GRUB multiboot2 doesn't support relocation for ELF, which means it can't
load acrn.32.out to other address other than the one specified in ELF
header. Thus we need to use the raw binary file acrn.bin, and add
address/entry address/relocatable tags to instruct multiboot2 loader
how to load the raw binary.
Tracked-On: #4441
Reviewed-by: Fengwei Yin <fengwei.yin@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
In direct boot mode, boot_context[] which is saved from cpu_primary_save_32()
is no longer used since commit 6beb34c3cb ("vm_load: update init gdt
preparation"). Thus, the call to it and the function itself can be removed.
Tracked-On: #4441
Reviewed-by: Fengwei Yin <fengwei.yin@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.com>
reduce the use of similar APIs (particularly the name confusion) for
CFG space read/write.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Make the name of the functions more accurate
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Add some pre-assumption and safety check for PCIe ECAM:
1) ACRN only support platforms with PCIe ECAM to access PCIe device CFG space;
2) Must not use ECAM to access PCIe device CFG space before
pci_switch_to_mmio_cfg_ops was called. (In release version, ACRN didn't support
IO port Mechanism. ECAM is the only way to access the PCIe device CFG space).
Tracked-On: #4371
Signed-off-by: Li Fei1 <fei1.li@intel.com>
One argument is missing for the function ptirq_alloc_entry.
This patch fixes the doc generation error.
Tracked-On: #3882
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
- change variable name from hpa to hva because in this function we are
dealing with hva, not hpa.
- can get the address of ld_text_end by directly referring to this symbol,
because relative addressing yields the correct hva, not the hva before
relocation.
Tracked-On: #4441
Signed-off-by: Zide Chen <zide.chen@intel.com>
The pci_read_cap and pci_read_ext_cap are used to enumerate PCI
legacy capability and extended capability.
Change the name pci_read_cap to pci_enumerate_cap
Change the name pci_read_ext_cap to pci_enumerate_ext_cap
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Fixed misspellings and rst formatting issues.
Added ptdev.h to the list of include file for doxygen
Tracked-On: #3882
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
Signed-off-by: David B. Kinder <david.b.kinder@intel.com>
Add doxygen style comments to ptdev public APIs.
Add these API descriptions to group acrn_passthrough.
Tracked-On: #3882
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
This patch updates board.c files for RDT MBA on existing
platforms. Also, fixes setting RDT flag in WHL config file.
Tracked-On: #3725
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This patch adds RDT MBA support to detect, configure and
and setup MBA throttle registers based on VM configuration.
Tracked-On: #3725
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Currently config tool generated 'console=/dev/ttySn' in boot cmdline
for logical_partiton scenario, need to strip '/dev/' to avoid kernel
boot issue.
Tracked-On: #4451
Signed-off-by: Wei Liu <weix.w.liu@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
create new pdev and vdev structures for a SRIOV VF device initialization
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Add a new parameter pf_vdev for function vpci_init_vdev to support SRIOV
VF vdev initializaiton.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The init_one_dev_config is used to initialize a acrn_vm_pci_dev_config
SRIOV needs a explicit acrn_vm_pci_dev_config to create a VF vdev,so
refine it to return acrn_vm_pci_dev_config.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Due to SRIOV VF physical device needs to be initialized when
VF_ENABLE is set and a SRIOV VF physical device initialization
is same with standard PCIe physical device, so expose the
init_pdev for SRIOV VF physical device initialization.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
All SRIOV VF physical devices don't have bars in configuration space,
they are from the VF associated PF's VF_BAR registers of SRIOV capability.
Adding a vbars data structure in pci_cap_sriov data structure to store
SRIOV VF_BAR information, so that each VF bars can be initialized directly
through the vbars instead multiple accessing of the PF VF_BAR registers.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
To support SRIOV capability initialization, add a new parameter
is_sriov_pf_vdev for init_vdev_pt function.
If parameter is_sriov_pf_vdev of function init_vdev_pt is true,
then function init_vdev_pt initializes the vdev's SRIOV capability.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Devices that support SR-IOV can expose their capabilities in
lspci -vv command as below. The offline tool, instead of picking
up the bios exposed memory region(bc000000) for the devices ends
picking up the SR-IOV memory region(00000000c0000000) and generates
VBAR address (in pci_devices.h). This is incorrect. This patch
fixes the offline tool to take the right memory region as the
VBAR address.
