Many instances of __emit_inst(x) expand to a directive. In a few places
it is used as a macro argument, e.g.
arch/arm64/include/asm/sysreg.h
#define __emit_inst(x) .inst (x)
arch/arm64/include/asm/sysreg.h
#define SET_PSTATE_PAN(x) __emit_inst(0xd500401f | PSTATE_PAN | ((!!x) << PSTATE_Imm_shift))
arch/arm64/kvm/hyp/entry.S
ALTERNATIVE(nop, SET_PSTATE_PAN(1), ARM64_HAS_PAN, CONFIG_ARM64_PAN)
Clang integrated assembler parses `.inst (x)` as two arguments passing
to a macro. We delete the space separator so that `.inst(x)` will be
parsed as one argument.
Note, GNU as parsing `.inst (x)` as one argument is unintentional (for
example the x86 backend will parse the construct as two arguments).
See https://sourceware.org/bugzilla/show_bug.cgi?id=25750#c10
Link: https://github.com/ClangBuiltLinux/linux/issues/939
Cc: [email protected]
Signed-off-by: Fangrui Song <[email protected]>
---
arch/arm64/include/asm/sysreg.h | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h
index ebc622432831..af21e2ec5e3e 100644
--- a/arch/arm64/include/asm/sysreg.h
+++ b/arch/arm64/include/asm/sysreg.h
@@ -49,7 +49,9 @@
#ifndef CONFIG_BROKEN_GAS_INST
#ifdef __ASSEMBLY__
-#define __emit_inst(x) .inst (x)
+// The space separator is omitted so that __emit_inst(x) can be parsed as
+// either a directive or a macro argument.
+#define __emit_inst(x) .inst(x)
#else
#define __emit_inst(x) ".inst " __stringify((x)) "\n\t"
#endif
--
2.26.0.110.g2183baf09c-goog
On Sun, Apr 12, 2020 at 8:38 PM 'Fangrui Song' via Clang Built Linux
<[email protected]> wrote:
>
> Many instances of __emit_inst(x) expand to a directive. In a few places
> it is used as a macro argument, e.g.
>
> arch/arm64/include/asm/sysreg.h
> #define __emit_inst(x) .inst (x)
>
> arch/arm64/include/asm/sysreg.h
> #define SET_PSTATE_PAN(x) __emit_inst(0xd500401f | PSTATE_PAN | ((!!x) << PSTATE_Imm_shift))
>
> arch/arm64/kvm/hyp/entry.S
> ALTERNATIVE(nop, SET_PSTATE_PAN(1), ARM64_HAS_PAN, CONFIG_ARM64_PAN)
>
> Clang integrated assembler parses `.inst (x)` as two arguments passing
> to a macro. We delete the space separator so that `.inst(x)` will be
> parsed as one argument.
>
> Note, GNU as parsing `.inst (x)` as one argument is unintentional (for
> example the x86 backend will parse the construct as two arguments).
> See https://sourceware.org/bugzilla/show_bug.cgi?id=25750#c10
Thanks for the patch and for leading the discussion with the binutils
developers on this curious parsing case.
>
> Link: https://github.com/ClangBuiltLinux/linux/issues/939
> Cc: [email protected]
> Signed-off-by: Fangrui Song <[email protected]>
Shouldn't this have:
Suggested-by: Ilie Halip <[email protected]>
Since Ilie sugguested this in
https://github.com/ClangBuiltLinux/linux/issues/939#issuecomment-601776123?
> ---
> arch/arm64/include/asm/sysreg.h | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h
> index ebc622432831..af21e2ec5e3e 100644
> --- a/arch/arm64/include/asm/sysreg.h
> +++ b/arch/arm64/include/asm/sysreg.h
> @@ -49,7 +49,9 @@
> #ifndef CONFIG_BROKEN_GAS_INST
>
> #ifdef __ASSEMBLY__
> -#define __emit_inst(x) .inst (x)
> +// The space separator is omitted so that __emit_inst(x) can be parsed as
> +// either a directive or a macro argument.
> +#define __emit_inst(x) .inst(x)
> #else
> #define __emit_inst(x) ".inst " __stringify((x)) "\n\t"
What happens if someone starts using `__emit_inst` from inline
assembly, and passes that subexpression to a macro? There are no
users today in arch/arm64/, but I think it's better to change both.
