For the IGD device the opregion addr is returned by reading the 0xFC config of
0:02.0 bdf. And the opregion addr is required by GPU driver.
The opregion_addr should be the GPA addr.
When the IGD is assigned to pre-launched VM, the value in 0xFC of igd_vdev is
programmed into with new GPA addr. In such case the prelaunched VM reads
the value from 0xFC of 0:02.0 vdev.
But for the Service VM, the IGD is initialized by using the same policy as other PCI
devices. We only initialize the vdev_head_conf(0x0-0x3F) by checking the
corresponding pbdf. The remaining pci_config_space will be read by
leveraging the corresponding pdev. But as the above code doesn't handle the
scenario for Service VM, it causes that the Service VM fails to
read the 0xFC config_space for IGD vdev.
Then the i915 GPU driver in SOS has some issues because of incorrect 0xFC
pci_conf_space.
This patch initializes offset 0xfc of CFG space of IGD for Service VM,
it is simple and can cover post-launched VM too.
Tracked-On: #6387
Signed-off-by: Liu,Junming <junming.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In the current hypervisor, only support at most two legacy vuarts
(COM1 and COM2) for a VM, COM1 is usually configured as VM console,
COM2 is configured as communication channel of S5 feature.
Hypervisor can support MAX_VUART_NUM_PER_VM(8) legacy vuart, but only
register handlers for two legacy vuart since the assumption (legacy
vuart is less than 2) is made.
In the current hypervisor configurtion, io port (2F8H) is always
allocated for virtual COM2, it will be not friendly if user wants to
assign this port to physical COM2.
Legacy vuart is common communication channel between service VM and
user VM, it can work in polling mode and its driver exits in each
guest OS. The channel can be used to send shutdown command to user VM
in S5 featuare, so need to config serval vuarts for service VM and one
vuart for each user VM.
The following changes will be made to support at most
MAX_VUART_NUM_PER_VM legacy vuarts:
- Refine legacy vuarts initialization to register PIO handler for
related vuart.
- Update assumption of legacy vuart number.
BTW, config tools updates about legacy vuarts will be made in other
patch.
v1-->v2:
Update commit message to make this patch's purpose clearer;
If vuart index is valid, register handler for it.
Tracked-On: #6652
Signed-off-by: Xiangyang Wu <xiangyang.wu@intel.com>
Acked-by: Eddie Dong <eddie.dong@Intel.com>
It's difficult to configure CONFIG_HV_RAM_SIZE properly at once. This patch
not only remove CONFIG_HV_RAM_SIZE, but also we use ld linker script to
dynamically get the size of HV RAM size.
Tracked-On: #6663
Signed-off-by: Fei Li <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This patch implements a separate path for L2 VMEntry in run_vcpu(),
which has several benefits:
- keep run_vcpu() clean, to reduce the number of is_vcpu_in_l2_guest()
statements:
- current code has three is_vcpu_in_l2_guest() already.
- supposed to have another 2 statement so that nested VMEntry won't
hit the "Starting vCPU" and "vCPU launched" pr_info() and a few
other statements in the VM launch path.
- save few other things in run_vcpu() that are not needed for nested.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@Intel.com>
In run time, one vCPU won't read or write a register on other vCPUs,
thus we don't need the LOCK prefixed instructions on reg_cached and
reg_updated.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
move the bitmap test call out of vcpu_inject_exception(), then we call
the expensive bitmap_test_and_clear_lock() only pending_req_bits is
non-zero and call vcpu_inject_exception() only if needed.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@Intel.com>
In very large number of VM extis, the VM-exit instruction length could be
zero, and it's no need to update VMX_GUEST_RIP.
Some examples:
- all external interrupt VM exits in non LAPIC passthru setup.
- for all the nested VM-exits that are reflecting to L1 hypervisor.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- wrap a new function exec_vmentry() to reduce code duplication.
- remove exec_vmread(VMX_GUEST_RSP) since ACRN doesn't need to know
guest RSP in run time.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Change "uint64_t vmcs_hva" to "void *vmcs_hva" in the input argument,
list, so that no type casting is needed when calling them from pointers.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
update board name from nuc7i7dnb to nuc11tnbi5 in makefile because
we have removed the nuc7i7dnb board folder, and also update the
scenario name from industry to shared to fix "make all" build issue.
Tracked-On: #6315
Signed-off-by: Kunhui-Li <kunhuix.li@intel.com>
In current design, when pass-thru dev,
for the PIO bar, need to ensure the guest PIO start address
equals to host PIO start address.
But malicious guest may reprogram the PIO bar,
then hv will pass-thru the reprogramed PIO address to guest.
This isn't safe behavior.
When guest tries to reprogram pass-thru dev PIO bar,
inject #GP to guest directly.
Tracked-On: #6508
Signed-off-by: Liu,Junming <junming.liu@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Fei Li <fei1.li@intel.com>
In current design, when pass-thru dev,
for the PIO bar, need to ensure the guest PIO start address
equals to host PIO start address.
Then set the VMCS io bitmap to pass-thru the corresponding
port io to guest for performance.
ACRN-DM and acrn-config should ensure the identical mapping of PIO bar.
If ACRN-DM or acrn-config failed to achieve this,
we should deny the launch of VM
Tracked-On: #6508
Signed-off-by: Liu,Junming <junming.liu@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Fei Li <fei1.li@intel.com>
In the commit of 4e1deab3d9, we changed the
init sequence that init paging first and then init e820 because we worried
about the efi memory map could be beyond 4GB space on some platform.
After we double checked multiboot2 spec, when system boot from multiboot2
protocol, the efi memory map info will be embedded in multiboot info so it
is guaranteed that the efi memory map must be under 4GB space. Consider that
the page table will be allocated in free memory space in future, we have
to change the init sequence back that init e820 first and then init paging.
If we need to support other boot protocol in future that the efi memory map
might be put beyond 4GB, we could have below options:
1. Request bootloader put efi memory map below 4GB;
2. Call EFI_BOOT_SERVICES.GetMemoryMap() before ExitBootServices();
3. Enable a early 64bit page table to get the efi memory map only;
Tracked-On: #5626
Signed-off-by: Victor Sun <victor.sun@intel.com>
This patch changes the size of vvmcs[] array from 1 to
PER_VCPU_ACTIVE_VVMCS_NUM, and actually enables multiple active VMCS12
support in ACRN. The basic operations:
- if L1 VMPTRLDs a VMCS12 without previously VMCLEAR the current
VMCS12, ACRN no longer unconditionally flushes the current VMCS12
back to L1. Instead, it tries to keep both the current and the newly
loaded VMCS12 in the nested->vvmcs[] array, unless:
- if there is no more available vvmcs[] entry, ACRN flushes one active
VMCS12 to make room for this new VMCS12.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Some processors don't support VMX_PROCBASED_CTLS_TERTIARY bit
and VMX_PROCBASED_CTLS2_UWAIT_PAUSE bit in MSRs
(IA32_VMX_PROCBASED_CTLS & IA32_VMX_PROCBASED_CTLS2),
HV will output error log which will cause confusion,
change the log level from pr_err to pr_info.
Tracked-On: #6397
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
This patch adds a new priority based scheduler to support
vCPU scheduling based on their pre-configured priorities.
A vCPU can be running only if there is no higher priority
vCPU running on the same pCPU.
Tracked-On: #6571
Signed-off-by: Jie Deng <jie.deng@intel.com>
The current config.mk uses the variable BOARD_FILE as the path to the board
XML when generating an unmodified copy of configuration files for
comparison, which is incorrect. The right variable is HV_BOARD_XML which is
the path to the copy of board XML that is actually used for the build.
This patch corrects the bug above.
In addition, this patch also skips binary files (which are not meant to be
edited manually) when calculating the differences.
Tracked-On: #6592
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
For local APIC passthrough case, EOI would not trigger VM-exit. So virtual
'Remote IRR' would not be updated. Needs to read physical IOxAPIC RTE to
update virtual 'Remote IRR' field each time when guest wants to read I/O
REDIRECTION TABLE REGISTERS
Tracked-On: #5923
Signed-off-by: Fei Li <fei1.li@intel.com>
In local APIC passthrough case, when devices triggered a INTx interrupt, this
interrupt would be delivered to vCPU directly. For this case, need to set the
virtual vector in
the 'Interrupt Vector' field of physical IOxAPIC I/O REDIRECTION TABLE REGISTER
(bits 7:0) and 'Vector' field of vt-d Interrupt Remapping Table Entry (IRTE)
for Remapped Interrupts.
Assumption:
(a) IOAPIC pins won't be shared between LAPIC PT guest and other guests;
(b) The guest would not trigger this IRQ before it switched to x2 APIC mode.
Tracked-On: #5923
Signed-off-by: Zide Chen <zide.chen@intel.com>
check some essential vmx capablility,
will panic if processor doesn't support it.
Tracked-On: #6584
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In current RDT code, if CDP is configured, L2/L3 resources' num_closids calculation
is wrong:
res_cap_info[res].num_closids = (uint16_t)((edx & 0xffffU) >> 1U) + 1U;
Should be:
res_cap_info[res].num_closids = (uint16_t)((edx & 0xffffU) >> 1U + 1) >> 1U;
Aslo, in order to enable CDP system-wide, need to enable the CDP bit (bit 0) on all pcpus,
not just on pcpu 0.
Tracked-On: #5917
Signed-off-by: dongshen <dongsheng.x.zhang@intel.com>
Rename the clos_max field in struct rdt_info to num_closids
Rename variable valid_clos_num to common_num_closids and make it static
Tracked-On: #5917
Signed-off-by: dongshen <dongsheng.x.zhang@intel.com>
Normalize hypervisor help command output format, remove the 10 lines
limit for one screen, fix the misspelled words.
Tracked-On: #5112
Signed-off-by: Liu Long <long.liu@intel.com>
Reviewed-by: VanCutsem, Geoffroy <geoffroy.vancutsem@intel.com>
These dirty flags are supposed to be per VMCS12, so move them from the
per vCPU acrn_nested struct to the newly added acrn_vvmcs struct.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This variable represents the L1 GPA of the current VMCS12. But it's
no longer needed in the multiple active VMCS12 case, which uses the
following variables for this purpose.
