interface (the legacy hypercalls would break if subcommands with large
argument structures were added, as it would grow the size of the
union of all argument structures).
Also, based on a patch from Kevin Tian, add a new physdev op to
signal EOI for a particular irq.
Signed-off-by: Keir Fraser <keir@xensource.com>
Add support to physdev ops, and thus give IOSAPIC RTEs
managed by Xen now. Dom0 now issues hypercall to r/w
RTE entry. Another change is the irq vector allocation
which is also owned by xen now.
After this change, the IOSAPIC is almost owned by xen
with only exception as IOSAPIC EOI which is still issued
by dom0 directly. But that's OK since currently dom0
owns all external physical devices. Later full event
channel mechanism will provide necessary support for
driver domain, and at that time, dom0 instead issues
physdev_op (PHYSDEVOP_IRQ_UNMASK_NOTIFY) naturally as
replace of IOSAPIC EOI.
Signed-off-by Kevin Tian <kevin.tian@intel.com>
xen/ia64 with the P2M/VP model pseudo physical address(gpaddr) is
fully virtualized so it defines
xen_features(XENFEAT_auto_translated_physmap) = 1. In this case
page_to_phys(page) should return pseudo physical address like
pfn_to_mfn() and its families. However dma is not virtualized, it
can't be used for pci-dma-xen.c, swiotlb.c.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
callback address.
This new hypercall incorporates the functionality of the existing set_callbacks
hypercall in a more flexible manner. set_callbacks is retained for compatibility.
Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
Largely this involves making the logic symmetrical: for example,
not only should netfront be able to tell netback that a packet has
an empty protocol checksum field, but the reverse must also be true.
Another change is that the drivers only advertise IP checksum
offload functionality. There is currently no information
propagated across the device channel about the offset of the
protocol-specific checksum field. Therefore it is not safe to
defer checksum calculation for protocols the remote end may not
understand -- it will end up dropping having to drop the packet.
Yet another change is to allow netback to disable tx checksum
offload, just as we already could for netfront. Currently there is
no support for disabling rx checksum offload -- that would seem
to require some way of propagating the checksum-offload advertisement
(or lack of it) across the device channel, as it really ought to be
the transmitter that acts on it.
Thanks to Ian Jackson for pointing out some of the problems with
our checksum-offload handling. Several of the changes here are
due to his comments.
Signed-off-by: Keir Fraser <keir@xensource.com>
switch out of a transaction, in the few cases where it is inside one.
In order to behave properly, it is necessary for a driver to know its own
xenbus state (see changeset 9469:b3cb19d2b07f, for example). This
value is stored as xenbus_device.state and updated by xenbus_switch_state.
If xenbus_switch_state occurs within a transaction, then there is a possibility
that the transaction would be aborted, leaving the state field dangerously out
of sync with the value currently in the store.
This fixes recent problems seen whereby bringing multiple devices up at the
same time results in some devices not coming up (often all of the even-numbered
ones, because of the pattern of transaction conflict).
Signed-off-by: Ewan Mellor <ewan@xensource.com>
Per agreement on the summit, xen/ia64 will move to event channel
model same as xen/x86, under which event is the layer under pirq
(external interrupt), virq, and ipi with the latter three bound
to event ports. Within that model, no external interrupt will be
injected directly and evtchn_upcall_mask is the flag to control
whether events are deliverable.
So xenlinux needs to operate evtchn_upcall_mask at all places
where it originally operates vpsr.i. However these two flags are
presented at different shared area, and thus xenlinux can't ensure
atomical update on two flags which leaves severe stability issues.
One severe bug comes for this reason where some hypercall may be
restarted infinitely when events pending.
Actually based on description of future model, events become the
superset of external interrupts and thus evtchn_upcall_mask super-
set of vpsr.i (interrupt_delivery_enable). We can merge two flags
into one by removing the latter. By this way, we ensure correctness
and most importantly conform to common code which always assumes
upon evtchn_upcall_mask.
Signed-off-by Kevin Tian <kevin.tian@intel.com>
Define interface version in Kconfig and define it through CPPFLAGS for
C and assembly file compilation.
Add HYPERVISOR_{yield,block,shutdown} to remove direct use of sched_op
from guest code.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
1. No longer hold xenbus_lock while taking down VCPUs
in SMP suspend path. This allows block device hotplug
to continue working and so we will not deadlock on
paging in userspace hotplug code.
2. Track xenbus and local-admin permitted cpumasks for
VCPUs to bring online. So, if a local admin takes a
CPU down, that won't surprisingly get overridden next
time the kernel interrogates xenstore.
Signed-off-by: Keir Fraser <keir@xensource.com>
to GPL modules. Rename pte_page parameter to pmd_page (following the
Linux idiom). apply_to_pte_range() should take the pte spinlock.
Signed-off-by: Keir Fraser <keir@xensource.com>
1. Make it clear in all dual-licensed files that the
license when distributed with Linux is GPLv2.
2. Add MODULE_LICENSE(dual) in a few places.
3. Change many of our EXPORT_SYMBOL() decls to EXPORT_SYMBOL_GPL().
This is the preference of the Linux maintainers, and avoids any
trouble down the road if our EXPORT_SYMBOL'ed functions use Linux
internal interfaces that are GPL only.
Signed-off-by: Keir Fraser <keir@xensource.com>
module has a tri-state configuration so it can be a loadable module.
Views the hypervisor as hardware device, uses sysfs as a scripting
interface for user space tools that need these attributes.
Some user space apps, particularly for systems management, need to
know if their kernel is running in a virtual machine and if so in
what type of virtual machine. This property is contained in the
file /sys/hypervisor/type.
The file hypervisor_sysfs.c creates a generic hypervisor subsystem
that can be linked to by any hypervisor. The file xen_sysfs.c exports
the xen-specific attributes.
Signed-off-by: Mike D. Day <ncmike@us.ibm.com>