With this applied on -next, and testing via:
$ ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- make LLVM=1 LLVM_IAS=1
-j71 defconfig
$ qemu-system-aarch64 -kernel arch/arm64/boot/Image.gz -machine virt
-cpu cortex-a57 -nographic --append "console=ttyAMA0" -m 2048 -initrd
<path to my buildroot.cpio>
I was able to build and boot, modulo:
https://github.com/ClangBuiltLinux/linux/issues/988 (new)
https://github.com/ClangBuiltLinux/linux/issues/716
https://github.com/ClangBuiltLinux/linux/issues/510 (more specific to
LLD than AS)
so we're pretty close to being able to assemble an arm64 defconfig with Clang.
--
Thanks,
~Nick Desaulniers
On 2020-04-14, Mark Rutland wrote:
>Hi Fangrui,
>
>On Sun, Apr 12, 2020 at 08:38:11PM -0700, Fangrui Song wrote:
>> Many instances of __emit_inst(x) expand to a directive. In a few places
>> it is used as a macro argument, e.g.
>>
>> arch/arm64/include/asm/sysreg.h
>> #define __emit_inst(x) .inst (x)
>>
>> arch/arm64/include/asm/sysreg.h
>> #define SET_PSTATE_PAN(x) __emit_inst(0xd500401f | PSTATE_PAN | ((!!x) << PSTATE_Imm_shift))
>>
>> arch/arm64/kvm/hyp/entry.S
>> ALTERNATIVE(nop, SET_PSTATE_PAN(1), ARM64_HAS_PAN, CONFIG_ARM64_PAN)
>>
>> Clang integrated assembler parses `.inst (x)` as two arguments passing
>> to a macro. We delete the space separator so that `.inst(x)` will be
>> parsed as one argument.
>
>I'm a little confused by the above; sorry if the below sounds stupid or
>pedantic, but I just want to make sure I've understood the problem
>correctly.
>
>For the above, ALTERNATIVE() and SET_PSTATE_PAN() are both preprocessor
>macros, so I would expect those to be expanded before either the
>integrated assembler or an external assembler consumes any of the
>assembly (and both would see the same expanded text). Given that, I'm a
>bit confused as to why the integrated assembly would have an impact on
>preprocessing.
>
>Does compiling the pre-processed source using the integrated assembler
>result in the same behaviour? Can we see the expanded text to make that
>clear?
>
>... at what stage exactly does this go wrong?
>
>Thanks,
>Mark.
Hi Mark,
The C preprocessor expands arch/arm64/kvm/hyp/entry.S
ALTERNATIVE(nop, SET_PSTATE_PAN(1), ARM64_HAS_PAN, CONFIG_ARM64_PAN)
to:
alternative_insn nop, .inst (0xd500401f | ((0) << 16 | (4) << 5) | ((!!1) << 8)), 4, 1
`alternative_insn` is an assembler macro, not handled by the C preprocessor.
Both comma and space are separators, with an exception that content
inside a pair of parentheses/quotes is not split, so clang -cc1as or GNU
as x86 splits arguments this way:
nop, .inst, (0xd500401f | ((0) << 16 | (4) << 5) | ((!!1) << 8)), 4, 1
I actually feel that GNU as arm64's behavior is a little bit buggy. It
works just because GNU as has another preprocessing step `do_scrub_chars`
and its arm64 backend deletes the space before '('
alternative_insn nop,.inst(0xd500401f|((0)<<16|(4)<<5)|((!!1)<<8)),4,1
The x86 backend keeps the space before the outmost '('
alternative_insn nop,.inst (0xd500401f|((0)<<16|(4)<<5)|((!!1)<<8)),4,1
By reading its state machine, I think keeping the spaces will be the
most reasonable behavior.
If .inst were only used as arguments,
alternative_insn nop, ".inst (0xd500401f | ((0) << 16 | (4) << 5) | ((!!1) << 8))", 4, 1
would be the best to avoid parsing issues.
>>
>> Note, GNU as parsing `.inst (x)` as one argument is unintentional (for
>> example the x86 backend will parse the construct as two arguments).
