Xen Developer Email Discussion

As part of the new emphasis on exposing developer activities to the broader community, I plan to highlight a few xen-devel discussions every week on this blog to allow the community a better view of developer interaction. If you have any thoughts on this new activity please let me know.

1. State of gdbsx in xen-4.0-testing.hg and debugger documents – http://lists.xensource.com/archives/html/xen-devel/2010-07/msg00043.html; picking up this discussion near the beginning, Bruce Edge has questions about the gdbsx debugger put together by Mukesh Rathor as the current documentation has not been updated since the Xen 4.0 release. Mukesh responds with the information to build xen with “make gdbsx=y” which will enable the debugger to be part of the xen kernel and to see more information at tools/debugger/gdbsx/README. Mukesh also recommends removing tools/debugger/gdb as it is not being maintained or used. Keir Fraser agrees on the gdb removal and pushes the final decision to Ian Jackson or Stefano Stabellini. The discussion then continues wiht questions from other developers on using the gdbsx debugger and will leave that to the reader to follow-up if you need more details.
2. S3 sleep in dom0 breaks dom0 <->domU wallclock synchronization – http://lists.xensource.com/archives/html/xen-devel/2010-07/msg00037.html; Rafal Wojtczuk reports the following issue – “after resume from S3 sleep in Dom0, the wall clock in DomU is desynchronized from Dom0’s one (the delta is the length of the S3 sleep). It does not seem that adj_timex is in progress (the delta is constant in time).” Keir Fraser responds with a carbon copy to Jeremy Fitzhardinge on how the pvops DomU handles the wall clock and what a possible solution is using NTP. Joanna Rutkowska replies that NTP is not a valid solution to this issue due to not wanting every DomU to have networking access. Keir replies that testing is needed to indeed confirm if the issue is pvops DomU related than a simple solution can be found. The thread is still ongoing at the time of this posting so this issue is not yet closed.