ACRN hypervisor needs to kick vCPU off VMX non-root mode to do some
operations in hypervisor, such as interrupt/exception injection, EPT
flush etc. For non lapic-pt vCPUs, we can use IPI to do so. But, it
doesn't work for lapic-pt vCPUs as the IPI will be injected to VMs
directly without vmexit.
Without the way to kick the vCPU off VMX non-root mode to handle pending
request on time, there may be fatal errors triggered.
1). Certain operation may not be carried out on time which may further
lead to fatal errors. Taking the EPT flush request as an example, once we
don't flush the EPT on time and the guest access the out-of-date EPT,
fatal error happens.
2). ACRN now will send an IPI with vector 0xF0 to target vCPU to kick the vCPU
off VMX non-root mode if it wants to do some operations on target vCPU.
However, this way doesn't work for lapic-pt vCPUs. The IPI will be delivered
to the guest directly without vmexit and the guest will receive a unexpected
interrupt. Consequently, if the guest can't handle this interrupt properly,
fatal error may happen.
The NMI can be used as the notification signal to kick the vCPU off VMX
non-root mode for lapic-pt vCPUs. So, this patch uses NMI as notification signal
to address the above issues for lapic-pt vCPUs.
Tracked-On: #3886
Acked-by: Eddie Dong <eddie.dong@intel.com>
Signed-off-by: Kaige Fu <kaige.fu@intel.com>
For now, we set NOOP scheduler as default. User can choose IORR scheduler as needed.
Tracked-On: #4178
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Add yield support for schedule, which can give up pcpu proactively.
Tracked-On: #4178
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
kick means to notify one thread_object. If the target thread object is
running, send a IPI to notify it; if the target thread object is
runnable, make reschedule on it.
Also add kick_vcpu API in vcpu layer to notify vcpu.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
After moving softirq to following interrupt path, softirq handler might
break in the schedule spinlock context and try to grab the lock again,
then deadlock.
Disable interrupt with schedule spinlock context.
For the IRQ disable/restore operations:
CPU_INT_ALL_DISABLE(&rflag)
CPU_INT_ALL_RESTORE(rflag)
each takes 50~60 cycles.
renaming: get_schedule_lock -> obtain_schedule_lock
Tracked-On: #3813
Signed-off-by: Conghui Chen <conghui.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Clean up do_swtich and do switch related things in schedule().
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This patch decouple some scheduling logic and abstract into a scheduler.
Then we have scheduler, schedule framework. From modulization
perspective, schedule framework provides some APIs for other layers to
use, also interact with scheduler through scheduler interaces.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
To get pcpu_id from sched_control quickly and easier.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Let init thread end with run_idle_thread(), then idle thread take over and
start to do scheduling.
Change enter_guest_mode() to init_guest_mode() as run_idle_thread() is removed
out of it. Also add run_thread() in schedule module to run
thread_object's thread loop directly.
rename: switch_to_idle -> run_idle_thread
Tracked-On: #3813
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
sleep one thread_object means to prevent it from being scheduled.
wake one thread_object is an opposite operation of sleep.
This patch also add notify_mode in thread_object to indicate how to
deliver the request.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Now, we have three valid status for thread_object:
THREAD_STS_RUNNING,
THREAD_STS_RUNNABLE,
THREAD_STS_BLOCKED.
This patch also provide several helpers to check the thread's status and
a status set wrapper function.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
schedule infrastructure is per pcpu, so move its initialization to each
pcpu's initialization.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
With cpu sharing enabled, per_cpu vcpu cannot work properly as we might
has multiple vcpus running on one pcpu.
Add a schedule API sched_get_current to get current thread_object on
specific pcpu, also add a vcpu API get_running_vcpu to get corresponding
vcpu of the thread_object.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
With cpu sharing enabled, we will map acrn_vcpu to thread_object
in scheduling. From modulization perspective, we'd better hide the
pcpu_id in acrn_vcpu and move it to thread_object.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Currently we are using a 1:1 mapping logic for pcpu:vcpu. So don't need
a runqueue for it. Removing it as preparation work to abstract scheduler
framework.
Tracked-On: #3813
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
As we introduced vcpu_affinity[] to assign vcpus to different pcpus, the
old policy and functions are not needed. Remove them.
Tracked-On: #3663
Signed-off-by: Conghui Chen <conghui.chen@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
From modulization perspective, it's not suitable to put pcpu and vm
related request operations in schedule. So move them to pcpu and vm
module respectively. Also change need_offline return value to bool.
