--- Comment #5 from Gerald Combs <> ---
It looks like we're ending up with an optimized strlen, for which Valgrind
doesn't care[1]. If I change the "read_buf[0] = '\0'" at line 126 to
"memset(read_buf, 0, 64)" Valgrind stops complaining. Passing "-fno-builtin" to
clang doesn't fix it unfortunately:

==25583== Conditional jump or move depends on uninitialised value(s)
==25583==    at 0x4C32D08: strlen (in
==25583==    by 0x751C832: process_mmdbr_stdout (maxmind_db.c:135)
==25583==    by 0x7579828: uat_load (uat_load.l:372)
==25583==    by 0x7569114: uat_load_all (uat.c:532)
==25583==    by 0x752FE07: init_prefs (prefs.c:3961)
==25583==    by 0x752FE07: read_prefs (prefs.c:4314)
==25583==    by 0x75170EA: epan_load_settings (epan.c:283)
==25583==    by 0x119CC9: main (tshark.c:1016)

[1] (It's for gcc, but I'm assuming
that the same applies for clang).

You are receiving this mail because:
You are watching all bug changes.
Sent via:    Wireshark-bugs mailing list <>

Reply via email to