[PATCH 0/8] objtool, x86/alternative: Convert objtool annotations and x86 alternatives to assembler macros

Josh Poimboeuf posted 8 patches 1 week, 5 days ago
arch/x86/include/asm/alternative.h    | 196 +++++++++++++-------------
arch/x86/include/asm/asm.h            |   8 +-
arch/x86/include/asm/bug.h            |   2 +-
arch/x86/include/asm/irq_stack.h      |   2 +-
arch/x86/include/asm/jump_label.h     |   1 +
arch/x86/include/asm/nospec-branch.h  |   8 +-
arch/x86/include/asm/paravirt_types.h |   2 +-
arch/x86/include/asm/processor.h      |   2 +-
arch/x86/include/asm/smap.h           |   8 +-
arch/x86/include/asm/vmware.h         |   2 +-
arch/x86/platform/efi/efi_stub_64.S   |   2 +-
include/linux/annotate.h              | 133 ++++++++---------
include/linux/objtool_types.h         |   5 +-
tools/include/linux/objtool_types.h   |   5 +-
tools/objtool/check.c                 |   4 +-
15 files changed, 196 insertions(+), 184 deletions(-)
[PATCH 0/8] objtool, x86/alternative: Convert objtool annotations and x86 alternatives to assembler macros
Posted by Josh Poimboeuf 1 week, 5 days ago
Convert objtool annotations and x86 alternatives to assembler macros.
This creates a shared C/asm alternative implementation and makes the
code generation significantly more readable.

Patches 1-5: convert objtool
Patches 6-8: convert alternatives

Based on tip/master.

Before:
-----
  # ./arch/x86/include/asm/smap.h:47: 	alternative("", "stac", X86_FEATURE_SMAP);
  # 47 "./arch/x86/include/asm/smap.h" 1
  	# ALT: oldinstr
  771:

  772:
  # ALT: padding
  .skip -(((775f-774f)-(772b-771b)) > 0) * ((775f-774f)-(772b-771b)),0x90
  773:
  .pushsection .altinstructions, "aM", @progbits, 14
   .long 771b - .
   .long 774f - .
   .4byte ( 9*32+20)
   .byte 773b-771b
   .byte 775f-774f
  .popsection
  .pushsection .altinstr_replacement, "ax"
  .pushsection .discard.annotate_data, "M", @progbits, 8; .long 1b - ., 1; .popsection
  # ALT: replacement
  774:
  	stac
  775:
  .popsection
-----

After:
-----
  # ./arch/x86/include/asm/smap.h:47: 	alternative("", "stac", X86_FEATURE_SMAP);
  # 47 "./arch/x86/include/asm/smap.h" 1
	ALTERNATIVE "", "stac", "( 9*32+20)"
-----

Future work:
  - BUG()/WARN()
  - static_cpu_has()
  - static calls/branches
  - exceptions
  - smp locks
  - unwind hints

Josh Poimboeuf (8):
  objtool: Rename ANNOTYPE_IGNORE_ALTS -> ANNOTYPE_IGNORE_ALTERNATIVE
  objtool: Make ANNOTYPE_DATA_SPECIAL unique across all annotations
  objtool: Rename C ANNOTATE_REACHABLE to ANNOTATE_REACHABLE_LABEL
  objtool: Rename asm ANNOTATE_NOCFI_SYM to ANNOTATE_NOCFI
  objtool: Convert annotations to assembler macros
  x86/asm: Use unique code labels in __FILL_RETURN_BUFFER
  x86/asm: Remove newlines in alternatives
  x86/alternative: Convert alternatives to assembler macros

