Re: [kvm-devel] [PATCH] KVM: Avoid using vmx instruction directly

[Avi Kivity <avi@qumranet.com>]

H. Peter Anvin wrote:
> Jeremy Fitzhardinge wrote:
>> Avi Kivity wrote:
>>>> Or gcc
>>>> might move the assignment of phys_addr to after the inline assembly.
>>>>   
>>> "asm volatile" prevents that (and I'm not 100% sure it's necessary).
>>
>> No, it won't necessarily.  "asm volatile" simply forces gcc to emit the
>> assembler, even if it thinks its output doesn't get used.  It makes no
>> ordering guarantees with respect to other code (or even other "asm
>> volatiles").   The "memory" clobbers should fix the ordering of the asms
>> though.
>>
>
> I think you're wrong about that; in particular, I'm pretty sure "asm 
> volatiles" are ordered among themselves.  What the "volatile" means is 
> "this has side effects you (the compiler) don't understand", and gcc 
> can't assume that it can reorder such side effects.

The gcc manual has this to say:

   Similarly, you can't expect a sequence of volatile `asm' instructions
  to remain perfectly consecutive.  If you want consecutive output, use a
  single `asm'.  Also, GCC will perform some optimizations across a
  volatile `asm' instruction; GCC does not "forget everything" when it
  encounters a volatile `asm' instruction the way some other compilers do.

I wonder how we are supposed to code the following sequence:


    asm volatile ("blah")  /* sets funky processor mode */

    some_c_code();

    asm volatile ("unblah");

Let's say "blah" disables floating point exceptions, and some_c_code() 
must run without exceptions.  Is is possible to code this in gcc without 
putting functions in another translation unit?  Is a memory clobber 
sufficient?  I'd certainly hate to use it.

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.