>> See https://sourceware.org/bugzilla/show_bug.cgi?id=25750#c10
>>
>> Link: https://github.com/ClangBuiltLinux/linux/issues/939
>> Cc: [email protected]
>> Signed-off-by: Fangrui Song <[email protected]>
>> ---
>> arch/arm64/include/asm/sysreg.h | 4 +++-
>> 1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h
>> index ebc622432831..af21e2ec5e3e 100644
>> --- a/arch/arm64/include/asm/sysreg.h
>> +++ b/arch/arm64/include/asm/sysreg.h
>> @@ -49,7 +49,9 @@
>> #ifndef CONFIG_BROKEN_GAS_INST
>>
>> #ifdef __ASSEMBLY__
>> -#define __emit_inst(x) .inst (x)
>> +// The space separator is omitted so that __emit_inst(x) can be parsed as
>> +// either a directive or a macro argument.
>> +#define __emit_inst(x) .inst(x)
>> #else
>> #define __emit_inst(x) ".inst " __stringify((x)) "\n\t"
>> #endif
>> --
>> 2.26.0.110.g2183baf09c-goog
>>
Hi Fangrui,
On Sun, Apr 12, 2020 at 08:38:11PM -0700, Fangrui Song wrote:
> Many instances of __emit_inst(x) expand to a directive. In a few places
> it is used as a macro argument, e.g.
>
> arch/arm64/include/asm/sysreg.h
> #define __emit_inst(x) .inst (x)
>
> arch/arm64/include/asm/sysreg.h
> #define SET_PSTATE_PAN(x) __emit_inst(0xd500401f | PSTATE_PAN | ((!!x) << PSTATE_Imm_shift))
>
> arch/arm64/kvm/hyp/entry.S
> ALTERNATIVE(nop, SET_PSTATE_PAN(1), ARM64_HAS_PAN, CONFIG_ARM64_PAN)
>
> Clang integrated assembler parses `.inst (x)` as two arguments passing
> to a macro. We delete the space separator so that `.inst(x)` will be
> parsed as one argument.
I'm a little confused by the above; sorry if the below sounds stupid or
pedantic, but I just want to make sure I've understood the problem
correctly.
For the above, ALTERNATIVE() and SET_PSTATE_PAN() are both preprocessor
macros, so I would expect those to be expanded before either the
integrated assembler or an external assembler consumes any of the
assembly (and both would see the same expanded text). Given that, I'm a
bit confused as to why the integrated assembly would have an impact on
preprocessing.
Does compiling the pre-processed source using the integrated assembler
result in the same behaviour? Can we see the expanded text to make that
clear?
... at what stage exactly does this go wrong?
Thanks,
Mark.
>
> Note, GNU as parsing `.inst (x)` as one argument is unintentional (for
> example the x86 backend will parse the construct as two arguments).
> See https://sourceware.org/bugzilla/show_bug.cgi?id=25750#c10
>
> Link: https://github.com/ClangBuiltLinux/linux/issues/939
> Cc: [email protected]
> Signed-off-by: Fangrui Song <[email protected]>
> ---
> arch/arm64/include/asm/sysreg.h | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h
> index ebc622432831..af21e2ec5e3e 100644
> --- a/arch/arm64/include/asm/sysreg.h
> +++ b/arch/arm64/include/asm/sysreg.h
> @@ -49,7 +49,9 @@
> #ifndef CONFIG_BROKEN_GAS_INST
>
> #ifdef __ASSEMBLY__
> -#define __emit_inst(x) .inst (x)
> +// The space separator is omitted so that __emit_inst(x) can be parsed as
> +// either a directive or a macro argument.
> +#define __emit_inst(x) .inst(x)
> #else
> #define __emit_inst(x) ".inst " __stringify((x)) "\n\t"
> #endif
> --
> 2.26.0.110.g2183baf09c-goog
>
On Tue, Apr 14, 2020 at 08:43:07AM -0700, Fangrui Song wrote:
>
> On 2020-04-14, Mark Rutland wrote:
> > Hi Fangrui,
> >
> > On Sun, Apr 12, 2020 at 08:38:11PM -0700, Fangrui Song wrote:
> > > Many instances of __emit_inst(x) expand to a directive. In a few places
> > > it is used as a macro argument, e.g.
> > >
> > > arch/arm64/include/asm/sysreg.h
> > > #define __emit_inst(x) .inst (x)
> > >
> > > arch/arm64/include/asm/sysreg.h
> > > #define SET_PSTATE_PAN(x) __emit_inst(0xd500401f | PSTATE_PAN | ((!!x) << PSTATE_Imm_shift))
> > >
> > > arch/arm64/kvm/hyp/entry.S
> > > ALTERNATIVE(nop, SET_PSTATE_PAN(1), ARM64_HAS_PAN, CONFIG_ARM64_PAN)
> > >
> > > Clang integrated assembler parses `.inst (x)` as two arguments passing
> > > to a macro. We delete the space separator so that `.inst(x)` will be
> > > parsed as one argument.
