Am Samstag, 29. Dezember 2018, 19:34:42 CET schrieb Mojtaba:
In msg_translator.c source file, goes to check_via_address function in line 147, In that function, the length of both ip in host(name) and host(ip) is just checked in ipv6 (if condition section). the code are here:
/* check if name->s is an ipv6 address or an ipv6 address ref. */ if ((ip->af==AF_INET6) && ( ((len==name->len)&&(strncasecmp(name->s, s, name->len)==0))
((len==(name->len-2))&&(name->s[0]=='[')&& (name->s[name->len-1]==']')&& (strncasecmp(name->s+1, s, len)==0)) )) { return 0; } else { if (unlikely(name->s==NULL)) { LM_CRIT("invalid Via host name\n"); return -1; }
if (strncmp(name->s, s, name->len)==0) //I think the length would be checked return 0; }
Hello Mojtaba,
thank you for the report. This looks indeed not correct. I fixed this today in git master (commit 98ee014a) and will also backport it. I wonder why this was not found earlier, this is a really old function. But most people probably use DNS names in their VIAs.
Best regards,
Henning
But in else section, the length does not check. I think it would be nice to check . like this:
if (len == name->len && strncmp(name->s, s, name->len)==0)
Not actually, i am not referring to git master yet. I just want to know do it would be nice or not? With Regards.Mojtaba Esfandiari.S
Does anybody know why the lenght of both ip-address (s and name) in check_via_address function in msg_translator.c source file dose not check first? Actually i think it's better do. For example if via:192.168.122.1 and src:192.168.122.100, the result returned true, as they are not the same! With.Regards. Mojtaba
Hello Mojtaba,
can you give more details, which line in e.g. git master are you referring to?