--- /dev/null
+From 67219f0130ec7c876ac0b299046460fad31caabf Mon Sep 17 00:00:00 2001
+From: Rich Felker <dalias@aerifal.cx>
+Date: Mon, 30 Mar 2026 16:00:50 -0400
+Subject: fix pathological slowness & incorrect mappings in iconv gb18030
+ decoder
+
+in order to implement the "UTF" aspect of gb18030 (ability to
+represent arbitrary unicode characters not present in the 2-byte
+mapping), we have to apply the index obtained from the encoded 4-byte
+sequence into the set of unmapped characters. this was done by
+scanning repeatedly over the table of mapped characters and counting
+off mapped characters below a running index by which to adjust the
+running index by on each iteration. this iterative process eventually
+leaves us with the value of the Nth unmapped character replacing the
+index, but depending on which particular character that is, the number
+of iterations needed to find it can be in the tens of thousands, and
+each iteration traverses the whole 126x190 table in the inner loop.
+this can lead to run times exceeding an entire second per character on
+moderate-speed machines.
+
+on top of that, the transformation logic produced wrong results for
+BMP characters above the the surrogate range, as a result of not
+correctly accounting for it being excluded, and for characters outside
+the BMP, as a result of a misunderstanding of how gb18030 encodes
+them.
+
+this patch replaces the unmapped character lookup with a single linear
+search of a list of unmapped ranges. there are only 206 such ranges,
+and these are permanently assigned and unchangeable as a consequence
+of the character encoding having to be stable, so a simple array of
+16-bit start/length values for each range consumes only 824 bytes, a
+very reasonable size cost here.
+
+this new table accounts for the previously-incorrect surrogate
+handling, and non-BMP characters are handled correctly by a single
+offset, without the need for any unmapped-range search.
+
+there are still a small number of mappings that are incorrect due to
+late changes made in the definition of gb18030, swapping PUA
+codepoints with proper Unicode characters. correcting these requires a
+postprocessing step that will be added later.
+---
+ src/locale/gb18030utf.h | 206 ++++++++++++++++++++++++++++++++++++++++
+ src/locale/iconv.c | 33 +++++--
+ 2 files changed, 230 insertions(+), 9 deletions(-)
+ create mode 100644 src/locale/gb18030utf.h
+
+--- /dev/null
++++ b/src/locale/gb18030utf.h
+@@ -0,0 +1,206 @@
++{ 0x80, 36 },
++{ 0xa5, 2 },
++{ 0xa9, 7 },
++{ 0xb2, 5 },
++{ 0xb8, 31 },
++{ 0xd8, 8 },
++{ 0xe2, 6 },
++{ 0xeb, 1 },
++{ 0xee, 4 },
++{ 0xf4, 3 },
++{ 0xf8, 1 },
++{ 0xfb, 1 },
++{ 0xfd, 4 },
++{ 0x102, 17 },
++{ 0x114, 7 },
++{ 0x11c, 15 },
