Re: [PATCH 04/24] x86,nospec: Annotate indirect calls/jumps

[Peter Zijlstra]

On Fri, Jan 26, 2018 at 10:19:47AM +0000, David Woodhouse wrote:
> On Tue, 2018-01-23 at 16:25 +0100, Peter Zijlstra wrote:
> > Annotate the indirect calls/jumps in the CALL_NOSPEC/JUMP_NOSPEC
> > alternatives.
> > 
> > 
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> 
> Reviewed-by: David Woodhouse <dwmw@xxxxxxxxxxxx>
> 
> However...
> 
> 
> >  /*
> > + * This should be used immediately before an indirect jump/call. It tells
> > + * objtool the subsequent indirect jump/call is vouched safe for retpoline
> > + * builds.
> > + */
> > +.macro ANNOTATE_RETPOLINE_SAFE
> > +	.Lannotate_\@:
> > +	.pushsection .discard.retpoline_safe
> > +	_ASM_PTR .Lannotate_\@
> > +	.popsection
> > +.endm
> 
> Didn't I just see one of those in patch 3? So this makes two...
> 
> 
> 
> > @@ -143,6 +155,12 @@
> >  	".long 999b - .\n\t"					\
> >  	".popsection\n\t"
> >  
> > +#define ANNOTATE_RETPOLINE_SAFE					\
> > +	"999:\n\t"						\
> > +	".pushsection .discard.retpoline_safe\n\t"		\
> > +	_ASM_PTR " 999b\n\t"					\
> > +	".popsection\n\t"
> > +
> >  #if defined(CONFIG_X86_64) && defined(RETPOLINE)
> 
> ... three.
> 
> Now, I did briefly toy with the idea of using a .macro from both
> __ASSEMBLY__ and inline asm, making the latter work by means of 
> asm(".include \"asm/nospec-branch.h\");
> 
> In the end I just ended up with the __FILL_RETURN_BUFFER CPP macro
> which is used from both by other tricks.
> 
> Can we look at doing something like that, please?


I'll try. The paravirt one might be tricky, I always end in header-hell
with that thing.