- nested->current_vvmcs refers to the vvmcs[] entry which contains the
cached current VMCS12, its associated VMCS02, and other context info.
- nested->current_vvmcs->vmcs12_gpa refers to the L1 GPA of this
current VMCS12.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Add an array of struct acrn_vvmcs to struct acrn_nested, so it is
possible to cache multiple active VMCS12s.
This patch declares the size of this array to 1, meaning that there is
only one active VMCS12. This is to minimize the logical code changes.
Add pointer current_vvmcs to struct acrn_nested, which refers to the
current vvmcs[] entry. In this patch, if any VMCS12 is active, it
always points to vvmcs[0].
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
No any logical changes, this patch is preparing for multiple active
VMCS12 support.
- currently it's easy to get the vmcs12 pointer from the vcpu pointer.
In multiple active vmcs12 case, we need to explicitly add "struct
acrn_vmcs12 *vmcs12" to certain APIs' input argument list, in order to
get the desired vmcs12 pointer.
- merge flush_current_vmcs12() into clear_vmcs02() for multiple reasons:
a) it's called only once; b) we don't wrap the opposite operation
(loading vmcs12) in an API; c) this API has simple and clear logic.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
By changing the way to assign L1 VPID from bottom-up to top-down,
the possibilities for VPID conflicts between L1 and L2 guests are
small.
Then we can flush VPID just in case of conflicting.
Tracked-On: #6289
Signed-off-by: Anthony Xu <anthony.xu@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Before checking whether a PCI device is a Multi-Function Device or not, we need
make sure this PCI device is a valid PCI device. For a valid PCI device, the
'Header Layout' field in Header Type Register must be 000 0000b (Type 0 PCI device)
or 000 0001b (Type 1 PCI device).
So for a valid PCI device, the Header Type can't be 0xff.
Tracked-On: #4134
Signed-off-by: Fei Li <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In run time, it's rare for L1 to write to the intercepted non host-state
VMCS fields, and using multiple dirty flags is not necessary.
This patch uses one single dirty flag to manage all non host-state VMCS
fields. This helps to simplify current code and in the future we may
not need to declare new dirty flags when we intercept more VMCS fields.
Tracked-On: #5923
Signed-off-by: Zide Chen <zide.chen@intel.com>
Currently vmptrld_vmexit_handler() doesn't sync VMX_EPT_POINTER_FULL
from vmcs12 to vmcs02, instead it sets gpa_field_dirty and relies on
nested_vmentry() to sync EPTP in next nested VMentry.
This creates readability issue since all other intercepted VMCS fields
are synced in sync_vmcs12_to_vmcs02(). Another issue is that other
VMCS fields managed by gpa_field_dirty are repeatedly synced in both
vmptrld and nested vmentry handler.
This patch moves get_nept_desc() ahead of sync_vmcs12_to_vmcs02(), such
that shadow_eptp is allocated before sync_vmcs12_to_vmcs02() which
can sync EPTP properly.
BTW, in nested_vmexit_handler(), don't need to read from VMCS to get
the exit reason, since vcpu->arch.exit_reason has it already.
Tracked-On: #5923
Signed-off-by: Zide Chen <zide.chen@intel.com>
Add a couple of missing dependencies in the ACRN Makefiles:
1. 'acrn.bin' is required before the hypervisor can be installed
2. The 'acrn_mngr.h' needs to be installed ('tools-install') in
the build folder.
Tracked-On: #6360
Signed-off-by: Geoffroy Van Cutsem <geoffroy.vancutsem@intel.com>
While we have default values of configuration entries stated in the schema
of scenario XMLs, today we still require user-given scenario XMLs to
contain literally ALL XML nodes. Missing of a single node will cause schema
validation errors even though we can use its default value defined in the
schema.
This patch allows user-given scenario XMLs to ignore nodes with default
values. It is done by adding the missing nodes, all containing the defined
default values, to the input scenario XML when copying it to the build
directory. This approach imposes no changes to either the schema or
subsequent scripts in the build system.
Tracked-On: #6292
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Previously it is (falsely) assumed that the major_ver of 32-bit SMBIOS
entry point structure (which is called SMBIOS 2.1 in spec, or SMBIOS2 in code)
will have a value of 2 and major_ver of 64-bit SMBIOS (which is called SMBIOS
3.0 in spec, and SMBIOS3 in code) will have a value of 3. This turned out to be
wrong. This major_ver refers to the implemented doc revision, and 32-bit SMBIOS2
can have its major_ver to be 3 (current most recent implementation).
This patch removes the use of major_ver to distinguish between
SMBIOS2/3, and use a doc-defined anchor string instead.
Tracked-On: #6528
Signed-off-by: Yifan Liu <yifan1.liu@intel.com>
The locked btr instruction is expensive. This patch changes the
logic to ensure that the bitmap is non-zero before executing
bitmap_test_and_clear_lock().
The VMX transition time gets significant improvement. SOS running
on TGL, the CPUID roundtrip reduces from ~2400 cycles to ~2000 cycles.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Currently ACRN assumes firmware setup IA32_PAT correctly. This patch
explicitly initializes host IA32_PAT MSR according to ISDM Table 11-12.