++{ 0x12c, 24 },
++{ 0x145, 3 },
++{ 0x149, 4 },
++{ 0x14e, 29 },
++{ 0x16c, 98 },
++{ 0x1cf, 1 },
++{ 0x1d1, 1 },
++{ 0x1d3, 1 },
++{ 0x1d5, 1 },
++{ 0x1d7, 1 },
++{ 0x1d9, 1 },
++{ 0x1db, 1 },
++{ 0x1dd, 28 },
++{ 0x1fa, 87 },
++{ 0x252, 15 },
++{ 0x262, 101 },
++{ 0x2c8, 1 },
++{ 0x2cc, 13 },
++{ 0x2da, 183 },
++{ 0x3a2, 1 },
++{ 0x3aa, 7 },
++{ 0x3c2, 1 },
++{ 0x3ca, 55 },
++{ 0x402, 14 },
++{ 0x450, 1 },
++{ 0x452, 7102 },
++{ 0x2011, 2 },
++{ 0x2017, 1 },
++{ 0x201a, 2 },
++{ 0x201e, 7 },
++{ 0x2027, 9 },
++{ 0x2031, 1 },
++{ 0x2034, 1 },
++{ 0x2036, 5 },
++{ 0x203c, 112 },
++{ 0x20ad, 86 },
++{ 0x2104, 1 },
++{ 0x2106, 3 },
++{ 0x210a, 12 },
++{ 0x2117, 10 },
++{ 0x2122, 62 },
++{ 0x216c, 4 },
++{ 0x217a, 22 },
++{ 0x2194, 2 },
++{ 0x219a, 110 },
++{ 0x2209, 6 },
++{ 0x2210, 1 },
++{ 0x2212, 3 },
++{ 0x2216, 4 },
++{ 0x221b, 2 },
++{ 0x2221, 2 },
++{ 0x2224, 1 },
++{ 0x2226, 1 },
++{ 0x222c, 2 },
++{ 0x222f, 5 },
++{ 0x2238, 5 },
++{ 0x223e, 10 },
++{ 0x2249, 3 },
++{ 0x224d, 5 },
++{ 0x2253, 13 },
++{ 0x2262, 2 },
++{ 0x2268, 6 },
++{ 0x2270, 37 },
++{ 0x2296, 3 },
++{ 0x229a, 11 },
++{ 0x22a6, 25 },
++{ 0x22c0, 82 },
++{ 0x2313, 333 },
++{ 0x246a, 10 },
++{ 0x249c, 100 },
++{ 0x254c, 4 },
++{ 0x2574, 13 },
++{ 0x2590, 3 },
++{ 0x2596, 10 },
++{ 0x25a2, 16 },
++{ 0x25b4, 8 },
++{ 0x25be, 8 },
++{ 0x25c8, 3 },
++{ 0x25cc, 2 },
++{ 0x25d0, 18 },
++{ 0x25e6, 31 },
++{ 0x2607, 2 },
++{ 0x260a, 54 },
++{ 0x2641, 1 },
++{ 0x2643, 2110 },
++{ 0x2e82, 2 },
++{ 0x2e85, 3 },
++{ 0x2e89, 2 },
++{ 0x2e8d, 10 },
++{ 0x2e98, 15 },
++{ 0x2ea8, 2 },
++{ 0x2eab, 3 },
++{ 0x2eaf, 4 },
++{ 0x2eb4, 2 },
++{ 0x2eb8, 3 },
++{ 0x2ebc, 14 },
++{ 0x2ecb, 293 },
++{ 0x2ffc, 4 },
++{ 0x3004, 1 },
++{ 0x3018, 5 },
++{ 0x301f, 2 },
++{ 0x302a, 20 },
++{ 0x303f, 2 },
++{ 0x3094, 7 },
++{ 0x309f, 2 },
++{ 0x30f7, 5 },
++{ 0x30ff, 6 },
++{ 0x312a, 246 },
++{ 0x322a, 7 },
++{ 0x3232, 113 },
++{ 0x32a4, 234 },
++{ 0x3390, 12 },
++{ 0x339f, 2 },
++{ 0x33a2, 34 },
++{ 0x33c5, 9 },
++{ 0x33cf, 2 },
++{ 0x33d3, 2 },
++{ 0x33d6, 113 },
++{ 0x3448, 43 },
++{ 0x3474, 298 },
++{ 0x359f, 111 },
++{ 0x360f, 11 },
++{ 0x361b, 765 },
++{ 0x3919, 85 },
++{ 0x396f, 96 },
++{ 0x39d1, 14 },
++{ 0x39e0, 147 },
++{ 0x3a74, 218 },
++{ 0x3b4f, 287 },
++{ 0x3c6f, 113 },
++{ 0x3ce1, 885 },
++{ 0x4057, 264 },
++{ 0x4160, 471 },
++{ 0x4338, 116 },
++{ 0x43ad, 4 },
++{ 0x43b2, 43 },
++{ 0x43de, 248 },
++{ 0x44d7, 373 },
++{ 0x464d, 20 },
++{ 0x4662, 193 },
++{ 0x4724, 5 },
++{ 0x472a, 82 },
++{ 0x477d, 16 },
++{ 0x478e, 441 },
++{ 0x4948, 50 },
++{ 0x497b, 2 },
++{ 0x497e, 4 },
++{ 0x4984, 1 },
++{ 0x4987, 20 },
++{ 0x499c, 3 },
++{ 0x49a0, 22 },
++{ 0x49b8, 703 },
++{ 0x4c78, 39 },
++{ 0x4ca4, 111 },
++{ 0x4d1a, 148 },
