Do away with memcpy/memset/memcmp

Now everything is either explicitly a rep thing, or explicitly some
other small run of reasonably efficient instructions. No library calls
or function call overhead of any kind. Code size remains nice and small.

Ban a few other bad libc functions while we're at it. More can be added
later if we decide that's a useful idea.

This will require some discipline and workarounds to cope with Clang's
tendency to shove mem* calls in places you don't want them even when you
tell it not to. Well, c'est la vie.

I'm a little nervous that the __builtin_memmove() usage in shuntvars()
could still one day emit actual library calls. It appears we're not
importing memmove() from anywhere at the moment, though. So I guess it's
good enough for now.

Also I had to get rid of alloca() in con_.c but that was kind of bad
anyway, so that's fine.

1 files changed, 17 insertions, 1 deletions

diff --git a/src/os.c b/src/os.c
index 5fb84bc..2bc0e45 100644
--- a/src/os.c
+++ b/src/os.c
@@ -32,6 +32,22 @@
 
 #include "intdefs.h"
 #include "langext.h"
+#include "os.h"
+
+// HACK: host tools link in os.c as well. if cross-compiling, we can't use our
+// inline asm mem_copy implementation. just use memcpy if not on 32-bit x86.
+// the host tools will have the C runtime available, unlike SST itself
+#if defined(__i386__) || defined(_M_IX86)
+#include "mem.h"
+#endif
+
+void os_spancopy(os_char *restrict dest, const os_char *restrict src, int n) {
+#if defined(__i386__) || defined(_M_IX86)
+	mem_copy(dest, src, n * sizeof(os_char));
+#else
+	memcpy(dest, src, n * sizeof(os_char));
+#endif
+}
 
 #ifdef _WIN32
 
@@ -196,7 +212,7 @@ int os_dlfile(void *lib, char *buf, int sz) {
 	struct link_map *lm = lib;
 	int ssz = strlen(lm->l_name) + 1;
 	if_cold (ssz > sz) { errno = ENAMETOOLONG; return -1; }
-	memcpy(buf, lm->l_name, ssz);
+	mem_copy(buf, lm->l_name, ssz);
 	return ssz;
 }
 #endif