diff mbox series

[v2,3/4] Kbuild: make DWARF version a choice

Message ID 20201104005343.4192504-4-ndesaulniers@google.com (mailing list archive)
State New
Headers show
Series Kbuild: DWARF v5 support | expand

Commit Message

Nick Desaulniers Nov. 4, 2020, 12:53 a.m. UTC
Modifies CONFIG_DEBUG_INFO_DWARF4 to be a member of a choice. Adds an
explicit CONFIG_DEBUG_INFO_DWARF2, which is the default. Does so in a
way that's forward compatible with existing configs, and makes adding
future versions more straightforward.

Suggested-by: Fangrui Song <maskray@google.com>
Suggested-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
---
 Makefile          | 14 ++++++++------
 lib/Kconfig.debug | 19 +++++++++++++++----
 2 files changed, 23 insertions(+), 10 deletions(-)

Comments

kernel test robot Nov. 5, 2020, 5:58 a.m. UTC | #1
Hi Nick,

I love your patch! Yet something to improve:

[auto build test ERROR on kbuild/for-next]
[also build test ERROR on asm-generic/master linus/master v5.10-rc2 next-20201104]
[cannot apply to tip/x86/core]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Nick-Desaulniers/Kbuild-DWARF-v5-support/20201104-085610
base:   https://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild.git for-next
config: ia64-randconfig-r034-20201104 (attached as .config)
compiler: ia64-linux-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/f8b8c21feb85f4422b79a96e0b56cf4e8ff53274
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Nick-Desaulniers/Kbuild-DWARF-v5-support/20201104-085610
        git checkout f8b8c21feb85f4422b79a96e0b56cf4e8ff53274
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=ia64 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   {standard input}: Assembler messages:
>> {standard input}:34: Error: file number 1 already allocated

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
Arvind Sankar Nov. 23, 2020, 11:22 p.m. UTC | #2
On Tue, Nov 03, 2020 at 04:53:42PM -0800, Nick Desaulniers wrote:
> Modifies CONFIG_DEBUG_INFO_DWARF4 to be a member of a choice. Adds an
> explicit CONFIG_DEBUG_INFO_DWARF2, which is the default. Does so in a
> way that's forward compatible with existing configs, and makes adding
> future versions more straightforward.
> 
> Suggested-by: Fangrui Song <maskray@google.com>
> Suggested-by: Masahiro Yamada <masahiroy@kernel.org>
> Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
> ---
>  Makefile          | 14 ++++++++------
>  lib/Kconfig.debug | 19 +++++++++++++++----
>  2 files changed, 23 insertions(+), 10 deletions(-)
> 
> diff --git a/Makefile b/Makefile
> index 75b1a3dcbf30..e23786a4c1c7 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -826,12 +826,14 @@ else
>  DEBUG_CFLAGS	+= -g
>  endif
>  
> -ifndef LLVM_IAS
> -KBUILD_AFLAGS	+= -Wa,-gdwarf-2
> -endif
> -
> -ifdef CONFIG_DEBUG_INFO_DWARF4
> -DEBUG_CFLAGS	+= -gdwarf-4
> +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF2) := 2
> +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF4) := 4
> +DEBUG_CFLAGS	+= -gdwarf-$(dwarf-version-y)
> +ifneq ($(dwarf-version-y)$(LLVM_IAS),21)
> +# Binutils 2.35+ required for -gdwarf-4+ support.
> +dwarf-aflag	:= $(call as-option,-Wa$(comma)-gdwarf-$(dwarf-version-y))
> +DEBUG_CFLAGS	+= $(dwarf-aflag)
> +KBUILD_AFLAGS	+= $(dwarf-aflag)
>  endif
>  

For LLVM_IAS=1, adding dwarf-aflag to DEBUG_CFLAGS should not be
necessary, no? This seems to add it for dwarf-4.

The as-option check will only work on binutils 2.35.1 onwards: earlier
versions will silently accept any -gdwarf-N option. Do we care? I think
it'll just get dwarf-2 for assembly files even though dwarf-4 might have
been configured. The earlier versions only error if the double-hyphen
form --gdwarf-N is used.

More generally, do we want to force this option via -Wa or should we
leave it up to the compiler driver when we can? For both Clang/IAS and
gcc/binutils, passing -gdwarf-N in KBUILD_AFLAGS will allow the compiler
to pass on the appropriate option to the assembler (with gcc you only
get --gdwarf-2 for the assembler except on trunk though). The only case
that would absolutely require -Wa is Clang without IAS, might be worth
adding the ability to pass on the flag to the external assembler?

