> > no, it's a quality of implementation issue. Also it's a usability
> > issue, bugs detected should be reported back to user-space. It's
> > _not_ a valid excuse that if it's not defined explicitly in POSIX
> > or that other OSs are less careful.
>
> I don't disagree its a useful feature - but its behaves
> inconsistently between different OSs and different versions of linux.
the only issue we care about when something is already defined in another
OS.
> But... the behaviour isn't consistent right now, why should we expect
> it to be in ther future?
2.0 is definitely consistent. (or should be)
> We changed the semantics once already, they could change again,
> writing applications based upon on set on undefined semantics sounds
> a bit dodgey to me.
? we have not changed any semantics.
> Anyhow... we can do the above, (non-portably) in user-space my mmap
> PROT_NONE and pulling apart the stack frame we get in the segfault
> signal handler.
this is what i'm talking about. If the driver is buggy and doesnt return
-EFAULT then user-space has no chance to catch the fault on the PROT_NONE.
-- mingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/