Emulate VM0 resume from S3 state:
- reset BSP of VM0
- set the BSP entry to saved VM0 wakeup vec and set BSP to real mode
- start BSP
To match trampoline_spinlock release on ACRN Sx resume path, acquire
trampoline_spinlock if ACRN Sx enter fails.
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
enter_s3 is main function for ACRN to enter S3 state with following
process:
- pause vm0
- save the wakeup vec of vm0 and set wakeup vec of ACRN. So
resume from S3 will jump to ACRN wakeup
- offline APs
- update the main entry of trampoline to resume entry. After BSP
is resume from S3, it will jump to resume entry instead of AP
startup routine
- turn off vmx
- suspend devices
- enter S3.
exit S3 with following process:
- release trampoline_spinlock which is hold in trampoline code
- resume devices
- enable vmx
- update the main entry of trampoline to AP startup routine.
- online APs.
The following operations will be resume vm0 which will be added
in next patch.
Signed-off-by: Zheng Gen <gen.zheng@intel.com>
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Couple of small changes merged in this change:
- export main_entry, trampoline_spinlock and stop_cpus.
- change vm_resume() name to resume_vm()
- change resume_console_enable() name to resume_console()
- extend reset_vcpu to reset more fields of vcpu
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The S3 enter lowlevel routine saves the cpu context to memory
and enter S3 state
The S3 wakeup lowlevel routine restore cpu context and return.
Signed-off-by: Zheng Gen <gen.zheng@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
We will add host_pm.h for ACRN power manager header file.
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- add a commont head file include/common/irq.h, to include the common data
structure and APIs;
- move the common data struct and APIs from arch/x86/irq.h to the common header.
Signed-off-by: Yan, Like <like.yan@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
MISRA C required function return type should be consistented.
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
It is already assumed that ''long'' has 8-bytes, and thus there is no need to
use ULL to indicate a 8-byte unsigned constant.
This patch changes all ULL suffixes found in the hypervisor to UL.
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Snoop control is not supported in the graphics VT-d engine and so should
be disabled in the PTEs. Also enabled iommu for graphics in the
dmar_drhd array.
v2: removed disable iommu in handle_one_drhd and combined if cond.
v3: minor code review changes
v4: moved the snoop control change to ept.c to accomodate upstream changes
Signed-off-by: Anitha Chrisanthus <anitha.chrisanthus@intel.com>
Reviewed-by: Gong Zhipeng <zhipeng.gong@intel.com>
The for loop statement should work in both & and &&, but it seems
make more sense to have && when executing a condtion.
Signed-off-by: Yang, Yu-chu <yu-chu.yang@intel.com>
It could be also used by host as well. So we remove GUEST from
MACRO name
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <Eddie.dong@intel.com>
phys_cpu_num is more popular than phy_cpu_num, update them
through command.
Signed-off-by: Xiangyang Wu <xiangyang.wu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Update the structure definition to define the address type
(HVA vs HPA vs GPA) explicitly.
Convert address to HVA before access the GPA/HPA type of address.
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Before we set the page table, we should know the attribute. So
move configure the page table attribute outside of modify_paging.
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
MISRA C doesn't allowed negative shift, changed any potential signed value
to unsigned value.
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In the hypervisor, physical cpu id is defined as "int" or "uint32_t"
type in the hypervisor. So there are some sign conversion issues
about physical cpu id (pcpu_id) reported by static analysis tool.
Sign conversion violates the rules of MISRA C:2012.
In this patch, define physical cpu id as "uint16_t" type for all
modules in the hypervisor and change related codes. The valid
range of pcpu_id is 0~65534, INVALID_PCPU_ID is defined to the
invalid pcpu_id for error detection, BROADCAST_PCPU_ID is
broadcast pcpu_id used to notify all valid pcpu.
The type of pcpu_id in the struct vcpu and vcpu_id is "int" type,
this will be fixed in another patch.
V1-->V2:
* Change the type of pcpu_id from uint32_t to uint16_t;
* Define INVALID_PCPU_ID for error detection;
* Define BROADCAST_PCPU_ID to notify all valid pcpu.
V2-->V3:
* Update comments for INVALID_PCPU_ID and BROADCAST_PCPU_ID;
* Update addtional pcpu_id;
* Convert hexadecimals to unsigned to meet the type of pcpu_id;
* Clean up for MIN_PCPU_ID and MAX_PCPU_ID, they will be
defined by configuration.
Note: fix bug in the init_lapic(), the pcpu_id shall be less than 8,
this is constraint by implement in the init_lapic().
Signed-off-by: Xiangyang Wu <xiangyang.wu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Per MISRA C, operands to bit-wise operations should have unsigned
types. However, C99 prioritizes to use signed integers for hexadecimal constants
without the 'U' suffixes, leading to tons of bit operations on signed integers.
This patch series add the 'U' suffixes to the constants which are used in bit
operations, and add the intended width of these integers when applicable
(i.e. the target value is at least 32-bit wide) to avoid functional differences
due to signed vs. unsigned extensions. The rule of thumb is:
'0' for signed char/short/int
'0U' for unsigned char/short/int
'0L' for signed long (should be 64-bit)
'0UL' for unsigned long (should be 64-bit)
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
MISRA C explicit required expression should be boolean when
in branch statements (if,while...).
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- when init, cr0 & cr4 should read from VMCS
- when world switch, cr0/cr4 read shadow should also be save/restore
v2:
- use context->vmx_cr0/cr4 to save/restore VMX_GUEST_CR0/CR4
- use context->cr0/cr4 to save/restore VMX_CR0/CR4_READ_SHADOW
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
MISRA C required that return value should be used, missing for it should
add "(void)" prefix before the function call.
Some function can be declared without return value to avoid this problem.
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- this patch is to enable SMEP in hypervisor, SMEP protects
guests' memory from supervisor-mode instruction fetches,
in other words, hypervisor which operating in supervisor
mode can't fetch instructions from (guests' memory)
linear addresses that are accessible in user mode.
Signed-off-by: Yonghua Huang <yonghua.huang@intel.com>
Move common stuff, like ptdev entry and softirq, to new ptdev.c
Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This change is to fix a guest vm hang issue at vm reset, especially easy to
be seen when it's a watchdog timeout reset.
vlapic_init create and init vlapic.vlapic_timer without deleting the
timer from cpu_times list, which breaks the list, results in a timer remains
with callback points to an invalid location.
Acked-by: Eddie Dong <eddie.dong@intel.com>
Signed-off-by: Yan, Like <like.yan@intel.com>
These two functions will be called when ACRN enter/exit S3.
Signed-off-by: Zheng Gen <gen.zheng@intel.com>
Signed-off-by: Yan Like <like.yan@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
They will be called when acrn enter S3.
NOTE: it's only needed for native BSP because all APs are offline.
Signed-off-by: Zheng Gen <gen.zheng@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The hardcoded CPU Px Cx table should be read only, so set them to static
and const for safety.
Signed-off-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The pm S state data is from host ACPI info and needed for S3/S5
implementation.
Signed-off-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Currently irq and vector numbers are used inconsistently.
* Sometimes vector or irq ids is used in bit operations, indicating
that they should be unsigned (which is required by MISRA C).
* At the same time we use -1 to indicate an unknown irq (in
common_register_handler()) or unavailable irq (in
alloc_irq()). Also (irq < 0) or (vector < 0) are used for error
checking. These indicate that irq or vector ids should be signed.
This patch converts irq and vector numbers to unsigned 32-bit integers, and
replace the previous -1 with IRQ_INVALID or VECTOR_INVALID. The branch
conditions are updated accordingly.
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- Also update all the references that need the absolute HPA with the
actual load addresses
- Save the trampoline code address to trampline_start16_paddr
Signed-off-by: Zheng, Gen <gen.zheng@intel.com>
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Acked-by: Xu, Anthony <anthony.xu@intel.com>
- For UEFI boot, allocate memory for trampoline code in ACRN EFI,
and pass the pointer to HV through efi_ctx
- Correct LOW_RAM_SIZE and LOW_RAM_START in Kconfig and bsp_cfg.h
- use trampline_start16_paddr instead of the hardcoded
CONFIG_LOW_RAM_START for initial guest GDT and page tables
Signed-off-by: Zheng, Gen <gen.zheng@intel.com>
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Acked-by: Xu, Anthony <anthony.xu@intel.com>
in real mode part, add extra pointers for page tables and long jump buffer
so it's possible for HV code to patch the relocation offset
in long mode part, use absolute addressing when referring HV symbols,
and use relative addressing for symbols within trampoline code
Signed-off-by: Zheng, Gen <gen.zheng@intel.com>
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Acked-by: Xu, Anthony <anthony.xu@intel.com>
emalloc_for_low_mem() is used if CONFIG_EFI_STUB is defined.
e820_alloc_low_memory() is used for other cases
In either case, the allocated memory will be marked with E820_TYPE_RESERVED
Signed-off-by: Zheng, Gen <gen.zheng@intel.com>
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Acked-by: Xu, Anthony <anthony.xu@intel.com>
- vm_gva2gpa is same as gva2gpa, so replace it with gva2gpa directly.
- remove dead usage of vm_gva2gpa in emulate_movs.
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
there are data transfer between guest virtual space(GVA) & hv(HVA), for
example, guest rip fetching during instruction decoding.
GVA is address continuous, but its GPA could be only 4K page address
continuous, this patch adds copy_from_gva & copy_to_gva functions by
doing page walking of GVA to avoid address breaking during accessing GVA.
v2:
- modify API interface based on new gva2gpa function, err_code added
- combine similar code with inline function _copy_gpa
- change API name from vcopy_from/to_vm to copy_from/to_gva
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
This patch drops "#include <bsp_cfg.h>" and include the generated config.h in
CFLAGS for the configuration data.
Also make sure that all configuration data have the 'CONFIG_' prefix.
v4 -> v5:
* No changes.
v3 -> v4:
* Add '-include config.h' to hypervisor/bsp/uefi/efi/Makefile.
* Update comments mentioning bsp_cfg.h.
v2 -> v3:
* Include config.h on the command line instead of in any header or source to
avoid including config.h multiple times.
* Add config.h as an additional dependency for source compilation.
v1 -> v2:
* No changes.
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
Cleanup "cpu_secondary_xx" in the symbols/section/functions/variables
name in trampline code.
There is item left: the default C entry is Ap start c entry. Before
ACRN enter S3, the c entry will be updated to high level S3 C entry.
So s3 resume will go s3 resume path instead of AP startup path.
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Zheng Gen <gen.zheng@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Acked-by: Eddie Dong <Eddie.dong@intel.com>
Linux commit edfe63ec97ed ("x86/mtrr: Fix Xorg crashes in Qemu sessions")
disables PAT feature if MTRR is not enabled. This patch does partial
emulation of MTRR to prevent this from happening: enable fixed-range
MTRRs and disable virable range MTRRs
By default IA32_PAT MSR (SDM Vol3 11.12.4, Table 11-12) doesn't include
'WC' type. If MTRR is disabled from the guests, Linux doesn't allow
writing IA32_PAT MSR so WC type can't be enabled. This creates some
performance issues for certian applications that rely on WC memory type.
Implementation summary:
- Enable MTRR feature: MTRRdefType.E=1
- Enable fixed range MTRRs: MTRRCAP.fix=1, MTRRdefType.FE=1
- For simplicity, disable variable range MTRRs: MTRRCAP.vcnt=0.
It's expected that this bit is honored by the guests and they won't
change the guest memory type through variable MTRRs.
Signed-off-by: bliu11 <baohong.liu@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- Add PAGING_REQUEST_TYPE_MODIFY_MT memory map request type
- Update map_mem_region() to allow modifying the memory type related
fields in a page table entry
- Add ept_update_mt()
- add modify_mem_mt() for both EPT and MMU
Signed-off-by: Zide Chen <zide.chen@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- export start_cpus to start/online APs.
- Add stop_cpus to offline APs.
- Update cpu_dead to decrement running cpus number and do cleanup
for AP down
Signed-off-by: Yin Fegnwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <Eddie.dong@intel.com>
To handle cpu down/up dynamically, arcn needs to support vmx off/on
dynamically. Following changes is introduced:
vmx_off will be used when down AP. It does:
- vmclear the mapped vcpu
- off vmx.
exec_vmxon_instr is updated to handle start and up AP both. It does
- if vmx was on on AP, load the vmxon_region saved. Otherwise,
allocate vmxon_region.
- if there is mapped vcpu, vmptrld mapped vcpu.
Signed-off-by: Zheng Gen <gen.zheng@intel.com>
Signed-off-by: Yin Fegnwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <Eddie.dong@intel.com>
Per performance consideration, we don't flush vcpu context when doing
vcpu swithing (because it's only swithing between vcpu and idle).
But when enter S3, we need to call vmclear against all vcpus attached
to APs. We need to know which vcpu is attached with which pcpu.
This patch introduced API to get vcpu mapped to specific pcpu.
Signed-off-by: Yin Fegnwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <Eddie.dong@intel.com>