Tracked-On: #3663
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
ACRN Coding guidelines requires type conversion shall be explicity. However,
there's no need for this case since we could return bool directly.
Tracked-On: #1842
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Now sched_object and sched_context are protected by scheduler_lock. There's no
chance to use runqueue_lock to protect schedule runqueue if we have no plan to
support schedule migration.
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Schedule context flag may set out of schedule lock protection, like NEED_OFFLINE
or NEED_SHUTDOWN_VM. So NEED_RESCHEDULE flag should be set atomically too.
Tracked-On: #1842
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
for ctx->flags is protected by scheduler lock, so not need
to set lock again.
Tracked-On: #3130
Signed-off-by: Minggui Cao <minggui.cao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
For pre-launched VMs and SOS, VM shutdown should not be executed in the
current VM context.
- implement NEED_SHUTDOWN_VM request so that the BSP of the target VM can shut
down the guest in idle thread.
- implement shutdown_vm_from_idle() to shut down target VM.
Tracked-On: #2700
Signed-off-by: Zide Chen <zide.chen@intel.com>
Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This patch adds prefix 'p' before 'cpu' to physical cpu related functions.
And there is no code logic change.
Tracked-On: #2991
Signed-off-by: Kaige Fu <kaige.fu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This patch makes make_reschedule_request support for kicking
off vCPU using INIT.
Tracked-On: #2865
Signed-off-by: Kaige Fu <kaige.fu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
it removes hypervisor.h and just includes needed header files.
Tracked-On: #1842
Signed-off-by: Minggui Cao <minggui.cao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Panic should only be used when system booting. Once the system boot done,
it could never be used. While ASSERT could be used in some situations, such
as, there are some pre-assumption for some code, using ASSERT here for debug.
Tracked-On: #861
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@inte.com>
after compile, the compiled code could change rsp, so use pure asm code
to avoid such problem which will cause schedule switch failure.
Tracked-On: #2410
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
this patch added full context switch support for scheduling, it make sure
each VCPU has its own stack, and added host_sp field in struct sched_object
to record host stack pointer for each switch out object.
Arch related function arch_switch_to is added for context switch.
To benefit debugging, a name[] field is also added into struct sched_object.
Tracked-On: #2394
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Xu, Anthony <anthony.xu@intel.com>
To support full context switch scheduling, each physical
CPU must have its own idle sched_obj.
Tracked-On: #2394
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Xu, Anthony <anthony.xu@intel.com>
Misra C requires Function must have only 1 return entry.
Fixed it by use "if ... else ..." format.
Tracked-On: #861
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Misra C requires Function must have only 1 return entry.
Fixed it by use "if ... else ..." format.
Tracked-On: #861
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Now, need_reschedule will test_and_clear the bit NEED_RESCHEDULE in
schedule context, then call schedule. It is not a exact match with the
name.
This patch move the flag clearing into scheudle, and need_reschedule
just check and return.
Tracked-On: #1821
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
use struct sched_object as the main interface of scheduling, then
make scheduler as an independent module to vcpu:
- add struct sched_object as one field in struct vcpu
- define sched_object.thread for switch_to thread
- define sched_object.prepare_switch_out/in for prepare_switch before
switch_to
- move context_switch_out/context_switch_in into vcpu.c as
vcpu.sched_obj.prepare_switch_out/in
- make default_idle as global idle.thread for idle_thread
- make vcpu_thread as vcpu.sched_obj.thread for each vcpu thread
- simplify switch_to based on sched_object
Tracked-On: #1842
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <edide.dong@intel.com>
just use pcpu_id for make_reschedule_request is enough
Tracked-On: #1842
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Acked-by: Eddie Dong <edide.dong@intel.com>
add struct sched_object, and use it as input param instead of vcpu for
below functions:
- add_to_cpu_runqueue renamed from add_vcpu_to_runqueue
- remove_from_cpu_runqueue renamed from remove_vcpu_from_runqueue
- get_next_sched_obj added to get next sched object
Tracked-On: #1842
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <edide.dong@intel.com>
This patch introduce a new mode of IO request completion, polling mode.
Now, the sketch of ioreq process can be,
A. UOS vcpu0 generate PIO/MMIO ->
B. pcpu1(vcpu0 of UOS) trap into HV ->
C. pcpu1 build ioreq, send IPI and enter idle ->
D.1 pcpu0(vcpu0 of SOS) response IPI,
D.2 pcpu0 handle the ioreq in HV, kernel, DM,
D.3 pcpu0 mark ioreq as complete,
D.4 pcpu0 hypercall to enter HV ->
E.1 pcpu0 send IPI to wake pcpu1 up
E.2 UOS vcpu0 continue running
With this change, it skips D.4, E.1 steps. In step C, pcpu1 will enter a
polling ioreq state idle after send out the IPI.