> >
> > I'm a little confused by the above; sorry if the below sounds stupid or
> > pedantic, but I just want to make sure I've understood the problem
> > correctly.
> >
> > For the above, ALTERNATIVE() and SET_PSTATE_PAN() are both preprocessor
> > macros, so I would expect those to be expanded before either the
> > integrated assembler or an external assembler consumes any of the
> > assembly (and both would see the same expanded text). Given that, I'm a
> > bit confused as to why the integrated assembly would have an impact on
> > preprocessing.
> >
> > Does compiling the pre-processed source using the integrated assembler
> > result in the same behaviour? Can we see the expanded text to make that
> > clear?
> >
> > ... at what stage exactly does this go wrong?
> >
> > Thanks,
> > Mark.
>
> Hi Mark,
>
> The C preprocessor expands arch/arm64/kvm/hyp/entry.S
> ALTERNATIVE(nop, SET_PSTATE_PAN(1), ARM64_HAS_PAN, CONFIG_ARM64_PAN)
>
> to:
> alternative_insn nop, .inst (0xd500401f | ((0) << 16 | (4) << 5) | ((!!1) << 8)), 4, 1
>
> `alternative_insn` is an assembler macro, not handled by the C preprocessor.
>
> Both comma and space are separators, with an exception that content
> inside a pair of parentheses/quotes is not split, so clang -cc1as or GNU
> as x86 splits arguments this way:
>
> nop, .inst, (0xd500401f | ((0) << 16 | (4) << 5) | ((!!1) << 8)), 4, 1
Thanks for this; I understand now.
Could we fold that into the commit message? I think this is much clearer
than the current wording. The explicit description of separator
behaviour, the pre-expansion of the CPP macros, and the example of how
the assembler will split this really help.
> I actually feel that GNU as arm64's behavior is a little bit buggy. It
> works just because GNU as has another preprocessing step `do_scrub_chars`
> and its arm64 backend deletes the space before '('
>
> alternative_insn nop,.inst(0xd500401f|((0)<<16|(4)<<5)|((!!1)<<8)),4,1
>
> The x86 backend keeps the space before the outmost '('
>
> alternative_insn nop,.inst (0xd500401f|((0)<<16|(4)<<5)|((!!1)<<8)),4,1
>
> By reading its state machine, I think keeping the spaces will be the
> most reasonable behavior.
I think I agree. This deviation across architectures is unfortunate for
such a low-level but common tool.
> If .inst were only used as arguments,
>
> alternative_insn nop, ".inst (0xd500401f | ((0) << 16 | (4) << 5) | ((!!1) << 8))", 4, 1
>
> would be the best to avoid parsing issues.
>
> > >
> > > Note, GNU as parsing `.inst (x)` as one argument is unintentional (for
> > > example the x86 backend will parse the construct as two arguments).
> > > See https://sourceware.org/bugzilla/show_bug.cgi?id=25750#c10
> > >
> > > Link: https://github.com/ClangBuiltLinux/linux/issues/939
> > > Cc: [email protected]
> > > Signed-off-by: Fangrui Song <[email protected]>
> > > ---
> > > arch/arm64/include/asm/sysreg.h | 4 +++-
> > > 1 file changed, 3 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h
> > > index ebc622432831..af21e2ec5e3e 100644
> > > --- a/arch/arm64/include/asm/sysreg.h
> > > +++ b/arch/arm64/include/asm/sysreg.h
> > > @@ -49,7 +49,9 @@
> > > #ifndef CONFIG_BROKEN_GAS_INST
> > >
> > > #ifdef __ASSEMBLY__
> > > -#define __emit_inst(x) .inst (x)
> > > +// The space separator is omitted so that __emit_inst(x) can be parsed as
> > > +// either a directive or a macro argument.
> > > +#define __emit_inst(x) .inst(x)
Can we make this a bit more explicit and say "assembler macro argument"?
That way we can avoid any confusion with a CPP macro.
With that (and with the details above folded into the commit message):
Reviewed-by: Mark Rutland <[email protected]>
Thanks,
Mark.
> > > #else
> > > #define __emit_inst(x) ".inst " __stringify((x)) "\n\t"
> > > #endif
> > > --
> > > 2.26.0.110.g2183baf09c-goog
> > >