| 1 |
On Tue, 20 Jul 2004 01:32:56 +0200 |
| 2 |
pageexec@××××××××.hu wrote: |
| 3 |
|
| 4 |
> > same again, with CFLAGS="-fno-stack-protector-all |
| 5 |
> > -fno-stack-protector-fno-pie -g -ggdb -fno-omit-frame-pointer" |
| 6 |
> > FEATURES="nostrip" as suggested: |
| 7 |
> |
| 8 |
> i guess that's because the sigsegv occurs in glibc or ld.so, so you'd |
| 9 |
> have to emerge glibc again without stripping it (which is a good idea |
| 10 |
> anyway when you have to troubleshoot a system). |
| 11 |
> |
| 12 |
> > warning: exec file is newer than core file. |
| 13 |
> |
| 14 |
> looks like this core file was generated previously, so that can also |
| 15 |
> contribute for gdb not matching the addresses, try again with ulimit |
| 16 |
> -c unlimited. |
| 17 |
> |
| 18 |
> > (gdb) x/8i $pc |
| 19 |
> > 0x40014dce: Cannot access memory at address 0x40014dce |
| 20 |
> |
| 21 |
> hmm, did you run cpufreqd from within gdb as i suggested? |
| 22 |
> |
| 23 |
> |
| 24 |
> -- |
| 25 |
> gentoo-hardened@g.o mailing list |
| 26 |
> |
| 27 |
|
| 28 |
Sorry, bit confused, I've never used gdb before so bear with me. Erm, I |
| 29 |
was using cpufreqd with -fno-pie etc. so then it was working, and |
| 30 |
specifying an old core, guess that doesn't help... I've recompiled |
| 31 |
without -fno-stack-protector -no-pie etc. but with debugging and run it |
| 32 |
within gdb (without specifying a core file, starting with 'gdb |
| 33 |
/usr/sbin/cpufreqd'), hopefully this helps more: |
| 34 |
|
| 35 |
(gdb) run cpufreqd |
| 36 |
Starting program: /usr/sbin/cpufreqd cpufreqd |
| 37 |
warning: Unable to find dynamic linker breakpoint function. |
| 38 |
GDB will be unable to debug shared library initializers |
| 39 |
and track explicitly loaded dynamic code. |
| 40 |
Warning: |
| 41 |
Cannot insert breakpoint -2. |
| 42 |
Error accessing memory address 0x16b0: Input/output error. |
| 43 |
|
| 44 |
(gdb) bt |
| 45 |
#0 0x256b9800 in ?? () |
| 46 |
(gdb) info registers |
| 47 |
No selected frame. |
| 48 |
(gdb) x/8i $pc |
| 49 |
0x256b9800: mov %esp,%eax |
| 50 |
0x256b9802: call 0x256b983d |
| 51 |
0x256b9807: mov %eax,%edi |
| 52 |
0x256b9809: call 0x256b97f0 |
| 53 |
0x256b980e: add $0x12d72,%ebx |
| 54 |
0x256b9814: mov 0xfffffef0(%ebx),%eax |
| 55 |
0x256b981a: pop %edx |
| 56 |
0x256b981b: lea (%esp,%eax,4),%esp |
| 57 |
(gdb) x/16x $sp |
| 58 |
0x5d900310: 0x00000002 0x5d900431 0x5d900444 |
| 59 |
0x00000000 0x5d900320: 0x5d90044d 0x5d900475 0x5d9004de |
| 60 |
0x5d9004eb 0x5d900330: 0x5d9004f6 0x5d900506 0x5d900510 |
| 61 |
0x5d90051e 0x5d900340: 0x5d900532 0x5d900554 |
| 62 |
0x5d900573 0x5d9005e0 |
| 63 |
|
| 64 |
|
| 65 |
So it will help to recompile glibc with nostrip etc.? Thanks again, |
| 66 |
|
| 67 |
Barry |
| 68 |
|
| 69 |
-- |
| 70 |
gentoo-hardened@g.o mailing list |
Archives