DrSocalkwe3n/luajit
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Avoid int16_t widening for pt->uv elements.

Browse Source
This commit is contained in:
Mike Pall
2010-01-09 22:41:08 +01:00
parent a33204ae5e
commit 2cc554db0c
4 changed files with 13 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/lj_parse.c
View File

@@ -1014,7 +1014,7 @@ static uint32_t indexupvalue(FuncState *fs, GCstr *name, ExpDesc *v)
uint32_t i;
GCproto *pt = fs->pt;
for (i = 0; i < fs->nuv; i++) {
if (fs->upvalues[i].k == v->k && fs->upvalues[i].info == v->u.s.info) {
if (fs->upvalues[i].info == v->u.s.info && fs->upvalues[i].k == v->k) {
lua_assert(pt->uvname[i] == name);
return i;
}
@@ -1171,12 +1171,12 @@ static void collectk(FuncState *fs, GCproto *pt)
static void collectuv(FuncState *fs, GCproto *pt)
{
uint32_t i;
pt->uv = lj_mem_newvec(fs->L, fs->nuv, int16_t);
pt->uv = lj_mem_newvec(fs->L, fs->nuv, uint16_t);
pt->sizeuv = fs->nuv;
for (i = 0; i < pt->sizeuv; i++) {
uint32_t v = fs->upvalues[i].info;
if (fs->upvalues[i].k == VUPVAL) v = ~v;
pt->uv[i] = (int16_t)v;
if (fs->upvalues[i].k == VLOCAL) v |= 0x8000;
pt->uv[i] = (uint16_t)v;
}
}