Re: LOCK prefix on uni processor has its use (was Re: [BUG FIX] Make x86_32 uni-processor Atomic ops, Atomic)

["Michael S. Zick" <lkml@morethan.org>]

On Wed May 27 2009, Thomas Gleixner wrote:
> On Wed, 27 May 2009, Harald Welte wrote:
> > Here are some statements from the CPU logic guys at VIA/Centaur:
> > 
> > * A read-modify-write sequence cannot be interupted.
> > * All X86 instructions except rep-strings are atomic wrt interrupts.
> > * The lock prefix has uses on a UP processor: It keeps DMA devices from
> >   interfering with a read-modify-write sequence
> ...
> 
> > Now if I understand the issues correctly, it would mean that there is some
> > driver code that modifies a certain chunk of memory, while DMA of some
> > peripheral is also accessing that memory.  I suppose it would not have to be
> > the same actual address, but probably being within the same cache line is
> > already sufficient.
> > 
> > Now the question is: Is this a valid operation of a driver?  Should the driver
> > do such things, or is such a driver broken?  When would that occur?  I'm trying
> > to come up with a case, but typically you e.g. allocate some DMA buffer and
> > then don't touch it until the hardware has processed it.
> 
> Right, that would be more than stupid, but even then it would not
> explain any breakage of the kernel. Such a driver would not be
> functional anyway if it relies on some read/write modify operations in
> an active DMA buffer. That would also explode on any other system as
> you have no control whether the access to that memory happens before
> or after the DMA operation.
> 

IFF your DMA buffer is cache-line aligned and doesn't have an immediately
adjacent spin-lock (or some such thing) sharing the cache-line.

Mike
> Can you please ask them to clarify that DMA issue further ?
> 
> Thanks,
> 
> 	tglx
> 
>