As was already pointed out months ago (see
http://lists.xen.org/archives/html/xen-devel/2012-02/msg00826.html),
gcc 4.7.x (imo validly) refuses to take both -mcpu=cortex-a15 and
-march=armv7-a due to conflicting feature sets causing amibiguity in
instruction selection. Since the former implies the latter, just use
the former (and drop the -march=).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Fixes crashes with Windows guests when shadow-on-nested is used.
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
Acked-by: Tim Deegan <tim@xen.org>
2205914617cb does its job in correcting the "wrong domain being
considered" issue introduced by 9d1fd58ff602. Unfortunately, when
dealing (again!) with the sedf scheduler, it is required for the
vCPUs of a domain to have been allocated and setup already (in
the hypervisor), when the first call to libxl_domain_sched_params_get()
happens, and that is not true.
This fixes that by avoiding calling that function at all, as we
only need to know which scheduler the domain is running under,
and that is provided by libxl__domain_scheduler() which is safe
to be called there.
While at it, also improve a bit the comments about the whole
sedf parameter validation and mangling process.
Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>
When attempting to use AMD's extension to access the extended PCI config
space, only the low byte of the register number was being passed to XSM.
Include the correct value of the register if this feature is enabled;
otherwise, bits 24-30 of port cf8 are reserved, so disallow the invalid
access.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Don't fail the permission check except when the MSR can't be read.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
Committed-by: Jan Beulich <jbeulich@suse.com>
Recent Dom0 kernels want to disable PCI MSI on all devices, yet doing
so on AMD IOMMUs (which get represented by a PCI device) disables part
of the functionality set up by the hypervisor.
Add a mechanism to mark certain PCI devices as having write protected
config spaces (both through port based [method 1] accesses and, for
x86-64, mmconfig), and use that for AMD's IOMMUs.
Note that due to ptwr_do_page_fault() being run first, there'll be a
MEM_LOG() issued for each such mmconfig based write attempt. If that's
undesirable, the order of the calls in fixup_page_fault() would need
to be swapped.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Tested-by: Wei Wang <wei.wang2@amd.com>
Acked-by: Keir Fraser <keir@xen.org>
This was previously done by xl itself however the domain was not
created at that point so there was no domid to check. This happened to
work on first boot because xl's global domid was initialised to zero
so we would (incorrectly) validate the new domain to be against
domain0. On reboot though we would try to use the old domain's id and
fail.
sched_params_valid is moved and gains a gc+domid parameters and
s/ctx/CTX/. The call is placed after
libxl__domain_build_info_setdefault in the create path, because
set_defaults doesn't have access to the domid and there are other
callers which don't even have a domid to give it.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>
These additions did not fulfill their purpose - they checked hypervisor
config space accesses instead of guest (Dom0) ones.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
For a multi-byte aligned read, this so far resulted in 0x00ff to be
put in the guest's register rather than 0xffff or 0xffffffff, which in
turn could confuse bus scanning functions (which, when reading vendor
and/or device IDs, expect to get back all zeroes or all ones).
As the value gets masked to the read width when merging back into the
full result, setting the initial value to all ones should not harm any
or the other cases.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
25468:9d1fd58ff602 was bogus in not letting a new domain being created
if its scheduling parameters --when running under the sedf scheduler--
were not fully specified, making creation fail like in this example
here below:
2012-06-16 07:37:47 Z executing ssh ... root@10.80.248.105 xl create /etc/xen/debian.guest.osstest.cfg
libxl: error: libxl.c:3619:sched_sedf_domain_set: setting domain sched sedf: Invalid argument
libxl: error: libxl_create.c:710:domcreate_bootloader_done: cannot (re-)build domain: -3
Parsing config from /etc/xen/debian.guest.osstest.cfg
This is due to the fact the values for period, slice, weight and
extratime should be consistent among each others, and if not all
are explicitly specified, someone has to make that happen. That
was right the purpose of the change in question, but it was failing
at achieving so.
This commit fixes things by forcing unspecified parameters to
sensible values, depending on the ones the user provided.
Signed-off-by: Dario Faggioli <dario.faggioli@citix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
While putting together the workaround announced in
http://lists.xen.org/archives/html/xen-devel/2012-06/msg00709.html, I
found that mod_l1_entry(), upon encountering a set bit in
mmio_ro_ranges, would return 1 instead of 0 (the removal of the write
permission is supposed to be entirely transparent to the caller, even
more so to the calling guest).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
Rather than deferring the detection of these to the point where they
get actually used (the fix for XSA-7, 25480:76eaf5966c05, causing a #GP
to be raised by IRET, which invokes the guest's [fragile] fail-safe
callback), don't even allow such to be set.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
When command to modify turbo mode (CPB on AMD processors) comes
in the actual change happens later, when P-state transition is
requested. There is no time limit on when this transition will
occur and therefore change in CPB state may take long time from
the moment when command to toggle it is issued.
This patch makes CPB mode change happen immediately when request
is made.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@amd.com>
Committed-by: Keir Fraser <keir@xen.org>
Do not clear asid cleanbit unconditionally. This shaves off 100 cycles
from the VMRUN instruction.
