The vcpu state is initialized outside of init_guest_state:
- SOS BSP state is initialized in SOS loader
- UOS BSP state is initialized in UOS loader
- AP state is initialized during SIPI signal emulation
We could make init_guest_state only update the vcpu state
to VMCS structure.
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
NOTE: this patch is only workaround patch for UOS BSP state init.
Eventually, the DM will call hypercall to init UOS BSP state.
We use this workaround patch here to simplify the init_guest_state.
Will make the caller of init_guest_state calls init_guest_vmx
directly.
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
We move the SOS BSP state init to vm loader and drop
function init_guest_context_vm0_bsp.
Update the definition of vm0_boot_context to fix code
violations.
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
With reset_vcpu_regs as pre-condition, we only need to set
cs_selector and cs_base for AP.
We call set_ap_entry in two places:
1. When emulation AP SIPI
2. When sos BSP resume from S3. The BSP is resumed to real
mode with entry set to wakeup_vec. We call set_ap_entry
API here with entry twisted from wakeup_vec.
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This reset_vcpu_regs function will reset the vcpu registers to
default value: realmode with entry 0xFFFFFFF0
Make call to reset_vcpu_regs during create_vcpu and reset_vcpu
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
If defined CONFIG_VM0_DESC, HV will use predefined vm0_desc
to config VM0,now it is unneccessary, then remove these code.
Tracked-On: #861
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
Reviewed-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
TR selector initianlization using pre-defined HOST_GDT_RING0_CPU_TSS_SEL
rather than loading from register. Instead calculating real base address
of TSS based on TR selector and gdt, getting it from per cpu data.
Tracked-On: #1394
Signed-off-by: Yang, Yu-chu <yu-chu.yang@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
- add missing brackets for 'if/else' statements based on MISRA-C
requirements
v1 -> v2:
* add brackets for each conditions in 'if' statements to improve
the readability
* modify 'ptdev_init' to make the logic clearer
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
In the current hypervisor design, when HPA is not
found for the specified gpa by calling gpa2hpa or
local_gpa2hpa, 0 will be returned as a error code,
but 0 may be a valid HPA for vm0; error checking
is missed when invoking gpa2hpa or local_gpa2hpa;
when invoking lookup_address, the caller guarantees
that parameter pointer pml4_page and pointer pg_size
is not NULL.
If local_gpa2hpa/gpa2hpa returns a invalid HPA,
it means that this function fails to find the
HPA of the specified gpa of vm. If local_gpa2hpa/gpa2hpa
return value is a valid HPA, it means that this
function have found the HPA of the specified gpa of vm.
Each valid vm's EPTP is initialized during vm creating,
vm's EPTP is valid until this vm is destroyed. So the caller
can guarantee parameter pointer pml4_page is not NULL.
The caller uses a temporary variable to store page size.
So the caller can guarantee parameter pointer pg_size
is not NULL.
In this patch, define a invalid HPA for gpa2hpa and
local_gpa2hpa;add some error checking when invoking
local_gpa2hpa/gpa2hpa;add precondition for lookup_address
function and remove redundant error checking.
V1-->V2:
Define INVALID_HPA as a invalid HPA for gpa2hpa
and local_gpa2hpa;
Updated related error checking when invoking
gpa2hpa or local_gpa2hpa;
V2-->V3:
Add some debug information if specified gpa2hpa
mapping doesn't exit and ept_mr_del is called;
Update INVALID_HPA definition easier to be reviewed.
V3-->V4:
Add vm->id and gpa into pr_error;
Add precondition to ept_mr_del to cover [gpa,gpa+size)
unmapping case.
V4-->V5:
Update comments;
Update pr_error message.
Tracked-On: #1258
Signed-off-by: Xiangyang Wu <xiangyang.wu@linux.intel.com>
Reviewed-by: Li, Fei1 <fei1.li@intel.com>
- improve the readability of ept_cap_detect
right shift 32 bits of msr_val to check bits 63:32 of msr_val
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
1.Function return type inconsistent
2.cast on a constant value
V1->V2 add () to return type
V2->V3 keep the sbuf_get and sbuf_put return code
Tracked-On: #861
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Expression should be boolean immediate before 'if','while' key-words.
V1->V2 add () to bool expression
Tracked-On: #861
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Page table operation would not fault except:
1. the hypervisor it out of memory to allcate a page for page table operation
2. there is a bug with page table operation in hypervisor or devicemodle
While we assue that these would not happened in our platform when release, so
there is no need to check whether there is a fault with page table operation. However,
for debug version, we would panic the hypervisor if we can't meet the conditions really.
Tracked-On: #1124
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Sometimes we need to invalidate cached translation information for guest
when change some bits in CR0/CR4 which related to paging. Here're two cases:
1. If there change some bits to enable/disable paging (mode) or access rights.
For CR0: PG/WP/CD/NW; For CR4: PGE/PSE/PAE/SMEP/SMAP/PKE
2. When guest using PAE paging, we should reload the PDPTE registers sometimes,
detail in SDM Vol 3 Chap 4.4.1 and Chap 4.11.1
Tracked-On: #1379
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
add is_long_mode to check whether the processor is operating in IA-32e mode
add is_paging_enabled to check whether paging is enabled
add is_pae to check whether physical address extension is enabled.