Memory Type Setting of PAT Entries Following a Power-up or Reset.
ACRN creates host page tables based on PAT0 (WB) and PAT3 (UC).
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
is_lapic_pt_enabled() is called at least twice in one loop of the vCPU
thread, and it's called in vmexit_handler() frequently if LAPIC is not
pass-through. Thus the efficiency of this function has direct
impact to the system performance.
Since the LAPIC mode is not changed in run time, we don't have to
calculate it on the fly in is_lapic_pt_enabled().
BTW, removed the unused lapic_mask from struct acrn_vcpu_arch.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
For platforms that do not support XSAVES/XRSTORS instructions, like QEMU,
executing these instructions causes #UD.
This patch adds the check before the execution of XSAVES/XRSTORS instructions.
It also refines the logic inside rstore_xsave_area for the following reason:
If XSAVES/XRSTORS instructions are supported, restore XSAVE area if any of the
following conditions is met:
1. "vcpu->launched" is false (state initialization for guest)
2. "vcpu->arch.xsave_enabled" is true (state restoring for guest)
* Before vCPU is launched, condition 1 is satisfied.
* After vCPU is launched, condition 2 is satisfied because
is_valid_xsave_combination() guarantees that "vcpu->arch.xsave_enabled"
is consistent with pcpu_has_cap(X86_FEATURE_XSAVES).
Therefore, the check against "vcpu->launched" and "vcpu->arch.xsave_enabled"
can be eliminated here.
Tracked-On: #6481
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@Intel.com>
Use an unused MSR on host to save ACRN pcpu ID and avoid saving and
restoring TSC AUX MSR on VMX transitions.
Tracked-On: #6289
Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
Reviewed-by: Eddie Dong <eddie.dong@intel.com>
This feature is guarded under config CONFIG_SECURITY_VM_FIXUP, which
by default should be disabled.
This patch passthrough native SMBIOS information to prelaunched VM.
SMBIOS table contains a small entry point structure and a table, of which
the entry point structure will be put in 0xf0000-0xfffff region in guest
address space, and the table will be put in the ACPI_NVS region in guest
address space.
v2 -> v3:
uuid_is_equal moved to util.h as inline API
result -> pVendortable, in function efi_search_guid
recalc_checksum -> generate_checksum
efi_search_smbios -> efi_search_smbios_eps
scan_smbios_eps -> mem_search_smbios_eps
EFI GUID definition kept
Tracked-On: #6320
Signed-off-by: Yifan Liu <yifan1.liu@intel.com>
This patch renames the GUEST_FLAG_TPM2_FIXUP to
GUEST_FLAG_SECURITY_VM.
v2 -> v3:
The "FIXUP" suffix is removed.
Tracked-On: #6320
Signed-off-by: Yifan Liu <yifan1.liu@intel.com>
In ACRN RT VM if the lapic is passthrough to the guest, the ipi can't
trigger VM_EXIT and the vNMI is just for notification, it can't handle
the smp_call function. Modify vcpu_dumpreg function prompt user switch
to vLAPIC mode for vCPU register dump.
Tracked-On: #6473
Signed-off-by: Liu Long <long.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
- remove vcpu->arch.nrexits which is useless.
- record full 32 bits of exit_reason to TRACE_2L(). Make the code simpler.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
GSI of hcall_set_irqline should be checked against target_vm's
total GSI count instead of SOS's total GSI count.
Tracked-On: #6357
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This helps to improve performance:
- Don't need to execute VMREAD in vcpu_get_efer(), which is frequently
called.
- VMX_EXIT_CTLS_SAVE_EFER can be removed from VM-Exit Controls.
- If the value of IA32_EFER MSR is identical between the host and guest
(highly likely), adjust the VMX controls not to load IA32_EFER on
VMExit and VMEntry.
It's convenient to continue use the exiting vcpu_s/get_efer() APIs,
other than the common vcpu_s/get_guest_msr().
Tracked-On: #6289
Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
MAXIMUM_PA_WIDTH will be calculated from board information.
Tracked-On: #6357
Signed-off-by: Liang Yi <yi.liang@intel.com>
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Mask off support of 57-bit linear addresses and five-level paging.
ICX-D has LA57 but ACRN doesn't support 5-level paging yet.
Tracked-On: #6357
Signed-off-by: Liang Yi <yi.liang@intel.com>
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
It is used to specify the maximum number of EFI memmap entries.
On some platforms, like Tiger Lake, the number of EFI memmap entries
becomes 268 when the BIOS settings are changed.
The current value of MAX_EFI_MMAP_ENTRIES (256) defined in hypervisor
is not big enough to cover such cases.
As the number of EFI memmap entries depends on the platforms and the
BIOS settings, this patch introduces a new entry MAX_EFI_MMAP_ENTRIES
in configurations so that it can be adjusted for different cases.
Tracked-On: #6442
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
If SOS is using kernel 5.4, hypervisor got panic with #GP.
Here is an example on KBL showing how the panic occurs when kernel 5.4 is used:
Notes:
* Physical MSR_IA32_XSS[bit 8] is 1 when physical CPU boots up.
* vcpu_get_guest_msr(vcpu, MSR_IA32_XSS)[bit 8] is initialized to 0.
