Re: [dpdk-stable] [dpdk-dev] [PATCH 1/2] mk: fix unsupported flag error on armhf architercture

[Thomas Monjalon <thomas@monjalon.net>]

12/11/2019 07:40, Phil Yang (Arm Technology China):
> > -----Original Message-----
> > From: Thomas Monjalon <thomas@monjalon.net>
> > Sent: Tuesday, November 12, 2019 2:03 PM
> > To: Phil Yang (Arm Technology China) <Phil.Yang@arm.com>
> > Cc: Bruce Richardson <bruce.richardson@intel.com>; dev@dpdk.org;
> > david.marchand@redhat.com; jerinj@marvell.com; Honnappa Nagarahalli
> > <Honnappa.Nagarahalli@arm.com>; Gavin Hu (Arm Technology China)
> > <Gavin.Hu@arm.com>; Joyce Kong (Arm Technology China)
> > <Joyce.Kong@arm.com>; nd <nd@arm.com>; stable@dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH 1/2] mk: fix unsupported flag error on armhf
> > architercture
> > 
> > 12/11/2019 06:25, Phil Yang (Arm Technology China):
> > > From: Bruce Richardson <bruce.richardson@intel.com>
> > > > On Mon, Nov 11, 2019 at 06:34:20PM +0800, Phil Yang wrote:
> > > > > The older version (e.g. version 7.4.0 ) of GNU C compiler for the armhf
> > > > > architecture doesn't support the flag '-Wno-address-of-packed-
> > member',
> > > > > so remove this flag for aarch32.
> > > > >
> > > > > Fixes: a385972c3675 ("mk: disable warning for packed member pointer")
> > > > > Cc: stable@dpdk.org
> > > > >
> > > > > Signed-off-by: Phil Yang <phil.yang@arm.com>
> > > > > Reviewed-by: Gavin Hu <gavin.hu@arm.com>
> > > > > Tested-by: Joyce Kong <joyce.kong@arm.com>
> > > > >
> > > > > ---
> > > > >  mk/toolchain/gcc/rte.vars.mk | 2 ++
> > > > >  1 file changed, 2 insertions(+)
> > > > >
> > > > > diff --git a/mk/toolchain/gcc/rte.vars.mk
> > b/mk/toolchain/gcc/rte.vars.mk
> > > > > index 9fc7041..ec0cbbf 100644
> > > > > --- a/mk/toolchain/gcc/rte.vars.mk
> > > > > +++ b/mk/toolchain/gcc/rte.vars.mk
> > > > > @@ -100,7 +100,9 @@ WERROR_FLAGS += -Wno-format-truncation
> > > > >  endif
> > > > >
> > > > >  # disable packed member unalign warnings
> > > > > +ifneq ($(CONFIG_RTE_ARCH_ARM), y)
> > > > >  WERROR_FLAGS += -Wno-address-of-packed-member
> > > > > +endif
> > > > >
> > > > You don't need to do this, as gcc will not complain about this unknown
> > flag
> > > > unless you have other issues in your code.[1] I think it's better to keep
> > the
> > > > code clean in this case, otherwise we'll have the code littered with
> > > > conditionals for various flags.
> > > >
> > > > /Bruce
> > > >
> > > > [1] https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
> > > > "When an unrecognized warning option is requested (e.g.,
> > > > -Wunknown-warning), GCC emits a diagnostic stating that the option is
> > not
> > > > recognized. However, if the -Wno- form is used, the behavior is slightly
> > > > different: no diagnostic is produced for -Wno-unknown-warning unless
> > other
> > > > diagnostics are being produced. This allows the use of new -Wno- options
> > > > with old compilers, but if something goes wrong, the compiler warns that
> > an
> > > > unrecognized option is present."
> > >
> > > Thanks, Bruce.
> > >
> > > There are thousands of warnings with alignment when compiling for
> > ARMv7 (ARMv7 supports unaligned memory access).
> > > Without this new flag, the build system works fine for ARMv7.  So I add this
> > conditional for ARMv7 only.
> > > I agree with you we should clean up the code, but it needs a lot of effort to
> > achieve that.
> > > This patch is going to make the ARMv7 target build successfully during this
> > period.
> > 
> > I would go the opposite direction.
> > I think we should remove all warnings disablement like "-Wno-something",
> > and clean the code.
> 
> Thanks, Thomas.
> 
> I think it is a compiler issue rather than a code defect in this case, as It cannot recognize the '-Waddress-of-packed-member' flag. 
> "arm-linux-gnueabihf-gcc: error: unrecognized command line option '-Waddress-of-packed-member'"
> 
> > 
> > If you want to allow building ARMv7, you can just allow warnings as not
> > errors.
> > So we will still see something is wrong but it won't prevent from building.
> 
> Yes. We did this already, all the alignment issues are emitted as warnings on ARMv7. 
> Please see these two commits:
> 79fa5c2a9992 ("mk: ignore alignment errors for ARMv7")
> 6433c19105e8 ("mk: generalize strict alignment warning handling")

I don't know what to do with this patch.
I think it is not how it should work,
but you say your compiler has a real bug.

Can you confirm which compiler is it?
Does it prevent to compile even with -Wno-error ?

Any other opinion?