Consitently make such fields 64 bit values as used by dominfo.
It is not clear if ->video_memkb and/or shadow_memkb shouldn't be part of
->u.hvm but I have not changed that here.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
This serves no purpose. It relates to the old 8M to "account for backend
allocations" which we used to add. This leaves a bit of unpopulated space in
the Pseudo-physical address space which can be used by backends when mapping
foreign memory. However 8M is not representative of that any more and modern
kernels do not operate in this way anyway.
I suspect an argument could be made for removing this from the libxl API
altogether but instead lets just set the overhead to 0.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Some fields require further scaffolding before they can use the
_init/_setdefault scheme and are handled in later patches.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Some fields require further scaffolding before they can use the
_init/_setdefault scheme and are handled in later patches.
It doesn't seem reasonable for the library to pick a default for the domain
type so reject as invalid.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Don't special case types which we happen to know do not contain allocated data
such that in the future if this changes we do not need to add an API. Although
there is likely to be latent bugs in callers due to this having the API in old
libraries mean that when callers are fixed they will not need to make special
arrangements to handle old and new versions of the library.
Adds dispose functions for:
- libxl_dominfo
- libxl_cpupoolinfo
- libxl_physinfo
- libxl_sched_credit
- libxl_sched_credit2
I have attempted to fix any latent bugs in xl by inspection but have not
bothered with libxl (on the basis that internally the library is allowed to
make these sorts of assumptions and because it was looking like a very invasive
job and that more would only creep in anyway).
Several callsites use libxl_domain_info to check for the presence of a domain
and throw away the actual info. As a convenience accept a NULL info pointer and
just return the status.
Also fix a memory leak in libxl_domain_list.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Using sysctl.h is restricted to "node control tools only" and requires magic
defines. Therefore make its use internal to libxl.
Also removes an indirect include of domctl.h which has the same restrction.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
This isn't required because we also checkin the generated files. A
more complete fix might also allow the user to force regeneration but
I didn't bother here.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
24905:2b5cf6bde62d accidentally contained a change to
tools/hotplug/Linux/init.d/xencommons. That change is probably OK but
should not be in 24905 so remove it.
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
With the changeset 24529:87854d3bed93 we got 2 initialisation
functions for the arch specific stuff. This is no more needed.
Signed-off-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
Committed-by: Keir Fraser <keir@xen.org>
- struct msr_load_store_entry is unused
- struct pmumsr is only used in the vmx part
Signed-off-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
Committed-by: Keir Fraser <keir@xen.org>
This hypercall allows a domain to identify the security context of a
domain that it is communicating with using the interdomain event
channel that it is using for the communication. This can be used to
augment Xen's security permissions with intra-domain security checks.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Committed-by: Keir Fraser <keir@xen.org>
When multiple CPUs hit an AVC audit message, the resulting output in
the ring buffer and serial console is garbled due to the audit process
using many separate printk invocations for each message. Change the
AVC audit process to use a temporary buffer and output the contents
once the entire audit message is complete.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Committed-by: Keir Fraser <keir@xen.org>
Don't attempt to relabel the already-bound half of the event channel
pair created by an interdomain event channel. This relabeling also
performed an incorrect check that the destination domain is permitted
to create the reverse event channel, which may not be true if the
unbound channel was created by the domain builder (like the xenstore
channel).
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Committed-by: Keir Fraser <keir@xen.org>
Event channels created during the domain build process for HVM domains
did not call the XSM creation hook. Since these channels are used
internally by Xen, redirect them to DOMAIN_INVAID instead of failing
to create if the XSM hook fails the permissions check.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Committed-by: Keir Fraser <keir@xen.org>
Since the introduction of autoconf, builds with XSM enabled in .config
have been broken unless FLASK_ENABLE was explicitly set. Since the
setting in .config has apparently been deprecated in favor of an
autoconf --enable-xsm, add config/Xen.mk to export this to Xen. This
also makes --disable-debug and some paths to be pulled from the
configure process in the hypervisor build.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Committed-by: Keir Fraser <keir@xen.org>
- Remove the 15cpus hack from ovmf because it should be unnecessary on
nowadays windows/EFI supported.
Signed-off-by: Attilio Rao <attilio.rao@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Keir Fraser <keir@xen.org>
The symbols in it are wrong anyway and will only confuse people
who ought to be looking at xen-syms.
Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>
If memcpy() and memmove() were used with source and destination of
different alignment then the result would be all jumbled.
When their implementations were imported from Linux some macros for
big-endian platforms were taken instead of the correct little-endian
ones (specifically, the push and pull macros in assembler.h).
Fix this by taking Linux's arch/include/asm/assembler.h as-is and
making only the minimum changes necessary for Xen.