Sample lspic -vv log:
67:00.0 Ethernet controller: Intel Corporation Ethernet Connection
X722 for 10GbE backplane (rev 09)
Subsystem: Intel Corporation Device 0000
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr+
Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort-
Latency: 0
Interrupt: pin A routed to IRQ 61
Region 0: Memory at bc000000 (64-bit, prefetchable) [size=16M]
Region 3: Memory at c1000000 (64-bit, prefetchable) [size=32K]
Capabilities: [160 v1] Single Root I/O Virtualization (SR-IOV)
IOVCap: Migration-, Interrupt Message Number: 000
IOVCtl: Enable- Migration- Interrupt- MSE- ARIHierarchy+
IOVSta: Migration-
Initial VFs: 32, Total VFs: 32, Number of VFs: 0, Function Dependency Link: 00
VF offset: 16, stride: 1, Device ID: 37cd
Supported Page Size: 00000553, System Page Size: 00000001
Region 0: Memory at 00000000c0000000 (64-bit, prefetchable)
Region 3: Memory at 00000000c1020000 (64-bit, prefetchable)
VF Migration: offset: 00000000, BIR: 0
Tracked-On: #4443
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
This patches removes the unused CLOS_required flag from the
offline guest configuration tool.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
This patch updates existing <$BOARD$>.xml for RDT support.
If the platform supports RDT, the resource and MAX CLOS
value is generated, else the CLOS_INFO is set to empty.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
This patch checks if RDT feature is supported by platform and
updates the CONFIG_RDT_ENABLED flag in the <$BOARD$>.config
file accordingly.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
This patch adds support for as generating board.c file
with multiple RDT resources.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
and vm configuration
This patch adds support for,
1. Providing list of max supported CLOS value that is common
between all RDT resource to user when configuring scenario xml.
2. Validate and set the clos value in the vm_configuration file.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
value among RDT resources.
This patch identifies the least common supported clos value
from multiple RDT resource. This is done so as to have
consistent capabilities across all resource allocations.
From SDM, "The number of CLOS supported for the MBA feature
may or may not align with other resources such as L3 CAT. In
cases where the RDT features support different numbers of
CLOS the lowest numerical CLOS support the common set of
features, while higher CLOS may support a subset. For instance,
if L3 CAT supports 8 CLOS while MBA supports 4 CLOS, all 8 CLOS
would have L3 CAT masks available for cache control, but the
upper 4 CLOS would not offer MBA support. In this case the upper
4 CLOS would not be subject to any throttling control. Software
can manage supported resources / CLOS in order to either have
1) consistent capabilities across CLOS by using the common subset
or 2) enable more flexibility by selectively applying resource
control where needed based on careful CLOS and thread mapping".
We decided to go with option #1, as it will be more consistent
and less prone to user errorw hen programming the resource mask
MSRs.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
xml file
This patch adds support for parsing RDT resources and
its max clos, max supported mask values from the target xml file.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
This patch adds support for storing multiple RDT resource
and its max supported clos value in the target xml file under
clos subsection.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
If the guest is RTVM, dm process exit doesn't mean RTVM is
shutdown. Only shutdown_vm in hypervisor guarantee RTVM is
shutodwn. So we should avoid touch guest memory content
from DM if the guest is RTVM.
Tracked-On: #4428
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
Since this callback is only useful for pass through PCIe device and now pass through
PCIe device emulation has splited from DM to HV. we could remove this callback now.
Tracked-On: #4371
Signed-off-by: Li Fei1 <fei1.li@intel.com>
Current code avoid the rule 88 S in MISRA-C, so move xsaves and xrstors
assembler to individual functions.
Tracked-On: #4436
Signed-off-by: Conghui Chen <conghui.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
VF_ENABLE is one field of SRIOV capability that is used to create
or remove VF physical devices. If VF_ENABLE is set, hv can detect
if the VF physical devices are ready after waiting 100 ms.
v2: Add sanity check for writing NumVFs register, add precondition
and application constraints when VF_ENABLE is set and refine
code style.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Introduce SRIOV capability field for pci_vdev and add SRIOV capability
interception entries.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Make the SRIOV-Capable device invisible from SOS if there is
no room for its all virtual functions.
v2: fix a issue that if a PF has been dropped, the subsequent PF
will be dropped too even there is room for its VFs.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
if the device has PCIe capability, walks all PCIe extended
capabilities for SRIOV discovery.
v2: avoid type casting and refine naming.
Tracked-On: #4433
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
It puts the new line in the wrong place, and the logs are confusing.
For example, for these entries:
mmap[0] - type: 1, base: 0x00000, length: 0x9800
mmap[1] - type: 2, base: 0x98000, length: 0x8000
mmap[2] - type: 3, base: 0xc0000, length: 0x4000
Currently it prints them in this way:
mmap table: 0 type: 0x1
Base: 0x0000000000000000 length: 0x0000000000098000
mmap table: 1 type: 0x2
Base: 0x0000000000098000 length: 0x0000000000008000
mmap table: 2 type: 0x3
Base: 0x00000000000c0000 length: 0x0000000000040000
With this fix, it looks like the following, and now it's of same style
with how prepare_sos_vm_memmap() logs ve820 tables.
mmap table: 0 type: 0x1
Base: 0x0000000000000000 length: 0x0000000000098000
mmap table: 1 type: 0x2
Base: 0x0000000000098000 length: 0x0000000000008000
mmap table: 2 type: 0x3
Base: 0x00000000000c0000 length: 0x0000000000040000
Tracked-On: #1842
Signed-off-by: Zide Chen <zide.chen@intel.com>
add vpci bridge operations in hypervisor, to avoid SOS mis-operations
to affect other VM's PCI devices.
assumption: before hypervisor bootup, the physical pci-bridge shall be
configured correctly by BIOS or other bootloader; for ACS (Access
Control Service) capability, it is configured by BIOS to support the
devices under it to be isolated and allocated to different VMs.
to simplify the emulations of vpci bridge, set limitations as following:
1. expose all configure space registers, but readonly
2. BIST not support; by default is 0
3. not support interrupt, including INTx and MSI.
