Said hypercall for large BARs can take quite a while. As such
we can require that the hypercall MUST break up the request
in smaller values.
Another approach is to add preemption to it - whether we do the
preemption using hypercall_create_continuation or returning
EAGAIN to userspace (and have it re-invocate the call) - either
way the issue we cannot easily solve is that in 'map_mmio_regions'
if we encounter an error we MUST call 'unmap_mmio_regions' for the
whole BAR region.
Since the preemption would re-use input fields such as nr_mfns,
first_gfn, first_mfn - we would lose the original values -
and only undo what was done in the current round (i.e. ignoring
anything that was done prior to earlier preemptions).
Unless we re-used the return value as 'EAGAIN|nr_mfns_done<<10' but
that puts a limit (since the return value is a long) on the amount
of nr_mfns that can provided.
This patch sidesteps this problem by:
- Setting an hard limit of nr_mfns having to be 64 or less.
- Toolstack adjusts correspondingly to the nr_mfn limit.
- If the there is an error when adding the toolstack will call the
remove operation to remove the whole region.
The need to break this hypercall down is for large BARs can take
more than the guest (initial domain usually) time-slice. This has
the negative result in that the guest is locked out for a long
duration and is unable to act on any pending events.
We also augment the code to return zero if nr_mfns instead
of trying to the hypercall.
This is XSA-125 / CVE-2015-2752.
Suggested-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
On a Thinkpad T4440p with OpenSUSE tumbleweed with v3.15-rc4
and today's latest xen tip from the git tree strace -f reveals
we end up on a never ending wait shortly after
write(20, "backend/console/5\0", 18 <unfinished ...>
This is right before we just wait on the qemu process which we
had mmap'd for. Without this you'll end up getting stuck on a
loop if mmap() worked but madvise() did not. While at it I noticed
even the mmap() error fail was not being checked, fix that too.
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
(cherry picked from commit e86539a388)
Due to the layout of the guest physical address space we cannot support more
than 768M of RAM before overrunning the area set aside for the grant table. Due
to the presence of the magic pages at the end of the RAM region guests are
actually limited to 767M.
Catch this case during domain build and fail gracefully instead of obscurely
later on.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
(cherry picked from commit 5a959f44ed)
314c9815e2 "tools: implement initial ramdisk support for ARM." broke starting
guests with <= 128 MB ram by placing the boot modules (dtb and initrd)
immediately after the kernel in this case, running the risk of them being
overwritten. Instead place the modules at the end of RAM, as the hypervisor
does for dom0.
The hypervisor also falls back to placing things before the kernel as a last
resort before failing, so add that here too.
Tested with the Debian installer initrd and guests of 96MB, 128MB, 256MB and
1GB. All work, also tested with 64MB but the installer doesn't run with so
little RAM (but our placement of the initrd is correct).
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
(cherry picked from commit 6f4ff742a5)
The ramdisk is passed to the kernel as a property in the chosen node of the
device tree. This is somewhat tricky since in order to place the ramdisk and
dtb in ram we first need to know the size of the dtb. So we initially create a
DTB with placeholders for the ramdisk and finalise the value (which doesn't
change the size) once we know where everything is.
Rename libxl__arch_domain_configure to xl__arch_domain_init_hw_description to
better reflect its use and to be consistent with the new
libxl__arch_domain_finalise_hw_description.
The common xc_dom_build_image() function did not support explicit placement of
the ramdisk, instead passing 0 to xc_dom_alloc_segment, meaning "pick
somewhere". This change instead passes ramdisk_seg.vstart. If nothing has set
vstart then it will be zero because the entire dom struct is zeroed on
allocation in xc_dom_allocate(). Therefore there is no change to the behaviour
on x86. This is also consistent with how other segments (kernel, dtb) are
handled.
Furthermore if the ramdisk has been explicitly placed then xc_dom_build_image()
assumes that it is not to be decompressed (since that would muck up the sizings
used on placement).
With all that I'm able to boot a domain using the current Debian Jessie armhf
installer initrd and have it complete successfully.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
[ ijc -- s/itherwise/otherwise and dropped bogus emacs magic change ]
(cherry picked from commit 314c9815e2)
On error, PERROR calls the already destroyed logger, which can segfault.
Re-order the calls, so the logger is still available.
Signed-off-by: Jason Andryuk <andryuk@aero.org>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
(cherry picked from commit 86216963fd)
The code to check for an appended DTB was confusing and unnecessary. Since we
know the size of the kernel binary passed to us we should just load the entire
thing into guest RAM (subject to the limits checks). Removing this code avoids
a whole raft of overflow and alignment issues.
We also need to validate the limits of the segment where we intend to load the
kernel to avoid overflow issues.
For ARM32 we control the load address, but we need to validate the size. The
entry point is only relevant within the guest so we don't need to worry about
that.
For ARM64 we need to validate both the load address (which is the same as the
entry point) and the size.
This is XSA-95.
Reported-by: Thomas Leonard <talex5@gmail.com>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Avoid freeing info then returning it to the caller.
This is XSA-88.
