[tree-ssa] rfa soon: mudflap prototype

[Andreas Jaeger]

"Frank Ch. Eigler" <fche@redhat.com> writes:

> Hi -
>
> In a few minutes, I plan to submit an initial set of patches
> against the tree-ssa branch for our mudflap project.  This

Why against the tree-ssa branch?  Do you need infrastructure from it?

I would expect that a separate branch would make it easier for the
tree-ssa work, for your work - and for integration into GCC 3.3.

> Red Hat funded project aims to provide a new memory/bounds-checking
> extension for gcc, using concepts borrowed from Greg McGary's
> bounded-pointers work as well as Richard Jones' gcc-checker work.
>
> The goal is to provide reasonably high-performance bounds
> checking for C program modules, by instrumenting selected
> tree-level constructs like pointer-dereferencing and
> array-indexing.  This mechanism uses ordinary (ABI-preserving)
> pointers, and a separate interposition-style runtime library.
>
> The code is in prototype form, and much advice will be needed
> from more knowledgeable gcc folks to clean up many style issues
> and the mudflap hooks.  One of the early bits of advice sought
> would be regarding the source/build placement of the runtime
> library.  For now, the patches will include them as a plain
> hand-built module under gcc/libmudflap.
>
> Without the new "-fmudflap" option, the compiler does nothing
> new, and bootstraps fine on linux.

Excellent!  Thanks for the contribution,
Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj