qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Gordon Sim" <g...@redhat.com>
Subject Re: Review Request: C++ Broker Link channel numbers may collide
Date Thu, 25 Oct 2012 08:05:19 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/7727/#review12761
-----------------------------------------------------------

Ship it!


Ship It!

- Gordon Sim


On Oct. 24, 2012, 7:55 p.m., Chug Rolke wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/7727/
> -----------------------------------------------------------
> 
> (Updated Oct. 24, 2012, 7:55 p.m.)
> 
> 
> Review request for qpid.
> 
> 
> Description
> -------
> 
> Previously Link channel numbers are allocated by a simple incrementer. When the counter
wrapped then old channel numbers were blindly reused.
> 
> Now a pool of available channel numbers is kept in a RangeSet. Each bridge allocates
a channel number from the pool when it is created and returns the channel number to the pool
when it is destroyed.
> 
> Debug level messages log the allocation and freeing of channel numbers.
> 
> An exception is thrown if the pool runs empty.
> 
> 
> This addresses bug QPID-4392.
>     https://issues.apache.org/jira/browse/QPID-4392
> 
> 
> Diffs
> -----
> 
>   trunk/qpid/cpp/src/qpid/broker/Bridge.h 1400732 
>   trunk/qpid/cpp/src/qpid/broker/Link.h 1400732 
>   trunk/qpid/cpp/src/qpid/broker/Link.cpp 1400732 
>   trunk/qpid/cpp/src/qpid/broker/LinkRegistry.cpp 1400732 
> 
> Diff: https://reviews.apache.org/r/7727/diff/
> 
> 
> Testing
> -------
> 
> A test broker is built with a pool of only 15 (vs normal 32767) channel numbers. Simple
tests verify that 1) old code wraps and causes failures, 2) new code detects pool exhaustion,
and 3) new code logs channel numbers being properly reused.
> 
> Normal self tests pass.
> 
> 
> Thanks,
> 
> Chug Rolke
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message