Btw, is -gsplit-dwarf at all useful for assembler files?
Segher Boessenkool Nov. 24, 2020, 12:33 a.m. UTC | #3
On Mon, Nov 23, 2020 at 06:22:10PM -0500, Arvind Sankar wrote:
> Btw, is -gsplit-dwarf at all useful for assembler files?

If you invoke the assembler via the compiler, with that flag it still
creates separate .o and .dwo files (via objcopy invocations as usual).
Whether that is useful depends on if you have any debug info that can
be split :-)


Segher
Arvind Sankar Nov. 24, 2020, 4:56 p.m. UTC | #4
On Mon, Nov 23, 2020 at 06:33:57PM -0600, Segher Boessenkool wrote:
> On Mon, Nov 23, 2020 at 06:22:10PM -0500, Arvind Sankar wrote:
> > Btw, is -gsplit-dwarf at all useful for assembler files?
> 
> If you invoke the assembler via the compiler, with that flag it still
> creates separate .o and .dwo files (via objcopy invocations as usual).
> Whether that is useful depends on if you have any debug info that can
> be split :-)
> 
> 
> Segher

Right, the latter was what I was really asking :) We don't currently
pass -gsplit-dwarf for assembler and I was wondering if that mattered.

Thanks.
Segher Boessenkool Nov. 24, 2020, 5:46 p.m. UTC | #5
On Tue, Nov 24, 2020 at 11:56:02AM -0500, Arvind Sankar wrote:
> On Mon, Nov 23, 2020 at 06:33:57PM -0600, Segher Boessenkool wrote:
> > On Mon, Nov 23, 2020 at 06:22:10PM -0500, Arvind Sankar wrote:
> > > Btw, is -gsplit-dwarf at all useful for assembler files?
> > 
> > If you invoke the assembler via the compiler, with that flag it still
> > creates separate .o and .dwo files (via objcopy invocations as usual).
> > Whether that is useful depends on if you have any debug info that can
> > be split :-)
> 
> Right, the latter was what I was really asking :) We don't currently
> pass -gsplit-dwarf for assembler and I was wondering if that mattered.

If there is any debug info in the .s files, it will all end up in the .o
file, not a .dwo file.  That may matter aesthetically, and it can cost a
few bytes of disk space, but it doesn't matter functionally (GDB will
search in both places).


Segher
Masahiro Yamada Nov. 30, 2020, 6:04 p.m. UTC | #6
On Wed, Nov 4, 2020 at 9:53 AM 'Nick Desaulniers' via Clang Built
Linux <clang-built-linux@googlegroups.com> wrote:
>
> Modifies CONFIG_DEBUG_INFO_DWARF4 to be a member of a choice. Adds an
> explicit CONFIG_DEBUG_INFO_DWARF2, which is the default. Does so in a
> way that's forward compatible with existing configs, and makes adding
> future versions more straightforward.
>
> Suggested-by: Fangrui Song <maskray@google.com>
> Suggested-by: Masahiro Yamada <masahiroy@kernel.org>
> Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
> ---
>  Makefile          | 14 ++++++++------
>  lib/Kconfig.debug | 19 +++++++++++++++----
>  2 files changed, 23 insertions(+), 10 deletions(-)
>
> diff --git a/Makefile b/Makefile
> index 75b1a3dcbf30..e23786a4c1c7 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -826,12 +826,14 @@ else
>  DEBUG_CFLAGS   += -g
>  endif
>
> -ifndef LLVM_IAS
> -KBUILD_AFLAGS  += -Wa,-gdwarf-2
> -endif
> -
> -ifdef CONFIG_DEBUG_INFO_DWARF4
> -DEBUG_CFLAGS   += -gdwarf-4
> +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF2) := 2
> +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF4) := 4
> +DEBUG_CFLAGS   += -gdwarf-$(dwarf-version-y)
> +ifneq ($(dwarf-version-y)$(LLVM_IAS),21)
> +# Binutils 2.35+ required for -gdwarf-4+ support.
> +dwarf-aflag    := $(call as-option,-Wa$(comma)-gdwarf-$(dwarf-version-y))
> +DEBUG_CFLAGS   += $(dwarf-aflag)

This changes the behavior.

For the Dwarf-2 case,

Previously, -gdwarf-2 was passed to $(CC),
so the debug info was generated by gcc.

Now, -Wa,-gdwarf-2 is passed to $(CC).
-gdwarf-2 is handled by GNU as.
So, the source info points to /tmp/<hash>.s
instead of the original .c file.



