Re: [PATCH, RFC] x86: also CFI-annotate certain inline asm()s

[Jan Beulich]

>>> On 10.11.14 at 19:10, <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> Btw, the sane thing to do is to make your infrastructure just say "If
> my frame walker hits a push/pop without CFI information, I'll just add
> it myself".
> 
> Yes, that involved having to actuall ylook at the instruction. Tough
> shit. Just do it right. There aren't that many push/pop patterns.

Did you think this through? Inspecting instructions while unwinding
the stack would involve significant amounts of architecture specific
code, whereas the unwinder is largely architecture independent.
Apart from code to obtain machine state, only the annotations are
(necessarily) connected to the architecture since they accompany
machine instructions.

Did you ever write a disassembler capable of correctly dealing with
everything a compiler may generate (i.e. including data literals in the
middle of code)?

Anyway - I'm sure I won't convince you now or ever, this is too
religious a topic for you afaict, and hence an objective and fair
discussion is impossible.

Jan

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/