diff --git a/ChangeLog b/ChangeLog index 68f4fc4..6e75456 100644 --- a/ChangeLog +++ b/ChangeLog @@ -63,6 +63,9 @@ If the limit is reached, it just gives up trying for this optimization. 16. The control verb chain list must always be restored when exiting from a recurse function in JIT. +17. Fix a crash which occurs when the character type of an invalid UTF +character is decoded in JIT. + Version 10.34 21-November-2019 ------------------------------ diff --git a/src/pcre2_jit_compile.c b/src/pcre2_jit_compile.c index 10665a8..ef29a76 100644 --- a/src/pcre2_jit_compile.c +++ b/src/pcre2_jit_compile.c @@ -7224,7 +7224,13 @@ cc = ccbegin; if ((cc[-1] & XCL_NOT) != 0) read_char(common, min, max, backtracks, READ_CHAR_UPDATE_STR_PTR); else + { +#ifdef SUPPORT_UNICODE + read_char(common, min, max, (needstype || needsscript) ? backtracks : NULL, 0); +#else /* !SUPPORT_UNICODE */ read_char(common, min, max, NULL, 0); +#endif /* SUPPORT_UNICODE */ + } if ((cc[-1] & XCL_HASPROP) == 0) { diff --git a/src/pcre2_jit_test.c b/src/pcre2_jit_test.c index 187e565..619e738 100644 --- a/src/pcre2_jit_test.c +++ b/src/pcre2_jit_test.c @@ -1965,6 +1965,8 @@ static struct invalid_utf8_regression_test_case invalid_utf8_regression_test_cas { PCRE2_UTF, CI, 0, 0, 0, 4, 8, { "#\xc7\x85#", NULL }, "\x80\x80#\xc7#\xc7\x85#" }, { PCRE2_UTF, CI, 0, 0, 0, 7, 11, { "#\xc7\x85#", NULL }, "\x80\x80#\xc7\x80\x80\x80#\xc7\x85#" }, + { PCRE2_UTF | PCRE2_UCP, CI, 0, 0, 0, -1, -1, { "[\\s]", NULL }, "\xed\xa0\x80" }, + /* These two are not invalid UTF tests, but this infrastructure fits better for them. */ { 0, PCRE2_JIT_COMPLETE, 0, 0, 1, -1, -1, { "\\X{2}", NULL }, "\r\n\n" }, { 0, PCRE2_JIT_COMPLETE, 0, 0, 1, -1, -1, { "\\R{2}", NULL }, "\r\n\n" },