commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sebb (Updated) (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (NET-412) Apache Commons TFTP implementation subject to Sorcerer's Apprentice Syndrome
Date Thu, 05 Jan 2012 22:10:40 GMT

     [ https://issues.apache.org/jira/browse/NET-412?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Sebb updated NET-412:
---------------------

    Description: 
This is regarding the org.apache.commons.net.tftp.TFTPClient class.

In the sendFile() method, when a duplicate TFTP ACK is received, the "if (lastBlock == block)"
section of code fails and control moves to the "else" statement. The else block checks to
see if the ACK is for the previous DATA packet, and if so, sends control back to the _sendPacket
label where the DATA packet is resent.

This is in direct violation of Sorcerer's Apprentice Syndrome as defined in RFC 1123 section
4.2.3.1:

"the sender (i.e., the side originating the DATA packets) must never resend the current DATA
packet on receipt of a duplicate ACK"

The proper else block would look something like this:

{noformat}
{
    continue _receivePacket;
}
{noformat}

This enables the client to ignore the duplicated ACK and listen for ACK+1. If ACK+1 never
comes in, the listener should timeout and attempt to resend the current data packet. As described
in NET-410, the TFTP client also does not properly handle re-transmits either. It is recommended
that this issue and NET-410 be implemented in tandem.





  was:
This is regarding the org.apache.commons.net.tftp.TFTPClient class.

In the sendFile() method, when a duplicate TFTP ACK is received, the "if (lastBlock == block)"
section of code fails and control moves to the "else" statement. The else block checks to
see if the ACK is for the previous DATA packet, and if so, sends control back to the _sendPacket
label where the DATA packet is resent.

This is in direct violation of Sorcerer's Apprentice Syndrome as defined in RFC 1123 section
4.2.3.1:

"the sender (i.e., the side originating the DATA packets) must never resend the current DATA
packet on receipt of a duplicate ACK"

The proper else block would look something like this:

{
    continue _receivePacket;
}

This enables the client to ignore the duplicated ACK and listen for ACK+1. If ACK+1 never
comes in, the listener should timeout and attempt to resend the current data packet. As described
in NET-410, the TFTP client also does not properly handle re-transmits either. It is recommended
that this issue and NET-410 be implemented in tandem.





    
> Apache Commons TFTP implementation subject to Sorcerer's Apprentice Syndrome
> ----------------------------------------------------------------------------
>
>                 Key: NET-412
>                 URL: https://issues.apache.org/jira/browse/NET-412
>             Project: Commons Net
>          Issue Type: Bug
>          Components: TFTP
>    Affects Versions: 2.2, 3.0
>         Environment: Java 6 patch 20
>            Reporter: Chuck Wolber
>            Priority: Critical
>
> This is regarding the org.apache.commons.net.tftp.TFTPClient class.
> In the sendFile() method, when a duplicate TFTP ACK is received, the "if (lastBlock ==
block)" section of code fails and control moves to the "else" statement. The else block checks
to see if the ACK is for the previous DATA packet, and if so, sends control back to the _sendPacket
label where the DATA packet is resent.
> This is in direct violation of Sorcerer's Apprentice Syndrome as defined in RFC 1123
section 4.2.3.1:
> "the sender (i.e., the side originating the DATA packets) must never resend the current
DATA packet on receipt of a duplicate ACK"
> The proper else block would look something like this:
> {noformat}
> {
>     continue _receivePacket;
> }
> {noformat}
> This enables the client to ignore the duplicated ACK and listen for ACK+1. If ACK+1 never
comes in, the listener should timeout and attempt to resend the current data packet. As described
in NET-410, the TFTP client also does not properly handle re-transmits either. It is recommended
that this issue and NET-410 be implemented in tandem.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message