[ODE] ODE and Itanium processors
tanguy.fautre at spaceapplications.com
Thu Aug 3 01:53:06 MST 2006
Bram Stolk wrote:
> Rodrigo Hernandez wrote:
>> It complains, but continues compiling (the complaint is something along
>> the "all code is location independent on win32" line), originally, I had
>> the flag hard coded
>> into the compiler flags, but it was later conditionally removed because
>> it caused a compilation error on a platform, I don't remember which one
>> though, I wasn't
>> the one who made the change (though I might have applied the patch).
> Well.. for opcode is it hardcoded.
> So all platforms that can build opcode using autoconf can handle
> the flag:
> if OPCODE
> noinst_LIBRARIES+= libOPCODE.a
> libOPCODE_a_CPPFLAGS= -O2 -fno-strict-aliasing -fPIC
> if X86_64_SYSTEM
> libOPCODE_a_CPPFLAGS += -fPIC
> Note: the latter section makes no sense.
>>From the logs:
> r907 | tfautre | 2006-04-04 12:17:09 +0200 (Tue, 04 Apr 2006) | 2 lines
> - made the -fPIC compilation flag only active on X86_64_SYSTEM
> My guess is that the guard can be dropped, or reversed
> to 'if not win32' or something.
> Maybe tfautre has some additional insights?
From looking at svn logs,
"libOPCODE_a_CPPFLAGS= -O2 -fno-strict-aliasing -fPIC" is my mistake. I
added it in r906 while trying to get ODE to compile on x86_64 and I
forgot to remove it in r907 when adding the conditional "if
-fPIC was not present before r906/r907 and thus would not compile on
x86_64. I must admit that I do not know the -fPIC option very well,
hence I added it only to fix the compilation on x86_64 (since regular
x86 was already working), not knowing its implications on other platforms.
More information about the ODE