boot2

commit 34257e04a2beb2249af0ef36ac3cde0f2e3f86cb
parent 3d092914a70e99f5e3707e25f641e221b1695354
Author: Ryan Sepassi <rsepassi@gmail.com>
Date:   Mon, 27 Apr 2026 02:17:08 -0700

cc tests: lock in void* byte-arithmetic semantics

void *p; p + N is byte-arithmetic (GCC ext / tcc-mes assumption);
strict C99 disallows it. cg-binop's ptr-arith branch already does the
right thing because the (> sz 1) guard short-circuits scaling for
char* (size 1) and void* (size -1, pointee), naturally yielding byte
arithmetic for both. Lock-in fixture so the invariant doesn't regress.

tests/cc-parse/78-voidptr-arith.c casts char[] to void*, advances by
2, casts back, and dereferences; expects 'C' = 67.

Diffstat:
A
tests/cc-parse/78-voidptr-arith.c
 | 
17
+++++++++++++++++
A
tests/cc-parse/78-voidptr-arith.expected-exit
 | 
1
+

2 files changed, 18 insertions(+), 0 deletions(-)
diff --git a/tests/cc-parse/78-voidptr-arith.c b/tests/cc-parse/78-voidptr-arith.c
@@ -0,0 +1,17 @@
+/* void* arithmetic via the GCC byte-arithmetic extension.
+ *
+ * tcc.c relies on `void *p; p + N` being byte-stride pointer
+ * arithmetic. Strict C99 disallows it (no arithmetic on void*); we
+ * follow GCC and tcc-mes.
+ *
+ * Buffer: { 'A', 'B', 'C', 'D', 0 }. void* + 2 + cast to char* + deref
+ * yields 'C' = 67.
+ */
+
+char buf[] = "ABCD";
+
+int main(void) {
+    void *p = buf;
+    char *q = (char *)(p + 2);
+    return *q;
+}
diff --git a/tests/cc-parse/78-voidptr-arith.expected-exit b/tests/cc-parse/78-voidptr-arith.expected-exit
@@ -0,0 +1 @@
+67