tree-ssa -fmudflap problems

[Frank Ch. Eigler]

Eyal Lebedinsky <eyal@eyal.emu.id.au> writes:

> [...]  One way is to establish a registry of all valid memory areas,
> and then pointers are validated by checking them against this
> registry.  I have such a debug system implemeneted in our software
> [...]  In order to validate all memory I need to manually make calls
> to register static memory and auto memory. [...]  In short, a
> compiler can make better use of the pointer registry that we already
> have.

This is in fact how mudflap works.  The compiler emits instrumentation
that does the automatic static/auto memory registration and for
pointer/array checks; libmudflap intercepts str*/mem* functions and
provides checked versions.  I wouldn't be surprised if your object
tree looked a lot like the one in libmudflap.


> [...]
> I decided to try and get my system to first base. [...]
> My intention is to run some of my simple programs, which do not
> actually start another thread yet. I understand that this should
> be achievable at this point.

I just committed an additional libmudflap patch that works around the
infinite recursion you encountered during startup.  There is a good
chance that a program linked with -lpthread, but not actually spawning
any, will now work with mudflap.  Next come the mutexes.


- FChE