I have the core dump file. I kept the c binary application(Built with Debug ON option) in the GDB path which caused the core dump along with the core file.
But when I do GDB, I am getting the stack information as below:
sura#hyde:/abc/def/ghi/appln [css_absmin] gdb application_name core_1206
HP gdb 3.0 for PA-RISC 1.1 or 2.0 (narrow), HP-UX 11.00.
Copyright 1986 - 2001 Free Software Foundation, Inc.
Hewlett-Packard Wildebeest 3.0 (based on GDB) is covered by the
GNU General Public License. Type "show copying" to see the conditions to
change it and/or distribute copies. Type "show warranty" for warranty/support.
..
warning: exec file is newer than core file.
Core was generated by 'application_name'.
Program terminated with signal 11, Segmentation fault.
#0 0xc0199640 in ?? ()
(gdb) where
#0 0xc0199640 in ?? ()
(gdb) bt
#0 0xc0199640 in ?? ()
(gdb)
The C Binary is compiled in PA-RISC (unix).
How can I get the offending function name?
Sundar,
You posted a question about gdb but HP-UX gdb is not a GNU gdb, but rather gdb customized by HP. So it is called Hewlett-Packard Wildebeest (based on GDB) - wdb. Your problem is that you have extremely old version of wdb. Right now HP offers gdb 6.5 (http://h21007.www2.hp.com/portal/site/dspp/menuitem.863c3e4cbcdc3f3515b49c108973a801/?ciid=29080f1bace021100f1bace02110275d6e10RCRD) for HP 9000 systems running HP-UX 11i v1, 11i v2, or 11i v3 and your version is 3.0!
You seem to have HP-UX 11.00, so you can ask on HP-UX forum what is the latest wdb for HP 11.00. Ask here: http://h30499.www3.hp.com/t5/Languages-and-Scripting/bd-p/itrc-150#. Also after installing the latest wdb for your PA-RISC try to get backtrace again with it. I hope you will get a correct backtrace with the latest wdb.
Related
I came to know about the __libc_start_main function. I have been thinking that __libc_start_main call the main function like this, but when I checked ret of main function of my own program, it is the address of __libc_start_call_main. What's the diffrence between __libc_start_main and __libc_start_call_main?
source code of my program, test.c
#include <stdio.h>
int main(void)
{
puts("Sunghyeon Lee");
}
gdb output:
──(kali㉿kali)-[~]
└─$ gdb test
GNU gdb (Debian 12.1-3) 12.1
Copyright (C) 2022 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "x86_64-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<https://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from test...
(No debugging symbols found in test)
(gdb) b *main
Breakpoint 1 at 0x1139
(gdb) r
Starting program: /home/kali/test
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Breakpoint 1, 0x0000555555555139 in main ()
(gdb) x/a $rsp
0x7fffffffdec8: 0x7ffff7dd920a <__libc_start_call_main+122>
Thank you for your help!
I have searched about the diffrence between __libc_start_main and __ibc_start_call_main, I have never found the explanation about it.
I have never found the explanation about it.
Take a look at the commit which created __libc_start_call_main.
Effectively a chunk of __libc_start_main was split out into a separate routine.
I am new to GDB, and need to examine content of function using gdb. The program need to be debugged is perf and function name is ___fprintf_chk() , and tried some thing like:
perf stat -I 1000 -e branch-misses
time counts unit events
1.000222090 1746 branch-misses
2.000486444 1986 branch-misses
3.000712797 1783 branch-misses
sudo ./gdb --pid $(pidof perf)
GNU gdb (GDB) 10.2
Copyright (C) 2021 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "aarch64-none-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<https://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word".
Attaching to process 38977
Reading symbols from /usr/bin/perf...
warning: Unable to determine the number of hardware watchpoints available.
warning: Unable to determine the number of hardware breakpoints available.
Reading symbols from /lib/aarch64-linux-gnu/libpthread.so.0...
(No debugging symbols found in /lib/aarch64-linux-gnu/libpthread.so.0)
Error while reading shared library symbols for /lib/aarch64-linux-gnu/libpthread.so.0:
Cannot find user-level thread for LWP 38977: generic error
Reading symbols from /lib/aarch64-linux-gnu/librt.so.1...
(No debugging symbols found in /lib/aarch64-linux-gnu/librt.so.1)
Reading symbols from /lib/aarch64-linux-gnu/libm.so.6...
(No debugging symbols found in /lib/aarch64-linux-gnu/libm.so.6)
Reading symbols from /lib/aarch64-linux-gnu/libdl.so.2...
(No debugging symbols found in /lib/aarch64-linux-gnu/libdl.so.2)
Reading symbols from /lib/libopencsd_c_api.so.1...
--Type <RET> for more, q to quit, c to continue without paging--q
Quit
(gdb) b __fprintf_chk
Breakpoint 1 at 0xaaaae55b2614
(gdb) c
Continuing.
Breakpoint 1, 0x0000aaaae55b2614 in __fprintf_chk#plt ()
(gdb) info args
No symbol table info available.
Now, need to know when break point is put at __fprintf_chk, does it on __fprintf_chk that is part of perf binary or from other shared objects (.so files, as I see Reading symbols from /lib/aarch64-linux-gnu/libpthread.so.0...)
Also, info args gives nothing, does it mean perf need to be compiled with debugging info (how)?
I downloaded mingw-w64 so I can work on writing programs and debugging them from my home laptop, but for whatever reason, when ever I'm debugging a program with gdb and I wish to exit function using 'finish' command, gdb simply runs until it returns from function and simply exits and I return to the shell's prompt. I have no idea what is causing this bug, and it is especially annoying and make debugging very hard because it forces me to go through every line in a function before exiting (God forbids there's a loop). Has someone encountered this bug before? Is there a solution?
PS C:\Users\gamef\OneDrive\Desktop\CPE453\program4_tinyfs> gdb .\a.exe
GNU gdb (GDB) 8.1
Copyright (C) 2018 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-w64-mingw32".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from .\a.exe...done.
(gdb) br writeData
Breakpoint 1 at 0x402e44: file TinyFS.c, line 299.
(gdb) run
Starting program: C:\Users\gamef\OneDrive\Desktop\CPE453\program4_tinyfs\a.exe
[New Thread 40396.0x1cc8]
[New Thread 40396.0x6e30]
[New Thread 40396.0x3090]
[New Thread 40396.0x9e38]
Thread 1 hit Breakpoint 1, writeData (begBlock=0, data=0x61fab2 "\001_", 'ÿ' <repeats 31 times>, size=254,
type=DIRECTORY) at TinyFS.c:299
299 Byte buff[BLOCKSIZE] = {0};
(gdb) s
300 buff[MAGIC_OFFSET] = MAGIC_NUMBER;
(gdb) finish
Run till exit from #0 writeData (begBlock=0, data=0x61fab2 "\001_", 'ÿ' <repeats 31 times>, size=254, type=DIRECTORY)
at TinyFS.c:300
PS C:\Users\gamef\OneDrive\Desktop\CPE453\program4_tinyfs>
It seems that this problem is a caused by a bug inside the 8.1 version of GDB. So I reverted back to GDB 8.0 which seems to work perfectly.
I have a problem where one or more threads lock each other. I dont know what going on there. The debugger cannot break (thread 1), breaks but cannot get a backtrace (thread 2+5) or shows the backtrace (thread 3)
Gdb native shows the same.
I learned that this is case because libc imlements this in assembler an gdb cannot walt the stack correctly. Sometimes (i dont know when), i can do a few steps in the assembly, then i see the backtrace.
I just tried a x64 program and it works.
See my sample code:
#include <time.h>
int main()
{
while(1)
{
struct timespec ts;
ts.tv_sec = 1;
ts.tv_nsec = 0;
clock_nanosleep(CLOCK_MONOTONIC, 0, &ts, 0);
}
return 1;
}
gdb output 32 bit:
vagrant#PC41388-spvm-4650:/tmp$ gdb main32
GNU gdb (Ubuntu
7.7.1-0ubuntu5~14.04.2) 7.7.1 Copyright (C) 2014 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later
http://gnu.org/licenses/gpl.html This is free software: you are free
to change and redistribute it. There is NO WARRANTY, to the extent
permitted by law. Type "show copying" and "show warranty" for
details. This GDB was configured as "x86_64-linux-gnu". Type "show
configuration" for configuration details. For bug reporting
instructions, please see: http://www.gnu.org/software/gdb/bugs/.
Find the GDB manual and other documentation resources online at:
http://www.gnu.org/software/gdb/documentation/. For help, type
"help". Type "apropos word" to search for commands related to
"word"... Reading symbols from main32...(no debugging symbols
found)...done.
(gdb) r Starting program: /tmp/main32 [Thread
debugging using libthread_db enabled] Using host libthread_db library
"/lib/x86_64-linux-gnu/libthread_db.so.1". ^C Program received signal
SIGINT, Interrupt. 0x55579cd9 in ?? () (gdb) bt
#0 0x55579cd9 in ?? ()
#1 0x555b0af3 in __libc_start_main (main=0x80484dd , argc=1,
argv=0xffffcee4, init=0x8048520 <__libc_csu_init>,
fini=0x8048590 <__libc_csu_fini>, rtld_fini=0x55564160 <_dl_fini>,
stack_end=0xffffcedc) at libc-start.c:287
#2 0x08048401 in _start () (gdb)
gdb output 64 bit:
vagrant#PC41388-spvm-4650:/tmp$ gdb main64
GNU gdb (Ubuntu
7.7.1-0ubuntu5~14.04.2) 7.7.1 Copyright (C) 2014 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later
http://gnu.org/licenses/gpl.html This is free software: you are free
to change and redistribute it. There is NO WARRANTY, to the extent
permitted by law. Type "show copying" and "show warranty" for
details. This GDB was configured as "x86_64-linux-gnu". Type "show
configuration" for configuration details. For bug reporting
instructions, please see: http://www.gnu.org/software/gdb/bugs/.
