| 1 |
Mike Frysinger wrote: |
| 2 |
|
| 3 |
>On Monday 02 August 2004 04:32 am, Natanael Copa wrote: |
| 4 |
> |
| 5 |
> |
| 6 |
>>I just came back from a longweekend and wanted to check out what was new. |
| 7 |
>> |
| 8 |
>>But emerge sync segfaulted. |
| 9 |
>><snip> |
| 10 |
>>I guess there is a bug in libresolv in uclibc |
| 11 |
>> |
| 12 |
>> |
| 13 |
> |
| 14 |
>yeah, we've been looking into this ... |
| 15 |
>i dont know if solar found anything over the weekend (i was gone), but find |
| 16 |
>attached a small .c file i made to reproduce the error |
| 17 |
> |
| 18 |
> |
| 19 |
|
| 20 |
Thanks! |
| 21 |
|
| 22 |
The problem is that h->h_aliases is pointing at an unreadable nun-null |
| 23 |
location. |
| 24 |
|
| 25 |
I ran the code through gdb and it is strnlen that fails. |
| 26 |
|
| 27 |
I added those lines: |
| 28 |
|
| 29 |
char *p = h->h_aliases[hidx]; |
| 30 |
printf("Address of list is: %x\n", p); |
| 31 |
printf("Value of h->h_aliases[hidx] is: %i", (int)*p); |
| 32 |
|
| 33 |
it segfaults when reading *p. |
| 34 |
|
| 35 |
I also tried the code in my uclib environment based on ucrux and exactly |
| 36 |
the same thing happened there. So its definitively an uclibc bug. |
| 37 |
|
| 38 |
-- |
| 39 |
Natanael Copa |
| 40 |
|
| 41 |
|
| 42 |
-- |
| 43 |
gentoo-embedded@g.o mailing list |
Archives