++{ 0x4daf, 81 },
++{ 0x9fa6, 14426 },
++{ 0xe76c, 1 },
++{ 0xe7c8, 1 },
++{ 0xe7e7, 13 },
++{ 0xe815, 1 },
++{ 0xe819, 5 },
++{ 0xe81f, 7 },
++{ 0xe827, 4 },
++{ 0xe82d, 4 },
++{ 0xe833, 8 },
++{ 0xe83c, 7 },
++{ 0xe844, 16 },
++{ 0xe856, 14 },
++{ 0xe865, 4295 },
++{ 0xf92d, 76 },
++{ 0xf97a, 27 },
++{ 0xf996, 81 },
++{ 0xf9e8, 9 },
++{ 0xf9f2, 26 },
++{ 0xfa10, 1 },
++{ 0xfa12, 1 },
++{ 0xfa15, 3 },
++{ 0xfa19, 6 },
++{ 0xfa22, 1 },
++{ 0xfa25, 2 },
++{ 0xfa2a, 1030 },
++{ 0xfe32, 1 },
++{ 0xfe45, 4 },
++{ 0xfe53, 1 },
++{ 0xfe58, 1 },
++{ 0xfe67, 1 },
++{ 0xfe6c, 149 },
++{ 0xff5f, 129 },
++{ 0xffe6, 26 },
+--- a/src/locale/iconv.c
++++ b/src/locale/iconv.c
+@@ -74,6 +74,10 @@ static const unsigned short gb18030[126]
+ #include "gb18030.h"
+ };
+
++static const unsigned short gb18030utf[][2] = {
++#include "gb18030utf.h"
++};
++
+ static const unsigned short big5[89][157] = {
+ #include "big5.h"
+ };
+@@ -224,6 +228,8 @@ static unsigned uni_to_jis(unsigned c)
+ }
+ }
+
++#define countof(a) (sizeof (a) / sizeof *(a))
++
+ size_t iconv(iconv_t cd, char **restrict in, size_t *restrict inb, char **restrict out, size_t *restrict outb)
+ {
+ size_t x=0;
+@@ -423,15 +429,24 @@ size_t iconv(iconv_t cd, char **restrict
+ d = *((unsigned char *)*in + 3);
+ if (d-'0'>9) goto ilseq;
+ c += d-'0';
+- c += 128;
+- for (d=0; d<=c; ) {
+- k = 0;
+- for (int i=0; i<126; i++)
+- for (int j=0; j<190; j++)
+- if (gb18030[i][j]-d <= c-d)
+- k++;
+- d = c+1;
+- c += k;
++ /* Starting at 90 30 81 30 (189000), mapping is
++ * linear without gaps, to U+10000 and up. */
++ if (c >= 189000) {
++ c -= 189000;
++ c += 0x10000;
++ if (c >= 0x110000) goto ilseq;
++ break;
++ }
++ /* Otherwise we must process an index into set
++ * of characters unmapped by 2-byte table. */
++ for (int i=0; ; i++) {
++ if (i==countof(gb18030utf))
++ goto ilseq;
++ if (c<gb18030utf[i][1]) {
++ c += gb18030utf[i][0];
++ break;
++ }
++ c -= gb18030utf[i][1];
+ }
+ break;
+ }
--- /dev/null
+From 228da39e38c1cae13cbe637e771412c1984dba5d Mon Sep 17 00:00:00 2001
+From: Rich Felker <dalias@aerifal.cx>
+Date: Thu, 9 Apr 2026 22:51:30 -0400
+Subject: qsort: fix leonardo heap corruption from bug in doubleword ctz
+ primitive
+
+the pntz function, implementing a "count trailing zeros" variant for a
+bit vector consisting of two size_t words, erroneously returned zero
+rather than the number of bits in the low word when the first bit set
+was the low bit of the high word.
+
+as a result, a loop in the trinkle function which should have a
+guaranteed small bound on the number of iterations, could run
+unboundedly, thereby overflowing a stack-based working-space array
+which was sized for the bound.
+
+CVE-2026-40200 has been assigned for this issue.
+---
+ src/stdlib/qsort.c | 8 ++++----
+ 1 file changed, 4 insertions(+), 4 deletions(-)
+
+--- a/src/stdlib/qsort.c
++++ b/src/stdlib/qsort.c
+@@ -34,11 +34,11 @@
+
+ typedef int (*cmpfun)(const void *, const void *, void *);
+
++/* returns index of first bit set, excluding the low bit assumed to always
++ * be set, starting from low bit of p[0] up through high bit of p[1] */
+ static inline int pntz(size_t p[2]) {
+- int r = ntz(p[0] - 1);
+- if(r != 0 || (r = 8*sizeof(size_t) + ntz(p[1])) != 8*sizeof(size_t)) {
+- return r;
+- }
++ if (p[0] != 1) return ntz(p[0] - 1);
++ if (p[1]) return 8*sizeof(size_t) + ntz(p[1]);
+ return 0;
+ }
+
--- /dev/null
+From b3291b9a9f77f1f993d2b4f8c68a26cf09221ae7 Mon Sep 17 00:00:00 2001
+From: Rich Felker <dalias@aerifal.cx>
+Date: Thu, 9 Apr 2026 23:40:53 -0400
+Subject: qsort: hard-preclude oob array writes independent of any invariants
+
+while the root cause of CVE-2026-40200 was a faulty ctz primitive, the
+fallout of the bug would have been limited to erroneous sorting or
+infinite loop if not for the stores to a stack-based array that
+depended on trusting invariants in order not to go out of bounds.
+
+increase the size of the array to a power of two so that we can mask
+indices into it to force them into range. in the absence of any
+further bug, the masking is a no-op, but it does not have any
+measurable performance cost, and it makes spatial memory safety
+trivial to prove (and for readers not familiar with the algorithms to
+trust).
+---
+ src/stdlib/qsort.c | 20 +++++++++++++-------
+ 1 file changed, 13 insertions(+), 7 deletions(-)
+
+--- a/src/stdlib/qsort.c
++++ b/src/stdlib/qsort.c
+@@ -89,10 +89,16 @@ static inline void shr(size_t p[2], int
+ p[1] >>= n;
+ }
+
++/* power-of-two length for working array so that we can mask indices and
++ * not depend on any invariant of the algorithm for spatial memory safety.
++ * the original size was just 14*sizeof(size_t)+1 */
++#define AR_LEN (16 * sizeof(size_t))
++#define AR_MASK (AR_LEN - 1)
++
+ static void sift(unsigned char *head, size_t width, cmpfun cmp, void *arg, int pshift, size_t lp[])
+ {
+ unsigned char *rt, *lf;
+- unsigned char *ar[14 * sizeof(size_t) + 1];
++ unsigned char *ar[AR_LEN];
+ int i = 1;
+
+ ar[0] = head;
+@@ -104,16 +110,16 @@ static void sift(unsigned char *head, si
+ break;
+ }
+ if(cmp(lf, rt, arg) >= 0) {
+- ar[i++] = lf;
++ ar[i++ & AR_MASK] = lf;
+ head = lf;
+ pshift -= 1;
+ } else {
+- ar[i++] = rt;
++ ar[i++ & AR_MASK] = rt;
+ head = rt;
+ pshift -= 2;
+ }
+ }
+- cycle(width, ar, i);
++ cycle(width, ar, i & AR_MASK);
+ }
+
+ static void trinkle(unsigned char *head, size_t width, cmpfun cmp, void *arg, size_t pp[2], int pshift, int trusty, size_t lp[])
+@@ -121,7 +127,7 @@ static void trinkle(unsigned char *head,
+ unsigned char *stepson,
+ *rt, *lf;
+ size_t p[2];
+- unsigned char *ar[14 * sizeof(size_t) + 1];
++ unsigned char *ar[AR_LEN];
+ int i = 1;
+ int trail;
+
+@@ -142,7 +148,7 @@ static void trinkle(unsigned char *head,
+ }
+ }
+
+- ar[i++] = stepson;
++ ar[i++ & AR_MASK] = stepson;
+ head = stepson;
+ trail = pntz(p);
+ shr(p, trail);
+@@ -150,7 +156,7 @@ static void trinkle(unsigned char *head,
+ trusty = 0;
+ }
+ if(!trusty) {
+- cycle(width, ar, i);
++ cycle(width, ar, i & AR_MASK);
+ sift(head, width, cmp, arg, pshift, lp);
+ }
+ }
--- /dev/null
+From 5122f9f3c99fee366167c5de98b31546312921ab Mon Sep 17 00:00:00 2001
+From: Luca Kellermann <mailto.luca.kellermann@gmail.com>
+Date: Fri, 10 Apr 2026 03:03:22 +0200
+Subject: qsort: fix shift UB in shl and shr
+
+if shl() or shr() are called with n==8*sizeof(size_t), n is adjusted
+to 0. the shift by (sizeof(size_t) * 8 - n) that then follows will
+consequently shift by the width of size_t, which is UB and in practice
+produces an incorrect result.
