in _csched_cpu_pick(), as not doing so may result in the domain's
node-affinity mask (as retrieved by csched_balance_cpumask() )
and online mask (as retrieved by cpupool_scheduler_cpumask() )
having an empty intersection.
Therefore, when attempting a node-affinity load balancing step
and running this:
...
/* Pick an online CPU from the proper affinity mask */
csched_balance_cpumask(vc, balance_step, &cpus);
cpumask_and(&cpus, &cpus, online);
...
we end up with an empty cpumask (in cpus). At this point, in
the following code:
....
/* If present, prefer vc's current processor */
cpu = cpumask_test_cpu(vc->processor, &cpus)
? vc->processor
: cpumask_cycle(vc->processor, &cpus);
....
an ASSERT (from inside cpumask_cycle() ) triggers like this:
(XEN) Xen call trace:
(XEN) [<ffff82d08011b124>] _csched_cpu_pick+0x1d2/0x652
(XEN) [<ffff82d08011b5b2>] csched_cpu_pick+0xe/0x10
(XEN) [<ffff82d0801232de>] vcpu_migrate+0x167/0x31e
(XEN) [<ffff82d0801238cc>] cpu_disable_scheduler+0x1c8/0x287
(XEN) [<ffff82d080101b3f>] cpupool_unassign_cpu_helper+0x20/0xb4
(XEN) [<ffff82d08010544f>] continue_hypercall_tasklet_handler+0x4a/0xb1
(XEN) [<ffff82d080127793>] do_tasklet_work+0x78/0xab
(XEN) [<ffff82d080127a70>] do_tasklet+0x5f/0x8b
(XEN) [<ffff82d080158985>] idle_loop+0x57/0x5e
(XEN)
(XEN)
(XEN) ****************************************
(XEN) Panic on CPU 1:
(XEN) Assertion 'cpu < nr_cpu_ids' failed at /home/dario/Sources/xen/xen/xen.git/xen/include/xe:16481
It is for example sufficient to have a domain with node-affinity
to NUMA node 1 running, and issueing a `xl cpupool-numa-split'
would make the above happen. That is because, by default, all
the existing domains remain assigned to the first cpupool, and
it now (after the cpupool-numa-split) only includes NUMA node 0.
This change prevents that by generalizing the function used
for figuring out whether a node-affinity load balancing step
is legit or not. This way we can, in _csched_cpu_pick(),
figure out early enough that the mask would end up empty,
skip the step all together and avoid the splat.
Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: George Dunlap <george.dunlap@eu.citrix.com>
With HVM's MMIO operand handling now being capable of splitting large
reads, there's no need to issue at most machine word size reads when
we really need wider operands.
Not that this is not done everywhere - there are a couple of cases
where keeping the reads separate is more natural (and folding them
would complicate the code rather than simplifying it).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
We really need to check for a signed overflow of 8 bits, while the
previous check compared the sign-extended 8-bit result with the
zero-extended 16-bit one (which was wrong for all negative results).
Once at it
- also adjust the 16-bit comparison for symmetry
- improve the 8-bit multiplication (no need to zero-extend to 32-bits
the sign-extended to 16 bits original 8-bit value)
- fold both signed multiplication variants
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
... for the case of accessing MMIO.
Rather than doing the early operand type adjustment for just for that
case, do it for all of the 0xF6, 0xF7, and 0xFF groups (allowing some
other code to be dropped instead).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
... for the case of accessing MMIO.
Also streamline the ARPL emulation a little, and add tests for both
instructions (the MOVSXD one requires a few other adjustments, as we
now need to run in a mode where the emulator's mode_64bit() returns
true).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
Rather than mentioning the (linked) directory, mention the files thus
making sure things get rebuild as needed when the core emulator files
change.
Also enable debug info generation unconditionally, as this is testing
stuff only anyway.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
Just like real hardware we ought to split such accesses transparently
to the caller. With little extra effort we can at once even handle page
crossing accesses correctly.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
This helps avoid static analysis false-positives, and might lead to
better code density as the compiler knows it doesn't have to restore
spilled state &c.
Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Keir Fraser <keir@xen.org>
If the allocation fails, make sure to call vmx_vmcs_exit().
This is a candidate for backport.
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Signed-off-by: Mukesh Rathor <mukesh.rathor@oracle.com>
It looks like one of the failure cases in hvm_vcpu_initialise jumps to
the wrong label; this could lead to slow leaks if something isn't
cleaned up properly.
I will probably change these labels in a future patch, but I figured
it was better to have this fix separately.
This is also a candidate for backport.
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Signed-off-by: Mukesh Rathor <mukesh.rathor@oracle.com>
Add big warnings to configure, xend initscript and xm.
The big xm warning is displayed once (per boot, or per tmpreaper clean),
afterwards a single line warning is displayed.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
This was broken by 118104e5eaf2 "docs: Build docs for ARM as well as x86_64".
Move docs to hypercall/ARCH instead of hypercall-ARCH.
Support mulitple levels of subdirectories in gen-html-index tool.
This removes the need for a symlink hypercall->hypercall-x86_64 since there is
now a proper index at hypercall/index.html.
Update INDEX to human readable names for the architecture specific hypercalls.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
The lock's not used for anything, and AFAICT no locking is needed
since the IVRS tables are static after boot.
