Re: gwlib/http.c patch

[Vincent CHAVANIS]

We should in that case also patch *http_header_value()

Vincent.


Alexander Malysh a écrit :

hi,

yes, you are right here. it doesn't crash only due to the octstr_copy length adjust.

Patch applied to cvs.

Thanks,
Alex

Am 27.03.2009 um 18:30 schrieb Nikos Balkanas:

Hi,

A small patch that left as it is should lead to memory corruption. I have not tested the broken code but it seems quite obvious to me. Please vote & decide. BR,

Nikos
<http.diff>

--- /gateway-cvs/gwlib/http.c   2009-03-30 10:58:32.000000000 +0200
+++ /gateway/gwlib/http.c       2009-03-30 16:38:47.000000000 +0200
@@ -2850,7 +2850,7 @@
             current_name = octstr_copy(os, 0, colon);
         }
         if (octstr_case_compare(current_name, name) == 0) {
-            value = octstr_copy(os, colon + 1, octstr_len(os));
+            value = octstr_copy(os, colon + 1, octstr_len(os) - colon - 1);
             octstr_strip_blanks(value);
             octstr_destroy(current_name);
             return value;