Re: [patch 3/8] mutex subsystem, add atomic_*_call_if_*() to i386

[Keith Owens]

On Wed, 21 Dec 2005 15:54:13 -0500 (EST), 
Nicolas Pitre <nico@xxxxxxx> wrote:
>On Wed, 21 Dec 2005, Daniel Jacobowitz wrote:
>
>> This new macro is only going to be used in x86-specific files, right? 
>> There's no practical way to implement this on lots of other
>> architectures.
>
>The default implementation does the call in C.
>
>> Embedding a call in asm("") can break other things too - for instance,
>> unwind tables could become inaccurate.
>
>I doubt unwind tables are used at all for the kernel, are they?

Yes they are.  ia64 absolutely requires accurate unwind tables, it is
part of the ABI.  x86_64 is tending towards requiring accurate CFI
data.

Without valid unwind tables, backtraces are flakey at best.  The lack
of decent kernel unwind for i386 is one of the reasons that kdb
backtrace for i386 is so horrible.

-
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/