Handling the Dwarf capability is very complicated.

Are you still working for v3?



> +KBUILD_AFLAGS  += $(dwarf-aflag)
>  endif
>
>  ifdef CONFIG_DEBUG_INFO_REDUCED
> diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> index 83a860126897..03c494eefabd 100644
> --- a/lib/Kconfig.debug
> +++ b/lib/Kconfig.debug
> @@ -256,14 +256,25 @@ config DEBUG_INFO_SPLIT
>           to know about the .dwo files and include them.
>           Incompatible with older versions of ccache.
>
> +choice
> +       prompt "DWARF version"
> +       help
> +         Which version of DWARF debug info to emit.
> +
> +config DEBUG_INFO_DWARF2
> +       bool "Generate DWARF v2 debuginfo"
> +       help
> +         Generate DWARF v2 debug info.
> +
>  config DEBUG_INFO_DWARF4
>         bool "Generate dwarf4 debuginfo"
>         depends on $(cc-option,-gdwarf-4)
>         help
> -         Generate dwarf4 debug info. This requires recent versions
> -         of gcc and gdb. It makes the debug information larger.
> -         But it significantly improves the success of resolving
> -         variables in gdb on optimized code.
> +         Generate DWARF v4 debug info. This requires gcc 4.5+ and gdb 7.0+.
> +         It makes the debug information larger, but it significantly
> +         improves the success of resolving variables in gdb on optimized code.
> +
> +endchoice # "DWARF version"
>
>  config DEBUG_INFO_BTF
>         bool "Generate BTF typeinfo"
> --
> 2.29.1.341.ge80a0c044ae-goog
>
> --
> You received this message because you are subscribed to the Google Groups "Clang Built Linux" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to clang-built-linux+unsubscribe@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/clang-built-linux/20201104005343.4192504-4-ndesaulniers%40google.com.
Fāng-ruì Sòng Nov. 30, 2020, 8:27 p.m. UTC | #7
On Mon, Nov 30, 2020 at 10:05 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> On Wed, Nov 4, 2020 at 9:53 AM 'Nick Desaulniers' via Clang Built
> Linux <clang-built-linux@googlegroups.com> wrote:
> >
> > Modifies CONFIG_DEBUG_INFO_DWARF4 to be a member of a choice. Adds an
> > explicit CONFIG_DEBUG_INFO_DWARF2, which is the default. Does so in a
> > way that's forward compatible with existing configs, and makes adding
> > future versions more straightforward.
> >
> > Suggested-by: Fangrui Song <maskray@google.com>
> > Suggested-by: Masahiro Yamada <masahiroy@kernel.org>
> > Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
> > ---
> >  Makefile          | 14 ++++++++------
> >  lib/Kconfig.debug | 19 +++++++++++++++----
> >  2 files changed, 23 insertions(+), 10 deletions(-)
> >
> > diff --git a/Makefile b/Makefile
> > index 75b1a3dcbf30..e23786a4c1c7 100644
> > --- a/Makefile
> > +++ b/Makefile
> > @@ -826,12 +826,14 @@ else
> >  DEBUG_CFLAGS   += -g
> >  endif
> >
> > -ifndef LLVM_IAS
> > -KBUILD_AFLAGS  += -Wa,-gdwarf-2
> > -endif
> > -
> > -ifdef CONFIG_DEBUG_INFO_DWARF4
> > -DEBUG_CFLAGS   += -gdwarf-4
> > +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF2) := 2
> > +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF4) := 4
> > +DEBUG_CFLAGS   += -gdwarf-$(dwarf-version-y)
> > +ifneq ($(dwarf-version-y)$(LLVM_IAS),21)
> > +# Binutils 2.35+ required for -gdwarf-4+ support.
> > +dwarf-aflag    := $(call as-option,-Wa$(comma)-gdwarf-$(dwarf-version-y))
> > +DEBUG_CFLAGS   += $(dwarf-aflag)
>
> This changes the behavior.
>
> For the Dwarf-2 case,
>
> Previously, -gdwarf-2 was passed to $(CC),
> so the debug info was generated by gcc.
>
> Now, -Wa,-gdwarf-2 is passed to $(CC).
> -gdwarf-2 is handled by GNU as.
> So, the source info points to /tmp/<hash>.s
> instead of the original .c file.
>
>
>
> Handling the Dwarf capability is very complicated.
>
> Are you still working for v3?
>
>
>
> > +KBUILD_AFLAGS  += $(dwarf-aflag)
> >  endif
> >
> >  ifdef CONFIG_DEBUG_INFO_REDUCED
> > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> > index 83a860126897..03c494eefabd 100644
> > --- a/lib/Kconfig.debug
> > +++ b/lib/Kconfig.debug
> > @@ -256,14 +256,25 @@ config DEBUG_INFO_SPLIT
> >           to know about the .dwo files and include them.
> >           Incompatible with older versions of ccache.
> >
> > +choice
> > +       prompt "DWARF version"
> > +       help
> > +         Which version of DWARF debug info to emit.
> > +
> > +config DEBUG_INFO_DWARF2
> > +       bool "Generate DWARF v2 debuginfo"
> > +       help
> > +         Generate DWARF v2 debug info.
> > +
> >  config DEBUG_INFO_DWARF4
> >         bool "Generate dwarf4 debuginfo"
> >         depends on $(cc-option,-gdwarf-4)
> >         help
> > -         Generate dwarf4 debug info. This requires recent versions
> > -         of gcc and gdb. It makes the debug information larger.
> > -         But it significantly improves the success of resolving
> > -         variables in gdb on optimized code.
> > +         Generate DWARF v4 debug info. This requires gcc 4.5+ and gdb 7.0+.
> > +         It makes the debug information larger, but it significantly
> > +         improves the success of resolving variables in gdb on optimized code.
> > +
> > +endchoice # "DWARF version"
> >
> >  config DEBUG_INFO_BTF
> >         bool "Generate BTF typeinfo"
> > --
> > 2.29.1.341.ge80a0c044ae-goog