Coverity-ID: 1056192
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Guests are initially started with caches disabled and so we need to make sure
they see consistent data in RAM (requiring a cache clean) but also that they
do not have old stale data suddenly appear in the caches when they enable
their caches (requiring the invalidate).
This can be split into two halves. First we must flush each page as it is
allocated to the guest. It is not sufficient to do the flush at scrub time
since this will miss pages which are ballooned out by the guest (where the
guest must scrub if it cares about not leaking the pagecontent). We need to
clean as well as invalidate to make sure that any scrubbing which has occured
gets committed to real RAM. To achieve this add a new cacheflush_page function,
which is a stub on x86.
Secondly we need to flush anything which the domain builder touches, which we
do via a new domctl.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Cc: keir@xen.org
There have been ABI/API changes in libxc. Bump its MAJOR (which
affets libxenguest et al too.)
There have been ABI changes in libxl. Bump its MAJOR.
(The API changes have been dealt with as we go along - there is
already a LIBXL_API_VERSION 0x040400.)
None of the other libraries have changed their interfaces. I have
verified this by building the tools and searching the dist/install
tree for files matching *.so.*. For each library that showed up, I
did this:
git-diff RELEASE-4.3.0..staging -- `find tools/FOO/ -name \*.h`
where FOO is the corresponding source directory.
Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Acked-by: Ian Campbell <Ian.Campbell@citrix.com>
The variable 'rc' is set to 1 at the top of xc_domain_restore, and for the
most part is left alone until success, at which point it is set to 0.
There is a separate 'frc' which for the most part is used to check function
calls, keeping errors separate from 'rc'.
For a toolstack which sets callbacks->toolstack_restore(), and the function
returns 0, any subsequent error will end up with code flow going to "out;",
resulting in the migration being declared a success.
For consistency, update the callsites of xc_dom_gnttab{,_hvm}_seed() to use
'frc', even though their use of 'rc' is currently safe.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: George Dunlap <george.dunlap@eu.citrix.com>
ARM does not implemented PoD and so returns ENOSYS from XENMEM_set_pod_target.
The correct solution here would be to check for ENOSYS in libxl, unfortunately
xc_domain_set_pod_target suffers from the same broken error reporting as the
rest of libxc and throws away the errno.
So for now conditionally define xc_domain_set_pod_target to return success
(which is what PoD does if nothing needs doing). xc_domain_get_pod_target sets
errno==-1 and returns -1, which matches the broken error reporting of the
existing function. It appears to have no in tree callers in any case.
The conditional should be removed once libxc has been fixed.
This makes ballooning (xl mem-set) work for ARM domains.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Cc: george.dunlap@citrix.com
The original code is wrong because:
* claim mode wants to know the total number of pages needed while
original code provides the additional number of pages needed.
* if pod is enabled memory will already be allocated by the time we try
to claim memory.
So the fix would be:
* move claim mode before actual memory allocation.
* pass the right number of pages to hypervisor.
The "right number of pages" should be number of pages of target memory
minus VGA_HOLE_SIZE, regardless of whether PoD is enabled.
This fixes bug #32.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Tested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Both of these "goto finish;" statements are actually errors, and need to "goto
out;" instead, which will correctly destroy the domain and return an error,
rather than trying to finish the migration (and in at least one scenario,
return success).
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: George Dunlap <george.dunlap@eu.citrix.com>
"ret", being set to -1 early on, gets cleared by the first invocation
of lz4_decompress()/lz4_decompress_unknownoutputsize(), and hence
subsequent failures wouldn't be noticed by the caller without setting
it back to -1 right after those calls.
Linux commit: 2a1d689c9ba42a6066540fb221b6ecbd6298b728
Reported-by: Matthew Daley <mattjd@gmail.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This reverts commit a0035ecc0d.
Even with this fix there is a period between the flush and the unmap where
processor may speculate data into the cache. The solution is to map this
region uncached or to use the HCR.DC bit to mark all guest accesses cached.
89eb02c220 "xen: arm: force guest memory accesses to cacheable when MMU is
disabled" has arranged to do the latter.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
The errors have been incorrectly identifying their function since c/s
861aef6e15 which did a lot of error handling
cleanup.
Use __func__ to ensure the name remains correct in the future.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
Provide semantic documentation for how the libxc calls relate to the
hypervisor interface, and how they are to be used.
Also document the bug (present at least in Linux 3.12) that setting
the evtchn fd to nonblocking doesn't in fact make xc_evtchn_pending
nonblocking, and describe the appropriate workaround.
Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
CC: Jan Beulich <JBeulich@suse.com>
CC: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
c/s 2e82c18cd8 changed the error returns of
xc_get_{cpu,node}map_size() to now include returning -1. This invalidated the
error checks from callers, which expected 0 to be the only error case.
Coverity ID: 1135907 1135908 1135909 1135910 1135911 1135912 1135913 1135914
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: George Dunlap <george.dunlap@eu.citrix.com>
On ARM guest OSes are started with MMU and Caches disables (as they are on
native) however caching is enabled in the domain running the builder and
therefore we must flush the cache as we load the blobs, otherwise when the
guest starts running it may not see them. The dom0 build in the hypervisor has
the same requirements and already does the right thing.
