commit 0e3fd4df9990b240cab3754a7c3e809d081b56d9 Author: teor teor@torproject.org Date: Tue Jan 7 17:09:25 2020 +1000

string: Check UTF-8 string pointer and length

If they are inconsistent, output a raw bug log.

Part of 32845. --- src/lib/string/util_string.c | 21 +++++++++++++++++++-- 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/src/lib/string/util_string.c b/src/lib/string/util_string.c index f5061a11d..93b3eb09f 100644 --- a/src/lib/string/util_string.c +++ b/src/lib/string/util_string.c @@ -506,6 +506,23 @@ validate_char(const uint8_t *c, uint8_t len) int string_is_utf8(const char *str, size_t len) { + // If str is NULL, don't try to read it + if (!str) { + // We could test for this case, but the low-level logs would produce + // confusing test output. + // LCOV_EXCL_START + if (len) { + // Use the low-level logging function, so that the log module can + // validate UTF-8 (if needed in future code) + tor_log_err_sigsafe( + "BUG: string_is_utf8() called with NULL str but non-zero len."); + // Since it's a bug, we should probably reject this string + return false; + } + // LCOV_EXCL_STOP + return true; + } + for (size_t i = 0; i < len;) { uint8_t num_bytes = bytes_in_char(str[i]); if (num_bytes == 0) // Invalid leading byte found. @@ -530,8 +547,8 @@ string_is_utf8(const char *str, size_t len) int string_is_utf8_no_bom(const char *str, size_t len) { - if (len >= 3 && (!strcmpstart(str, "\uFEFF") || - !strcmpstart(str, "\uFFFE"))) { + if (str && len >= 3 && (!strcmpstart(str, "\uFEFF") || + !strcmpstart(str, "\uFFFE"))) { return false; } return string_is_utf8(str, len);