<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#ffffff">
On 05/27/2010 02:14 AM, Mikael Pettersson wrote:
<blockquote cite="mid:19453.25855.346911.194353@pilspetsen.it.uu.se"
type="cite">
<pre wrap="">Eric Liang wrote:
</pre>
<blockquote type="cite">
<pre wrap="">I've done a build of the source, but it just can't match the object. How
do you make it? I use the command: apt-get source to get the source, so
it does have the same version with the object.
</pre>
</blockquote>
<pre wrap="">
I did:
</pre>
<blockquote type="cite">
<pre wrap="">tar zxvf otp_src_R13B03.tar.gz
cd otp_src_R13B03
./configure; make
</pre>
</blockquote>
<pre wrap="">
The binary files of interest are bin/x86_64-unknown-linux-gnu/beam and
erts/emulator/obj/x86_64-unknown-linux-gnu/opt/plain/erl_goodfit_alloc.o.
</pre>
</blockquote>
Thanks Mikael, and sorry for replying you too late as the seg-fault is
not occured every time.<br>
<br>
I get the debug symbols by this:<br>
<blockquote><a class="moz-txt-link-freetext" href="http://forum.nginx.org/read.php?26,93440,94735">http://forum.nginx.org/read.php?26,93440,94735</a><br>
</blockquote>
<blockquote cite="mid:19453.25855.346911.194353@pilspetsen.it.uu.se"
type="cite">
<pre wrap=""></pre>
<blockquote type="cite">
<blockquote type="cite">
<pre wrap="">You can get a stack dump from the crash by attaching gdb to the
soon-to-crash beam process. Now instead of being terminated gdb will
get control of the process and you should be able to print a stack
trace with bt or where. (This does require that there's a sufficient
time window from the start of the application to the crash.)
=20
</pre>
</blockquote>
<pre wrap="">I've make a core dump 4 seconds before it crash, as mentioned above,=20
because don't get the right symbols, it just with some quesion-marks:
Core was generated by `/usr/lib/erlang/erts-5.7.2/bin/beam'.
#0 0x00007f0a28ecd5a9 in ?? ()
(gdb) whe
#0 0x00007f0a28ecd5a9 in ?? ()
#1 0x0000000000000000 in ?? ()
(gdb)
</pre>
</blockquote>
<pre wrap="">
A core dump from a time point before the crash is useless. Either get a
core dump from the crash itself (execute `ulimit -c unlimited' in bash
before running the test), or attach gdb, continue the process, and wait
for gdb to receive control when the crash occurs.
</pre>
</blockquote>
<br>
I do set the ulimit -c in /etc/profile and after I reboot it:<br>
<blockquote>sunny@dev-2:~$ ulimit -c<br>
unlimited<br>
sunny@dev-1:~/commands$ cat /proc/sys/kernel/core_pattern <br>
/tmp/core.%t.%e.%p<br>
</blockquote>
And I the test is ok:<br>
<blockquote>sunny@dev-2:~$ kill -s SIGSEGV $$<br>
Connection to dev-2 closed.<br>
sunny@dev-2:~$ ls /tmp/<br>
core.1275730620.bash.12566<br>
</blockquote>
But still no core file generated,when the error occurs.<br>
<br>
Anyway, I attatched the running process by gdb, and here is the result:<br>
<blockquote>Program received signal SIGSEGV, Segmentation fault.<br>
unlink_free_block (allctr=0x7ad480, block=0x0) at
beam/erl_goodfit_alloc.c:453<br>
453 Uint sz = BLK_SZ(blk);<br>
(gdb) whe<br>
#0 unlink_free_block (allctr=0x7ad480, block=0x0) at
beam/erl_goodfit_alloc.c:453<br>
#1 0x0000000000437fd6 in get_free_block (allctr=0x7ad480,
size=<value optimized out>, cand_blk=0x0, cand_size=0)<br>
at beam/erl_goodfit_alloc.c:421<br>
#2 0x00000000004322c6 in mbc_alloc_block (allctr=0x7ad480, size=72) at
beam/erl_alloc_util.c:631<br>
#3 mbc_alloc (allctr=0x7ad480, size=72) at beam/erl_alloc_util.c:758<br>
#4 0x00000000004b1697 in erts_alloc () at beam/erl_alloc.h:179<br>
#5 exit_async () at beam/erl_async.c:132<br>
#6 0x000000000043c13d in system_cleanup (exit_code=<value optimized
out>) at beam/erl_init.c:1306<br>
#7 0x000000000043c443 in erl_exit (n=0, fmt=0x54649c "") at
beam/erl_init.c:1380<br>
#8 0x000000000045d042 in halt_0 (A__p=<value optimized out>) at
beam/bif.c:3319<br>
#9 0x00000000004d081f in process_main () at beam/beam_emu.c:2008<br>
#10 0x000000000043d56c in erl_start (argc=34, argv=<value optimized
out>) at beam/erl_init.c:1233<br>
