Re: [Xen-devel] [PATCH 2/2] x86: fix indirect thunk usage of CONFIG_INDIRECT_THUNK

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Wed, Jan 24, 2018 at 04:52:17PM +0000, Roger Pau Monné wrote:
> On Wed, Jan 24, 2018 at 09:23:37AM -0700, Jan Beulich wrote:
> > >>> On 24.01.18 at 16:48, <roger.pau@xxxxxxxxxx> wrote:
> > > When indirect_thunk_asm.h is instantiated directly into assembly files
> > > CONFIG_INDIRECT_THUNK might not be defined, and thus using .if against
> > > it is wrong.
> > > 
> > > Add a check to define CONFIG_INDIRECT_THUNK to 0 if not defined, so
> > > that using .if CONFIG_INDIRECT_THUNK is always correct.
> > > 
> > > This suppresses the following clang error:
> > > 
> > > <instantiation>:8:9: error: expected absolute expression
> > >     .if CONFIG_INDIRECT_THUNK == 1
> > >         ^
> > > <instantiation>:1:1: note: while in macro instantiation
> > > INDIRECT_BRANCH call %rdx
> > > ^
> > > entry.S:589:9: note: while in macro instantiation
> > >         INDIRECT_CALL %rdx
> > >         ^
> > 
> > Why is the same no problem with gas? It wants constant
> > expressions with .if too, after all.
> 
> Right, and I cannot figure out why it works when using
> -no-integrated-as and not when not using it. The more that I cannot
> see CONFIG_INDIRECT_THUNK being unconditionally defined to either 0 or
> 1. I guess I'm missing something.

So given the following asm snippet:

# cat snip.S
.if FOO == 1
.error "FOO"
.endif

Attempting to compile it with different compilers/assemblers:

# clang -o snip.o -c snip.S
snip.S:1:5: error: expected absolute expression
.if FOO == 1
    ^
snip.S:2:1: error: FOO
.error "FOO"
^
# gcc -o snip.o -c snip.S
# as -o snip.o snip.S

So there's clearly something different in how as/gcc handle .if
directives when they reference undefined symbols versus clang.

I can expand the commit message to reflect/include the above, but I
think the change still stands.

Thanks, Roger.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel