[
https://issues.apache.org/jira/browse/TS-3222?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14253577#comment-14253577
]
Alan M. Carroll commented on TS-3222:
-------------------------------------
I was unable to replicate the problem - in fact, despite trying a variety of
host requests I was unable to get the sprint/snprintf in question to execute at
all. I now wonder if that's effectively dead code for 5.X. I also moved the
port buffer to the class instead of being a local var - how that hasn't failed
catastrophically puzzles me.
> Port number padded with 0 when Host header has explicit port specification
> --------------------------------------------------------------------------
>
> Key: TS-3222
> URL: https://issues.apache.org/jira/browse/TS-3222
> Project: Traffic Server
> Issue Type: Bug
> Reporter: Nakagome Tomoyuki
> Assignee: Alan M. Carroll
> Fix For: 5.3.0
>
>
> Due to this line in proxy/hdrs/HTTP.cc:1552
> {noformat}
> ui->m_len_port = sprintf(port_buff, "%.5d", hdr->m_port);
> {noformat}
> UrlPrintHack::UrlPrintHack() prepends unnecessary '0's so the port_buff will
> be 5 chars, like "00080". Suggestion is to remove .5 from the format string
> and change it to
> {noformat}
> ui->m_len_port = snprintf(port_buff, 6, "%d", hdr->m_port);
> {noformat}
> unless there are some specific reasons for the current code.
> To replicate the symptom, send a request with a Host header having a port
> number like:
> Host: www.apache.org:80
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
