Re: [edk2-devel] [Patch] BaseTools tools_def.template: Add back -fno-pie option in GCC49 tool chain

["Michael D Kinney" <michael.d.kinney@intel.com>]

Liming,

Can you please provide a few more details on the failure.

For the UnitTestFrameworkPkg patch set, I had to add the
following to get host based unit test applications to build
and run.  I was seeing link failures between FW libs and 
host libs when building the POSIX host application.

[BuildOptions]
  GCC:*_*_*_CC_FLAGS = -fno-pie

I think with the proposed change below, I could remove
this. Do you agree?

Thanks,

Mike


> -----Original Message-----
> From: devel@edk2.groups.io <devel@edk2.groups.io> On
> Behalf Of Liming Gao
> Sent: Tuesday, February 4, 2020 4:52 AM
> To: Laszlo Ersek <lersek@redhat.com>;
> devel@edk2.groups.io
> Cc: Feng, Bob C <bob.c.feng@intel.com>; Ard Biesheuvel
> <ard.biesheuvel@linaro.org>; Gao, Liming
> <liming.gao@intel.com>
> Subject: Re: [edk2-devel] [Patch] BaseTools
> tools_def.template: Add back -fno-pie option in GCC49
> tool chain
> 
> Laszlo:
> 
> > -----Original Message-----
> > From: Laszlo Ersek <lersek@redhat.com>
> > Sent: Tuesday, February 4, 2020 8:02 PM
> > To: devel@edk2.groups.io; Gao, Liming
> <liming.gao@intel.com>
> > Cc: Feng, Bob C <bob.c.feng@intel.com>; Ard
> Biesheuvel <ard.biesheuvel@linaro.org>
> > Subject: Re: [edk2-devel] [Patch] BaseTools
> tools_def.template: Add back -fno-pie option in GCC49
> tool chain
> >
> > (+Ard)
> >
> > On 02/04/20 05:54, Liming Gao wrote:
> > > BZ:
> https://bugzilla.tianocore.org/show_bug.cgi?id=2502
> > > This option is required to make GCC49 tool chain
> work with the high
> > > version GCC compiler.
> > >
> > > Cc: Bob Feng <bob.c.feng@intel.com>
> > > Signed-off-by: Liming Gao <liming.gao@intel.com>
> > > ---
> > >  BaseTools/Conf/tools_def.template | 4 ++--
> > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/BaseTools/Conf/tools_def.template
> b/BaseTools/Conf/tools_def.template
> > > index feee2bbf16..d02424ae44 100755
> > > --- a/BaseTools/Conf/tools_def.template
> > > +++ b/BaseTools/Conf/tools_def.template
> > > @@ -1974,7 +1974,7 @@ DEFINE
> GCC48_ARM_ASLDLINK_FLAGS      =
> DEF(GCC_ARM_ASLDLINK_FLAGS) -Wl,--oformat
> > >  DEFINE GCC48_AARCH64_ASLDLINK_FLAGS  =
> DEF(GCC_AARCH64_ASLDLINK_FLAGS)
> > >  DEFINE GCC48_ASLCC_FLAGS             =
> DEF(GCC_ASLCC_FLAGS)
> > >
> > > -DEFINE GCC49_IA32_CC_FLAGS           =
> DEF(GCC48_IA32_CC_FLAGS)
> > > +DEFINE GCC49_IA32_CC_FLAGS           =
> DEF(GCC48_IA32_CC_FLAGS) -fno-pic -fno-pie
> > >  DEFINE GCC49_X64_CC_FLAGS            =
> DEF(GCC48_X64_CC_FLAGS)
> > >  DEFINE GCC49_IA32_X64_DLINK_COMMON   = -nostdlib -
> Wl,-n,-q,--gc-sections -z common-page-size=0x40
> > >  DEFINE GCC49_IA32_X64_ASLDLINK_FLAGS =
> DEF(GCC49_IA32_X64_DLINK_COMMON) -Wl,--
> defsym=PECOFF_HEADER_SIZE=0
> > DEF(GCC_DLINK2_FLAGS_COMMON) -Wl,--
> entry,ReferenceAcpiTable -u ReferenceAcpiTable
> > > @@ -1997,7 +1997,7 @@ DEFINE
> GCC49_ARM_ASLDLINK_FLAGS      =
> DEF(GCC48_ARM_ASLDLINK_FLAGS)
> > >  DEFINE GCC49_AARCH64_ASLDLINK_FLAGS  =
> DEF(GCC48_AARCH64_ASLDLINK_FLAGS)
> > >  DEFINE GCC49_ASLCC_FLAGS             =
> DEF(GCC48_ASLCC_FLAGS)
> > >
> > > -DEFINE GCC5_IA32_CC_FLAGS            =
> DEF(GCC49_IA32_CC_FLAGS) -fno-pic -fno-pie
> > > +DEFINE GCC5_IA32_CC_FLAGS            =
> DEF(GCC49_IA32_CC_FLAGS)
> > >  DEFINE GCC5_X64_CC_FLAGS             =
> DEF(GCC49_X64_CC_FLAGS)
> > >  DEFINE GCC5_IA32_X64_DLINK_COMMON    =
> DEF(GCC49_IA32_X64_DLINK_COMMON)
> > >  DEFINE GCC5_IA32_X64_ASLDLINK_FLAGS  =
> DEF(GCC49_IA32_X64_ASLDLINK_FLAGS)
> > >
> >
> > - What has changed relative to commit 11d0cd23dd1b
> ("BaseTools/tools_def
> > IA32: drop -no-pie linker option for GCC49", 2018-06-
> 18)?
> >
> > - Also, if we are reverting one half of 11d0cd23dd1b
> (the compiler
> > flags), shouldn't we then revert the other half too
> (the linker flags)?
> 
> Yes. Half change is revert. CC_FLAGS is added back.
> DLINK flag is not,
> because GCC4.9 doesn't know the link option -no-pie.
> But, GCC 4.9 accepts the CC option -fno-pie.
> I verify this change. CC flags -fno-pie can resolve the
> build failure with GCC7.4. I also see -fno-pie option
> Is in GCC ARM and AARCH64 arch. So, I think this change
> is enough.
> 
> >
> > - The commit message says, "work with the high
> version GCC compiler".
> > What does that mean? If it is 4.9.x, with x>2, then I
> agree the patch is
> > justified (because commit 11d0cd23dd1b was apparently
> made for 4.9.2).
> > But if the phrase stands for gcc8 or so (just an
> example), then I don't
> > think the patch is a good idea; users of gcc8 can
> just specify the GCC5
> > toolchain.
> >
> > Ah, indeed, I need only look at TianoCore#2502:
> >
> > "GCC49 tool chain meets with the build failure when
> GCC7.4 compiler".
> >
> > So I think this approach is wrong. Unless there is a
> new gcc-4.9.x
> > release, i.e., after gcc-4.9.2, I think we still need
> commit
> > 11d0cd23dd1b in place. And, please use GCC5 for gcc-
> 7.4 -- is there a
> > problem with that?
> 
> By design, GCC49 can work with the high version GCC
> compiler like GCC5.
> GCC49 is the tool chain without LTO enable. GCC5 is the
> tool chain with LTO.
> So, they are for two different GCC setting. They should
> both support
> high version GCC compiler. GCC49 supported GCC compiler
> version is from GCC 4.9.
> GCC5 supported GCC compiler version is from GCC 5.0. I
> know GCC49 or GCC5 tool chain
> name brings a little confuse. I will add more detail
> info in tools_def.txt for them.
> 
> Thanks
> Liming
> >
> > Thanks
> > Laszlo
> 
> 
>