A credit reset basically means going through all the
vCPUs of a runqueue and altering their credits, as a
consequence of a 'scheduling epoch' having come to an
end.
Blocked or runnable vCPUs are fine, all the credits
they've spent running so far have been accounted to
them when they were scheduled out.
But if a vCPU is running on a pCPU, when a reset event
occurs (on another pCPU), that does not get properly
accounted. Let's therefore begin to do so, for better
accuracy and fairness.
In fact, after this patch, we see this in a trace:
csched2:schedule cpu 10, rq# 1, busy, not tickled
csched2:burn_credits d1v5, credit = 9998353, delta = 202996
runstate_continue d1v5 running->running
...
csched2:schedule cpu 12, rq# 1, busy, not tickled
csched2:burn_credits d1v6, credit = -1327, delta = 9999544
csched2:reset_credits d0v13, credit_start = 10500000, credit_end = 10500000, mult = 1
csched2:reset_credits d0v14, credit_start = 10500000, credit_end = 10500000, mult = 1
csched2:reset_credits d0v7, credit_start = 10500000, credit_end = 10500000, mult = 1
csched2:burn_credits d1v5, credit = 201805, delta = 9796548
csched2:reset_credits d1v5, credit_start = 201805, credit_end = 10201805, mult = 1
csched2:burn_credits d1v6, credit = -1327, delta = 0
csched2:reset_credits d1v6, credit_start = -1327, credit_end = 9998673, mult = 1
Which shows how d1v5 actually executed for ~9.796 ms,
on pCPU 10, when reset_credit() is executed, on pCPU
12, because of d1v6's credits going below 0.
Without this patch, this 9.796ms are not accounted
to anyone. With this patch, d1v5 is charged for that,
and its credits drop down from 9796548 to 201805.
And this is important, as it means that it will
begin the new epoch with 10201805 credits, instead
of 10500000 (which he would have, before this patch).
Basically, we were forgetting one round of accounting
in epoch x, for the vCPUs that are running at the time
the epoch ends. And this meant favouring a little bit
these same vCPUs, in epoch x+1, providing them with
the chance of execute longer than their fair share.
Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: George Dunlap <george.dunlap@citrix.com>
In fact, whether or not a pCPU has been tickled, and is
therefore about to re-schedule, is something we look at
and base decisions on in various places.
So, let's make sure that we do that basing on accurate
information.
While there, also tweak a little bit smt_idle_mask_clear()
(used for implementing SMT support), so that it only alter
the relevant cpumask when there is the actual need for this.
(This is only for reduced overhead, behavior remains the
same).
Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: George Dunlap <george.dunlap@eu.citrix.com>
When toolstack overrides Intel CPUID leaf 0xa's PMU version with an
invalid value VPMU should not be available to the guest.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
vpmu_enabled() (used by hvm/pv_cpuid() to properly report 0xa leaf
for Intel processors) is based on the value of VPMU_CONTEXT_ALLOCATED
bit. This is problematic:
* For HVM guests VPMU context is allocated lazily, during the first
access to VPMU MSRs. Since the leaf is typically queried before guest
attempts to read or write the MSRs it is likely that CPUID will report
no PMU support
* For PV guests the context is allocated eagerly but only in responce to
guest's XENPMU_init hypercall. There is a chance that the guest will
try to read CPUID before making this hypercall.
This patch introduces VPMU_AVAILABLE flag which is set (subject to vpmu_mode
constraints) during VCPU initialization for both PV and HVM guests. Since
this flag is expected to be managed together with vpmu_count, get/put_vpmu()
are added to simplify code.
vpmu_enabled() (renamed to vpmu_available()) can now use this new flag.
(As a side affect this patch also fixes a race in pvpmu_init() where we
increment vcpu_count in vpmu_initialise() after checking vpmu_mode)
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Freeing part of the BSS back for general use proves to be problematic. It is
not accounted for in xen_in_range(), causing errors when constructing the
IOMMU tables, resulting in a failure to boot.
Other smaller issues are that tboot treats the entire BSS as hypervisor data,
creating and checking a MAC of it on S3, and that, by being 1MB in size,
freeing it guarentees to shatter the hypervisor superpage mappings.
This is a stopgap fix to unblock master, while alternatives are discussed.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
For a transitional period (until we've managed to replace all
underscore prefixed instances), allow both names to co-exist.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
The current implementation of nested VMX cannot work without HAP.
Signed-off-by: Haozhong Zhang <haozhong.zhang@intel.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
For a HVM domain, if a vcpu is in the nested guest mode,
__raw_copy_to_guest(), __copy_to_guest() and __copy_field_to_guest()
used by update_runstate_area() and update_secondary_system_time() will
copy data to L2 guest rather than the L1 guest.
This commit temporally clears the nested guest flag before all guest
copies in update_runstate_area() and update_secondary_system_time(),
and restores the flag after those guest copy operations.
The flag clear/restore is combined with the existing
smap_policy_change() which is renamed to update_guest_memory_policy().
Signed-off-by: Haozhong Zhang <haozhong.zhang@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
As some users have suggested, elaborate the usage of RMRR specification
on the command line, and provide a usage example.
Also, always treat the specified page numbers as hexadecimal values.
Signed-off-by: Venu Busireddy <venu.busireddy@oracle.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
That is to say, don't support a domain assigns a device to itself or detachs
a device from itself.
Signed-off-by: Chao Gao <chao.gao@intel.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
The default dom0_mem is 128M which is not sufficient to boot a Ubuntu
based Dom0. It is not clear what a better default value could be.
Instead, loudly warn the user when dom0_mem is unspecified and wait 3
secs. Then use 512M.
Update the docs to specify that dom0_mem is required on ARM. (The
current xen-command-line document does not actually reflect the current
behavior of dom0_mem on ARM correctly.)
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
c/s b4cd59fe "x86: reorder .data and .init when linking" had an unintended
side effect, where xen_in_range() and the tboot S3 MAC were no longer correct.
In practice, it means that Xen's .data section is excluded from consideration,
which means:
1) Default IOMMU construction for the hardware domain could create mappings.
2) .data isn't included in the tboot MAC checked on resume from S3.
Adjust the comments and virtual address anchors used to define the regions.
Reported-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
As a memory report can now be triggered via XS_CONTROL support via
command line and signal handler is no longer needed. Remove it.
Signed-off-by: Juergen Gross <jgross@suse.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Add a XS_CONTROL command to xenstored for doing a talloc report to a
file. Right now this is supported by specifying a command line option
when starting xenstored and sending a signal to the daemon to trigger
the report.
To dump the report to the standard log file call:
xenstore-control memreport
To dump the report to a new file call:
xenstore-control memreport <file>
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Today Xenstore supports logging only if specified at start of the
Xenstore daemon. As it can't be disabled during runtime it is not
recommended to start xenstored with logging enabled.
Add support for switching logging on and off at runtime and to
specify a (new) logfile. This is done via the XS_CONTROL wire command
which can be sent with xenstore-control.
To switch logging on just use:
xenstore-control log on
To switch it off again:
xenstore-control log off
To specify a (new) logfile:
xenstore-control logfile <file>
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
The Xenstore protocol supports the XS_CONTROL command for triggering
various actions in the Xenstore daemon. Enhance that support by using
a command table and adding a help function.
Support multiple control commands in the associated xenstore-control
program used to issue XS_CONTROL commands.
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Move the XS_CONTROL handling of xenstored to a new source file
xenstored_control.c.
In order to avoid making get_string() in xenstored_core.c globally
visible use strlen() instead, which is save in this context due to
xs_count_strings() before returned a value > 1.
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
In preparation to support other than pure debug functionality via the
Xenstore XS_DEBUG wire command rename it to XS_CONTROL and make
XS_DEBUG an alias of it.
Add an alias xs_control_command for the associated xs_debug_command,
too.
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
After splitting out all the meaty bits, xl_cmdimpl.c doesn't contain
much. Merge the rest into xl.c and delete the file.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Include COLO / Remus code because they are built on top of the existing
migration protocol.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Add some function declarations to xl.h because they are now needed in
multiple files.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Including create, reboot, shutdown, pause, unpause and destroy.
Lift a bunch of core data structures and function declarations to xl.h
because they are needed in both xl_cmdimpl.c and xl_vmcontrol.c.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
A collections of functions that don't warrant their own files.
Moving main_devd there requires lifting do_daemonize to xl_utils.c.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
The new file also contains code for channel, which is just a console
in disguise.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
We will need to move main_vncviewer to a different file where it has no
access to the helper vncviewer.
Call libxl_vncviewer_exec directly.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
The array to map libxl_shutdown_action_to_shutdown to string is going to
be moved to a dedicated file.
Provide a function to do the translation so that we don't need to make
the array globally visible.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
We want to consistently prefix functions to parse input with "parse_".
No functional change.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Move some commonly used functions to a new file.
find_domain requires access to global variable common_domname. Make that
non-static.
No functional change.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Dario Faggioli
Boris Ostrovsky
Jan Beulich
Andrew Cooper
Haozhong Zhang
Venu Busireddy
Chao Gao
Stefano Stabellini
Roger Pau Monne
Juergen Gross
Wei Liu