#11 0x00000000004269b9 in main (argc=8049792, argv=0x0) at
sys/unix/erl_main.c:29<br>
(gdb) f 1<br>
#1 0x0000000000437fd6 in get_free_block (allctr=0x7ad480,
size=<value optimized out>, cand_blk=0x0, cand_size=0)<br>
at beam/erl_goodfit_alloc.c:421<br>
421 unlink_free_block(allctr, blk);<br>
(gdb) l 421<br>
416 /* We are guaranteed to find a block that fits in this
bucket */<br>
417 blk = search_bucket(allctr, min_bi, size);<br>
418 ASSERT(blk);<br>
419 if (cand_blk && cand_size <= BLK_SZ(blk))<br>
420 return NULL; /* cand_blk was better */<br>
421 unlink_free_block(allctr, blk);<br>
422 return blk;<br>
423 }<br>
424 <br>
425 <br>
(gdb) <br>
</blockquote>
As the running process use the no-debug symbol version beam, I guess
the ASSERT in line:418 does not work. So I dig in<br>
<blockquote>(gdb) p allctr<br>
$1 = (Allctr_t *) 0x7ad480<br>
(gdb) p min_bi<br>
$2 = <value optimized out><br>
(gdb) p size<br>
$3 = <value optimized out><br>
(gdb) p *allctr<br>
$4 = {name_prefix = 0x534227 "sl_", alloc_no = 3, name = {alloc = 0,
realloc = 0, free = 0}, <br>
vsn_str = 0x53602f "2.1", t = 0, ramv = 0, sbc_threshold = 524288,
sbc_move_threshold = 80, <br>
mbc_move_threshold = 50, main_carrier_size = 131072, max_mseg_sbcs =
256, max_mseg_mbcs = 5, <br>
largest_mbc_size = 10485760, smallest_mbc_size = 1048576,
mbc_growth_stages = 10, mseg_opt = {cache = 1, <br>
preserv = 1, abs_shrink_th = 4145152, rel_shrink_th = 80},
mbc_header_size = 32, sbc_header_size = 32, <br>
min_mbc_size = 16384, min_mbc_first_free_size = 4096, min_block_size
= 32, mbc_list = {first = 0x7f4f93a5d010, <br>
last = 0x7f4f93a5d010}, sbc_list = {first = 0x0, last = 0x0},
main_carrier = 0x7f4f93a5d010, <br>
get_free_block = 0x437f40 <get_free_block>, link_free_block =
0x437d00 <link_free_block>, <br>
unlink_free_block = 0x437e10 <unlink_free_block>, info_options
= 0x438480 <info_options>, <br>
get_next_mbc_size = 0x430e40 <get_next_mbc_size>, creating_mbc
= 0x438100 <update_last_aux_mbc>, <br>
destroying_mbc = 0x438100 <update_last_aux_mbc>, init_atoms =
0x4385c0 <init_atoms>, mutex = {mtx = {pt_mtx = {<br>
__data = {__lock = 0, __count = 0, __owner = 0, __nusers = 0,
__kind = 0, __spins = 0, __list = {<br>
__prev = 0x0, __next = 0x0}}, __size = '\000' <repeats
39 times>, __align = 0}, is_rec_mtx = 0, <br>
prev = 0x0, next = 0x0}}, thread_safe = 0, ts_list = {prev = 0x0,
next = 0x0}, atoms_initialized = 0, <br>
stopped = 0, calls = {this_alloc = {giga_no = 0, no = 2460},
this_free = {giga_no = 0, no = 2458}, <br>
this_realloc = {giga_no = 0, no = 0}, mseg_alloc = {giga_no = 0, no
= 0}, mseg_dealloc = {giga_no = 0, no = 0}, <br>
mseg_realloc = {giga_no = 0, no = 0}, sys_alloc = {giga_no = 0, no
= 1}, sys_free = {giga_no = 0, no = 0}, <br>
sys_realloc = {giga_no = 0, no = 0}}, sbcs = {curr_mseg = {no = 0,
size = 0}, curr_sys_alloc = {no = 0, <br>
size = 0}, max = {no = 0, size = 0}, max_ever = {no = 0, size =
0}, blocks = {curr = {no = 0, size = 0}, <br>
max = {no = 0, size = 0}, max_ever = {no = 0, size = 0}}}, mbcs =
{curr_mseg = {no = 0, size = 0}, <br>
curr_sys_alloc = {no = 1, size = 131112}, max = {no = 1, size =
131112}, max_ever = {no = 0, size = 0}, <br>
blocks = {curr = {no = 4, size = 384}, max = {no = 144, size =
13848}, max_ever = {no = 0, size = 0}}}}<br>
(gdb)<br>
</blockquote>
And stalled here, do you have any advices? and also, any other
suggestions would be appreciated. TIA.<br>
<br>
<blockquote cite="mid:19453.25855.346911.194353@pilspetsen.it.uu.se"
type="cite">
<pre wrap="">
</pre>
<blockquote type="cite">
<pre wrap="">Look back the initial error:
segfault at 0 ip 0000000000437e10 sp 00007fffce250948 error 4 in
beam[400000+174000]
Would you mind tell me what's the meaning of -ip-/-sp-, and what does
-error 4- means?
</pre>
</blockquote>
<pre wrap="">
IP is the Instruction Pointer aka Program Counter. SP is the stack pointer.
The 'error 4' is a low-level error code which is irrelevant for us.
</pre>
</blockquote>
Got it, thank you.<br>
<br>
Eric<br>
</body>
</html>