Charset: fix buffer over-read in recode_from_utf8().

When a multi-byte UTF-8 character was split across 3+ single-byte
buffers, the saved bytes continuation path had two related bugs:

ngx_utf8_decode() was called with the last saved-array index instead
of the byte count, causing it to report "incomplete" even when the
sequence was already complete.

The subsequent ngx_memcpy() used that same index as the copy length,
reading past the input buffer boundary.
This commit is contained in:
David Carlier 2026-04-12 07:13:23 +01:00 committed by Sergey Kandaurov
parent 5461e8bbc0
commit 54b7945961

View file

@ -689,7 +689,6 @@ ngx_http_charset_recode_from_utf8(ngx_pool_t *pool, ngx_buf_t *buf,
u_char c, *p, *src, *dst, *saved, **table;
uint32_t n;
ngx_buf_t *b;
ngx_uint_t i;
ngx_chain_t *out, *cl, **ll;
src = buf->pos;
@ -783,18 +782,12 @@ ngx_http_charset_recode_from_utf8(ngx_pool_t *pool, ngx_buf_t *buf,
ngx_log_debug1(NGX_LOG_DEBUG_HTTP, pool->log, 0,
"http charset utf saved: %z", ctx->saved_len);
p = src;
for (i = ctx->saved_len; i < NGX_UTF_LEN; i++) {
ctx->saved[i] = *p++;
if (p == buf->last) {
break;
}
}
len = ngx_min(NGX_UTF_LEN - ctx->saved_len, (size_t) (buf->last - src));
ngx_memcpy(&ctx->saved[ctx->saved_len], src, len);
len += ctx->saved_len;
saved = ctx->saved;
n = ngx_utf8_decode(&saved, i);
n = ngx_utf8_decode(&saved, len);
c = '\0';
@ -810,7 +803,7 @@ ngx_http_charset_recode_from_utf8(ngx_pool_t *pool, ngx_buf_t *buf,
/* incomplete UTF-8 symbol */
if (i < NGX_UTF_LEN) {
if (len < NGX_UTF_LEN) {
out = ngx_http_charset_get_buf(pool, ctx);
if (out == NULL) {
return NULL;
@ -823,8 +816,7 @@ ngx_http_charset_recode_from_utf8(ngx_pool_t *pool, ngx_buf_t *buf,
b->sync = 1;
b->shadow = buf;
ngx_memcpy(&ctx->saved[ctx->saved_len], src, i);
ctx->saved_len += i;
ctx->saved_len = len;
return out;
}