[mpich-discuss] mpich build error

Sayan Ghosh sayandeep52 at gmail.com
Wed Oct 21 17:00:13 CDT 2015


Thanks Rob, here is the specific part from my mpl config, almost looks like
your config:

configure:13058: checking for strncmp
configure:13058: icc -std=gnu99 -o conftest       conftest.c  >&5
configure:13058: $? = 0
configure:13058: result: yes
configure:13070: checking whether strncmp needs a declaration
configure:13089: icc -std=gnu99 -c      conftest.c >&5
configure:13089: $? = 0
configure:13097: result: no

I get build errors even with GCC 4.9 and clang with older versions of
mpich, so this is definitely a system issue as you mentioned.

sg at sg:~/sources/mpich-3.1.4/build$ cat /etc/debian_version
8.2
sg at sg:~/sources/mpich-3.1.4/build$ uname -a
Linux sg 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt11-1+deb8u5 (2015-10-09)
x86_64 GNU/Linux

On Wed, Oct 21, 2015 at 7:39 AM, Rob Latham <robl at mcs.anl.gov> wrote:

> The warnings are intel's preprocessor being a bit pickier than gnu's cpp
>
> the errors, though, are indeed odd.
> I will eat my sock if strncmp really truly is not on your system; far more
> likely the strncmp test failed for some unrelated reason.
>
> Look at your src/mpl/config.log.  On my system, the (successful) check for
> strncmp looks like this:
>
> configure:12723: checking for strncmp
> configure:12723: ccache gcc -o conftest -Wall -g -Wno-unused-label
> -fstack-usage -fPIC -ggdb3   -I /home/robl/work/soft/lustre/include
> conftest.c  >&5
> conftest.c:65:6: warning: conflicting types for built-in function 'strncmp'
>  char strncmp ();
>       ^
> configure:12723: $? = 0
> configure:12723: result: yes
> configure:12734: checking whether strncmp needs a declaration
> configure:12752: ccache gcc -c -Wall -g -Wno-unused-label -fstack-usage
> -fPIC -ggdb3   -I /home/robl/work/soft/lustre/include  conftest.c >&5
> configure:12752: $? = 0
> configure:12759: result: no
>
> Does yours look like this?
>
> ==rob
>
> Hi,
>>
>> I am getting an error while trying to compile the latest mpich (from the
>> git repo) with Intel 15.0.1 compiler...excerpt of the error:
>>
>> In file included from ../../.././src/mpl/include/mpl.h(10),
>>                   from ../../.././src/mpl/src/mplstr.c(7):
>> ./include/mplconfig.h(238): warning #14: extra text after expected end
>> of preprocessing directive
>>    #ifndef _mpl_                 ^
>>
>> In file included from ../../.././src/mpl/include/mpl.h(10),
>>                   from ../../.././src/mpl/src/mplstr.c(7):
>> ./include/mplconfig.h(239): warning #2217: white space is required
>> between the macro name "_mpl_" and its replacement text
>>    #define _mpl_                   ^
>>
>> In file included from ../../.././src/mpl/include/mpl.h(111),
>>                   from ../../.././src/mpl/src/mplstr.c(7):
>> ../../.././src/mpl/include/mplstr.h(48): error: #error directive:
>> "strncmp is required"
>>    #error "strncmp is required"
>>     ^
>>
>> ../../.././src/mpl/src/mplstr.c(41): warning #266: function "isdigit"
>> declared implicitly
>>                if (isdigit(nc)) {
>>                    ^
>>
>> Internal error: null pointer
>>
>>
>> Please suggest.
>>
>> --
>> Regards,
>> Sayan
>>
>>
>> _______________________________________________
>> discuss mailing list     discuss at mpich.org
>> To manage subscription options or unsubscribe:
>> https://lists.mpich.org/mailman/listinfo/discuss
>>
>>
> --
> Rob Latham
> Mathematics and Computer Science Division
> Argonne National Lab, IL USA
> _______________________________________________
> discuss mailing list     discuss at mpich.org
> To manage subscription options or unsubscribe:
> https://lists.mpich.org/mailman/listinfo/discuss
>



-- 
Regards,
Sayan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpich.org/pipermail/discuss/attachments/20151021/271ea7f4/attachment.html>
-------------- next part --------------
_______________________________________________
discuss mailing list     discuss at mpich.org
To manage subscription options or unsubscribe:
https://lists.mpich.org/mailman/listinfo/discuss


More information about the discuss mailing list