Tracked-On: #1379
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Loop the global vm_array[] instead of the vm_list.
Tracked-On: #861
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
-- Replace dynamic memory allocation with static memory
-- Remove the parameter check if the vm is NULL
Tracked-On: #861
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This patch is ready for next one,we will switch from
pointer to embedded structures such as structure vcpu,
some const qualifiers can be improper and raise compilation errors,
this patch remove const qualifier for struct vm.
Tracked-On: #861
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
add active_polarity to decide what signal to call: GSI_SET_LOW, GSI_SET_HIGH,
GSI_FALLING_PULSE or GSI_RAISING_PULSE.
Tracked-On: #1269
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
There are chances that names with leading underscore declared by
developers are conflict with the ones reserved for the compiler.
What this patch does:
- rename these functions/variables/macros starting with
underscore to avoid such unintentational mistakes.
- remove gpr.h without any contents
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Commit 6085781 replaced __emalloc() with a call to uefi allocate_page()
and allows UEFI FW to allocate memory for hypervisor from high to low
address below 4GB. However, this change triggers an issue (might be
memory corruption), in turn, PXE boot cannot work.
Since root cause the issue might take some time, the PXE boot issue
blocks auto-test, we disable hypervisor relocation by default for the
time being in config option, and users can enable it by themselves. In
the following weeks, if we root cause the issue, we can re-enable
relocation feature.`
Tracked-On: #1371
Signed-off-by: Chaohong guo <chaohong.guo@intel.com>
The patch is replace irq_static_mappings type from 2 dimension array
into structure and one dimension array.
Tracked-On: #1375
Signed-off-by: Wei Liu <weix.w.liu@intel.com>
Reviewed-by: Jason Chen <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
For SBL platform, struct acrn_vcpu_regs are used as boot context.
Now the boot_ctx is only for UEFI platform. Rename struct boot_ctx
to efi_context.
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Add struct acrn_vcpu_regs and make struct boot_ctx based on
struct acrn_vcpu_regs.
vm0_boot_context is also changed from struct boot_ctx to struct
acrn_vcpu_regs.
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
We will define the hypercall to set acrn vcpu registers and
move this struct to acrn_commmon.h for reference from hypercall
parameter definition.
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
struct seg_desc_vmcs is actually for segment descriptor vmcs fields.
Change its name to vmcs_seg_field
Tracked-On: #1231
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The printf related functions have been called by other various
functions, e.g. get_ptdev_info(), get_ioapic_info, etc. The patch
is used to clear up the usage.
Tracked-On: #861
Signed-off-by: Junjun Shan <junjun.shan@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Current free_irq sequence will release vector first, then use the
released vector to free irq number.It will cause irq leak for MSI IRQ.
At present, there is no one to free the irqs which in irq_static_mappings,
So this patch will only make sure free non-gsi irqs.
Tracked-On: #1359
Signed-off-by: Wei Liu <weix.w.liu@intel.com>
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Timer IRQ was static mapped IRQ.
It should not be cleaned up when one of the cpu dead.
The patch will retain the timer irq.
Tracked-On: #1359
Signed-off-by: Wei Liu <weix.w.liu@intel.com>
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
MISRA-C states that a void procedure used in expressions is dangerous.
This patch removes the improper 'return' when calling the void procedure
'send_dest_ipi'.
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- replace vpic/vioapic_xassert_irq() APIs
with vpic/vioapic_set_irq()
- unify the description of IRQ/PIN state in vpic. & vioapic.c
Tracked-On: #861
Signed-off-by: Yonghua Huang <yonghua.huang@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
1. The wrong operand size is assigned in instruction decode phase
if the operand size is 1 byte.
According to the SDM, the bit 0(w bit) of opcode should be checked
first to detect whether the operand size is 1 byte. Then, check
whether there is prefix to overwrite the default operand size.
The original instruction decode doesn't care about the operand
size. But do opsize fixup during instruction emulation phase.
With ACRN we need operand size packed to ioreq and send to DM
after instruction decode.
2. We should always touch the GPA by following opsize to avoid side
effect (especially when GPA is for a MMIO).
Tracked-On: #1337
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
This patch fixes the following issue pointed by Xiangyang and Junjie.
There are some macro arguments acting as formal parameter names.
Drop such arguments since they make no difference to the expanded
implementation and they might confuse some developers.
Here is an example.
'ptr' is dropped in this patch, which is acting as a formal parameter
name and make no difference to the expanded implementation.
-#define build_atomic_load(name, size, type, ptr) \
+#define build_atomic_load(name, size, type) \
static inline type name(const volatile type *ptr) \
{ \
type ret; \
asm volatile("mov" size " %1,%0" \
: "=r" (ret) \
: "m" (*ptr) \
: "cc", "memory"); \
return ret; \
}
Some minor coding style fixes are also included in this patch.
- use TAB for the alignment rather than mixing TAB with space
- fix some typo in the comments
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
merge hv_lib.h and hypervisor.h into one file
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Now, the warning is promoted to error. The warning break ACRN
release build.
Tracked-On: #1343
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Fix the compiler warning and turn on the flag to make
compiler warning as compiler error.
Tracked-On: #1343
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Currently, hypervisor doesn't have the information of pci devices
on the platform when init. In order to reduce boot time and memory
usage, hypervisor set a bus limitation for pci devices when init vtd.
However, maximum bus number varies on different platforms.
This patch add a config for the bus limitation.
Set 0xF as the default value for the maximum bus number on sbl platform.
Set 0xFF as the default value for the maximum bus number on uefi platfrom.
Tracked-On: #1339
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
MISRA-C requires that there should be no unused parameters in
functions.
This patch removes the unused parameters that is not being used
unconditionally.
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Currently the acrn-hypervisor is using the PLATFORM_IPI vector to notify
the sos_kernel. And then sos_kernel will handle the notification from acrn
hypervisor in PLATFORM_IPI ISR. But as the PLATFORM_IPI ISR can be registered
by the other modules, it will have the conflict when trying to register
acrn intr ISR. So the HYPERVISOR_CALLBACK_VECTOR will be used instead.
In order to switch the notification vector from PLATFORM_IPI to
HYPERVISOR_CALLBACK_VECTOR, one API is added so that sos can configure
the up-notifier interrrupt vector.
Tracked-On: https://github.com/projectacrn/acrn-hypervisor/issues/1325
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
MISRAC checks whether a variable is assigned a value not used in
all branches of a program. Var value which is unused on all paths
can be removed with a consequent improvement in the readability
and efficiency of the code. This patch is used to fix these
violations.
Tracked-On: #861
Signed-off-by: Junjun Shan <junjun.shan@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>
MISRA-C requires that every switch case shall be terminated with break
to avoid the unintentional fall through.
The code will become redundant if we enforce this rule.
So, we will keep the current implementation for the following two cases.
1. The fall through is intentional.
2. The function is returned in the switch case.
If we decide to eliminate the mutiple returns in one function later,
this case would be handled properly at that time.
What this patch does:
- add the mssing break for the default case
- add the pre condition for some functions and remove the corresponding
panic which will never happen since the function caller could guarantee
the pre condition based on the code implementation
v1 -> v2:
* remove the redundant cases above default in 'vlapic_get_lvtptr'
* add the similar pre condition for 'lvt_off_to_idx' as
'vlapic_get_lvtptr' since all the function callers could guarantee it
* remove the assertion in 'lvt_off_to_idx' since the pre condition
could guarantee that the assertion will never happen
* add the similar pre condition for 'vpic_set_irqstate' as
'vioapic_set_irqstate' since all the function callers could guarantee it
* remove the assertion in 'vpic_set_irqstate' since the pre condition
could guarantee that the assertion will never happen
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Reviewed-by: Junjie Mao <junjie.mao@intel.com>
Convert GSI_MASK_IRQ and GSI_UNMASK_IRQ to inline functions.
v1 -> v2:
After changing GSI_MASK_IRQ and GSI_UNMASK_IRQ from MACROs to functions,
'gsi_(mask|unmask)_irq' are the exposed APIs and 'irq_gsi_mask_unmask'
becomes internal.
In order to reflect this change,
- change 'irq_gsi_mask_unmask' as internal function in ioapic.c
- declare 'gsi_(mask|unmask)_irq' in ioapic.h
- define 'gsi_(mask|unmask)_irq' in ioapic.c
Tracked-On: #861
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Reviewed-by: Junjie Mao <junjie.mao@intel.com>
The hypercall can be used by SOS/DM to monitor UOS
pass-thru devices' interrupt data; and if an "interrupt
storm" happens in UOS, it can be used to delay the UOS
PTdev interrupt's injection for sometime.
The two functions are implemented by one hypercall with
two sub-commands, and with the data/params in the buffer.
Tracked-On: #866
Signed-off-by: Minggui Cao <minggui.cao@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
This patch is to enhance the feature of "interrupt storm
mitigation"; when interrupt storm happens on one UOS
it should keep as smaller effect to SOS as possible.
Add variables in PTdev entry & VM, used to record one
UOS's pass-thru devices' interrupt data; add a function
to collect the data.
Also add a timer used to delay UOS pass-thru devices'
interrupt injection if an "interrupt storm" detected
by SOS.
Tracked-On: #866
Signed-off-by: Minggui Cao <minggui.cao@intel.com>
Reviewed-by: Li Fei1 <fei1.li@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Current code enables IOMMU from init_iommu_vm0_domain which is called
from vm0/sos boot sequence. For partition mode VMs, this is not called as
VMs are numbered from 1.
This patch adds support to initialize root table pointer for each IOMMU
and enable all IOMMUs from init_iommu. Hence IOMMUs are enabled even
though ACRN does not boot vm0.
This patch also has changes to fix#1270.
Tracked-On: #1246#1270
Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.com>
