[Devel] Re: [Users] action.c:1: error: bad value (x86_64) for
-mtune= switch
Bogdan-Andrei Iancu
bogdan at voice-system.ro
Tue Feb 27 10:46:33 CET 2007
Hi Alex,
the CPU is not autodetected (the ARCH is, using uname), but
statistically defined based on ARCH and compiler. So the sed has no
sense - you can directly put whatever you want; now "opteron" is used.
If you want to change, just do:
"CPU_TYPE=nocona make all"
btw, what GCC versions supports nocona?
regards,
bogdan
Alexandr Dubovikov wrote:
> On Tue, Feb 27, 2007 at 10:19:33AM +0100, Alexandr Dubovikov wrote:
>
>>> I'd like to bring up this topic again before too late: I am using an
>>> openSuSE 10.2 system with
>>> -----------
>>> bastian at han:~> gcc --version
>>> gcc (GCC) 4.1.2 20061115 (prerelease) (SUSE Linux)
>>> -----------
>>> on a machine; compilation of a fresh SVN checkout still fails due to
>>> this reason. I guess this should be fixed before 1.2...
>>>
>> value x86_64 for gcc 4.x.x is not in use anymore
>> You should use x86-64 instead of x86_64.
>>
>>
>> Bogdan, I think better way to fix it in Makefile.defs:
>>
>>
>> --- openser-1.1.1-notls/Makefile.defs.orig 2007-01-22 13:16:11.000000000 +0100
>> +++ openser-1.1.1-notls/Makefile.defs 2007-02-27 10:13:32.000000000 +0100
>> @@ -569,7 +569,7 @@ ifeq ($(CC_SHORTVER), 4.x)
>> CFLAGS+=-minline-all-stringops \
>> -falign-loops \
>> -ftree-vectorize \
>> - -mtune=$(CPU)
>> + -mtune=$(shell echo $(CPU) |sed -e 's/x86_64/x86-64/' )
>> else
>> #if gcc 3.4
>> ifeq ($(CC_SHORTVER), 3.4)
>>
>>
>
> ooops sorry, for Intel processer should be "nocona"
>
> -mtune=$(shell echo $(CPU) |sed -e 's/x86_64/nocona/' )
>
>
>
>
> Wbr,
>
More information about the Devel
mailing list