Re: [Xen-devel] [PATCH 3/7] x86/alt: Clean up the assembly used to generate alternatives

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 23.02.18 at 17:24, <andrew.cooper3@xxxxxxxxxx> wrote:
> On 23/02/18 15:12, Jan Beulich wrote:
>>>>> On 23.02.18 at 15:03, <andrew.cooper3@xxxxxxxxxx> wrote:
>>> On 13/02/18 14:37, Jan Beulich wrote:
>>>>  >>> On 12.02.18 at 12:23, <andrew.cooper3@xxxxxxxxxx> wrote:
>>>>> --- a/xen/include/asm-x86/alternative-asm.h
>>>>> +++ b/xen/include/asm-x86/alternative-asm.h
>>>>> @@ -9,60 +9,67 @@
>>>>>   * enough information for the alternatives patching code to patch an
>>>>>   * instruction. See apply_alternatives().
>>>>>   */
>>>>> -.macro altinstruction_entry orig alt feature orig_len alt_len
>>>>> +.macro altinstruction_entry orig repl feature orig_len repl_len
>>>>>      .long \orig - .
>>>>> -    .long \alt - .
>>>>> +    .long \repl - .
>>>>>      .word \feature
>>>>>      .byte \orig_len
>>>>> -    .byte \alt_len
>>>>> +    .byte \repl_len
>>>>>  .endm
>>>>>  
>>>>> +#define orig_len               (.L\@_orig_e       -     .L\@_orig_s)
>>>>> +#define repl_len(nr)           (.L\@_repl_e\()nr  -     .L\@_repl_s\()nr)
>>>>> +#define decl_repl(insn, nr)     .L\@_repl_s\()nr: insn; .L\@_repl_e\()nr:
>>>> Wouldn't it work equally well but look slightly less odd if you used
>>>> \(nr) instead of \()nr?
>>> How would that work?  \() is the token separator.
>> When there's nothing inside the parentheses, this construct
>> can be used as a token separator, but that's not its main
>> purpose. Instead \(<text>) means to take <text> literally,
>> without e.g. expanding macro arguments inside it.
> 
> I've never come across it before, and I still can't find reference to it
> in the as manual.
> 
> As for why not to use it, Clang has no idea what \(nr) means, meaning
> that it doesn't expand the construct in the way you describe.  (Although
> I see that GCC/AS do behave as you describe).

Oh, if clang can't cope, then leave it as is.

Jan


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