It can save about ~5000 cpu cycles.
In polling mode, we do the polling in idle instead of pause cpu all the
time. It will consume more power. A better way is to use monitor/mwait
instructions which can put cpu into a sleep state with monitoring a
memory address. Unfortunately, APL has bug with monitor. We can gather
all ioreqs state into one monitorable memory and take advantage of
monitor/mwait for future platform.
The way polling or notification is per VM. We can config VMs in
different mode. By default, IO request completion will use notification
mode for all VMs. We can switch it by Kconfig.
Tracked-On: #1821
Signed-off-by: Shuo Liu <shuo.a.liu@intel.com>
Reviewed-by: Eddie Dong <eddie.dong@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
For data structure types "struct vcpu", its name is identical
with variable name in the same scope. This is a MISRA C violation.
Naming convention rule:If the data structure type is used by multi
modules, its corresponding logic resource is exposed to external
components (such as SOS, UOS), and its name meaning is simplistic
(such as vcpu, vm), its name needs prefix "acrn_".
The following udpates are made:
struct vcpu *vcpu-->struct acrn_vcpu *vcpu
Tracked-On: #861
Signed-off-by: Xiangyang Wu <xiangyang.wu@linux.intel.com>
Fix violations whose parameter can be read-only.
This patch only fix the parameter whose name is vcpu.
Tracked-On: #861
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
MISRA-C requires that there should be no unused parameters in
functions.
In some cases, we will keep the unused parameters.
vmexit handler is one example. It is used as function pointer.
Some of the vmexit handlers use the input parameter 'vcpu', some of
them don't. We still need to keep the unused parameters 'vcpu' for
those handlers don't use 'vcpu'.
This patch removes the unused parameters that is not being used
unconditionally.
v1 -> v2:
* remove the non-implemented API 'vlapic_id_write_handler'
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
to avoid send IPI to self, also improve the related code:
1. get_cpu_id is uint16_t now
2. MISRA-C requirement. like add {}
Signed-off-by: Minggui Cao <minggui.cao@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
add wrap function cpu_do_idle in default_idle to handle arch cpu specific
idle operation.
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
V1:
In order to remove MISRA C violations for rule
219S, rename functions, macros whose name
starts with "_".
Also removed some __mmio functions because they are
duplicates and are not used anywhere.
Renamed functions like __assert, to asm_assert,
because they were only wrappers around asm calls.
V2:
Over and above the changes on V1, modified bitmap
functions names to lock (corresponding to unlock)
introduced in V1
Signed-off-by: Arindam Roy <arindam.roy@intel.com>
This is the final cleanup of the integral type related issues, reported by the
static checker, under common/, mostly including
* make explicit the narrowings of vm_ids passed by register.
* work around the confusion of the static checker by abstracting
sub-expressions to local variables.
The remaining reports that are not trivial to suppress will be in the scope of a
separate document.
v1 -> v2:
* Instead of converting vm_ids inside hcall_xxx, update the prototypes of
these functions and do the conversion in vmcall_vmexit_handler.
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
There are many type conversions in the atomic
operations invoking reported by static analysis tool. These
type conversions voilate MISRA C.
To keep uniform naming convention, rename atomic operation
function names:
atomic_set_int/long --> atomic_set32/64;
atomic_clear_int/long --> atomic_clear32/64;
atomic_load --> atomic_load32;
atomic_store --> atomic_store32;
atomic_swap --> atomic_swap32;
atomic_readandclear --> atomic_readandclear32;
atomic_inc --> atomic_inc32;
atomic_dec --> atomic_dec32;
atomic_cmpxchg --> atomic_cmpxchg32;
atomic_xadd --> atomic_xadd32.
Update the type of atomic_load32/64, atomic_store32/64,
atomic_swap32/64, atomic_cmpxchg32/6.
Update related variables and callers.
Note: the type of return value and parameters of atomic_xadd32/64
still keep signed int/long since caller pass
negative variable to atomic_xadd32/64;
V1-->V2:
Add comments for atomic_set/clear to differ from
bitmap_set/clear.
Signed-off-by: Xiangyang Wu <xiangyang.wu@intel.com>
Reviewed-by: Junjie.Mao <junjie.mao@intel.com>
