mirror of
https://github.com/vim/vim.git
synced 2026-05-28 00:21:37 +02:00
patch 9.2.0513: [security]: memory safety issues in spellfile.c
Problem: [security]: memory safety issues in spellfile.c
(tacdm)
Solution: Add recursion limit to read_tree_node(), add length limit
check in tree_count_words(), use alloc_clear() in
spell_read_tree().
Github Security Advisory:
https://github.com/vim/vim/security/advisories/GHSA-3h95-3962-mmvf
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
This commit is contained in:
@@ -119,6 +119,7 @@ struct slang_S
|
||||
// Info from the .sug file. Loaded on demand.
|
||||
time_t sl_sugtime; // timestamp for .sug file
|
||||
char_u *sl_sbyts; // soundfolded word bytes
|
||||
long sl_sbyts_len; // length of sl_sbyts
|
||||
idx_T *sl_sidxs; // soundfolded word indexes
|
||||
buf_T *sl_sugbuf; // buffer with word number table
|
||||
int sl_sugloaded; // TRUE when .sug file was loaded or failed to
|
||||
|
||||
+28
-14
@@ -313,7 +313,7 @@ static int set_sofo(slang_T *lp, char_u *from, char_u *to);
|
||||
static void set_sal_first(slang_T *lp);
|
||||
static int *mb_str2wide(char_u *s);
|
||||
static int spell_read_tree(FILE *fd, char_u **bytsp, long *bytsp_len, idx_T **idxsp, int prefixtree, int prefixcnt);
|
||||
static idx_T read_tree_node(FILE *fd, char_u *byts, idx_T *idxs, int maxidx, idx_T startidx, int prefixtree, int maxprefcondnr);
|
||||
static idx_T read_tree_node(FILE *fd, char_u *byts, idx_T *idxs, int maxidx, idx_T startidx, int prefixtree, int maxprefcondnr, int depth);
|
||||
static void set_spell_charflags(char_u *flags, int cnt, char_u *upp);
|
||||
static int set_spell_chartab(char_u *fol, char_u *low, char_u *upp);
|
||||
static void set_map_str(slang_T *lp, char_u *map);
|
||||
@@ -597,7 +597,7 @@ endOK:
|
||||
* Returns the total number of words.
|
||||
*/
|
||||
static void
|
||||
tree_count_words(char_u *byts, idx_T *idxs)
|
||||
tree_count_words(char_u *byts, long byts_len, idx_T *idxs)
|
||||
{
|
||||
int depth;
|
||||
idx_T arridx[MAXWLEN];
|
||||
@@ -635,8 +635,8 @@ tree_count_words(char_u *byts, idx_T *idxs)
|
||||
++wordcount[depth];
|
||||
|
||||
// Skip over any other NUL bytes (same word with different
|
||||
// flags).
|
||||
while (byts[n + 1] == 0)
|
||||
// flags). But don't go over the end
|
||||
while (n + 1 < byts_len && byts[n + 1] == 0)
|
||||
{
|
||||
++n;
|
||||
++curi[depth];
|
||||
@@ -732,8 +732,8 @@ suggest_load_files(void)
|
||||
* <SUGWORDTREE>: <wordtree>
|
||||
* Read the trie with the soundfolded words.
|
||||
*/
|
||||
if (spell_read_tree(fd, &slang->sl_sbyts, NULL, &slang->sl_sidxs,
|
||||
FALSE, 0) != 0)
|
||||
if (spell_read_tree(fd, &slang->sl_sbyts, &slang->sl_sbyts_len,
|
||||
&slang->sl_sidxs, FALSE, 0) != 0)
|
||||
{
|
||||
someerror:
|
||||
semsg(_(e_error_while_reading_sug_file_str),
|
||||
@@ -782,8 +782,10 @@ someerror:
|
||||
* Need to put word counts in the word tries, so that we can find
|
||||
* a word by its number.
|
||||
*/
|
||||
tree_count_words(slang->sl_fbyts, slang->sl_fidxs);
|
||||
tree_count_words(slang->sl_sbyts, slang->sl_sidxs);
|
||||
tree_count_words(slang->sl_fbyts, slang->sl_fbyts_len,
|
||||
slang->sl_fidxs);
|
||||
tree_count_words(slang->sl_sbyts, slang->sl_sbyts_len,
|
||||
slang->sl_sidxs);
|
||||
|
||||
nextone:
|
||||
if (fd != NULL)
|
||||
@@ -1603,8 +1605,11 @@ spell_read_tree(
|
||||
if (len <= 0)
|
||||
return 0;
|
||||
|
||||
// Allocate the byte array.
|
||||
bp = alloc(len);
|
||||
// Allocate the byte array. Zero-initialize so that any position the
|
||||
// tree does not visit reads as 0; a stray BY_INDEX shared reference
|
||||
// into such a slot then behaves as end-of-word in spellsuggest()
|
||||
// instead of consuming an arbitrary heap byte as a siblingcount.
|
||||
bp = alloc_clear(len);
|
||||
if (bp == NULL)
|
||||
return SP_OTHERERROR;
|
||||
*bytsp = bp;
|
||||
@@ -1618,9 +1623,11 @@ spell_read_tree(
|
||||
*idxsp = ip;
|
||||
|
||||
// Recursively read the tree and store it in the array.
