Now the DM has adopted the new VHM request state transitions and
REQ_STATE_FAILED is obsolete since neither VHM nor kernel mediators will set the
state to FAILED.
This patch drops the definition to REQ_STATE_FAILED in the hypervisor, makes
''processed'' unsigned to make the compiler happy about typing and simplifies
error handling in the following ways.
* (dm_)emulate_(pio|mmio)_post no longer returns an error code, by introducing a
constraint that these functions must be called after an I/O request
completes (which is the case in the current design) and assuming
handlers/VHM/DM will always give a value for reads (typically all 1's if the
requested address is invalid).
* emulate_io() now returns a positive value IOREQ_PENDING to indicate that the
request is sent to VHM. This mitigates a potential race between
dm_emulate_pio() and pio_instr_vmexit_handler() which can cause
emulate_pio_post() being called twice for the same request.
* Remove the ''processed'' member in io_request. Previously this mirrors the
state of the VHM request which terminates at either COMPLETE or FAILED. After
the FAILED state is removed, the terminal state will always be constantly
COMPLETE. Thus the mirrored ''processed'' member is no longer useful.
Note that emulate_instruction() will always succeed after a reshuffle, and this
patch takes that assumption in advance. This does not hurt as that returned
value is not currently handled.
This patch makes it explicit that I/O emulation is not expected to fail. One
issue remains, though, which occurs when a non-aligned cross-boundary access
happens. Currently the hypervisor, VHM and DM adopts different policy:
* Hypervisor: inject #GP if it detects that the access crossed boundary
* VHM: deliver to DM if the access does not complete falls in the range of a
client
* DM: a handler covering part of the to-be-accessed region is picked and
assertion failure can be triggered.
A high-level design covering all these components (in addition to instruction
emulation) is needed for this. Thus this patch does not yet cover the issue.
Tracked-On: #875
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The bracket is required when the level of precedence of
the operators is less than 13. Add the bracket to logical
conjunctions. The commit applys the rule to the files under
Signed-off-by: Yang, Yu-chu <yu-chu.yang@intel.com>
Reviewed-by: Junjie Mao <junjie.mao@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Current EPT check runs after EPT enabling in init_exec_ctrl. This
patch fixes wrong order.
Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
-- add clear trusty memory flag
In some cases such as UOS power off or UOS full reset,
need to clear trusty memory,no need to clear memory such as
UOS S3 or UOS system reset,then add a flag to distinguish it
when destroy secure world.
-- Restore trusty memory to guest normal world.
-- Moved free trusty EPT inside destroy_secure_world
In some cases such as UOS S3 or UOS system reset,
only need to free trusty EPT, this patch move free
trusty EPT inside destroy_secure_world.
Because PD/PT are shared in both secure world's EPT
and normal world's EPT,before freeing trusty EPT,
it will memset all PDPTEs except trusty memory,
then call 'free_ept_mem', it can only free trusty EPT,
and does't affect shared normal world EPT.
v2-->v3:
-- Used new mmu api ept_mr_add when restore trusty memory
to SOS and normal world
-- Dropped this patch "Removed reverted page tables for trusty memory"
because map_mem will be removed in future
It will have a patch, need to update this api(ept_mr_add),
it will not create inverted page tables for trusty memory.
v1-->v2:
-- free trusty ept
still use free_ept_mem, not add a new api,but need to
memset pdptes except trusty memory
-- Removed reverted page tables for trusty memory.
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Define Bitmap flag to indicate secure world's state:
supported: 0(not supported), 1(supported)
active: 0(inactive), 1(active)
Refine secure_world_memory:
base_gpa_in_sos: base_gpa from SOS's view
base_gpa_in_uos: base_gpa from UOS's view, this is the original base_gpa
allocated by bootloader.
Recording above GPA is for usage of trusty EPT destroy and re-create.
There is an assumption: the secure world's memory address is contiguous
in both SOS and physical side.
Signed-off-by: Qi Yadong <yadong.qi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Originally, there is cross-references between instr_emul.h and instr_emul_wrapper.h.
User must include both of them when calling instruction emulation functions. This
will raise up some confusion and inconvenience.
So we rearrange the logic of instruction emulation code as following:
- External API -- defined in instr_emul.h
* decode_instruction(struct vcpu *vcpu)
* emulate_instruction(struct vcpu *vcpu)
- Make all other functions as static in instr_emul.c
- Remove instr_emul_wrapper.c/h
No functional change.
