From 302e24671942051d70707586cf8c605a5815edac Mon Sep 17 00:00:00 2001
From: lemon <lsof@mailbox.org>
Date: Mon, 15 Dec 2025 22:39:52 +0100
Subject: create distinct interned string type

Interned strings are used pervasively, so it's a good idea to add a
layer of type safety to differentiate them from general cstrs and avoid
potential bugs from comparing non-interned and interned strings. Not
that that's happened so far that I can remember, but it could.

I'm 90% sure it's legal to alias `struct {char c;}` pointers with `char`
pointers. This specific typedef gives type safety but with a simple
one-way `internstr -> const char *` typecast (with `&istr->c`).
Converting the other way around is more intentional: a straight up cast
`(internstr)cstr` which sticks out as unchecked and probably wrong, or
calling the intern(cstr) function, which is the right way.
---
 ir/intrin.c |  2 +-
 ir/ir.c     |  6 +++---
 ir/ir.h     | 12 ++++++------
 3 files changed, 10 insertions(+), 10 deletions(-)

(limited to 'ir')

diff --git a/ir/intrin.c b/ir/intrin.c
index 2ef826c..53e29fc 100644
--- a/ir/intrin.c
+++ b/ir/intrin.c
@@ -24,7 +24,7 @@ intrin(struct block *blk, int *curi, enum intrin in, struct arg *args, int narg,
          /* memcpy */
          *args[1].ty = *args[0].ty = mktyperef(cls2type(KPTR));
          insertinstr(blk, (*curi)++, mkarginstr(cls2type(cls), mkintcon(cls, td->siz)));
-         *this = mkinstr(Ocall, 0, mksymref("memcpy", 1), this->r);
+         *this = mkinstr(Ocall, 0, mksymref(intern("memcpy"), 1), this->r);
          calltab.p[this->r.i].narg = 3;
          calltab.p[this->r.i].ret = cls2type(0);
          return 0;
diff --git a/ir/ir.c b/ir/ir.c
index d932121..883e728 100644
--- a/ir/ir.c
+++ b/ir/ir.c
@@ -138,14 +138,14 @@ mkfltcon(enum irclass k, double f)
 }
 
 union ref
-mksymref(const char *s, bool isfunc)
+mksymref(internstr s, bool isfunc)
 {
    struct xcon con = { .issym = 1, .sym = s, .isfunc = isfunc };
    return mkref(RXCON, addcon(&con));
 }
 
 union ref
-mkdatref(const char *name, union type ctype, uint siz, uint align, const void *bytes, uint n, bool deref)
+mkdatref(internstr name, union type ctype, uint siz, uint align, const void *bytes, uint n, bool deref)
 {
    struct irdat dat = { .ctype = ctype, .align = align, .siz = siz, .name = name };
    dat.section = objout.code && align >= 4 && align <= targ_primsizes[TYPTR] && siz <= 16 ? Stext : Srodata;
@@ -168,7 +168,7 @@ mkdatref(const char *name, union type ctype, uint siz, uint align, const void *b
    return mkref(RXCON, addcon(&(struct xcon){.isdat = 1, .deref = deref, .dat = dattab.n - 1}));
 }
 
-const char *
+internstr
 xcon2sym(int ref)
 {
    struct xcon con = conht[ref];
diff --git a/ir/ir.h b/ir/ir.h
index 1a1d907..8c2990e 100644
--- a/ir/ir.h
+++ b/ir/ir.h
@@ -22,14 +22,14 @@ struct irdat {
    union type ctype;
    uint siz;
    uint off;
-   const char *name;
+   internstr name;
 };
 
 struct xcon {
    bool issym, isdat, isfunc, deref;
    uchar cls;
    union {
-      const char *sym;
+      internstr sym;
       int dat;
       vlong i;
       double f;
@@ -163,7 +163,7 @@ enum {
 };
 struct function {
    struct arena **arena;
-   const char *name;
+   internstr name;
    struct block *entry, *curblk;
    struct use *use;
    short *nuse;
@@ -250,9 +250,9 @@ union ref mkfltcon(enum irclass, double);
 #define isaddrcon(r,derefok) ((r).t == RXCON && !conht[(r).i].cls && (derefok || !conht[(r).i].deref))
 #define intconval(r) ((r).t == RICON ? (r).i : conht[(r).i].i)
 #define fltconval(r) ((r).t == RICON ? (r).i : conht[(r).i].f)
-union ref mksymref(const char *, bool isfunc);
-union ref mkdatref(const char *name, union type ctype, uint siz, uint align, const void *, uint n, bool deref);
-const char *xcon2sym(int ref);
+union ref mksymref(internstr, bool isfunc);
+union ref mkdatref(internstr name, union type ctype, uint siz, uint align, const void *, uint n, bool deref);
+internstr xcon2sym(int ref);
 struct instr mkalloca(uint siz, uint align);
 union ref mkcallarg(union irtype ret, uint narg, int vararg);
 #define mkintrin(B, C, N) mkinstr(Ointrin, C, {{.t=RICON,B}}, mkcallarg((union irtype){{0}},N,-1))
-- 
cgit v1.2.3