Following thread switches would happen at run time:
1. idle thread -> vcpu thread
context_switch_in happens and rstore_xsave_area is called.
At this moment, vcpu->arch.xsave_enabled is false as vcpu is not launched yet
and init_vmcs is not called yet (where xsave_enabled is set to true).
Thus, physical MSR_IA32_XSS is not updated with the value of guest MSR_IA32_XSS.
States at this point:
* Physical MSR_IA32_XSS[bit 8] is 1.
* vcpu_get_guest_msr(vcpu, MSR_IA32_XSS)[bit 8] is 0.
2. vcpu thread -> idle thread
context_switch_out happens and save_xsave_area is called.
At this moment, vcpu->arch.xsave_enabled is true. Processor state is saved
to memory with XSAVES instruction. As physical MSR_IA32_XSS[bit 8] is 1,
ectx->xs_area.xsave_hdr.hdr.xcomp_bv[bit 8] is set to 1 after the execution
of XSAVES instruction.
States at this point:
* Physical MSR_IA32_XSS[bit 8] is 1.
* vcpu_get_guest_msr(vcpu, MSR_IA32_XSS)[bit 8] is 0.
* ectx->xs_area.xsave_hdr.hdr.xcomp_bv[bit 8] is 1.
3. idle thread -> vcpu thread
context_switch_in happens and rstore_xsave_area is called.
At this moment, vcpu->arch.xsave_enabled is true. Physical MSR_IA32_XSS is
updated with the value of guest MSR_IA32_XSS, which is 0.
States at this point:
* Physical MSR_IA32_XSS[bit 8] is 0.
* vcpu_get_guest_msr(vcpu, MSR_IA32_XSS)[bit 8] is 0.
* ectx->xs_area.xsave_hdr.hdr.xcomp_bv[bit 8] is 1.
Processor state is restored from memory with XRSTORS instruction afterwards.
According to SDM Vol1 13.12 OPERATION OF XRSTORS, a #GP occurs if XCOMP_BV
sets a bit in the range 62:0 that is not set in XCR0 | IA32_XSS.
So, #GP occurs once XRSTORS instruction is executed.
Such issue does not happen with kernel 5.10. Because kernel 5.10 writes to
MSR_IA32_XSS during initialization, while kernel 5.4 does not do such write.
Once guest writes to MSR_IA32_XSS, it would be trapped to hypervisor, then,
physical MSR_IA32_XSS and the value of MSR_IA32_XSS in vcpu->arch.guest_msrs
are updated with the value specified by guest. So, in the point 2 above,
correct processor state is saved. And #GP would not happen in the point 3.
This patch initializes the XSAVE related processor state for guest.
If vcpu is not launched yet, the processor state is initialized according to
the initial value of vcpu_get_guest_msr(vcpu, MSR_IA32_XSS), ectx->xcr0,
and ectx->xs_area. With this approach, the physical processor state is
consistent with the one presented to guest.
Tracked-On: #6434
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Reviewed-by: Li Fei1 <fei1.li@intel.com>
Currently init_vmx_msrs() emulates same value for the IA32_VMX_xxx_CTLS
and IA32_VMX_TRUE_xxx_CTLS MSRs.
But the value of physical MSRs could be different between the pair,
and we need to adjust the emulated value accordingly.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
check_vmx_permission() is called in vmresume_vmexit_handler() and
vmlaunch_vmexit_handler() already.
Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Currently the sched event handling may encounter data race problem, and
as a result some vcpus might be stalled forever.
One example can be wbinvd handling where more than 1 vcpus are doing
wbinvd concurrently. The following is a possible execution of 3 vcpus:
-------
0 1 2
req [Note: 0]
req bit0 set [Note: 1]
IPI -> 0
req bit2 set
IPI -> 2
VMExit
req bit2 cleared
wait
vcpu2 descheduled
VMExit
req bit0 cleared
wait
vcpu0 descheduled
signal 0
event0->set=true
wake 0
signal 2
event2->set=true [Note: 3]
wake 2
vcpu2 scheduled
event2->set=false
resume
req
req bit0 set
IPI -> 0
req bit1 set
IPI -> 1
(doesn't matter)
vcpu0 scheduled [Note: 4]
signal 0
event0->set=true
(no wake) [Note: 2]
event0->set=false (the rest doesn't matter)
resume
Any VMExit
req bit0 cleared
wait
idle running
(blocked forever)
Notes:
0: req: vcpu_make_request(vcpu, ACRN_REQUEST_WAIT_WBINVD).
1: req bit: Bit in pending_req_bits. Bit0 stands for bit for vcpu0.
2: In function signal_event, At this time the event->waiting_thread
is not NULL, so wake_thread will not execute
3: eventX: struct sched_event of vcpuX.
4: In function wait_event, the lock does not strictly cover the execution between
schedule() and event->set=false, so other threads may kick in.
-----
As shown in above example, before the last random VMExit, vcpu0 ended up
with request bit set but event->set==false, so blocked forever.
This patch proposes to change event->set from a boolean variable to an
integer. The semantic is very similar to a semaphore. The wait_event
will add 1 to this value, and block when this value is > 0, whereas signal_event
will decrease this value by 1.