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
Committed-by: Jan Beulich <jbeulich@suse.com>
Processors with this erratum are subject to a DoS attack by unprivileged
guest users.
This is XSA-9 / CVE-2012-2934.
Signed-off-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
Check for non-canonical guest RIP before attempting to execute sysret.
If sysret is executed with a non-canonical value in RCX, Intel CPUs
take the fault in ring0, but we will necessarily already have switched
to the the user's stack pointer.
This is a security vulnerability, XSA-7 / CVE-2012-0217.
Signed-off-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: Ian Campbell <Ian.Campbell@citrix.com>
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>
Tested-by: Ian Campbell <Ian.Campbell@citrix.com>
Acked-by: Keir Fraser <keir.xen@gmail.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
We recently saw a machine that has the EBDA extending as low as
0x7c000, so that Xen fails to boot after relocating the trampoline.
To fix this, I removed BOOT_TRAMPOLINE and bootsym_phys completely.
Here are the parts:
1) the trampoline segment is set to 64k below the EBDA. head.S grows
the ability to relocate the trampoline segment
2) reloc.c is made position-independent. It allocates data below the
trampoline, whose address is passed in _eax.
3) cmdline.S is called before relocating, so all bootsym_phys there
become sym_phys.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
jb: - fall back to low memory size (instead of segment 0x7c00) if EBDA
value is out of range
- also add upper limit check on EBDA value
- fix and simplify inline assembly operands in reloc_mbi_struct()
- use lret instead of retf
- renamed early_stack to wakeup_stack, defined and used now only
in wakeup.S
- aligned reloc.bin's end of .text to 16 bytes, so that checking
__bss_start == end works reliably
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Committed-by: Keir Fraser <keir@xen.org>
Additionally, correct the text description to reflect what is being
done, and make use of fatal_trap() in preference to kexec_crash() in
case an unknown NMI occurs before a kdump kernel has been loaded.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Committed-by: Keir Fraser <keir@xen.org>
gdbsx got broken along the way. During domain pause, don't send
VIRQ_DEBUGGER to guest if gdbsx is active on that guest.
Signed-off-by: Mukesh Rathor <mukesh.rathor@oracle.com>
Committed-by: Keir Fraser <keir@xen.org>
NetBSD doesn't have a gntdev, so libvchan is unable to build due to
the lack of the header files gntalloc.h.
This is the error:
gcc -O1 -fno-omit-frame-pointer -m64 -g -fno-strict-aliasing
-std=gnu99 -Wall -Wstrict-prototypes -Wdeclaration-after-statement
-D__XEN_TOOLS__ -MMD -MF .init.o.d -fno-optimize-sibling-calls
-I../include -I.
-I/root/xen/xen-netbsd/tools/libvchan/../../tools/xenstore
-I/root/xen/xen-netbsd/tools/libvchan/../../tools/include
-I/root/xen/xen-netbsd/tools/libvchan/../../tools/libxc
-I/root/xen/xen-netbsd/tools/libvchan/../../tools/include -c -o init.o init.c
init.c:45:30: fatal error: xen/sys/gntalloc.h: No such file or directory
compilation terminated.
gmake[3]: *** [init.opic] Error 1
gmake[3]: *** Waiting for unfinished jobs....
init.c:45:30: fatal error: xen/sys/gntalloc.h: No such file or directory
compilation terminated.
Signed-off-by: Roger Pau Monne <roger.pau@citrix.com>
Acked-by: Christoph Egger <Christoph.Egger@amd.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
Up until the point where we think this is a bzImage there is no point in
printing panicy messages -- some other loader will have a go (probably the
compressed ELF one)
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>
Currently qemu-xen will be compiled with CFLAGS only if CFLAGS was
already in the environment during make invocation. If CFLAGS is in
environment then make will append all of the various flags specified in
xen Makefiles to this environment variable, which is then used in qemu
configure. Since qemu-xen is not ready for compiler flags like
"-std=gnu99" compilation will fail. If CFLAGS is not in environment,
then configure will use just its own "-O2 -g" because make does not
export its own CFLAGS variable.
>From a distro perspective, it is required to build libraries and
binaries with certain global cflags (arbitrary gcc options). Up to the
point when qemu-xen was imported it worked as expected by exporting
CFLAGS before 'make tools'. Now qemu-upstream reuses these CFLAGS, but
it cant deal with the result.
This patch extends the tools Makefiles so that three new environment
variables are recognized:
EXTRA_CFLAGS_XEN_TOOLS= specifies CFLAGS for the tools build.
EXTRA_CFLAGS_QEMU_TRADITIONAL= specifies CFLAGS for old qemu.
EXTRA_CFLAGS_QEMU_XEN= specifies CFLAGS for new qemu.
Special care needs to be taken in tools/firmware because the resulting
binaries are not linked with the hosts runtime libraries. These binaries
run in guest context. To avoid build errors from gcc options like
-fstack-protector, reuse existing practice to unset the new
EXTRA_CFLAGS_XEN_TOOLS for the firmware dirs.
