I think we should make sure the "universal driver" can deal with this, and remove the functionality then.
I'm actively working on this now (I was pre-February at which point I had a brief hiatus for the LLVM conference. Back on again.), and have a patch outstanding on the driver. The problem with testing is a difficult one, but one I'm going to try and solve by building up a database of known driver behaviour - inputs and outputs - for both Clang in its current state and GCC on as many different platforms as I can find (I will be asking the community to help with this at some point). >From there I can create a test suite that should run on any platform and test >all platforms. The ideal goal being that the driver is "known to be as correct >as GCC" on supported platforms, and easy to extend for nonsupported platforms. For the moment, I think this commit should stay in place as-is, there's no point expending extra effort on it I don't think! Cheers, James -----Original Message----- From: Evgeniy Stepanov [mailto:[email protected]] Sent: 24 April 2012 08:59 To: James Molloy Cc: Hal Finkel; [email protected] Subject: Re: [cfe-commits] r154389 - in /cfe/trunk: lib/Driver/Tools.cpp test/Driver/linux-as.c Not a hack, but unimplemented functionality. Ideally, every other platform should also have a set of forwarded (and maybe transformed) options, but I'm reluctant to do this now. I can't test it easily, and I don't know enough about most of them to be sure that I don't screw something up. WDYT? On Mon, Apr 23, 2012 at 6:12 PM, James Molloy <[email protected]> wrote: > It still sounds like a pretty dirty hack to get something to work in the > interim. > > People will wonder why passing -mcpu only works for ARM and not for any other > platform... > > -----Original Message----- > From: [email protected] > [mailto:[email protected]] On Behalf Of Hal Finkel > Sent: 23 April 2012 15:08 > To: Evgeniy Stepanov > Cc: [email protected] > Subject: Re: [cfe-commits] r154389 - in /cfe/trunk: lib/Driver/Tools.cpp > test/Driver/linux-as.c > > LGTM, thanks! > > -Hal > > On Mon, 23 Apr 2012 16:53:41 +0400 > Evgeniy Stepanov <[email protected]> wrote: > >> Ok, I moved this options under if(arm). >> Please take a look at the patch. >> >> On Mon, Apr 23, 2012 at 4:35 PM, Hal Finkel <[email protected]> wrote: >> > On Mon, 23 Apr 2012 07:14:19 -0500 >> > Hal Finkel <[email protected]> wrote: >> > >> >> On Mon, 23 Apr 2012 11:14:04 +0400 >> >> Evgeniy Stepanov <[email protected]> wrote: >> >> >> >> > On Sun, Apr 22, 2012 at 5:41 AM, Hal Finkel <[email protected]> >> >> > wrote: >> >> > > Evgeniy, >> >> > > >> >> > > Unfortunately, we need to be much more careful here. While gas >> >> > > for ARM does support -mcpu, -march and -mfpu, this is not true >> >> > > for other ISAs. Looking at the man page, it seems that the >> >> > > supported flags are: >> >> > > >> >> > > ARM: -mcpu=, -march=, -mfpu= >> >> > > i386: -march= (-mtune=: should we map -mcpu= to -mtune=?) >> >> > > MIPS: (same as i386) >> >> > > PowerPC: supports cpu, but as -m<cpu> (meaning >> >> > > -m403|-m405|-mppc64|...) SPARC: supports arch, but as -xarch= >> >> > > >> >> > > How do you think we should fix this? >> >> > >> >> > Well, we can make the set of forwarded options dependent on the >> >> > target. >> >> >> >> That makes sense. >> >> >> >> > >> >> > But what's the failing use case? I mean, why do you pass these >> >> > options to the compiler in the first place? >> >> >> >> This patch broke building on PPC with a non-default cpu type. I >> >> pass -mcpu=, as you do, to select appropriate cpu features and >> >> itineraries. The default clang behavior on PPC is to pass -many to >> >> the assembler. As you might imagine, this enables instructions for >> >> all known cpu types. With this patch, the assembler also gets the >> >> -mcpu=<cpu> flag provided to the driver. This is incorrect, it >> >> would need to be translated as -m<cpu> for PPC. I think it is >> >> unfortunate that gas cannot accept a common set of command-line >> >> parameters across platforms, but that I cannot fix ;) >> > >> > I should also add that it is quite possible for LLVM to have >> > itineraries for cpus that gas knows nothing about in particular. >> > >> > -Hal >> > >> >> >> >> Thanks again, >> >> Hal >> >> >> >> > >> >> > > >> >> > > -Hal >> >> > > >> >> > > On Tue, 10 Apr 2012 09:05:41 -0000 >> >> > > Evgeniy Stepanov <[email protected]> wrote: >> >> > > >> >> > >> Author: eugenis >> >> > >> Date: Tue Apr 10 04:05:40 2012 >> >> > >> New Revision: 154389 >> >> > >> >> >> > >> URL: http://llvm.org/viewvc/llvm-project?rev=154389&view=rev >> >> > >> Log: >> >> > >> Pass -march, -mcpu, -mfpu to linuxtools assembler. >> >> > >> >> >> > >> Added: >> >> > >> cfe/trunk/test/Driver/linux-as.c >> >> > >> Modified: >> >> > >> cfe/trunk/lib/Driver/Tools.cpp >> >> > >> >> >> > >> Modified: cfe/trunk/lib/Driver/Tools.cpp >> >> > >> URL: >> >> > >> http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Driver/Tools.cpp?rev=154389&r1=154388&r2=154389&view=diff >> >> > >> ============================================================================== >> >> > >> --- cfe/trunk/lib/Driver/Tools.cpp (original) +++ >> >> > >> cfe/trunk/lib/Driver/Tools.cpp Tue Apr 10 04:05:40 2012 @@ >> >> > >> -5089,6 +5089,10 @@ CmdArgs.push_back("-EL"); >> >> > >> } >> >> > >> >> >> > >> + Args.AddLastArg(CmdArgs, options::OPT_march_EQ); >> >> > >> + Args.AddLastArg(CmdArgs, options::OPT_mcpu_EQ); >> >> > >> + Args.AddLastArg(CmdArgs, options::OPT_mfpu_EQ); >> >> > >> + >> >> > >> Args.AddAllArgValues(CmdArgs, options::OPT_Wa_COMMA, >> >> > >> options::OPT_Xassembler); >> >> > >> >> >> > >> >> >> > >> Added: cfe/trunk/test/Driver/linux-as.c >> >> > >> URL: >> >> > >> http://llvm.org/viewvc/llvm-project/cfe/trunk/test/Driver/linux-as.c?rev=154389&view=auto >> >> > >> ============================================================================== >> >> > >> --- cfe/trunk/test/Driver/linux-as.c (added) +++ >> >> > >> cfe/trunk/test/Driver/linux-as.c Tue Apr 10 04:05:40 2012 @@ >> >> > >> -0,0 +1,31 @@ +// Check passing options to the assembler for >> >> > >> ARM targets. +// >> >> > >> +// RUN: %clang -target arm-linux -### \ >> >> > >> +// RUN: -no-integrated-as -c %s 2>&1 \ >> >> > >> +// RUN: | FileCheck -check-prefix=ARM %s >> >> > >> +// CHECK-ARM: as{{(.exe)?}}" >> >> > >> +// >> >> > >> +// RUN: %clang -target arm-linux -mcpu=cortex-a8 -### \ >> >> > >> +// RUN: -no-integrated-as -c %s 2>&1 \ >> >> > >> +// RUN: | FileCheck -check-prefix=ARM-MCPU %s >> >> > >> +// CHECK-ARM-MCPU: as{{(.exe)?}}" "-mcpu=cortex-a8" >> >> > >> +// >> >> > >> +// RUN: %clang -target arm-linux -mfpu=neon -### \ >> >> > >> +// RUN: -no-integrated-as -c %s 2>&1 \ >> >> > >> +// RUN: | FileCheck -check-prefix=ARM-MFPU %s >> >> > >> +// CHECK-ARM-MFPU: as{{(.exe)?}}" "-mfpu=neon" >> >> > >> +// >> >> > >> +// RUN: %clang -target arm-linux -march=armv7-a -### \ >> >> > >> +// RUN: -no-integrated-as -c %s 2>&1 \ >> >> > >> +// RUN: | FileCheck -check-prefix=ARM-MARCH %s >> >> > >> +// CHECK-ARM-MARCH: as{{(.exe)?}}" "-march=armv7-a" >> >> > >> +// >> >> > >> +// RUN: %clang -target arm-linux -mcpu=cortex-a8 -mfpu=neon >> >> > >> -march=armv7-a -### \ +// RUN: -no-integrated-as -c %s 2>&1 >> >> > >> \ +// RUN: | FileCheck -check-prefix=ARM-ALL %s >> >> > >> +// CHECK-ARM-ALL: as{{(.exe)?}}" "-march=armv7-a" >> >> > >> "-mcpu=cortex-a8" "-mfpu=neon" +// >> >> > >> +// RUN: %clang -target armv7-linux -mcpu=cortex-a8 -### \ >> >> > >> +// RUN: -no-integrated-as -c %s 2>&1 \ >> >> > >> +// RUN: | FileCheck -check-prefix=ARM-TARGET %s >> >> > >> +// CHECK-ARM-TARGET: as{{(.exe)?}}" "-mfpu=neon" >> >> > >> "-mcpu=cortex-a8" >> >> > >> >> >> > >> >> >> > >> _______________________________________________ >> >> > >> cfe-commits mailing list >> >> > >> [email protected] >> >> > >> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits >> >> > > >> >> > > >> >> > > >> >> > > -- >> >> > > Hal Finkel >> >> > > Postdoctoral Appointee >> >> > > Leadership Computing Facility >> >> > > Argonne National Laboratory >> >> >> >> >> >> >> > >> > >> > >> > -- >> > Hal Finkel >> > Postdoctoral Appointee >> > Leadership Computing Facility >> > Argonne National Laboratory > > > > -- > Hal Finkel > Postdoctoral Appointee > Leadership Computing Facility > Argonne National Laboratory > > _______________________________________________ > cfe-commits mailing list > [email protected] > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits > > _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