TODO:
1. configure tool can select whether a PCI bridge is emulated or pass
through.
Open:
1. SOS how to reset PCI device under the PCI bridge?
Tracked-On: #3381
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Minggui Cao <minggui.cao@intel.com>
Acked-by: Eddie Dong <eddie.dong@Intel.com>
The init value for XCR0 and XSS should be the same with spec:
In SDM Vol1 13.3:
XCR0[0] is associated with x87 state (see Section 13.5.1). XCR0[0] is
always 1. The other bits in XCR0 are all 0 coming out of RESET.
The IA32_XSS MSR (with MSR index DA0H) is zero coming out of RESET.
The previous code try to fix the xsave area leak to other VMs during init
phase, but bring the error to linux. Besides, it cannot avoid the
possible leak in running phase. Need find a better solution.
Tracked-On: #4430
Signed-off-by: Conghui Chen <conghui.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The dedicated DMAR unit for Intel integrated GPU
shall be available on the physical platform.
So remove the assert and add application constraint
in handle_one_drhd func.
Tracked-On: #4405
Signed-off-by: Junming Liu <junming.liu@intel.com>
Reviewed-by: Wu Binbin <binbin.wu@intel.com>
Reviewed-by: Wu Xiangyang <xiangyang.wu@linux.intel.com>
The slot in launch config would be conflicted when the same bdf is
selected from webUI.
This patch do the below condition to avoid such case:
1. Add the bdf check for pass-through devices.
2. Use the bus:dev:func as passthru device option in launch config.
v1-v2:
1. If BUS of PCI device for passthru is not 0, then alloc a virtual
slot for it.
Tracked-On: #4312
Signed-off-by: Wei Liu <weix.w.liu@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
KATA VM is not supported on dual-core systems override the
MAX_KATA_VM_NUM to 0.
Tracked-On: #4319
Signed-off-by: Wei Liu <weix.w.liu@intel.com>
Acked-by: Victor Sun <victor.sun@intel.com>
is not set
This patch does the following,
1. Removes RDT code if CONFIG_RDT_ENABLED flag is
not set.
2. Set the CONFIG_RDT_ENABLED flag only on platforms
that support RDT so that build scripts will automatically
reflect the config.
Tracked-On: #3715
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
cache configuration.
This patch creates a generic infrastructure for
RDT resources instead of just L2 or L3 cache. This
patch also fixes L3 CAT config overwrite by L2 in
cases where both L2 and L3 CAT are supported.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
There can be times when user unknowinlgy enables
CONFIG_CAT_ENBALED SW flag, but the hardware might
not support L3 or L2 CAT. In such case software can
end up writing to the CAT MSRs which can cause
undefined results. The patch fixes the issue by
enabling CAT only when both HW as well software
via the CONFIG_CAT_ENABLED supports CAT.
The patch also address typo with "clos2prq_msr"
function name. It should be "clos2pqr_msr" instead.
PQR stands for platform qos register.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Upcoming intel platforms can support both L2 and L3
but our current code only supports either L2 or L3 CAT.
So split the MSRs so that we can support allocation
for both L2 and L3.
This patch does the following,
1. splits programming of L2 and L3 cache resource
based on the resource ID.
2. Replace generic platform_clos_array struct with resource
specific struct in all the existing board.c files.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
As part of rdt cat refactoring, goal is to combine all rdt
specific features such as CAT under one module. So renaming
rdt resouce specific files such as cat.c/.h to generic rdt.c/.h
files.
Tracked-On: #3715
Signed-off-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Per Section 4.4 Speculation Barriers, in
"Retpoline: A Branch Target Inject Mitigation" white paper,
"LFENCE instruction limits the speculative execution that
a processor implementation can perform around the LFENCE,
possibly impacting processor performance,but also creating
a tool with which to mitigate speculative-execution
side-channel attacks."
Tracked-On: #4424
Signed-off-by: Yonghua Huang <yonghua.huang@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
Initialize efi info of acrn mbi when boot from multiboot2 protocol, with
this patch hypervisor could get host efi info and pass it to Linux zeropage,
then make guest Linux possible to boot with efi environment;
Tracked-On: #4419
Signed-off-by: Victor Sun <victor.sun@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>