We can bump -Wa,-gdwarf-2 to -Wa,-gdwarf-3 since GNU actually emits
DWARF v3 DW_AT_ranges (see
https://sourceware.org/bugzilla/show_bug.cgi?id=26850 )
This can avoid the `warning: DWARF2 only supports one section per
compilation unit` warning for Clang.

Deleting -Wa,-gdwarf-2 also sounds good to me if people can verify
their debugging experience is not regressed (I believe it is useless).
Nick Desaulniers Nov. 30, 2020, 8:45 p.m. UTC | #8
On Mon, Nov 30, 2020 at 10:05 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> On Wed, Nov 4, 2020 at 9:53 AM 'Nick Desaulniers' via Clang Built
> Linux <clang-built-linux@googlegroups.com> wrote:
> >
> > Modifies CONFIG_DEBUG_INFO_DWARF4 to be a member of a choice. Adds an
> > explicit CONFIG_DEBUG_INFO_DWARF2, which is the default. Does so in a
> > way that's forward compatible with existing configs, and makes adding
> > future versions more straightforward.
> >
> > Suggested-by: Fangrui Song <maskray@google.com>
> > Suggested-by: Masahiro Yamada <masahiroy@kernel.org>
> > Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
> > ---
> >  Makefile          | 14 ++++++++------
> >  lib/Kconfig.debug | 19 +++++++++++++++----
> >  2 files changed, 23 insertions(+), 10 deletions(-)
> >
> > diff --git a/Makefile b/Makefile
> > index 75b1a3dcbf30..e23786a4c1c7 100644
> > --- a/Makefile
> > +++ b/Makefile
> > @@ -826,12 +826,14 @@ else
> >  DEBUG_CFLAGS   += -g
> >  endif
> >
> > -ifndef LLVM_IAS
> > -KBUILD_AFLAGS  += -Wa,-gdwarf-2
> > -endif
> > -
> > -ifdef CONFIG_DEBUG_INFO_DWARF4
> > -DEBUG_CFLAGS   += -gdwarf-4
> > +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF2) := 2
> > +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF4) := 4
> > +DEBUG_CFLAGS   += -gdwarf-$(dwarf-version-y)
> > +ifneq ($(dwarf-version-y)$(LLVM_IAS),21)
> > +# Binutils 2.35+ required for -gdwarf-4+ support.
> > +dwarf-aflag    := $(call as-option,-Wa$(comma)-gdwarf-$(dwarf-version-y))
> > +DEBUG_CFLAGS   += $(dwarf-aflag)
>
> This changes the behavior.
>
> For the Dwarf-2 case,
>
> Previously, -gdwarf-2 was passed to $(CC),
> so the debug info was generated by gcc.
>
> Now, -Wa,-gdwarf-2 is passed to $(CC).
> -gdwarf-2 is handled by GNU as.
> So, the source info points to /tmp/<hash>.s
> instead of the original .c file.
>
>
>
> Handling the Dwarf capability is very complicated.
>
> Are you still working for v3?

