00:00:00 mkdir09
1 D 1001 566 564 0 80 0 - 405 - pts/0 00:00:00 mkdir09
1 D 1001 567 564 0 80 0 - 405 - pts/0 00:00:00 mkdir09
1 D 1001 568 564 0 80 0 - 405 - pts/0 00:00:00 mkdir09
1 D 1001 569 564 0 80 0 - 405 - pts/0 00:00:00 mkdir09
1 D 1001 570 564 1 80 0 - 405 - pts/0 00:00:01 mkdir09
palantir13:/opt# ps sa
UID PID PENDING BLOCKED IGNORED CAUGHT STAT TTY TIME COMMAND
0 459 00000000 00000000 00000002 00002000 Ss ttyS0 0:00 /bin/lo
0 460 00000000 00080000 00324004 7b883efb S ttyS0 0:00 -bash
1001 469 00000000 00080000 00324004 7b883efb Ss pts/0 0:06 -bash
1001 564 00000000 00000000 00004000 <7ff2beff S+ pts/0 0:00 ./mkdir
1001 565 00004000 00000000 00000000 <7ffafeff D+ pts/0 0:00 ./mkdir
1001 566 00004000 00000000 00000000 <7ffafeff D+ pts/0 0:00 ./mkdir
1001 567 00004000 00000000 00000000 <7ffafeff D+ pts/0 0:00 ./mkdir
1001 568 00004000 00000000 00000000 <7ffafeff D+ pts/0 0:00 ./mkdir
1001 569 00004000 00000000 00000000 <7ffafeff D+ pts/0 0:00 ./mkdir
1001 570 00004000 00000000 00000000 <7ffafeff D+ pts/0 0:01 ./mkdir
The 2nd ps shows that main thread (564) has send a SIGTERM to all other threads,
but they are blocked on something. No further investigation done on this yet.
2) rename14
Running this test causes the following console output:
EXT3-fs warning (device sdb4): ext3_rename: Deleting old file (162881), 2, error=-2
EXT3-fs warning (device sdb4): ext3_rename: Deleting old file (162881), 2, error=-2
EXT3-fs warning (device sdb4): ext3_rename: Deleting old file (162881), 2, error=-2
EXT3-fs warning (device sdb4): ext3_rename: Deleting old file (162881), 2, error=-2
EXT3-fs warning (device sdb4): ext3_unlink: Deleting nonexistent file (162885), 0
and corrupts the FS. Looks like a endian problem in ext3.
Issue being discussed here:
http://article.gmane.org/gmane.comp.file-systems.ext4/3772
3) pipe04
Not investigated yet:
palantir13:~# kernel BUG at mm/highmem.c:196!
\|/ ____ \|/
"@'/ ,. \`@"
/_| \__/ |_\
\__U_/
pipe04(3665): Kernel bad trap [#1]
PSR: 400010c0 PC: f006b010 NPC: f006b014 Y: 00000000 Not tainted
PC: <kunmap_high+0xac/0xd8>
%G: 000000c4 400010e7 f01f5084 f01f5078 0054d000 f14ba000 f0a82000 f076fdd8
%O: 00000023 f01c9cc0 000000c4 efb4a9ec 00000000 00000000 f0a83c90 f006b008
RPC: <kunmap_high+0xa4/0xd8>
%L: 00000074 f06f90c0 f019aea4 00000002 00000080 00000000 f0a82000 efb49002
%I: f06f90c0 00000003 fcfe1000 f0224000 00000090 f00f8b70 f0a83cf8 f0087da0
Caller[f0087da0]: pipe_read+0x24c/0x47c
Caller[f0081998]: do_sync_read+0xac/0xe4
Caller[f0081a6c]: vfs_read+0x9c/0xcc
Caller[f0081c84]: sys_read+0x38/0x64
Caller[f001541c]: syscall_is_too_hard+0x3c/0x40
Caller[00010ce8]: 0x10cf0
Instruction DUMP: 921020ce 7ffead1e 01000000 <91d02005> 7fff1ac6 81e80000 82184005 80a00001 10bfffec
note: pipe04[3665] exited with preempt_count 1
BUG: scheduling while atomic: pipe04/0x04000001/3665
[f019a9cc : cond_resched+0x40/0x48 ] [f0038704 : put_files_struct+0xa4/0xec ] [f0038f98 : do_exit+0x184/0x8f0 ] [f00166f8 : die_if_kernel+0x13c/0x148 ] [f0016754 : do_hw_interrupt+0x50/0x8c ] [f00146a0 : bad_trap_handler+0x28/0x30 ] [f006b008 : kunmap_high+0xa4/0xd8 ] [f0087da0 : pipe_read+0x24c/0x47c ] [f0081998 : do_sync_read+0xac/0xe4 ] [f0081a6c : vfs_read+0x9c/0xcc ] [f0081c84 : sys_read+0x38/0x64 ] [f001541c : syscall_is_too_hard+0x3c/0x40 ] [00010ce8 : 0x10cf0 ]
BUG: soft lockup - CPU#2 stuck for 11s! [pan:3668]
PSR: 400000c3 PC: f019c0f4 NPC: f019c0ec Y: 00000012 Tainted: G D
PC: <_spin_trylock_bh+0x40/0x70>
%G: 00000000 00000002 000000ff f0965720 f0abe1d0 efffe000 f0af0000 00000040
%O: f02241a0 f14559e0 efffffc3 00000000 00000001 00000007 f0af1d18 f006b04c
RPC: <kmap_high+0x10/0x234>
%L: f098ade0 f0081aa0 f0085f98 00000040 00000080 00000000 f0af0000 00000006
%I: f071fea0 efffefc3 00000001 80808080 f0224000 00000008 f0af1db0 f0085b5c
--
Martin
-
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