It may happen that this value was decreased to a negative number but that
is OK. As long as the wait_event and signal_event are paired and
program order is observed (that is, wait_event always happens-before signal_event
on a single vcpu), this value will eventually be 0.
Tracked-On: #6405
Signed-off-by: Yifan Liu <yifan1.liu@intel.com>
This is a simply implement for the 32bit and 64bit elf loader.
The loading function first reads the image header, and finds the program
entries that are marked as PT_LOAD, then loads segments from elf file to
guest ram. After that, it finds the bss section in the elf section entries, and
clear the ram area it points to.
Limitations:
1. The e_type of the elf image must be ET_EXEC(executable). Relocatable or
dynamic code is not supported.
2. The loader only copies program segments that has a p_type of
PT_LOAD(loadable segment). Other segments are ignored.
3. The loader doesn’t support Sections that are relocatable
(sh_type is SHT_REL or SHT_RELA)
4. The 64bit elf’s entry address must below 4G.
5. The elf is assumed to be able to put segments to valid guest memory.
Tracked-On: #6323
Signed-off-by: Zhou, Wu <wu.zhou@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This patch adds a function elf_loader() to load elf image.
It checks the elf header, get its 32/64 bit type, then calls
the corresponding loading routines, which are empty, and
will be realized later.
Tracked-On: #6323
Signed-off-by: Zhou, Wu <wu.zhou@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Source: https://github.com/freebsd/freebsd-src/blob/main/sys/sys/elf_common.h
Trimed to meet the minimal requirements for the Zephyr elf file to be loaded
Also added elf file header data struct and program/section entry data structs.
Tracked-On: #6323
Signed-off-by: Zhou, Wu <wu.zhou@intel.com>
Reviewed-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In order to make better sense, vm_elf_loader, vm_bzimage_loader and
vm_rawimage_loader are changed to elf_loaer, bzimage_loaer and
rawimage_loader.
Tracked-On: #6323
Signed-off-by: Zhou, Wu <wu.zhou@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Remove the acpi loading function from elf_loader, rawimage_loaer and
bzimage_loader, and call it together in vm_sw_loader.
Now the vm_sw_loader's job is not just loading sw, so we rename it to
prepare_os_image.
Tracked-On: #6323
Signed-off-by: Zhou, Wu <wu.zhou@intel.com>
Reviewed-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
For the guest OS loaders, prapare_loading_xxx are not accurate for
what those functions actually do. Now they are changed to load_xxx:
load_rawimage, load_bzimage.
And the 'bsp' expression is confusing in the comments for
init_vcpu_protect_mode_regs, changed to a better way.
Tracked-On: #6323
Signed-off-by: Zhou, Wu <wu.zhou@intel.com>
Reviewed-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
vboot_info.h declares vm loader function also, so rename the file name to
vboot.h;
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The patch splits the vm_load.c to three parts, the loader function of bzImage
kernel is moved to bzimage_loader.c, the loader function of raw image kernel
is moved to rawimage_loader.c, the stub is still stayed in vm_load.c to load
the corresponding kernel loader function. Each loader function could be
isolated by CONFIG_GUEST_KERNEL_XXX macro which generated by config tool.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Change if condition to switch in vm_sw_loader() so that the sw loader
could be compiled conditionally.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Rename KERNEL_ZEPHYR to KERNEL_RAWIMAGE. Added new type "KERNEL_ELF".
Add CONFIG_GUEST_KERNEL_RAWIMAGE, CONFIG_GUEST_KERNEL_ELF and/or
CONFIG_GUEST_KERNEL_BZIMAGE to config.h if it's configured.
Tracked-On: #6323
Signed-off-by: Yang,Yu-chu <yu-chu.yang@intel.com>
Reviewed-by: Victor Sun <victor.sun@intel.com>
Previously we only support loading raw format of zephyr image as prelaunched
Zephyr VM, this would cause guest F segment overridden issue because the zephyr
raw image covers memory space from 0x1000 to 0x100000 upper. To fix this issue,
we should support ELF format image loading so that parse and load the multiple
segments from ELF image directly.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
When pass-thru GPU to pre-launched Linux guest,
need to pass GPU OpRegion to the guest.
Here's the detailed steps:
1. reserve a memory region in ve820 table for GPU OpRegion
2. build EPT mapping for GPU OpRegion to pass-thru OpRegion to guest
3. emulate the pci config register for OpRegion
For the third step, here's detailed description:
The address of OpRegion locates on PCI config space offset 0xFC,
Normal Linux guest won't write this register,
so we can regard this register as read-only.
When guest reads this register, return the emulated value.
When guest writes this register, ignore the operation.
Tracked-On: #6387
Signed-off-by: Liu,Junming <junming.liu@intel.com>
ACRN does not support the variable range vMTRR. The default
memory type of vMTRR is UC. With this vMTRR emulation guest VM
such as Linux refuses to map the MMIO address space as WB. In
order to get better performance SHM BAR of ivshmem is mapped
with PAT ignored and memory type of SHM BAR is fixed to WB.
Tracked-On: #6389
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Create virtual acpi table of tpm2 based on the raw data if the TPM2
device is presented and the passthrough tpm2 is enabled.
