Pragma Inspection_Point and Caching

Pragma Inspection_Point and Caching

[Ken Garlington]

An engineer asked me a question today, and I didn't feel I had a
particularly good answer, so I'm trolling for another one. Here's the
question: Based on the definition of pragma Inspection_Point, does the
compiler have any obligation to attempt to defeat caching (assuming the
underlying ISA allows some or all of the cache to be flushed), if the
debugging system does not have access to the cache?

I mumbled something about "even the ARM has limits to its scope, and caching
is outside it," but given the discussion in the Annex and in the Rationale,
this didn't seem to satisfy her (or me).

Re: Pragma Inspection_Point and Caching

[Robert Dewar]

In article <88dfak$kjm$1@epimetheus.cswireless.net>,
  "Ken Garlington" <Ken.Garlington@computer.org> wrote:
> I mumbled something about "even the ARM has limits to its
> scope, and caching is outside it," but given the discussion in
> the Annex

Nothing in the rationale has any effect whatsoever on the RM.
And you response regarding the RM was reasonable. This is
definitely in the realm of implementation defined behavior.

Mind you, a debugger that does not have access to cached data
on a machine with a non-write through cache is pretty much
useless if you ask me!


Sent via Deja.com http://www.deja.com/
Before you buy.