In blktap's qcow we need split up read/write requests if the requests
span multiple clusters. However, with our MAX_AIO_REQUESTS define we
assume that there is only ever a single aio request per tapdisk
request and under heavy i/o we can run out of room causing us to
cancel requests.
Dynamically allocate (based on cluster_bits) the various io request
queues the driver maintains.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
In XendDomainInfo, BOOTLOADER_LOOPBACK_DEVICE is defined to be
'/dev/xvdp'. However, there is still one place in this file that has
'/dev/xvdp' hard-coded.
Signed-off-by: Charles Coffing <ccoffing@novell.com>
1. XendDomainInfo._createDevices() gets a list of devices to be
created from XendConfig.ordered_device_refs().
On a simple guest, this has 4 devices, vfb, vbd, vif, vkbd - in
that order.
2. It iterates over those devices, creating the appropriate
DevController subclass instance, and then calling createDevice()
on that object.
3. When createDevice() is called on the vfb device, it spawns
xen-vncfb daemon.
4. During startup xen-vncfb writes into the backend paths
/local/domain/0/backend/vfb/0
And
/local/domain/0/backend/vkbd/0
5. When createDevice() is called on the vkbd device in XenD, if the
2nd xenstore path write from step 4 has occurred, then you'll hit
the 'Device 0 (vkbd) is already connected' error. If the 2nd path
write didn't complete yet then everything is fine.
I think the reason it often works once after boot is that loading
xen-vncfb from disk the first time around is just enough of a slow
down to ensure step 5 occurs before the 2nd xenstore write in step 4
has occurred.
The key seems to be to ensure the vkbd device is initialized in
xenstore before the vfb device - this ensures all the xenstored setup
from XenD is complete before the xen-vncfb daemon starts. I'm now able
to create & destroy a domain many times over with this patch & never
hit the error message any more.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Allow a policy to be updated on a running system and domains to be
relabeled. The updating of a policy is happening in several steps:
relabeling the domains, testing whether the system would be in a valid
state after the relabeling (according to the policy), committing the
changes if state is determined to be valid.
I have followed Keir's suggestion of building a 2nd linked list
parallel to the domain list. That 2nd list holds security information
related to the running domains. Each entry is pointed to by its domain
structure. The list is protected by its own read/write-lock. I have
moved nearly all ACM-related code that was traversing the domain list
previously to traverse this list instead and not hold onto the domain
list lock.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
This patch changes the XML schema of the ACM policy to require a
version and that every conflict set have a name. Every VM label must
have one Chinese Wall Type and every resource label one Simple Type
Enforcement Type. As a consequence of this some example policies
needed to be changed. Also not offering that many configuration
options for compiling xen anymore to make things simpler.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
The file /var/log/xen/xend-debug.log is currently being created with
executable permission bits set. This is because the os.open() method
defaults to using a mode of 0777 if no third parameter is provided.
The attached patch changes the mode to 0600 to ensure that the file
permissions come out as -rw------- instead of -rwxr-xr-x
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
When a domain is created, the function acm_domain_create() in
domain_create() is called that does what previously the pre- and
post_domain_create functions were doing. Similarly there's a function
acm_domain_destroy() in domain_kill() that reverts changes to state
when destroying a domain. There's no more separate initialization
necessary for domain-0, but domain_create takes one more parameter,
the ssidref. It is usually passed through the hypercall when a domain
is created.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
If one vcpu is executing ptc.g, while the other vcpu is executing itc,
the VHPT(VTLB) entry which should be purged may unexpectedly survive.
Then issue happens.
Signed-off-by: Anthony Xu <anthony.xu@intel.com>
Scale the memory parameter for domains using the XenAPIDomain to
create a domain. Those domains use the normal configuration where the
memory parameter is given in 'Mb', so it needs to be adapted to
'bytes'.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
extensions. mdelay() can more easily be implemented by polling the
DRAM refresh bit in port 0x61. And this does not mess with state that
a guest may be relying on.
Signed-off-by: Keir Fraser <keir@xensource.com>
kfraser@localhost.localdomain
Keir Fraser
Tom Wilkie
Steven Smith
Ian Campbell
Christian Limpach
Alex Williamson
Tim Deegan