Refine the arguments of bin_gen.py. The --board and --scenario take the
path to the XMLs as the argument. The allocation.xml is needed for
bin_gen.py to generate tpm2 acpi table.
Refine the condition of tpm2_acpi_gen. The tpm2 device "MSFT0101" can be
present in device id or compatible_id(CID). Check both attributes and
child node of tpm2 device.
Tracked-On: #6320
Signed-off-by: Yang,Yu-chu <yu-chu.yang@intel.com>
Relocate ACPI address to 0x7fe00000 and ACPI NVS to 0x7ff00000 correspondingly.
In this case, we could include TPM event log region [0x7ffb0000, 0x80000000)
into ACPI NVS.
Tracked-On: #6320
Signed-off-by: Fei Li <fei1.li@intel.com>
ACRN used to prepare the vTPM2 ACPI Table for pre-launched VM at the build stage
using config tools. This is OK if the TPM2 ACPI Table never changes. However,
TPM2 ACPI Table may be changed in some conditions: change BIOS configuration or
update BIOS.
This patch do TPM2 fixup to update the vTPM2 ACPI Table and TPM2 MMIO resource
configuration according to the physical TPM2 ACPI Table.
Tracked-On: #6366
Signed-off-by: Tao Yuhong <yuhong.tao@intel.com>
Signed-off-by: Fei Li <fei1.li@intel.com>
1. add a name field to indicate what the MMIO Device is.
2. add two more MMIO resource to the acrn_mmiodev data structure.
Tracked-On: #6366
Signed-off-by: Tao Yuhong <yuhong.tao@intel.com>
Signed-off-by: Fei Li <fei1.li@intel.com>
ACRN could run without XSAVE Capability. So remove XSAVE dependence to support
more (hardware or virtual) platforms.
Tracked-On: #6287
Signed-off-by: Fei Li <fei1.li@intel.com>
Check whether condition is met before check whether time is out after iommu_read32.
This is because iommu_read32 would cause time out on some virtual platform in
spite of the current DMAR status meets the pre_condition.
Tracked-On: #6371
Signed-off-by: Fei Li <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
If HV enable trigger #GP for uc-lock, and is about to emulate guest uc-lock
instructions, should trap guest #GP. Guest uc-lock instrucction trigger #GP,
cause vmexit for #GP, HV handle this vmexit and emulate uc-lock
instruction.
Tracked-On: #6299
Signed-off-by: Tao Yuhong <yuhong.tao@intel.com>
User could use make targz-pkg command to generate tar package in
build directory,which could help user simplify the process
of installing acrn hypervisor in target board. user need to copy the
tarball package to target board,and extract it to "/" directory.
Tracked-On: #6355
Signed-off-by: liu hang1 <hang1.liu@intel.com>
Reviewed-by: VanCutsem, Geoffroy <geoffroy.vancutsem@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
Currently the HV console does not support PCI UART with 64bit BAR, but in the
case that the BAR is in 64bit and the BAR space is below 4GB (i.e. the high
32bit address of the 64bit BAR is zero), HV should be able to support it.
Tracked-On: #6334
Signed-off-by: Victor Sun <victor.sun@intel.com>
When guest kernel has multiple loading segments like ELF format image, just
define one load address in sw_kernel_info struct is meaningless.
The patch removes kernel_load_addr member in struct sw_kernel_info, the load
address should be parsed in each specified format image processing.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
The previous code did not load bzImage start from protected mode part, result
in the protected mode part un-align with kernel_alignment field and then cause
kernel decompression start from a later aligned address. In this case we had
to enlarge the needed size of bzImage kernel to kernel_init_size plus double
size of kernel_alignment.
With loading issue of bzImage protected mode part fixed, the kernel needed size
is corrected in this patch.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
When LaaG boots with bzImage module file, only protected mode code need
to be loaded to guest space since the VM will boot from protected mode
directly. Futhermore, per Linux boot protocol the protected mode code
better to be aligned with kernel_alignment field in zeropage, otherwise
kernel will take time to do "rep movs" to the aligned address.
In previous code, the bzImage is loaded to the address where aligned with
kernel_alignment, this would make the protected mode code unalign with
kernel_alignment. If the kernel is configured with CONFIG_RELOCATABLE=n,
the guest would not boot. This patch fixed this issue.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
This patch moves get_bzimage_kernel_load_addr() from init_vm_sw_load() to
vm_sw_loader() stage so will set kernel load address of bzImage type kernel
in vm_bzimage_loader() in vm_load.c.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
This patch moves get_initrd_load_addr() API from init_vm_sw_load() to
vm_sw_loader() stage. The patch assumes that the kernel image have been
loaded to guest space already.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
In load_sw_modules() implementation, we always assuming the guest kernel
module has one load address and then the whole kernel image would be loaded
to guest space from its load address. This is not true when guest kernel
has multiple load addresses like ELF format kernel image.
