Re-indent xc_linux_restore, and add code to force PAE
root pgdirs below 4GB on restore.

Signed-off-by: Ian Pratt <ian@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

merge?

Always fire watches, even on the connection which caused the watch to fire.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Xen_version hypercalls takes two args, not one.

Signed-off-by: Ian Pratt <ian@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

mlock argument to version hypercall in libxc.

Signed-off-by: Ian Pratt <ian@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

Improve debug tracing from HYPERVISOR_memory_op().

Signed-off-by: Ian Pratt <ian@xensource.com>

removing XEN_LOCALVERSION for a smaller alternative

Signed-off-by: Vincent Hanquez <vincent@xensource.com>

Re-indent vmx code.

Signed-off-by: Keir Fraser <keir@xensource.com>

Yes, that fixes the problem, our patches crossed. I saw Keir already
integrated my patch and while it is harmless, it is redundant and
here is the diff to take it out.

The patch also addresses some indent problems and aligns instructions
up with the rest of the block.

Signed-Off-By: Leendert van Doorn <leendert@watson.ibm.com>

add XEN_LOCALVERSION to the dist building process

Signed-off-by: Vincent Hanquez <vincent@xensource.com>

An FC4/i386 install inside VMX on an x86_64 system fails because byte
size is not handled by __set_reg_value. This patch adds that support.

This patch also reindents Chengyuan Li cmpb patch so that is aligns
with the rest of the code in that block.

Signed-Off-By: Leendert van Doorn <leendert@watson.ibm.com>

Restore NULL checks before calling kfree().
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

add a simple usage string to xenconsoled

Signed-off-by: Vincent Hanquez <vincent@xensource.com>

fix ballooning out logic to handle dying domain freeing memory.

Signed-off-by: Vincent Hanquez <vincent@xensource.com>

use XEN_FULLVERSION variable where possible

Signed-off-by: Vincent Hanquez <vincent@xensource.com>

remove pointless NULL check before calling kfree

Signed-off-by: Vincent Hanquez <vincent@xensource.com>

convert initializers to C99 initializers

Signed-off-by: Vincent Hanquez <vincent@xensource.com>

With this patch, 32-bit binary can work on 64-bit VMX guest.

Signed-off-by: Chengyuan Li <chengyuan.li@intel.com>
Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>
Signed-off-by: Asit Mallick <asit.k.mallick@intel.com>

This patch fixes xenlinux timer interrupt.

In xenlinux timer interrupt variable cpu is sometimes uesd in a for
loop, but later it is used to access per cpu data i.e.
per_cpu(processed_system_time, cpu), which causes an invalid pointer.

Signed-off-by: Xin Li <xin.b.li@intel.com>

This patch fix a bug in set_reg_value() for x86_64 VMX guest.

Signed-off-by: Chengyuan Li <chengyuan.li@intel.com>

One more instruction for the VMX MMIO decoder.
This patch is to handle the instruction with opcode 0x80.
Without this patch, the 64-bit VMX guest can't boot.

Signed-off-by: Chengyuan Li <chengyuan.li@intel.com>

Fix bug that service os & vmx guest can't communicate with
each other. The bug was physical packets smaller than
minimal packet size of 60 bytes were gettign thrown away.
Hence ARP traffic for example was dropped.

Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Signed-off-by: Xiaofeng Ling <xiaofeng.ling@intel.com>

Fix xm info handling of empty strings (fix bugzilla bug #216)

Signed-off-by: Steven Hand <steven@xensource.com>

Fix 'xm info' (sizeof(param) doesn't do that one might hope).

Signed-off-by: Steven Hand <steven@xensource.com>

merge?

Make vif cleanup use xstransact and use classmethods for setup.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Add remove and list support.  Also make all class methods "safe".
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Only retry transactions which fail from timeout.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

common/kernel.o contains changeset/compiler info.
Blow away on every build.

cleanups

Setup network devices in xenstore using xstransact.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Remove dependency between domain db and path/uuid.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Always start transactions on an existing path.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

The patch allows the shadow mode code to support >4GB physical memory. I
tested VMX domains on an x86_64 machine with >4GB; it fails without this
change. This should be applicable to PAE as well.

This patch is good for up to 32GB RAM.

Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>

Add more version and capability reporting to Xen. Print the results with 'xm info'.
Signed-off-by: ian@xensource.com

pagefault handler fixes.

Add simple transactional read/write python xenstore interface.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Make error reporting consistent.
Also note addition of Py_BEGIN_ALLOW_THREADS/Py_END_ALLOW_THREADS
which slipped in in previous merge changeset:  allow other threads
to run during communication with xenstored.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

merge?

Xenbus implementation ported from Linux to Mini-os, simple thread support introduced
to simplify the porting. 64 bit version of Mini-os now compiles, but does not work
because of the pagetables and some bits of scheduler not being written.

Signed-off-by: Grzegorz Milos <gm281@cam.ac.uk>

Fix blkfront request completion.

