| 1 |
Glenn Enright wrote: |
| 2 |
|
| 3 |
>On Thu, 21 Jul 2005 17:45, Zac Medico wrote: |
| 4 |
> |
| 5 |
> |
| 6 |
>>Richard Fish wrote: |
| 7 |
>> |
| 8 |
>> |
| 9 |
>>>I will take a look through the gzip source and see if anything |
| 10 |
>>>interesting shows up. |
| 11 |
>>> |
| 12 |
>>> |
| 13 |
>>Richard, you are a saint ;-). But remember, he said that gunzip only |
| 14 |
>>segfaults with files that he compressed himself which suggests that the |
| 15 |
>>files are being corrupted when he compresses them. You'd thing that gunzip |
| 16 |
>>would be able to handle the corruption better that segfaulting though. |
| 17 |
>> |
| 18 |
>>Zac |
| 19 |
>> |
| 20 |
>> |
| 21 |
> |
| 22 |
>Ok just ran gcc-config and found |
| 23 |
>lib # gcc-config -l |
| 24 |
>/usr/bin/gcc-config: line |
| 25 |
>583: /etc/env.d/gcc/i686-pc-linux-gnu-3.4.3-20050110: No such file or |
| 26 |
>directory |
| 27 |
> * /usr/bin/gcc-config: Profile does not exist or invalid setting |
| 28 |
>for /etc/env.d/gcc/i686-pc-linux-gnu-3.4.3-20050110 |
| 29 |
>[1] i686-pc-linux-gnu-3.3.5 |
| 30 |
>[2] i686-pc-linux-gnu-3.3.5-hardened |
| 31 |
>[3] i686-pc-linux-gnu-3.3.5-hardenednopie |
| 32 |
>[4] i686-pc-linux-gnu-3.3.5-hardenednossp |
| 33 |
>[5] i686-pc-linux-gnu-3.4.4 |
| 34 |
>[6] i686-pc-linux-gnu-3.4.4-hardened |
| 35 |
>[7] i686-pc-linux-gnu-3.4.4-hardenednopie |
| 36 |
>[8] i686-pc-linux-gnu-3.4.4-hardenednopiessp |
| 37 |
>[9] i686-pc-linux-gnu-3.4.4-hardenednossp |
| 38 |
> |
| 39 |
>Thats not right. gonna select 5 and take it through glibc (which was failing |
| 40 |
>the 'make test') and gzip, see what happens :P |
| 41 |
> |
| 42 |
> |
| 43 |
|
| 44 |
This seems like a very sane plan. |
| 45 |
|
| 46 |
FYI, regarding the points I made earlier: |
| 47 |
|
| 48 |
The "trampolin" stuff is related to gzip looking to see if the |
| 49 |
filesystem automatically shortened the file name that gzip creates. I |
| 50 |
didn't analyze it in detail, but it seems to be normal and intended. |
| 51 |
|
| 52 |
Regarding the lstat64 vs stat64, this is more interesting. It basically |
| 53 |
means that the configure step probably didn't find an lstat call on your |
| 54 |
system. This suggests to me a problem in your system headers or |
| 55 |
compiler toolchain. |
| 56 |
|
| 57 |
If fixing the gcc-config doesn't produce sane results, what I suggest is |
| 58 |
the following commands: |
| 59 |
|
| 60 |
# cd /usr/local/src |
| 61 |
# tar -xzvf /usr/portage/distfiles/gzip-1.3.5.tar.gz |
| 62 |
# cd gzip-1.3.5 |
| 63 |
# ./configure |
| 64 |
# grep HAVE_LSTAT config.h |
| 65 |
|
| 66 |
If HAVE_LSTAT is commented out, then create a "testlstat.c" file with |
| 67 |
the following: |
| 68 |
|
| 69 |
char lstat(); |
| 70 |
int main() { |
| 71 |
char (*f)(); |
| 72 |
f = lstat; |
| 73 |
return 0; |
| 74 |
} |
| 75 |
|
| 76 |
Compile the above with "gcc -o testlstat testlstat.c". It should fail |
| 77 |
to compile, and the resulting error should give us a good place to look |
| 78 |
for the real problem. |
| 79 |
|
| 80 |
-Richard |
| 81 |
|
| 82 |
-- |
| 83 |
gentoo-user@g.o mailing list |
Archives