This patch removes load_sw_modules() API, and the loading method of each
format of kernel image could be specified in prepare_loading_xxximage() API.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
As the previous commit said the kernel load address should be moved
from init_vm_sw_load() to vm_sw_loader() stage. This patch refines
the API of get_bzimage_kernel_load_addr() in init_vm_kernel_info()
for later use.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
Currently the guest kernel load address and ramdisk load address are
initialized during init_vm_sw_load() stage, this is meaningless when
guest kernel has multiple segments with different loading addresses.
In that case, the kernel load addresses should be parsed and loaded
in vm_sw_loader() stage, the ramdisk load address should be set in
that stage also because it is depended on kernel load address.
This patch refines the API of get_initrd_load_addr() which will set
proper initrd load address of bzImage type kernel for later use.
Tracked-On: #6323
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
According to PCIe Spec, for a RW register bits, If the optional feature
that is associated with the bits is not implemented, the bits are permitted
to be hardwired to 0b. However Zephyr would use INTx Line Register as writable
even this PCI device has no INTx, so emulate INTx Line Register as writable.
Tracked-On: #6330
Signed-off-by: Fei Li <fei1.li@intel.com>
There're some virtual platform which doesn't meet this constraint. So remove
this constraint.
Tracked-On: #6329
Signed-off-by: Fei Li <fei1.li@intel.com>
for core partition VM (like RTVM), PMC is always used for performance
profiling / tuning, so expose PMC capability and pass-through its MSRs
to the VM.
Tracked-On: #6307
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>
if one array just used in local only, and its size not used extern,
use ARRAY_SIZE macro to calculate its size.
Tracked-On: #6307
Signed-off-by: Minggui Cao <minggui.cao@intel.com>
Reviewed-by: Junjie Mao <junjie.mao@intel.com>
In some scenarios (e.g., nested) where lapic-pt is enabled for a vcpu
running on a pcpu hosting console timer, the hv console will be
inaccessible.
This patch adds the console callback to every VM-exit event so that the
console can still be somewhat functional under such circumstance.
Since this is VM-exit driven, the VM-exit/second can be low in certain
cases (e.g., idle or running stress workload). In extreme cases where
the guest panics/hangs, there will be no VM-exits at all.
In most cases, the shell is laggy but functional (probably enough for
debugging purpose).
Tracked-On: #6312
Signed-off-by: Yifan Liu <yifan1.liu@intel.com>
For an atomic operation using bus locking, it would generate LOCK# bus
signal, if it has Non-WB memory operand. This is an UC lock. It will
ruin the RT behavior of the system.
If MSR_IA32_CORE_CAPABILITIES[bit4] is 1, then CPU can trigger #GP
for instructions which cause UC lock. This feature is controlled by
MSR_TEST_CTL[bit28].
This patch enables #GP for guest UC lock.
Tracked-On: #6299
Signed-off-by: Tao Yuhong <yuhong.tao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Because the emulation code is for both split-lock and uc-lock,
rename splitlock.c/splitlock.h to lock_instr_emul.c/lock_instr_emul.h
Tracked-On: #6299
Signed-off-by: Tao Yuhong <yuhong.tao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Because the emulation code is for both split-lock and uc-lock, Changed
these API names:
vcpu_kick_splitlock_emulation() -> vcpu_kick_lock_instr_emulation()
vcpu_complete_splitlock_emulation() -> vcpu_complete_lock_instr_emulation()
emulate_splitlock() -> emulate_lock_instr()
Tracked-On: #6299
Signed-off-by: Tao Yuhong <yuhong.tao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Split-lock emulation can be re-used for uc-lock. In emulate_splitlock(),
it only work if this vmexit is for #AC trap and guest do not handle
split-lock and HV enable #AC for splitlock.
Add another condition to let emulate_splitlock() also work for #GP trap
and guest do not handle uc-lock and HV enable #GP for uc-lock.
Tracked-On: #6299
Signed-off-by: Tao Yuhong <yuhong.tao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
When ACRN uses decode_instruction to emulate split-lock/uc-lock
instruction, It is actually a try-decode to see if it is XCHG.
If the instruction is XCHG instruction, ACRN must emulate it
(inject #PF if it is triggered) with peer VCPUs paused, and advance
the guest IP. If the instruction is a LOCK prefixed instruction
with accessing the UC memory, ACRN Halted the peer VCPUs, and
advance the IP to skip the LOCK prefix, and then let the VCPU
Executes one instruction by enabling IRQ Windows vm-exit. For
other cases, ACRN injects the exception back to VCPU without
emulating it.
So change the API to decode_instruction(vcpu, bool full_decode),
when full_decode is true, the API does same thing as before. When
full_decode is false, the different is if decode_instruction() meet unknown
instruction, will keep return = -1 and do not inject #UD. We can use
this to distinguish that an #UD has been skipped, and need inject #AC/#GP back.
Tracked-On: #6299
Signed-off-by: Tao Yuhong <yuhong.tao@intel.com>
To add brakets for '(char *)(ptr)' in MACRO
container_of(), which may be used recursively.
Tracked-On: #6284
Signed-off-by: Yonghua Huang <yonghua.huang@intel.com>
Common structures are used by DM, kernel, HV. Aligned attribute might
caused structures size mismatch between DM/HV and kernel, as kernel uses
default GCC alignment.
So, make DM/HV also use the default GCC alignment.
Tracked-On: #6282
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