Signed-off-by: Kaige Fu <kaige.fu@intel.com>
Reviewed-by: Eddie Dong <eddie.dong@intel.com>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Remove structure vm_attr
Wrap two APIs alloc_vm_id and free_vm_id
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In order to comply with MISRA C rules, renamed vairables
and function names starting with "_".
The major changes invloves mostly static function
names, as they are being called inside the same file
by a wrapper function.
Signed-off-by: Arindam Roy <arindam.roy@intel.com>
For data structure types "enum vpic_wire_mode, struct stack_canary",
its name is identical with variable name in the same scope.
This MISRA C violation is detected by static analysis tool.
For variables "segment_override, pde_index", its name is identical
with function name. This MISRA C violation is detected.
Naming convention rule:Variable name can be shortened from
its data structure type name.
The following udpates are made:
enum vpic_wire_mode vpic_wire_mode-->enum vpic_wire_mode wire_mode
struct stack_canary stack_canary-->struct stack_canary stk_canary
uint8_t segment_override:1 --> uint8_t seg_override:1
uint32_t pde_index--> uint32_t pde_idx
V1-->V2:
Remove update "enum cpu_state cpu_state-->enum cpu_state state"
and "enum irqstate irqstate-->enum irq_ops_mode ops_mode", other
patch will cover it.
V2-->V3:
Update "uint32_t pde_index--> uint32_t pde_idx".
Signed-off-by: Xiangyang Wu <xiangyang.wu@linux.intel.com>
This patch solely moves MMIO handler registration APIs from ept.c to io.c as it
is related more to I/O request handling.
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This is the counterpart to the PIO emulation side.
1. ept_violation_vmexit_handler (entry point for handling vmexit on EPT instruction):
Extract mmio address, size, direction and value (for write only), fill in an
I/O request, invoke do_io to handle that and emulate_pio_post for
post-processing.
2. emulate_io
Handle the given I/O request, either completed by registered MMIO handlers
or sent to VHM.
3. emulate_mmio_post:
Update guest registers after the emulation is done.
v2 -> v3:
* Rename: emulate_mmio_by_handler -> hv_emulate_mmio.
* Inline the original hv_emulate_mmio.
* No longer check alignment. The handlers are responsible for handling
unaligned accesses.
v1 -> v2:
* Rename: do_io -> emulate_io.
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
For data structure types "struct key_info, struct pir_desc,
struct map_params", its name is identical with variable name
in the same scope. This MISRA C violation is detected by
static analysis tool.
Naming convention rule:If the data structure type is used by only one
module and its name meaning is simplistic, its name needs prefix
shorten module name.
The following udpates are made:
struct key_info-->struct trusty_key_info
struct pir_desc-->struct vlapic_pir_desc
struct map_params-->struct mem_map_params
Signed-off-by: Xiangyang Wu <xiangyang.wu@linux.intel.com>
Reviewed-by: Junjie Mao <junjie.mao@intel.com>
The current struct vcpu has two members, namely 'struct vhm_request req' and
'struct mem_io mmio', that hold similar info, including the address, direction, size,
value and status of mmio reqeusts.
As a step towards a unified framework for both MMIO/PIO, this patch unifies
these two members by a tailored version of vhm_reqeust, mostly with the reserved
fields dropped. The definitions to request types, directions and process status
are reused.
Handling errors during emulations will be revisited after the I/O emulation
paths are unified. Thus for this patch the mmio.mmio_status in inherited by
io_req.processed which is not yet properly processed.
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Add lookup_address to lokup the page table for a virtual address
of primary page table or a physical address of extended page table.
The remaining obtain_last_page_table_entry could be removed with their
called function later.
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
The ept_mr_add still use the old map_mem. The old API will
keep for a peroid until the SOS the same refine be merged.
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Merge mmu_modify with mmu_del to mmu_modify_or_del(..., type).
While type is MR_MODIFY, the actual action is doing mmu_modify; while
type is MR_DEL, the actual action is doing mmu_del.
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In the function scope,the parameter should not be
changed as Misra required.
V1->V2 recover some violations because of ldra's false positive.
V2->V3 sync local variable' type to parameter's type with the prefix of const.