The mechanism for performing a cache flush from userspace is OS specific, so
implement this as a new osdep hook:
- On 32-bit ARM Linux provides a system call to flush the cache.
- On 64-bit ARM Linux the processor is configured to allow cache flushes
directly from userspace.
- Non-Linux platforms will need to provide their own implementation. If
similar mechanisms are not available then a new privcmd ioctl should be a
suitable alternative.
No cache maintenance is required on x86, so provide a stub for all non-Linux
platforms which returns success on x86 only and log an error otherwise.
This fixes guest building on Xgene which has a very large L3 cache and so is
particularly susceptible to this problem. It has also been observed
sporadically on midway.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Andre Przywara <andre.przywara@calxeda.com>
Cc: Pranavkumar Sawargaonkar <psawargaonkar@apm.com>
Cc: Anup Patel <apatel@apm.com>
In libxc, make xc_get_max_{cpus,node}() always return either a
positive number or -1, and change all the callers to deal with
that.
In libxl, make libxl_get_max_{cpus,nodes}() always return either a
positive number or a libxl error code. Thanks to that, it is also
possible to fix loggig for libxl_{cpu,node}_bitmap_alloc(), which
now happens inside the functions themselves, more accurately
reporting what happened.
Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
xc_interface_close() crashes if it gets NULL as an argument. However,
it just calls xc_interface_close_common() which is called by many
others functions. It means that they are also vulnerable. So fix above
mentioned issue by adding NULL check in xc_interface_close_common().
This way we fix similar issue in other functions which calls
xc_interface_close_common() too.
Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The ptr calculation shall take the offset into the page into account
when ptr is valid.
Reported regression on NetBSD's port-xen with last known working libxen
being rev 2.9. This corrupts the kernel symbol table when the table is
not loaded on a page boundary.
Issue was tracked down by FastIce and Jeff Rizzo. See also
http://mail-index.netbsd.org/port-xen/2013/10/16/msg008088.html
Signed-off-by: Jean-Yves Migeon <jym@NetBSD.org>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Coverity ID 1055563
In the original function, mmap() could be called with a length of -1 if the
second lseek failed and the caller had not provided max_size.
While fixing up this error, improve the logging of other error paths. I know
from personal experience that debugging failures function is rather difficult
given only "xc_dom_malloc_filemap: failed (on file <somefile>)" in the logs.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Ian Campbell <Ian.Campbell@citrix.com>
Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Making room for a 64 bit implementation.
Also fix a typo and stop refering to it as a bzImage, which is an x86-ism.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
The tools and the hypervisor need to agree on various aspects of the guest
environment, such as interrupt numbers, memory layout, initial register values
for registers which are implementation defined etc. Therefore move the
associated defines into the public interface headers, or create them as
necessary.
This just exposes the current de-facto standard guest layout, which may be
subject to change in the future. This deliberately does not make the guest
layout dynamic since there is currently no need.
These values should not be exposed to guests, they should find these things
out via device tree or should not be relying on implementation defined
defaults.
Various bits of the hypervisor needed to change to configure dom0 with the real
platform values while using the virtual platform configuration for guests.
Arrange for this where appropriate and plumb through as needed.
We also need to expose some 64-bit values (e.g. PSR_GUEST64_INIT) for the
benefit of 32 bit toolstacks building 64 bit guests.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
In order to set up the grant table for HVM guests, libxc needs to map
the grant table temporarily. At the moment, it does this by adding the
grant page to the HVM guest's p2m table in the MMIO hole (at gfn 0xFFFFE),
then mapping that gfn, setting up the table, then unmapping the gfn and
removing it from the p2m table.
This breaks with PVH guests with 4G or more of ram, because there is
no MMIO hole; so it ends up clobbering a valid RAM p2m entry, then
leaving a "hole" when it removes the grant map from the p2m table.
Since the guest thinks this is normal ram, when it maps it and tries
to access the page, it crashes.
This patch maps the page at max_gfn+1 instead.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Tim Deegan <tim@xen.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Keir Fraser <keir@xen.org>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This error case can only be triggered by gzread returning 0 (and having
not read anything), so move it there.
Coverity-ID: 1056076
Signed-off-by: Matthew Daley <mattjd@gmail.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
We don't check for null pointers in similar functions, and it's silly to
do so for these arguments here anyway; they should fail noisily if null
is passed.
While at it, remove pointless brackets.
Coverity-ID: 1055938
Coverity-ID: 1055939
Signed-off-by: Matthew Daley <mattjd@gmail.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Andrew Cooper
Ian Jackson
Konrad Rzeszutek Wilk
Luis R. Rodriguez
Ian Campbell
Jason Andryuk
Wei Liu
Jan Beulich
Dario Faggioli
Daniel Kiper
Nathan Studer
Jean-Yves Migeon
Mukesh Rathor
Roger Pau Monné
David Vrabel
Matthew Daley