with a multi-page head buffer. The new skbuff will not have a
multi-page head buffer, which is a necessary restriction until this is
supported by the internals of the netback driver.
Signed-off-by: Steven Smith <ssmith@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
This reverts the preparation for removing the sparse tree, since these
modified scripts are not yet ready:
1. XENGUEST suffix was lost yet depended on still by sparse tree
Makefiles (hacked around by 14883; also reverted).
2. XEN_TARGET_X86_PAE default value seems to be ignored
(out-of-the-box 'make dist' will build non-PAE kernels).
Signed-off-by: Keir Fraser <keir@xensource.com>
Netfront's use of nh.raw and h.raw for storing page+offset is a bit
hinky, and it breaks with upcoming network stack updates which reduce
these fields to sub-pointer sizes. Fortunately, skb offers the "cb"
field specifically for stashing this kind of info, so use it.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Chris Wright <chrisw@sous-sol.org>
Cc: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
1. rx_lock is a normal spinlock, and tx_lock is an irq spinlock. This
means that in normal use, tx_lock may be taken by an interrupt
routine while rx_lock is held. However, netif_disconnect_backend
takes them in the order tx_lock->rx_lock, which could lead to a
deadlock. Reverse them.
2. rx_lock can also be used in softirq context, so it should be
taken/released with spin_(un)lock_bh.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
drivers are not present. The feature-foo mechanism does not actually
work after save/restore since the fields do not get rewritten. And it
doesn't work when we actually need the guest to be shutdown through
Xen (e.g., on suspend). Since we can quite reasonably assume that the
only entity to screw with control/shutdown is xend, and since the one
place that does that now explicitly handles unenlightened HVM guests,
we can simply strip all this stuff out.
This has the nice benefit of making HVM save/restore with PV drivers
much more reliable.
Signed-off-by: Keir Fraser <keir@xensource.com>
xen-platform-pci fails to load w/ undefined symbols for some of the
sched_op hypercalls in machine_reboot.c. This adds them to xencomm
mini to allow them to be called from modules.
Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Time interpolator remembers last status. When domain is restored, it
should be forgotten. Otherwise the interpolator causes bad timer value.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
This effectively automates passing the nomca flag to domUs,
preventing them from setting up CMC/CPE interrupts. This saves
3 dynamic IRQs per processor and allows us to get domUs up to 64
vCPUs without increasing the number of dynamic IRQs available.
Signed-off-by: Alex Williamson <alex.williamson@hp.com>
after balloon operations in an HVM guest.
This removes the I/O port hack from the guest OS, and from ioemu.
Also we flush on reservation *increases* as well as decreases. This is
necessary until qemu-dm can demand-fault page mappings into existing
valid buckets.
Signed-off-by: Steven Hand <steven@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
Where possible,
- use hypercalls instead of writing to read-only pages
- fold TLB flushes into page table update hypercalls
- on PAE, use single-access updates instead of two-access ones
The single change to PAE pte_clear() yields a 25-30% boost for kernel
builds on a 4x2x2 CPUs, 8Gb box; the other changes together yield
improvements of 2-5%.
Also, adjust backward compatibility handling in a few more places.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
A character device for accessing (in user-space) pages that have been
granted by other domains.
Signed-off-by: Derek Murray <Derek.Murray@cl.cam.ac.uk>
Fix ioctl interface to be 32/64-bit invariant. Move xen_class
mechanism to common util.c.
Signed-off-by: Keir Fraser <keir@xensource.com>
they may not (and even if they try, can not) restore the BARs. Since
Linux does not do this for us, pciback must do it for itself manually.
Possible todo: Restore all first 64 bytes of the config space?
Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