|
||||
idx = read_tree_node(fd, bp, ip, len, 0, prefixtree, prefixcnt);
|
||||
idx = read_tree_node(fd, bp, ip, len, 0, prefixtree, prefixcnt, 0);
|
||||
if (idx < 0)
|
||||
return idx;
|
||||
if (idx != len)
|
||||
return SP_FORMERROR;
|
||||
return 0;
|
||||
}
|
||||
|
||||
@@ -1642,7 +1649,8 @@ read_tree_node(
|
||||
int maxidx, // size of arrays
|
||||
idx_T startidx, // current index in "byts" and "idxs"
|
||||
int prefixtree, // TRUE for reading PREFIXTREE
|
||||
int maxprefcondnr) // maximum for <prefcondnr>
|
||||
int maxprefcondnr, // maximum for <prefcondnr>
|
||||
int depth) // recursiong level
|
||||
{
|
||||
int len;
|
||||
int i;
|
||||
@@ -1652,6 +1660,12 @@ read_tree_node(
|
||||
int c2;
|
||||
#define SHARED_MASK 0x8000000
|
||||
|
||||
// Bail out on a crafted .spl whose tree recurses beyond the maximum
|
||||
// word length: each tree level corresponds to one byte of a word, so
|
||||
// any well-formed file has depth <= MAXWLEN.
|
||||
if (depth > MAXWLEN)
|
||||
return SP_FORMERROR;
|
||||
|
||||
len = getc(fd); // <siblingcount>
|
||||
if (len <= 0)
|
||||
return SP_TRUNCERROR;
|
||||
@@ -1737,7 +1751,7 @@ read_tree_node(
|
||||
{
|
||||
idxs[startidx + i] = idx;
|
||||
idx = read_tree_node(fd, byts, idxs, maxidx, idx,
|
||||
prefixtree, maxprefcondnr);
|
||||
prefixtree, maxprefcondnr, depth + 1);
|
||||
if (idx < 0)
|
||||
break;
|
||||
}
|
||||
@@ -5649,7 +5663,7 @@ sug_filltree(spellinfo_T *spin, slang_T *slang)
|
||||
spin->si_blocks_cnt = 0;
|
||||
|
||||
// Skip over any other NUL bytes (same word with different
|
||||
// flags). But don't go over the end.
|
||||
// flags). But don't go over the end
|
||||
while (n + 1 < slang->sl_fbyts_len && byts[n + 1] == 0)
|
||||
{
|
||||
++n;
|
||||
|
||||
@@ -912,7 +912,10 @@ func Test_spellsuggest_too_deep()
|
||||
" This was incrementing "depth" over MAXWLEN.
|
||||
new
|
||||
norm s000G00ý000000000000
|
||||
sil norm ..vzG................vvzG0 v z=
|
||||
try
|
||||
sil norm ..vzG................vvzG0 v z=
|
||||
catch /E759:/
|
||||
endtry
|
||||
bwipe!
|
||||
endfunc
|
||||
|
||||
|
||||
@@ -372,6 +372,24 @@ func Test_spellfile_format_error()
|
||||
" LWORDTREE: incorrect sibling node count
|
||||
call Spellfile_Test(0zFF00000001040000000000000000, 'E759:')
|
||||
|
||||
" LWORDTREE: declared nodecount larger than the tree actually fills.
|
||||
" Root has two siblings: 'x' (recurses into an end-of-word at idx 3..4)
|
||||
" and BY_INDEX targeting position 9. Tree fills positions 0..4, leaving
|
||||
" 5..9 unwritten — byts[9] would be uninitialized without the fix.
|
||||
call Spellfile_Test(0zFF0000000A02780100000979010000000000000000000000, 'E759:')
|
||||
|
||||
" LWORDTREE: recursion depth past MAXWLEN. A linear chain of 254
|
||||
" (siblingcount=1, byte='a') frames drives read_tree_node to depth
|
||||
" MAXWLEN where the new guard rejects. The trailing (01 00) gives the
|
||||
" chain a clean end-of-word so an *unguarded* parser would accept the
|
||||
" file silently — that's what makes this a meaningful regression test
|
||||
" for the depth check specifically (a deeper chain would also crash
|
||||
" unguarded builds via stack overflow, which we don't want in CI).
|
||||
let v = eval('0zFF00000200' .. repeat('0161', 255)
|
||||
\ .. '0100' .. repeat('00', 8))
|
||||
|
||||
call Spellfile_Test(v, 'E759:')
|
||||
|
||||
" KWORDTREE: missing tree node
|
||||
call Spellfile_Test(0zFF0000000000000004, 'E758:')
|
||||
|
||||
|
||||
@@ -729,6 +729,8 @@ static char *(features[]) =
|
||||
|
||||
static int included_patches[] =
|
||||
{ /* Add new patch number below this line */
|
||||
/**/
|
||||
513,
|
||||
/**/
|
||||
512,
|
||||
/**/
|
||||
|
||||
Reference in New Issue
Block a user