Add USEMALLOC and enable it by default. You can switch back to brk()
by removing `#define USEMALLOC` in memory.c.
USEMALLOC tells the allocator to use malloc() and realloc(), not
brk(). This might help systems where brk() doesn't work, or where
malloc() can allocate outside the brk area.
My build shows no changes in share/ack/examples (except hilo_bas.*).
Option -u was passing an offset from modulptr(0) in ALLOMODL to the
string in argv. If entername() would move ALLOMODL to make room in
ALLOGCHR, then the offset would become invalid, so the string would
get lost. This fix copies the string into ALLOMODL.
This was often not a problem because the initial size of ALLOGCHR in
mach.h is probably large enough for -u. This became a problem when I
caused the initial allocations to fail, and then only because the B
runtime uses -u.
Also move the declarations of `incore` and `core_alloc` to "memory.h".
Also correct SYMDEBUG to SYMDBUG. (I don't know if SYMDBUG works
because our build system never defines it.)
ind_t becomes an alias of size_t. ind_t becomes unsigned, so I edit
some code that was using negative ind_t. Some casts disappear, like
(long)sizeof(...) because the size is already a size_t. There are
changes to overflow checks. Callers with a size too big for size_t
must check it before calling the memory allocator. An overflow check
of BASE + incr in memory.c sbreak() now happens on all platforms, not
only when a pointer is smaller than a long.
My build shows no changes in share/ack/examples (except hilo_bas.*
changing with every build).
edge splitting can cause new basic blocks to be added to the graph, but while
the graph itself gets properly rewritten the descriptor tables can't be updated
to take these into account, so they end up pointing at the wrong blocks. This
causes really hard-to-debug problems.
The new approach is to parse the descriptor blocks and then generate a
comparison chain. Brute force, but much easier for the compiler to reason
about.
need flt_arith any more. (And also generates them correctly on little-endian
systems.) as now parses numbers properly, doesn't trash memory all over the
place, and can handle negative numbers.