Yes, I plan to revisit the series based on all of the feedback thus
far.  Lately I'm focused on enabling LLVM_IAS=1 for Android; but I
would like to see this land so that the Linux kernel may provide
coverage and feedback to the toolchain developers for DWARF v5 (as
well as reduced binary image sizes).  Maybe later this week I'll have
time to revisit.
Masahiro Yamada Dec. 1, 2020, 3:38 a.m. UTC | #9
On Tue, Dec 1, 2020 at 5:27 AM 'Fāng-ruì Sòng' via Clang Built Linux
<clang-built-linux@googlegroups.com> wrote:
>
> On Mon, Nov 30, 2020 at 10:05 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
> >
> > On Wed, Nov 4, 2020 at 9:53 AM 'Nick Desaulniers' via Clang Built
> > Linux <clang-built-linux@googlegroups.com> wrote:
> > >
> > > Modifies CONFIG_DEBUG_INFO_DWARF4 to be a member of a choice. Adds an
> > > explicit CONFIG_DEBUG_INFO_DWARF2, which is the default. Does so in a
> > > way that's forward compatible with existing configs, and makes adding
> > > future versions more straightforward.
> > >
> > > Suggested-by: Fangrui Song <maskray@google.com>
> > > Suggested-by: Masahiro Yamada <masahiroy@kernel.org>
> > > Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
> > > ---
> > >  Makefile          | 14 ++++++++------
> > >  lib/Kconfig.debug | 19 +++++++++++++++----
> > >  2 files changed, 23 insertions(+), 10 deletions(-)
> > >
> > > diff --git a/Makefile b/Makefile
> > > index 75b1a3dcbf30..e23786a4c1c7 100644
> > > --- a/Makefile
> > > +++ b/Makefile
> > > @@ -826,12 +826,14 @@ else
> > >  DEBUG_CFLAGS   += -g
> > >  endif
> > >
> > > -ifndef LLVM_IAS
> > > -KBUILD_AFLAGS  += -Wa,-gdwarf-2
> > > -endif
> > > -
> > > -ifdef CONFIG_DEBUG_INFO_DWARF4
> > > -DEBUG_CFLAGS   += -gdwarf-4
> > > +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF2) := 2
> > > +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF4) := 4
> > > +DEBUG_CFLAGS   += -gdwarf-$(dwarf-version-y)
> > > +ifneq ($(dwarf-version-y)$(LLVM_IAS),21)
> > > +# Binutils 2.35+ required for -gdwarf-4+ support.
> > > +dwarf-aflag    := $(call as-option,-Wa$(comma)-gdwarf-$(dwarf-version-y))
> > > +DEBUG_CFLAGS   += $(dwarf-aflag)
> >
> > This changes the behavior.
> >
> > For the Dwarf-2 case,
> >
> > Previously, -gdwarf-2 was passed to $(CC),
> > so the debug info was generated by gcc.
> >
> > Now, -Wa,-gdwarf-2 is passed to $(CC).
> > -gdwarf-2 is handled by GNU as.
> > So, the source info points to /tmp/<hash>.s
> > instead of the original .c file.
> >
> >
> >
> > Handling the Dwarf capability is very complicated.
> >
> > Are you still working for v3?
> >
> >
> >
> > > +KBUILD_AFLAGS  += $(dwarf-aflag)
> > >  endif
> > >
> > >  ifdef CONFIG_DEBUG_INFO_REDUCED
> > > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> > > index 83a860126897..03c494eefabd 100644
> > > --- a/lib/Kconfig.debug
> > > +++ b/lib/Kconfig.debug
> > > @@ -256,14 +256,25 @@ config DEBUG_INFO_SPLIT
> > >           to know about the .dwo files and include them.
> > >           Incompatible with older versions of ccache.
> > >
> > > +choice
> > > +       prompt "DWARF version"
> > > +       help
> > > +         Which version of DWARF debug info to emit.
> > > +
> > > +config DEBUG_INFO_DWARF2
> > > +       bool "Generate DWARF v2 debuginfo"
> > > +       help
> > > +         Generate DWARF v2 debug info.
> > > +
> > >  config DEBUG_INFO_DWARF4
> > >         bool "Generate dwarf4 debuginfo"
> > >         depends on $(cc-option,-gdwarf-4)
> > >         help
> > > -         Generate dwarf4 debug info. This requires recent versions
> > > -         of gcc and gdb. It makes the debug information larger.
> > > -         But it significantly improves the success of resolving
> > > -         variables in gdb on optimized code.
> > > +         Generate DWARF v4 debug info. This requires gcc 4.5+ and gdb 7.0+.
> > > +         It makes the debug information larger, but it significantly
> > > +         improves the success of resolving variables in gdb on optimized code.
> > > +
> > > +endchoice # "DWARF version"
> > >
> > >  config DEBUG_INFO_BTF
> > >         bool "Generate BTF typeinfo"
> > > --
> > > 2.29.1.341.ge80a0c044ae-goog
>
>
> We can bump -Wa,-gdwarf-2 to -Wa,-gdwarf-3 since GNU actually emits
> DWARF v3 DW_AT_ranges (see
> https://sourceware.org/bugzilla/show_bug.cgi?id=26850 )
> This can avoid the `warning: DWARF2 only supports one section per
> compilation unit` warning for Clang.


I am not a DWARF spec expert.

Please teach me.

In my understanding, "DWARF2 only supports one section ..."
is warned only when building .S files with LLVM_IAS=1

If this is due to the limitation of DWARF v2, why is it OK to
build .c files with LLVM_IAS?



>
> Deleting -Wa,-gdwarf-2 also sounds good to me if people can verify
> their debugging experience is not regressed (I believe it is useless).
>
>
> --
> 宋方睿
>
> --
> You received this message because you are subscribed to the Google Groups "Clang Built Linux" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to clang-built-linux+unsubscribe@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/clang-built-linux/CAFP8O3Ki9HoqcV450fn29fBOWAbmuGAdB6USLz8pGsW4Vzf7sg%40mail.gmail.com.
Masahiro Yamada Dec. 1, 2020, 4:17 a.m. UTC | #10
On Tue, Dec 1, 2020 at 5:45 AM 'Nick Desaulniers' via Clang Built
Linux <clang-built-linux@googlegroups.com> wrote:
>
> On Mon, Nov 30, 2020 at 10:05 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
> >
> > On Wed, Nov 4, 2020 at 9:53 AM 'Nick Desaulniers' via Clang Built
> > Linux <clang-built-linux@googlegroups.com> wrote:
> > >
> > > Modifies CONFIG_DEBUG_INFO_DWARF4 to be a member of a choice. Adds an
> > > explicit CONFIG_DEBUG_INFO_DWARF2, which is the default. Does so in a
> > > way that's forward compatible with existing configs, and makes adding
> > > future versions more straightforward.
> > >
> > > Suggested-by: Fangrui Song <maskray@google.com>
> > > Suggested-by: Masahiro Yamada <masahiroy@kernel.org>
> > > Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
> > > ---
> > >  Makefile          | 14 ++++++++------
> > >  lib/Kconfig.debug | 19 +++++++++++++++----
> > >  2 files changed, 23 insertions(+), 10 deletions(-)
> > >
> > > diff --git a/Makefile b/Makefile
> > > index 75b1a3dcbf30..e23786a4c1c7 100644
> > > --- a/Makefile
> > > +++ b/Makefile
> > > @@ -826,12 +826,14 @@ else
> > >  DEBUG_CFLAGS   += -g
> > >  endif
> > >
> > > -ifndef LLVM_IAS
> > > -KBUILD_AFLAGS  += -Wa,-gdwarf-2
> > > -endif
> > > -
> > > -ifdef CONFIG_DEBUG_INFO_DWARF4
> > > -DEBUG_CFLAGS   += -gdwarf-4
> > > +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF2) := 2
> > > +dwarf-version-$(CONFIG_DEBUG_INFO_DWARF4) := 4
> > > +DEBUG_CFLAGS   += -gdwarf-$(dwarf-version-y)
> > > +ifneq ($(dwarf-version-y)$(LLVM_IAS),21)
> > > +# Binutils 2.35+ required for -gdwarf-4+ support.
> > > +dwarf-aflag    := $(call as-option,-Wa$(comma)-gdwarf-$(dwarf-version-y))
> > > +DEBUG_CFLAGS   += $(dwarf-aflag)
> >
> > This changes the behavior.
> >
> > For the Dwarf-2 case,
> >
> > Previously, -gdwarf-2 was passed to $(CC),
> > so the debug info was generated by gcc.
> >
> > Now, -Wa,-gdwarf-2 is passed to $(CC).
> > -gdwarf-2 is handled by GNU as.
> > So, the source info points to /tmp/<hash>.s
> > instead of the original .c file.
> >
> >
> >
> > Handling the Dwarf capability is very complicated.
> >
> > Are you still working for v3?
>
> Yes, I plan to revisit the series based on all of the feedback thus
> far.  Lately I'm focused on enabling LLVM_IAS=1 for Android; but I
> would like to see this land so that the Linux kernel may provide
> coverage and feedback to the toolchain developers for DWARF v5 (as
> well as reduced binary image sizes).  Maybe later this week I'll have
> time to revisit.
> --
> Thanks,
> ~Nick Desaulniers
>
> --
> You received this message because you are subscribed to the Google Groups "Clang Built Linux" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to clang-built-linux+unsubscribe@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/clang-built-linux/CAKwvOdmPeOEA4dfODCKLE4A_M-SF5RBVFEf-NuiTkUTXAbh-5w%40mail.gmail.com.