Find the GDB manual and other documentation resources online at:
http://www.gnu.org/software/gdb/documentation/. For help, type
"help". Type "apropos word" to search for commands related to
"word"... Reading symbols from main64...(no debugging symbols
found)...done.
(gdb) r Starting program: /tmp/main64 [Thread
debugging using libthread_db enabled] Using host libthread_db library
"/lib/x86_64-linux-gnu/libthread_db.so.1". b ^C Program received
signal SIGINT, Interrupt. 0x00002aaaaafe092a in __clock_nanosleep
(clock_id=1, flags=0,
req=0x7fffffffdc10, rem=0x2aaaaafe092a <__clock_nanosleep+58>)
at ../sysdeps/unix/sysv/linux/clock_nanosleep.c:41 41 ../sysdeps/unix/sysv/linux/clock_nanosleep.c: No such file or
directory.
(gdb) bt
#0 0x00002aaaaafe092a in __clock_nanosleep (clock_id=1, flags=0,
req=0x7fffffffdc10, rem=0x2aaaaafe092a <__clock_nanosleep+58>)
at ../sysdeps/unix/sysv/linux/clock_nanosleep.c:41
#1 0x0000000000400630 in main () (gdb)
set architecture i386 does not help either.
More news: info proc mapp shows the x32 app is in [vvar] whereas the x64 app is at libc. This would explain why gdb cant find the backtrace.
So my question is: Is there a different version of the libc, where this works? I am using ubuntu14.04.
I updated to a newer gdb version (currently the latest, 7.12.1). This fixed the problem.
Note that gbd:i386 did not work either on lubuntu x64, whereas it worked fine under lubuntu x32. Also note that both main32 and libc are binary identical on lubuntu x64 and x32.
Consider this gist. I have checked and double checked this piece of code for defects and can't find any apparent flaws in the code. It also compiles fine when I use g++ -g -std=c++11 -Wall dynlibtest.cc -ldl -lffi -lstdc++ -odynlibtest && ./dynlibtest (the -ldl and -lffi switches are for the dynamic loading and FFI libraries, respectively).
However, when the highlighted line (l.96) executes it segfaults.
I have also tried pulling it through gdb, and after installing the libc debugging symbols it spits this message out when the ./dynlibtest bin segfaults:
(gdb) next
Program received signal SIGSEGV, Segmentation fault.
__memcpy_sse2_unaligned () at ../sysdeps/x86_64/multiarch/memcpy-sse2-unaligned.S:157
157 ../sysdeps/x86_64/multiarch/memcpy-sse2-unaligned.S: No such file or directory.
Who can help me understand why this segfaults? Is it a bug of some kind or am I using one of the API's wrong?
For reference: the first part of the code calls gettimeofday directly to show that the code can indeed find it, and that even the args are correct when it is called directly.
EDIT: I have added the gdb output when the code segfaults with the output of bt also attached:
$ gdb ./dynlibtest
GNU gdb (Ubuntu 7.7.1-0ubuntu5~14.04.2) 7.7.1
Copyright (C) 2014 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later http://gnu.org/licenses/gpl.html
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
http://www.gnu.org/software/gdb/bugs/.
Find the GDB manual and other documentation resources online at:
http://www.gnu.org/software/gdb/documentation/.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from ./dynlibtest...done.
(gdb) break 96
Breakpoint 1 at 0x401032: file dynlibtest.cc, line 96.
(gdb) run
Starting program: /home/j/dev/elisp-ffi/dynlibtest
Test started...
Got main program handle
pre-alloc: tv.tv_sec = 140737340592552
Sleeping for 1 second
post-alloc: tv.tv_sec = 1432058412
Sleeping for 1 second
Fn ptr call: tv.tv_sec = 1432058413
FFI CIF preparation is OK
Sleeping for 1 second
Breakpoint 1, main () at dynlibtest.cc:96
96 ffi_call(&cif, FFI_FN(gettimeofday), &result, args);
(gdb) next
Program received signal SIGSEGV, Segmentation fault.
__memcpy_sse2_unaligned () at ../sysdeps/x86_64/multiarch/memcpy-sse2-unaligned.S:157
157 ../sysdeps/x86_64/multiarch/memcpy-sse2-unaligned.S: No such file or directory.
(gdb) bt
#0 __memcpy_sse2_unaligned () at ../sysdeps/x86_64/multiarch/memcpy-sse2-unaligned.S:157
#1 0x00007ffff79d34c2 in memcpy (__len=8, __src=0x0, __dest=0x7fffffffda48) at /usr/include/x86_64-linux-gnu/bits/string3.h:51
#2 ffi_call (cif=0x7fffffffdca0, fn=0x400ab0 , rvalue=0x7fffffffdc40, avalue=0x7fffffffdc00) at ../src/x86/ffi64.c:504
#3 0x000000000040104e in main () at dynlibtest.cc:96
(gdb)