(from cvs) sensors binary compile probs on rh8

Philip Edelbrock phil at netroedge.com
Wed Dec 18 20:08:39 CET 2002



(I can't remember if I've replied to this yet)

I didn't have to add /usr/local/lib to it (I've added it a long time 
ago), but it looks in /usr/lib first.  In fact, I think /lib and 
/usr/lib are built into the lib search path and don't need to be in 
ld.so.conf?

I wonder if we can add a checking function in the include header to test 
for the lib version (and call it when sensors is run).  The header 
version and the lib version should always be indentical, right?


Phil

Mark Studebaker wrote:

>> From reading 'man ld.so' and 'man ldconfig', it's apparent that 
>> /usr/lib and /lib are searched _after_ all
>
> the libraries listed in /etc/ld.so.conf. So the only problem is if 
> /usr/local/lib is not listed
> in /etc/ld.so.conf at all.
>
> Did you have /usr/local/lib/ in /etc/ld.so.conf or did you have to add 
> it?
>
> If Redhat doesn't have /usr/local/lib in there, I think that's the 
> check we need to have.
>
>
> Philip Edelbrock wrote:
>
>>
>> Ah, I found the trouble. I've got too many libsensors installed, and 
>> when sensors is compiled, it will link against an old one. At some 
>> point, we changed the default lib installation directory from 
>> /usr/lib to /usr/local/lib. Bins get linked against 
>> /usr/lib/libsensors.so instead of the newer /usr/local/libsensors.so.
>>
>> So, in essence, a newer include for the lib is used to link against 
>> an old lib.
>>
>> Once I deleted the old libs from /usr/lib and got it linking to the 
>> new lib, it worked fine.
>>
>> Should we have a check for an old lib? Or explicitly link against 
>> libsensors.1.4.0.so (if this is possible?).
>>
>>
>> Phil
>>
>>
>> Mark Studebaker wrote:
>>
>>> guess you saw the mail that it's a libc 2.3 thing...
>>> Can you recompile w/ -g (sensors and libsensors) and get a better bt?
>>> could be something I did to libsensors...
>>>
>>> Philip Edelbrock wrote:
>>>
>>>> :'(
>>>>
>>>> [phil at drtheopolis phil]$ sensors -v
>>>> sensors version 2.7.0
>>>>
>>>> [phil at drtheopolis phil]$ sensors
>>>> Segmentation fault (core dumped)
>>>>
>>>> [phil at drtheopolis phil]$ gdb sensors
>>>> GNU gdb Red Hat Linux (5.2.1-4)
>>>> Copyright 2002 Free Software Foundation, Inc.
>>>> GDB is free software, covered by the GNU General Public License, 
>>>> and you are
>>>> welcome to change it and/or distribute copies of it under certain 
>>>> conditions.
>>>> Type "show copying" to see the conditions.
>>>> There is absolutely no warranty for GDB. Type "show warranty" for 
>>>> details.
>>>> This GDB was configured as "i386-redhat-linux"...
>>>> (gdb) run
>>>> Starting program: /usr/local/bin/sensors
>>>>
>>>> Program received signal SIGSEGV, Segmentation fault.
>>>> 0x4207931a in strcmp () from /lib/i686/libc.so.6
>>>> (gdb) bt
>>>> #0 0x4207931a in strcmp () from /lib/i686/libc.so.6
>>>> #1 0x40028bd9 in sensors_match_chip () from /usr/lib/libsensors.so.1
>>>> #2 0x08049171 in do_the_real_work ()
>>>> #3 0x08049073 in main ()
>>>> #4 0x420158d4 in __libc_start_main () from /lib/i686/libc.so.6
>>>>
>>>>
>>>> It's possible that it's just my strange computer. This isn't a RH8 
>>>> install from scratch, but an upgrade. There could be a legacy lib 
>>>> or something causing trouble which may not appear on another Athlon 
>>>> running RH8.... has anybody else tried on an Athlon?
>>>>
>>>>
>>>> Phil
>>>>
>>>>
>>>> Mark Studebaker wrote:
>>>>
>>>>> Phil, what processor are you compiling for? Or what's wrong with 
>>>>> <math.h>?
>>>>> It works for me, but gcc is generating floating point assembler 
>>>>> instructions
>>>>> and so it doesn't need -lm.
>>>>> If we do add a -lm it should be added to lib/Module.mk for the 
>>>>> library,
>>>>> not in prog/sensors/Module.mk for sensors.
>>>>>
>>>>>
>>>>> Jean Delvare wrote:
>>>>>
>>>>>>> FYI-
>>>>>>>
>>>>>>> gcc -o prog/sensors/sensors prog/sensors/main.ro 
>>>>>>> prog/sensors/chips.ro
>>>>>>> -Llib -lsensors
>>>>>>> lib/libsensors.so: undefined reference to `log'
>>>>>>> lib/libsensors.so: undefined reference to `exp'
>>>>>>> collect2: ld returned 1 exit stat
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Looks like we need an extra "-lm" on the command line.
>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>
>



More information about the lm-sensors mailing list