In my understanding, the complexity comes from the fact
we are mixing up the $(CC) capability and $(AS) capability.

They are orthogonal if I understand correctly.


When building *.c files, the .debug* sections are generated by
gcc (or clang), and embedded into the intermediate *.s files.
The assembler (GAS or clang's IAS) simply transforms it
into byte stream in *.o. So we do not care about the assembler capability.


When building *.S files, the .debug* sections are generated by
the assembler. Here, the assembler capability is important.
Unless we use binutils 2.35+ or clang IAS,
DWARF v2 is the only possible choice.



So, we need two separate choices to handle this properly, I think.

The following is the rough sketch.





# The value is 2, 3, 4, or 5 depending on the assembler in use.
# Unfortunately, we cannot check this by $(cc-option, -Wa,-gdwarf-4)
# because GAS <= 2.34 accepts any -gdwarf-<N>.
# readelf --debug-dump=info and grep or something?
config AS_SUPPORTS_DWARF_VERSION
        int $(shell scripts/as_dwarf_support.sh)



choice
            "DWARF version for C code debugging"

config CC_DEBUG_INFO_DWARF2
            bool "..."

config CC_DEBUG_INFO_DWARF4
            bool "..."

config CC_DEBUG_INFO_DWARF5
            bool "..."
            depends on GCC_VERSION >= 700000 || CC_IS_CLANG
            depends on AS_SUPPORTS_DWARF_VERSION >= 5
            help
              gcc7+ or clang supports this.
              Unfortunately, we also need to check assembler capability
              because GAS <= 2.34 do not understand ".file 0"

endchoice



choice
            "DWARF version for assembly code debugging"

config AS_DEBUG_INFO_DWARF2
            bool "..."

config AS_DEBUG_INFO_DWARF4
            bool "..."
            depends on AS_SUPPORTS_DWARF_VERSION >= 4

config AS_DEBUG_INFO_DWARF5
            bool "..."
            depends on AS_SUPPORTS_DWARF_VERSION >= 5

endchoice
Segher Boessenkool Dec. 1, 2020, 9:32 a.m. UTC | #11
On Tue, Dec 01, 2020 at 12:38:16PM +0900, Masahiro Yamada wrote:
> > We can bump -Wa,-gdwarf-2 to -Wa,-gdwarf-3 since GNU actually emits
> > DWARF v3 DW_AT_ranges (see
> > https://sourceware.org/bugzilla/show_bug.cgi?id=26850 )
> > This can avoid the `warning: DWARF2 only supports one section per
> > compilation unit` warning for Clang.

That warning should be "there can be only one section with executable
code per translation unit", or similar.

> I am not a DWARF spec expert.

Neither am I.

> Please teach me.
> 
> In my understanding, "DWARF2 only supports one section ..."
> is warned only when building .S files with LLVM_IAS=1

.S files are simply run through the C preprocessor first, and then given
to the assembler.  The only difference there should be wrt debug info is
you could have some macros that expand to assembler debug statements.

> If this is due to the limitation of DWARF v2, why is it OK to
> build .c files with LLVM_IAS?

The compiler can of course make sure not to use certain constructs in
its generated assembler code, while the assembler will have to swallow
whatever the user wrote.