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <Ian.Campbell@citrix.com>
At least with xend, where there's not even a tool stack side attempt to
de-assign devices during domain shutdown, this allows immediate re-
starts of a domain to work reliably. (There's no apparent reason why
c/s 18010:c1577f094ae4 chose to put this in the asynchronous part of
domain destruction).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
This allows migration to a host with less MCA banks than the source
host had, while without this patch accesses to the excess banks' MSRs
caused #GP-s in the guest after migration (and it depended on the guest
kernel whether this would be fatal).
A fundamental question is whether we should also save/restore MCG_CTL
and MCi_CTL, as the HVM save record would better be defined to the
complete state that needs saving from the beginning (I'm unsure whether
the save/restore logic allows for future extension of an existing
record).
Of course, this change is expected to make migration from new to older
Xen impossible (again I'm unsure what the save/restore logic does with
records it doesn't even know about).
The (trivial) tools side change may seem unrelated, but the code should
have been that way from the beginning to allow the hypervisor to look
at currently unused ext_vcpucontext fields without risking to read
garbage when those fields get a meaning assigned in the future. This
isn't being enforced here - should it be? (Obviously, for backwards
compatibility, the hypervisor must assume these fields to be clear only
when the extended context's size exceeds the old original one.)
A future addition to this change might be to allow configuration of the
number of banks and other MCA capabilities for a guest before it starts
(i.e. to not inherits the values seen on the first host it runs on).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
... or even know of. Apart from CMCI, which was masked off already,
this now also suppresses the advertising of extended state registers
(reading of which would likely be meaningless in a guest and represent
an information leak).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
Use C99 initializers for mem_event_request_t users to make sure req is
always cleared, even with local debug patches that shuffle code around
to have a single exit point.
The common case is to use and send req, so it does not add significant
overhead to always clear req.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>
When mem_event was added the type flag should indicate who the consumer
is. But the concept of a single ring buffer for multiple event types can
not work for two reasons. One is that no multiplexer exists which
provides individual event types to the final consumer, and second is
that even if such multiplexer can not work reliable because a request
needs to be answered with a response. The response should be sent
roughly in the order of received events. But with multiple consumers one
of them can so stall all the others.
For that reason the single mem_event buffer for all types of events was
split into individual ring buffers with commit 23842:483c5f8319ad. This
commit made the type member already obsolete because the meaning of each
buffer is now obvious.
This change removes the type member and increases the flags field.
Even though this is an ABI incompatible change, it will have no practical
impact on existing binaries because the changeset referenced above already
bumped the SONAME. So these binaries have to be recompiled anyway for the
upcoming major release.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>
If _PAGE_PRESENT is clean in a pagetable entry, any pattern of bits
is valid in the rest of the entry. OSes that special-case
PFEC_invalid_bits (since it should never happen) will be confused
by our setting it in this way.
Signed-off-by: Tim Deegan <tim@xen.org>
In preparation for introducing a schedule parameter-based structure,
rename libxl_sched_{credit,credit2,sedf} to libxl_sched_{}_domain.
No functional changes.
v2: Wrap long lines while I'm at it
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Keir Fraser <keir@xen.org>
...when specified in the guest configuration file.
This work is somewhat based on Bei Guan effort during the SoC 2011 and
relies on upstream edk2/ovmf Tianocore ROM to be built separately and
manually copied as:
Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ->
tools/firmware/ovmf/ovmf-x64.bin
Build/OvmfIa32/DEBUG_GCC44/FV/OVMF.fd ->
toolf/firmware/ovmf/ovmf-ia32.bin
A way to integrate OVMF build directly into XEN has still be discussed
on the mailing list appropriately.
Signed-off-by: Attilio Rao <attilio.rao@citrix.com>
Disable CONFIG_OVMF by default as ovmf is not integrated into the
build.
Signed-off-by: Keir Fraser <keir@xen.org>
Committed-by: Keir Fraser <keir@xen.org>
Note: As of __XEN_INTERFACE_VERSION__ 0x00040201 the definition of
BLKIF_MAX_SEGMENTS_PER_REQUEST has changed. Drivers must be
updated to, at minimum, use BLKIF_MAX_SEGMENTS_PER_HEADER_BLOCK,
before being recompiled with a __XEN_INTERFACE_VERSION greater
than or equal to this value.
This extension first appeared in the FreeBSD Operating System.
Signed-off-by: Justin T. Gibbs <justing@spectralogic.com>
Committed-by: Keir Fraser <keir@xen.org>
Ian Campbell
Attilio Rao
Keir Fraser
Ian Jackson
Roger Pau Monne
Olaf Hering
Dietmar Hahn
Daniel De Graaf
Tim Deegan
Liu, Jinsong
David Vrabel
Jan Beulich
George Dunlap
Justin T. Gibbs