 arch/x86/include/asm/alternative.h    | 196 +++++++++++++-------------
 arch/x86/include/asm/asm.h            |   8 +-
 arch/x86/include/asm/bug.h            |   2 +-
 arch/x86/include/asm/irq_stack.h      |   2 +-
 arch/x86/include/asm/jump_label.h     |   1 +
 arch/x86/include/asm/nospec-branch.h  |   8 +-
 arch/x86/include/asm/paravirt_types.h |   2 +-
 arch/x86/include/asm/processor.h      |   2 +-
 arch/x86/include/asm/smap.h           |   8 +-
 arch/x86/include/asm/vmware.h         |   2 +-
 arch/x86/platform/efi/efi_stub_64.S   |   2 +-
 include/linux/annotate.h              | 133 ++++++++---------
 include/linux/objtool_types.h         |   5 +-
 tools/include/linux/objtool_types.h   |   5 +-
 tools/objtool/check.c                 |   4 +-
 15 files changed, 196 insertions(+), 184 deletions(-)

-- 
2.52.0
Re: [PATCH 0/8] objtool, x86/alternative: Convert objtool annotations and x86 alternatives to assembler macros
Posted by Borislav Petkov 1 week, 1 day ago
On Sat, Dec 06, 2025 at 01:41:07PM -0800, Josh Poimboeuf wrote:
> Convert objtool annotations and x86 alternatives to assembler macros.
> This creates a shared C/asm alternative implementation and makes the
> code generation significantly more readable.

I don't understand: all this churn for that?

So that .s output - which is looked at by 3 people on this planet - is "readable"?

Dunno, it was readable before too...

> Patches 1-5: convert objtool
> Patches 6-8: convert alternatives
> 
> Based on tip/master.
> 
> Before:
> -----
>   # ./arch/x86/include/asm/smap.h:47: 	alternative("", "stac", X86_FEATURE_SMAP);
>   # 47 "./arch/x86/include/asm/smap.h" 1
>   	# ALT: oldinstr
>   771:
> 
>   772:
>   # ALT: padding

... especially with those helpful comments...

>   .skip -(((775f-774f)-(772b-771b)) > 0) * ((775f-774f)-(772b-771b)),0x90
>   773:
>   .pushsection .altinstructions, "aM", @progbits, 14
>    .long 771b - .
>    .long 774f - .
>    .4byte ( 9*32+20)
>    .byte 773b-771b
>    .byte 775f-774f
>   .popsection
>   .pushsection .altinstr_replacement, "ax"
>   .pushsection .discard.annotate_data, "M", @progbits, 8; .long 1b - ., 1; .popsection
>   # ALT: replacement
>   774:
>   	stac
>   775:
>   .popsection
> -----
> 
> After:
> -----
>   # ./arch/x86/include/asm/smap.h:47: 	alternative("", "stac", X86_FEATURE_SMAP);
>   # 47 "./arch/x86/include/asm/smap.h" 1
> 	ALTERNATIVE "", "stac", "( 9*32+20)"

Well, that's "readable" but one still needs to go see that ALTERNATIVE
macro definition now, right?

So you move the gory details somewhere else.

Sorry, I fail to see why we're doing this...

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette
Re: [PATCH 0/8] objtool, x86/alternative: Convert objtool annotations and x86 alternatives to assembler macros
Posted by Josh Poimboeuf 1 week, 1 day ago
On Thu, Dec 11, 2025 at 02:06:06AM +0100, Borislav Petkov wrote:
> On Sat, Dec 06, 2025 at 01:41:07PM -0800, Josh Poimboeuf wrote:
> >   # ./arch/x86/include/asm/smap.h:47: 	alternative("", "stac", X86_FEATURE_SMAP);
> >   # 47 "./arch/x86/include/asm/smap.h" 1
> > 	ALTERNATIVE "", "stac", "( 9*32+20)"
> 
> Well, that's "readable" but one still needs to go see that ALTERNATIVE
> macro definition now, right?
> 
> So you move the gory details somewhere else.
> 
> Sorry, I fail to see why we're doing this...

Partially because Linus complains every time I mess up the code gen ;-)
But I like this change too.

Anybody reading the .s file should already know how alternatives work.
If not... go read alternatives.h!

-- 
Josh