Segher
Fāng-ruì Sòng Dec. 2, 2020, 1:08 a.m. UTC | #12
On 2020-12-01, Segher Boessenkool wrote:
>On Tue, Dec 01, 2020 at 12:38:16PM +0900, Masahiro Yamada wrote:
>> > We can bump -Wa,-gdwarf-2 to -Wa,-gdwarf-3 since GNU actually emits
>> > DWARF v3 DW_AT_ranges (see
>> > https://sourceware.org/bugzilla/show_bug.cgi?id=26850 )
>> > This can avoid the `warning: DWARF2 only supports one section per
>> > compilation unit` warning for Clang.
>
>That warning should be "there can be only one section with executable
>code per translation unit", or similar.
>
>> I am not a DWARF spec expert.
>
>Neither am I.
>
>> Please teach me.
>>
>> In my understanding, "DWARF2 only supports one section ..."
>> is warned only when building .S files with LLVM_IAS=1
>
>.S files are simply run through the C preprocessor first, and then given
>to the assembler.  The only difference there should be wrt debug info is
>you could have some macros that expand to assembler debug statements.
>
>> If this is due to the limitation of DWARF v2, why is it OK to
>> build .c files with LLVM_IAS?
>
>The compiler can of course make sure not to use certain constructs in
>its generated assembler code, while the assembler will have to swallow
>whatever the user wrote.
>

These are all correct. You can use `llvm-dwarfdump a.o` to dump a .o file.
It has one DW_TAG_compile_unit. If the translation unit has a single
contiguous address range, the assembler can emit a pair of
DW_AT_low_pc/DW_AT_high_pc (available in DWARF v2). In the case of
multiple executable sections, it is not guaranteed that in the final
linked image the sections will be contiguous, so the assembler has to
assume there may be non-contiguous address ranges and use DW_AT_ranges.

Unfortunately DW_AT_ranges was introduced in DWARF v3 and technically
not available in DWARF v2. But GNU as ignores this and emits
DW_AT_ranges anyway (this is probably fine - like using a GNU extension).

If -Wa,-gdwarf-2 -> -Wa,-gdwarf-3 can eliminate the LLVM integrated
assembler's warning, we should do it. If people think -Wa,-gdwarf-2 is
not useful and want to delete it, I'll be happier. Whether it is
necessary to use -Wa,-gdwarf-2/-Wa,-gdwarf-5? Personally I would think
this is unnecessary, but I won't mind if people don't mind the
additional complexity in Makefile. (I implemented the -gdwarf-5 address
range stuff for the integrated assembler).
diff mbox series

Patch

diff --git a/Makefile b/Makefile
index 75b1a3dcbf30..e23786a4c1c7 100644
--- a/Makefile
+++ b/Makefile
@@ -826,12 +826,14 @@  else
 DEBUG_CFLAGS	+= -g
 endif
 
-ifndef LLVM_IAS
-KBUILD_AFLAGS	+= -Wa,-gdwarf-2
-endif
-
-ifdef CONFIG_DEBUG_INFO_DWARF4
-DEBUG_CFLAGS	+= -gdwarf-4
+dwarf-version-$(CONFIG_DEBUG_INFO_DWARF2) := 2
+dwarf-version-$(CONFIG_DEBUG_INFO_DWARF4) := 4
+DEBUG_CFLAGS	+= -gdwarf-$(dwarf-version-y)
+ifneq ($(dwarf-version-y)$(LLVM_IAS),21)
+# Binutils 2.35+ required for -gdwarf-4+ support.
+dwarf-aflag	:= $(call as-option,-Wa$(comma)-gdwarf-$(dwarf-version-y))
+DEBUG_CFLAGS	+= $(dwarf-aflag)
+KBUILD_AFLAGS	+= $(dwarf-aflag)
 endif
 
 ifdef CONFIG_DEBUG_INFO_REDUCED
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 83a860126897..03c494eefabd 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -256,14 +256,25 @@  config DEBUG_INFO_SPLIT
 	  to know about the .dwo files and include them.
 	  Incompatible with older versions of ccache.
 
+choice
+	prompt "DWARF version"
+	help
+	  Which version of DWARF debug info to emit.
+
+config DEBUG_INFO_DWARF2
+	bool "Generate DWARF v2 debuginfo"
+	help
+	  Generate DWARF v2 debug info.
+
 config DEBUG_INFO_DWARF4
 	bool "Generate dwarf4 debuginfo"
 	depends on $(cc-option,-gdwarf-4)
 	help
-	  Generate dwarf4 debug info. This requires recent versions
-	  of gcc and gdb. It makes the debug information larger.
-	  But it significantly improves the success of resolving
-	  variables in gdb on optimized code.
+	  Generate DWARF v4 debug info. This requires gcc 4.5+ and gdb 7.0+.
+	  It makes the debug information larger, but it significantly
+	  improves the success of resolving variables in gdb on optimized code.
+
+endchoice # "DWARF version"
 
 config DEBUG_INFO_BTF
 	bool "Generate BTF typeinfo"