I'm getting the following times when doing a savevm and delvm in current
HEAD eb47d7c5d (time in minutes, each time corresponds to a savevm/delvm run):
savevm: 5:28m, 11:00m, 11:10m
delvm: 4:30m, 4:40m, > 15m
Now, trying with qemu 0.13.0 I get:
savevm: < 1:00m, 4:00m, 4:34m
delvm: few seconds for all cases
Yes, you read it correctly, I tried with 0.13.0 because 0.14.0 also has the
bug. This is the pattern I see when I run strace against HEAD while running
the savevm command:
pwrite(7,
"\0\1\0\1\0\1\0\1\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2"..., 65536,
196608) = 65536
fdatasync(7) = 0
Ie. a fdatasync() follows every single pwrite(). Something similar also happens
with delvm. I don't see this pattern with 0.13.0.
The good news is that I've tracked it down and Mr. git bisect says that:
29c1a7301af752de6721e031d31faa48887204bd is the first bad commit
commit 29c1a7301af752de6721e031d31faa48887204bd
Author: Kevin Wolf <kw...@redhat.com>
Date: Mon Jan 10 17:17:28 2011 +0100
qcow2: Use QcowCache
Use the new functions of qcow2-cache.c for everything that works on refcount
block and L2 tables.
Signed-off-by: Kevin Wolf <kw...@redhat.com>
:040000 040000 83e364185d37845bb27f1dccd1249d14cc7a9a1e
0c91964a52b5869333d4fb2cb0fa83104151359e M block
