walkero
/
pcre2
1
0
Fork 0
Code Issues Pull Requests Projects Releases 1 Wiki Activity

Re-arrange valgrind code.

This commit is contained in:
Philip.Hazel 2015-11-15 18:55:20 +00:00
parent 9006c26361
commit 7cc0597baa
2 changed files with 17 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
ChangeLog
View File

@ -309,6 +309,9 @@ about JIT compile failures in pcre2test.
92. Implement PCRE2_INFO_HASBACKSLASHC.
93. Re-arrange valgrind support code in pcre2test to avoid spurious reports
with JIT (possibly caused by SSE2?).
Version 10.20 30-June-2015
--------------------------

33
src/pcre2test.c
View File

@ -5369,7 +5369,7 @@ if (dbuffer != NULL)
the number of code units that will be needed (though the buffer may have to be
extended if replication is involved). */
needlen = (size_t)(len * code_unit_size);
needlen = (size_t)((len+1) * code_unit_size);
if (dbuffer == NULL || needlen >= dbuffer_size)
{
while (needlen >= dbuffer_size) dbuffer_size *= 2;
@ -5658,28 +5658,23 @@ if (pat_patctl.replacement[0] != 0 &&
return PR_OK;
}
/* If we have explicit valgrind support, mark the data from after its end to
the end of the buffer as unaddressable, so that a read over the end of the
buffer will be seen by valgrind, even if it doesn't cause a crash. If we're not
building with valgrind support, at least move the data to the end of the buffer
so that it might at least cause a crash. If we are using the POSIX interface,
or testing zero-termination, we must include the terminating zero. */
/* We now have the subject in dbuffer, with len containing the byte length, and
ulen containing the code unit length. Move the data to the end of the buffer so
that a read over the end can be caught by valgrind or other means. If we have
explicit valgrind support, mark the unused start of the buffer unaddressable.
If we are using the POSIX interface, or testing zero-termination, we must
include the terminating zero in the usable data. */
pp = dbuffer;
c = code_unit_size * (((pat_patctl.control & CTL_POSIX) +
(dat_datctl.control & CTL_ZERO_TERMINATE) != 0)? 1:0);
pp = memmove(dbuffer + dbuffer_size - len - c, dbuffer, len + c);
#ifdef SUPPORT_VALGRIND
VALGRIND_MAKE_MEM_NOACCESS(dbuffer + len + c, dbuffer_size - (len + c));
#else
pp = memmove(pp + dbuffer_size - len - c, pp, len + c);
VALGRIND_MAKE_MEM_NOACCESS(dbuffer, dbuffer_size - (len + c));
#endif
/* We now have len containing the byte length, ulen containing the code unit
length, and pp pointing to the subject string. POSIX matching is only possible
in 8-bit mode, and it does not support timing or other fancy features. Some
were checked at compile time, but we need to check the match-time settings
here. */
/* Now pp points to the subject string. POSIX matching is only possible in
8-bit mode, and it does not support timing or other fancy features. Some were
checked at compile time, but we need to check the match-time settings here. */
#ifdef SUPPORT_PCRE2_8
if ((pat_patctl.control & CTL_POSIX) != 0)
@ -5736,14 +5731,14 @@ if ((pat_patctl.control & CTL_POSIX) != 0)
if (pmatch[i].rm_so >= 0)
{
fprintf(outfile, "%2d: ", (int)i);
PCHARSV(dbuffer, pmatch[i].rm_so,
PCHARSV(pp, pmatch[i].rm_so,
pmatch[i].rm_eo - pmatch[i].rm_so, utf, outfile);
fprintf(outfile, "\n");
if ((i == 0 && (dat_datctl.control & CTL_AFTERTEXT) != 0) ||
(dat_datctl.control & CTL_ALLAFTERTEXT) != 0)
{
fprintf(outfile, "%2d+ ", (int)i);
PCHARSV(dbuffer, pmatch[i].rm_eo, len - pmatch[i].rm_eo,
PCHARSV(pp, pmatch[i].rm_eo, len - pmatch[i].rm_eo,
utf, outfile);
fprintf(outfile, "\n");
}