The new feature can be used like this in a rpm xen.spec file:
export EXTRA_CFLAGS_XEN_TOOLS="${RPM_OPT_FLAGS}"
export EXTRA_CFLAGS_QEMU_TRADITIONAL="${RPM_OPT_FLAGS}"
export EXTRA_CFLAGS_QEMU_XEN="${RPM_OPT_FLAGS}"
./configure \
--libdir=%{_libdir} \
--prefix=/usr
make
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
The current vif-hotplug-common.sh for renaming the tap device is failing
because it is racing with this script and therefore the device is unexpectedly
up when we come to rename it.
Fix this in the same way as libxl does, by disabling the script (only on
Linux).
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>
xs_watch() creates a thread to wake watchers using default attributes. The
stacksize can be quite large (8 MB on Linux), applications that link against
xenstore end up having a larger memory footprint than necessary.
Signed-off-by: Simon Rowe <simon.rowe@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
The thread created to wakeup watchers is not intended to handle signals
(and a later patch will reduce it's stack size which makes it unsuitable
for doing so).
Signed-off-by: Simon Rowe <simon.rowe@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
Other than in __list_splice(), the first element's prev pointer
doesn't need adjustment here - it already is PAGE_LIST_NULL. Rather
than fixing the assignment (to formally match __list_splice()), simply
assert that this assignment is really unnecessary.
Reported-by: Jisoo Yang <jisooy@gmail.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Also assert that the prev pointers are both PAGE_LIST_NULL.
Signed-off-by: Keir Fraser <keir@xen.org>
Committed-by: Keir Fraser <keir@xen.org>
If CPU stays in the same idle state for the full duration of
xenpm sample then average residency may not be reported correctly
since usage counter will not be incremented.
In addition, in order to calculate averages correctly residence
time and usage counter should be read and written atomically.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@amd.com>
Committed-by: Keir Fraser <keir@xen.org>
To avoid recent gcc complaining about:
libxl.c: In function 'libxl_primary_console_exec':
libxl.c:1233:9: error: case value '4294967295' not in enumerated type 'libxl_domain_type' [-Werror=switch]
Also:
- have all the call sites of libxl__domain_type() return with error in
case the function returns LIBXL_DOMAIN_TYPE_INVALID;
- adjust all other code segments where -Wswitch makes would claim that
LIBXL_DOMAIN_TYPE_INVALID is not handled by adding a "default: abort();"
clause.
Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- use LIBXL_DOMAIN_TYPE_INVALID instead of -1 for
libxl_domain_build_info.type's keyvar_init_val.
-- what used to be libxl_primary_console_exec is now in
libxl__primary_console_find so resolve the rejected hunk
accordingly ]
Committed-by: Ian Campbell <ian.campbell@citrix.com>
This api retrieve domain console from xenstore. With this new api, it is easy
to implement "virsh console" command in libvirt libxl driver.
Signed-off-by: Bamvor Jian Zhang <bjzhang@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- use NOGC instead of 0 to allow improvements to this infrastructure
in the future ]
Committed-by: Ian Campbell <ian.campbell@citrix.com>
And ditch the error handling in libxl_get_cpu_topology()
Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- define and use NOGC instead of NULL to allow improvements to this
infrastructure in the future ]
Committed-by: Ian Campbell <ian.campbell@citrix.com>
These renamed a type and the associated functions and the ocaml bindings were
not updated to suit.
This also highlighted that libxl_domain_sched_params should not be DIR_IN since
it is also use as an output struct.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>
_paths.h needs to be in AUTOINCS. That arranges for it to be an
explicit dependency of all object files. This is necessary so that it
is made before any compilation is attempted.
Making it a dependency of xl.h (as in 25426:e53a1d3c212c) is harmless,
but not sufficient because that only takes effect if there is already
an autogenerated .d file naming xl.h as a dependency of relevant
object files.
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
Fix typo in comment.
Fix cpuid_flags array init, use correct number of arguments for empty
array entry.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>
This is consistent with having a single struct for all parameters.
Effectively renames and exports libxl__sched_set_params as
libxl_domain_sched_params_set. libxl_domain_sched_params_get is new.
Improve const correctness of the setters while I'm here.
Use shorter LOG macros when touching a line anyway.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>
To do so we define a discriminating value which is never a valid real value for
each parameter.
While there:
- removed libxl_sched_*_domain in favour of libxl_domain_sched_params.
- use this new functionality for the various xl commands which set sched
parameters, which saves an explicit read-modify-write in xl.
- removed call of xc_domain_getinfolist from a few functions which weren't
actually using the result (looks like a cut and paste error)
- fix xl which was setting period for a variety of different config keys.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>
Jan Beulich
Christoph Egger
Dario Faggioli
Daniel De Graaf
Ian Campbell
Dario Faggioli
Boris Ostrovsky
Andrew Cooper
Mukesh Rathor
Olaf Hering
Anthony PERARD
Juergen Gross
Roger Pau Monne
Xuesen Guo
Ian Jackson
David Vrabel
Simon Rowe
Bamvor Jian Zhang