Dump uses varints also for integer constants

Unlike sizes, these constants can be negative, so it encodes those
integers into unsigned integers in a way that keeps small numbers
small.
This commit is contained in:
Roberto Ierusalimschy 2025-06-13 14:14:50 -03:00
parent e657a48ea5
commit 0cecf1ab6d
3 changed files with 48 additions and 18 deletions

27
ldump.c
View file

@ -31,7 +31,7 @@ typedef struct {
int strip;
int status;
Table *h; /* table to track saved strings */
lua_Integer nstr; /* counter for counting saved strings */
lua_Unsigned nstr; /* counter for counting saved strings */
} DumpState;
@ -87,12 +87,12 @@ static void dumpByte (DumpState *D, int y) {
** size for 'dumpVarint' buffer: each byte can store up to 7 bits.
** (The "+6" rounds up the division.)
*/
#define DIBS ((l_numbits(size_t) + 6) / 7)
#define DIBS ((l_numbits(lua_Unsigned) + 6) / 7)
/*
** Dumps an unsigned integer using the MSB Varint encoding
*/
static void dumpVarint (DumpState *D, size_t x) {
static void dumpVarint (DumpState *D, lua_Unsigned x) {
lu_byte buff[DIBS];
unsigned n = 1;
buff[DIBS - 1] = x & 0x7f; /* fill least-significant byte */
@ -103,12 +103,13 @@ static void dumpVarint (DumpState *D, size_t x) {
static void dumpSize (DumpState *D, size_t sz) {
dumpVarint(D, sz);
dumpVarint(D, cast(lua_Unsigned, sz));
}
static void dumpInt (DumpState *D, int x) {
lua_assert(x >= 0);
dumpVarint(D, cast_sizet(x));
dumpVarint(D, cast_uint(x));
}
@ -117,8 +118,16 @@ static void dumpNumber (DumpState *D, lua_Number x) {
}
/*
** Signed integers are coded to keep small values small. (Coding -1 as
** 0xfff...fff would use too many bytes to save a quite common value.)
** A non-negative x is coded as 2x; a negative x is coded as -2x - 1.
** (0 => 0; -1 => 1; 1 => 2; -2 => 3; 2 => 4; ...)
*/
static void dumpInteger (DumpState *D, lua_Integer x) {
dumpVar(D, x);
lua_Unsigned cx = (x >= 0) ? 2u * l_castS2U(x)
: (2u * ~l_castS2U(x)) + 1;
dumpVarint(D, cx);
}
@ -136,8 +145,8 @@ static void dumpString (DumpState *D, TString *ts) {
TValue idx;
int tag = luaH_getstr(D->h, ts, &idx);
if (!tagisempty(tag)) { /* string already saved? */
dumpSize(D, 1); /* reuse a saved string */
dumpSize(D, cast_sizet(ivalue(&idx))); /* index of saved string */
dumpVarint(D, 1); /* reuse a saved string */
dumpVarint(D, l_castS2U(ivalue(&idx))); /* index of saved string */
}
else { /* must write and save the string */
TValue key, value; /* to save the string in the hash */
@ -147,7 +156,7 @@ static void dumpString (DumpState *D, TString *ts) {
dumpVector(D, s, size + 1); /* include ending '\0' */
D->nstr++; /* one more saved string */
setsvalue(D->L, &key, ts); /* the string is the key */
setivalue(&value, D->nstr); /* its index is the value */
setivalue(&value, l_castU2S(D->nstr)); /* its index is the value */
luaH_set(D->L, D->h, &key, &value); /* h[ts] = nstr */
/* integer value does not need barrier */
}

View file

@ -37,7 +37,7 @@ typedef struct {
const char *name;
Table *h; /* list for string reuse */
size_t offset; /* current position relative to beginning of dump */
lua_Integer nstr; /* number of strings in the list */
lua_Unsigned nstr; /* number of strings in the list */
lu_byte fixed; /* dump is fixed in memory */
} LoadState;
@ -94,8 +94,8 @@ static lu_byte loadByte (LoadState *S) {
}
static size_t loadVarint (LoadState *S, size_t limit) {
size_t x = 0;
static lua_Unsigned loadVarint (LoadState *S, lua_Unsigned limit) {
lua_Unsigned x = 0;
int b;
limit >>= 7;
do {
@ -127,9 +127,12 @@ static lua_Number loadNumber (LoadState *S) {
static lua_Integer loadInteger (LoadState *S) {
lua_Integer x;
loadVar(S, x);
return x;
lua_Unsigned cx = loadVarint(S, LUA_MAXUNSIGNED);
/* decode unsigned to signed */
if ((cx & 1) != 0)
return l_castU2S(~(cx >> 1));
else
return l_castU2S(cx >> 1);
}
@ -149,9 +152,9 @@ static void loadString (LoadState *S, Proto *p, TString **sl) {
return;
}
else if (size == 1) { /* previously saved string? */
lua_Integer idx = cast_st2S(loadSize(S)); /* get its index */
lua_Unsigned idx = loadVarint(S, LUA_MAXUNSIGNED); /* get its index */
TValue stv;
luaH_getint(S->h, idx, &stv); /* get its value */
luaH_getint(S->h, l_castU2S(idx), &stv); /* get its value */
*sl = ts = tsvalue(&stv);
luaC_objbarrier(L, p, ts);
return; /* do not save it again */
@ -175,7 +178,7 @@ static void loadString (LoadState *S, Proto *p, TString **sl) {
/* add string to list of saved strings */
S->nstr++;
setsvalue(L, &sv, ts);
luaH_setint(L, S->h, S->nstr, &sv);
luaH_setint(L, S->h, l_castU2S(S->nstr), &sv);
luaC_objbarrierback(L, obj2gco(S->h), ts);
}

View file

@ -482,5 +482,23 @@ do -- basic check for SETLIST
assert(count == 1)
end
do print("testing code for integer limits")
local function checkints (n)
local source = string.format(
"local a = {[true] = 0X%x}; return a[true]", n)
local f = assert(load(source))
checkKlist(f, {n})
assert(f() == n)
f = load(string.dump(f))
assert(f() == n)
end
checkints(math.maxinteger)
checkints(math.mininteger)
checkints(-1)
end
print 'OK'