Coverity CID 1087199
Signed-off-by: Tim Deegan <tim@xen.org>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Currently we use the chosen/bootargs property as the Xen commandline
and rely on xen,dom0-bootargs for Dom0. However this brings issues
with bootloaders, which usually build bootargs by bootscripts for a
Linux kernel - and not for the entirely different Xen hypervisor.
Introduce a new possible device tree property "xen,xen-bootargs"
explicitly for the Xen hypervisor and make the selection of which to
use more fine grained:
- If xen,xen-bootargs is present, it will be used for Xen.
- If xen,dom0-bootargs is present, it will be used for Dom0.
- If xen,xen-bootargs is _not_ present, but xen,dom0-bootargs is,
bootargs will be used for Xen. Like the current situation.
- If no Xen specific properties are present, bootargs is for Dom0.
- If xen,xen-bootargs is present, but xen,dom0-bootargs is missing,
bootargs will be used for Dom0.
The aim is to allow common bootscripts to boot both Xen and native
Linux with the same device tree blob. If needed, one could hard-code
the Xen commandline into the DTB, leaving bootargs for Dom0 to be set
by the (non Xen-aware) bootloader.
I will send out a appropriate u-boot patch, which writes the content
of the "xen_bootargs" environment variable into the xen,xen-bootargs
dtb property.
Signed-off-by: Andre Przywara <andre.przywara@linaro.org>
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
It's possible to have a device description in the DTS but the device is not
wired.
device_init must check if the device is available before doing anything with
it.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Device tree cells are 32-bit big endian value. Use __be32 to avoid confusion
later.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
All Xen code has been converted to the new device tree API that uses a tree
structure to describe the DTS.
The Flat Device tree is still used by Xen during early boot stage, but only in
internal. Remove entirely unneeded functions or move to a static function.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The Multi Core Timer (MCT) is a Samsung specific device.
This device tries to route IRQ in non-boot CPU which is not yet handled by Xen.
The user will see randomly dom0 hang, but I'm not sure that is the real reason.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
On Versatile there are a bunch of devices which must not be pass-through to any
guest (power management and cache coherency devices).
This commit also blacklists the HDLCD device because Xen is unable to correctly
map the framebuffer into dom0. Therefore, when Linux will try to access to the
framebuffer, Xen will receive a non-handled data access.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Devices used by Xen should not be pass-through to dom0. If the device is really
usefull for dom0 (for instance the timer and the GIC), it will recreate the
node.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Each platform code will list the device that must not pass-through to a guest.
Theses devices are used for: power management, timer,...
When theses devices are given to DOM0, it can controls the hardware and then
break the whole platform.
This callback is enough until we will start to care about power performance.
For this purpose, we may need to extend this interface to implement per-device
MMIO filtering to allow dom0 to continue to control devices which it owns which
happen to share e.g. a clock controller with Xen.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Recreate the GIC node and remove hypervisor specific ranges (vgic and hypervisor
controls).
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The number of cpus in dom0 can be different compare to the real number of
physical cpus.
For the moment, Xen assumes that the cpus are identical.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Remove the usage of the FDT in benefit of the device tree structure.
The latter is easier to use and can embedded meta-data for Xen (ie: is the
device is used by Xen...).
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
There is some place in Xen ARM code where multiple if conditions is used
check the presence of a node or find a node.
These pieces of code can be replace by an array and using proper device tree
helpers.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The video driver is initialized before the console is correctly set up.
Therefore, printk will never output if there is no serial configured.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
On some board CPU IDs are not contiguous (for instance the Versatile Express
with big.LITTLE supports). If the CPU ID is greater than NR_CPUS Xen will hang
without any message. This is because console driver is not yet initialized and
hypervisor data abort uses printk.
For the moment check the CPU ID and print an warning if an error occured.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The device tree compiler (dtc) will only warn if the "reg" property doesn't
match #address-cells and #size-cells size. It won't update the different
property. Therefore, Xen needs to check if the size match both properties,
otherwise Xen can retrieve a wrong range.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The "reg" property is only composed of one uint32. device_get_reg can be
replaced by dt_read_number.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
When of_node_cmp and of_compat_cmp was introduced in commit fb97eb6
"xen/arm: Create a hierarchical device tree", they were copied from the wrong
Linux header.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
On new Flat Device Tree version, the property "name" may not exist.
The property is never used in Xen code except to set the field "name" of
dt_device_node.
For convenience, remove the fake property. It will save space during the
creation of the dom0 FDT.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
There is 2 macros: for_each_device_node and for_each_property_of_node
with a too generic name.
Also replace all call-site with the new function names.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
With changeset 22885 support was added for setting the MTU in the vif-bridge
script for when a vif interface was set to 'online'. The was not done for the
'add' operation. The 'add' operation was added to the script for when tap
devices were specified (c/s 21944). With the setting of the MTU for the
'online' case was there a reason for omitting the 'add'?
This patch sets the MTU for both 'online' and 'add' in the vif-bridge script.
Signed-off-by: Charles Arnold <carnold@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Previously the ARINC653 scheduler was indirectly maintained by
Kathy Hadley and Josh Holtrop being cc'd on changes by George
Dunlap. Maintenance is currently being done by Robert VanVossen
and myself. To avoid potential communication problems in the
future, make Robbie and I the official maintainers.
Signed-off-by: Nathan Studer <nate.studer@dornerworks.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Matthew Daley
Dario Faggioli
Jan Beulich
Ian Campbell
Tim Deegan
George Dunlap
Andre Przywara
Julien Grall
Charles Arnold
Nathan Studer
Suravee Suthikulpanit