It was suggested on the xen-users list that it would be useful if the
loopback driver could instantiate an arbitrary number of interfaces, so
the attached patch does that.

[Also retab at the same time]

merge?

tpmback compilation fix.
A small patch to fix some compilation problems and debugging output.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Bug fix on x86_64 increase_reservation
Signed-off-by: ian@xensource.com

Test CD needs devfs
Signed-off-by: ian@xensource.com

No need to qsort the pfn array in domu builder.

merge

manual merge

Build fix.

Improved "PGDs must be under 4GB" handling for PAE.
Signed-off-by: ian@xensource.com

Fixes to sub 4GB allocator
Signed-off-by: ian@xensource.com

grant_table.c reindent.
Signed-off-by: Keir Fraser <keir@xensource.com>

Rename get_order() to get_order_from_bytes() and add
new function get_order_from_pages(). Fix
HYPERVISOR_memory_op(), properly this time.

Signed-off-by: Keir Fraser <keir@xensource.com>

Cleanups, and fix allocating DMA memory via
HYPERVISOR_memory_op().
Signed-off-by: Keir Fraser <keir@xensource.com>

Give each domain some memory below 4GB. This solves the "PGD's must be below 4GB" for the initial page tables. I'm not sure we'll stick with this approach, but this is good enough for the time being.

PAE should be a *lot* more robust on systems that actually have more than 4GB thanks to all the various patches that went in today. I find it astounding that it ever appeared to work at all!

Signed-off-by: ian@xensource.com

Yet more PAE fixes, this time in the paravirt drivers.
Signed-off-by: ian@xensource.com

More tools PAE fixes to avoid loosing high-order pte bits.
Signed-off-by: ian@xensource.com

Replace direct_remap_area_pages with direct_remap_pfn_range to help fix PAE domain building.
Signed-off-by: ian@xensource.com

Fix typo.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

merge?

Move console tty/limit information into console directory in domain dir.
Also only store port number of console event channel.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Disable device info getting written to store when xend restarts.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Disable sync dbmap saves -- they remove nodes added by other tools.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Create /dev/xen/evtchn if it doesn't exist.

Signed-off-by: Steven Hand <steven@xensource.com>

Don't leak memory when realloc fails.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

merge?

g/c unused xcs.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

g/c unused xu extension.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

g/c unused control message code.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Remove control interface in kernels and domain builder.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

merge

Trivial fix for a nasty PAE bug.
Signed-off-by: ian@xensource.com

Fix vif scripts for domains with more than one ip.
As reported in bug 218 by christian@prehl.de.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Remove debug printk.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Always bind dom0 virqs to vcpu 0.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Make xenstored bind to domain exception virq directly, instead of via xcs.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Remove python virq code.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Trigger domain cleanup on @releaseDomain watch-event instead of domain exc. virq.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Fix and cleanup error handling.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Add copyright notice and g/c some unused code.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

merge?

Add python bindings for watches.
Uses a watcher thread with per-watch registered callbacks.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Be more sensitive to losing sync with platform timer.
Signed-off-by: Keir Fraser <keir@xensource.com>

merge?

Small plan9 loader patch from Tim Newsham.

Extend save/restore interface to cope with domains >4GB by adding a level of indirection.
Signed-off-by: ian@xensource.com

Fix dependencies.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

merge?

Use @releaseDomain watch-events.
This replaces listening to the domain exception virq through xcs.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Tiny printk info cleanup.

The attached patch fixes two problems I ran into with the swiotlb code
in unstable (changeset aeaa3c83f6e5). Tested on a dual Opteron x86-64
machine with 4GB of memory and a tg3 modified to only DMA below 2GB.

- swiotlb_dma_supported() checked that the device DMA mask was equal
or bigger than 4GB, when in fact all that's needed is that the device
be able to DMA into the swiotlb aperture (1GB on my AMD x86-64
machine). Some device are actually only 31-bit DMA capable, so this
would've tripped them.

- On some platforms, PCI unmaps and syncs are nops, so there's no need to
keep track of the dma_addr they need after the initial mapping. The
DMA API supports this via the DECLARE_PCI_UNMAP_ADDR macros (see
Documentation/DMA-mapping.txt). Since the swiotlb code does make us of
the dma_addr for swiotlb_unmap_xxx and (more importantly)
swiotlb_dma_sync_xxx, we need to define them to something
meaningful.

Signed-Off-By: Muli Ben-Yehuda <mulix@mulix.org>

Exit if xen_setup fails.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Add @releaseDomain watch-event.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Destroy domain before cleaning it up.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

phys_to_mach and mach_to_phys tables contain long entries, not
32-bit entries.

Signed-off-by: Keir Fraser <keir@xensource.com>

Add xcs to include path for xcs_proto.h.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

Disable xcs stop from xend script since it kills xenstored now.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

merge?

Make xenstored listen to domain exception virqs.
The virq triggers a scan for domains which have gone away, and then
removes those domains.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>