+
+return early in this case. the bitvector p was already shifted by the
+required amount.
+---
+ src/stdlib/qsort.c | 2 ++
+ 1 file changed, 2 insertions(+)
+
+--- a/src/stdlib/qsort.c
++++ b/src/stdlib/qsort.c
+@@ -71,6 +71,7 @@ static inline void shl(size_t p[2], int
+ n -= 8 * sizeof(size_t);
+ p[1] = p[0];
+ p[0] = 0;
++ if (!n) return;
+ }
+ p[1] <<= n;
+ p[1] |= p[0] >> (sizeof(size_t) * 8 - n);
+@@ -83,6 +84,7 @@ static inline void shr(size_t p[2], int
+ n -= 8 * sizeof(size_t);
+ p[0] = p[1];
+ p[1] = 0;
++ if (!n) return;
+ }
+ p[0] >>= n;
+ p[0] |= p[1] << (sizeof(size_t) * 8 - n);
static const unsigned short jis0208[84][94] = {
#include "jis0208.h"
};
-@@ -85,6 +88,7 @@ static const unsigned short hkscs[] = {
+@@ -89,6 +92,7 @@ static const unsigned short hkscs[] = {
static const unsigned short ksc[93][94] = {
#include "ksc.h"
};
static const unsigned short rev_jis[] = {
#include "revjis.h"
-@@ -205,6 +209,7 @@ static unsigned legacy_map(const unsigne
+@@ -209,6 +213,7 @@ static unsigned legacy_map(const unsigne
return x < 256 ? x : legacy_chars[x-256];
}
static unsigned uni_to_jis(unsigned c)
{
unsigned nel = sizeof rev_jis / sizeof *rev_jis;
-@@ -223,6 +228,7 @@ static unsigned uni_to_jis(unsigned c)
+@@ -227,6 +232,7 @@ static unsigned uni_to_jis(unsigned c)
}
}
}
+#endif
- size_t iconv(iconv_t cd, char **restrict in, size_t *restrict inb, char **restrict out, size_t *restrict outb)
- {
-@@ -319,6 +325,7 @@ size_t iconv(iconv_t cd, char **restrict
+ #define countof(a) (sizeof (a) / sizeof *(a))
+
+@@ -325,6 +331,7 @@ size_t iconv(iconv_t cd, char **restrict
}
type = scd->state;
continue;
case SHIFT_JIS:
if (c < 128) break;
if (c-0xa1 <= 0xdf-0xa1) {
-@@ -518,6 +525,7 @@ size_t iconv(iconv_t cd, char **restrict
+@@ -533,6 +540,7 @@ size_t iconv(iconv_t cd, char **restrict
c = ksc[c][d];
if (!c) goto ilseq;
break;
default:
if (!c) break;
c = legacy_map(map, c);
-@@ -559,6 +567,7 @@ size_t iconv(iconv_t cd, char **restrict
+@@ -574,6 +582,7 @@ size_t iconv(iconv_t cd, char **restrict
}
}
goto subst;
case SHIFT_JIS:
if (c < 128) goto revout;
if (c == 0xa5) {
-@@ -632,6 +641,7 @@ size_t iconv(iconv_t cd, char **restrict
+@@ -647,6 +656,7 @@ size_t iconv(iconv_t cd, char **restrict
*(*out)++ = 'B';
*outb -= 8;
break;
projects / thirdparty / openwrt.git / commitdiff
