[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-devel] [PATCH 0/6] Refactor autostart of machines
From: |
Juan Quintela |
Subject: |
[Qemu-devel] [PATCH 0/6] Refactor autostart of machines |
Date: |
Wed, 19 Aug 2009 04:07:43 +0200 |
Hi
This series add several things:
- delay sighandler_setup(). Current code hangs until you are using
telnet:....,server for instance. With this change, ^C while waiting
for telnet program to conect will kill the vm.
- refactor autostart behaviour:
* incoming migration: depend of -S parameter
* loadvm from command line: depend of -S parameter
* loadvm from monitor: same state that the vm it replaces
* no loadvm no incoming: depend of -S parameter
I tested, and all combinations works this way.
- refactor do_loadvm(). do_loadvm() don't return errors, if there were
any error while loading a vm, it would obey autostart value independently of
the errors. Now it will never start a vm that had erros while loading.
- if you type "cont" in the monitor of one machine just migrated, qemu will
proceed with the start, with the consequent corruption. Last patch forbides
to continue a vm that has just been migrated with an error message. If you
load a new vm, flag is reset. This "bug/feature" was reported as a bug
by users.
One comment:
- Is there really a good idea to allow loadvm in the monitor even if the current
vm is running? I think that allowing loadvm command to only works when vm is
stopped is a better idea. what does everybody else thinks?
Comments?
Later, Juan.
Juan Quintela (6):
Delay sighandler_setup()
We want to autostart on incoming conections Once there, indent
surrounded code in qemu style
split do_loadvm() into do_loadvm() and load_vmstate()
move do_loadvm() to monitor.c
make load_vmstate() return errors
Continue a migrated vm is a bad idea
migration.c | 1 +
monitor.c | 16 ++++++++++++++++
savevm.c | 19 +++++++------------
sysemu.h | 3 ++-
vl.c | 24 +++++++++++++-----------
5 files changed, 39 insertions(+), 24 deletions(-)