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
1. fix some description for hv mmu_modify
2. add pml4_page input parameter for ept_mr_modify to keep align with
ept_mr_add and ept_mr_del which will support add or delete MR for trusty.
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
1. rename set_vm_memmap to set_vm_memory_region
2. split ept_mmap into ept_mr_add and ept_mr_del
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Added brackets for expression to make it easy to understand and
reduce the mistake of precedence. The rule is applied to the
mixed same level of prevedence opeartors, high level presedence
operators and logical expression.
Signed-off-by: Yang, Yu-chu <yu-chu.yang@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Most of the time, we use the virtual address of EPT PMl4 table,
not physical address.
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
There is chance to destroy vm whose ept table was not created, for
example, dm fails to setup mem for vm. In this case, it's better
to check the value of eptp before calling free_ept_mem() rather than
relying on assertion in the function.
Signed-off-by: Yan, Like <like.yan@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
In the hypervisor, VMCS fields include 16-bit fields,
32-bit fields, 64-bit fields and natural-width fields.
In the current implement, there are exec_vmread/exec_vmwrite
used for accessing 32-bit fields, 64-bit field and
natural-width fields. This usage will confue developer.
So there are many type casting for the return value and
parameters vmread/vmwrite operations.
Since exec_vmread/exec_vmwrite and exec_vmread64/exec_vmwrite64
are the same, update current exec_vmread/exec_vmwrite
implement into exec_vmread64/exec_vmwrite64 implement
and add MACRO define for exec_vmread/exec_vmwrite in
head file;
To access 64-bit fields in VMCS, callers use
exec_vmread64/exec_vmwrite64;
Update related variables type for vmread/vmwrite operations;
Update related caller according to VMCS fields size.
Note:Natural-width fields have 64 bits on processors
that support Intel 64 architecture.To access natural-width
fields in VMCS, callers still use exec_vmread/exec_vmwrite,
keep the current implementation.
V1--V2:
This is new part of this patch serial to only
update 64-bit vmread/vmread opertions and related
caller, for netural width fields, still use exec_vmread
or exec_vmwrite.
V2-->V3:
Fix few mistake updations for netural fields in VMCS,
just keep exec_vmread/exec_vmwrite to access them;
Fix few mistake updations for 64-bit fields in VMCS.
V3--V4:
Add "016ll" for 64-bit variable in log function;
Few updates for coding style;
Rename lssd32_idx as tr_sel in VMX module.
V4-->V5:
Use CPU_NATURAL_LAST in the vm_get_register and
vm_set_register to make condition statement more
understandable.
Signed-off-by: Xiangyang Wu <xiangyang.wu@intel.com>
Reviewed-by: Junjie Mao <junjie.mao@intel.com>
Includes header file of non-static function, and declare the
in-file use function static.
Signed-off-by: Yang, Yu-chu <yu-chu.yang@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
In the current memory module, there are many constatn value
without U/UL suffix, it is reported as MISRA C violations by
static analysis tool.
Add 'U/UL' suffix for unsigned contant value in memory module
as needed.
Note:In the most case, CPU_PAGE_SIZE(0x1000) is used as
unsigned integer contant value, so CPU_PAGE_SIZE is defined
as unsigned integer contant value, and it is safety converted
into unsigned long type according to MISRA C standard.
Signed-off-by: Xiangyang Wu <xiangyang.wu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
transfer num_vcpus,exp_num_vcpus to uint16_t.
transfer vm_hw_num_cores to uint16_t.
Signed-off-by: Huihuang Shi <huihuang.shi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
To be consistant with inline function mem_write, modifying the
mem_read* and using inline function instead.
Signed-off-by: Yang, Yu-chu <yu-chu.yang@intel.com>
No need to check the return value for memset
code like this:
int a(void) {
return 0;
}
int b(void){
a();
}
fix as follow:
int a(void) {
return 0;
}
int b(void){
(void)a();
}
Signed-off-by: Mingqiang Chi <mingqiang.chi@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In the hypervisor, virtual cpu id is defined as "int" or "uint32_t"
type in the hypervisor. So there are some sign conversion issues
about virtual cpu id (vcpu_id) reported by static analysis tool.
Sign conversion violates the rules of MISRA C:2012.
BTW, virtual cpu id has different names (vcpu_id, cpu_id, logical_id)
for different modules of HV, its type is defined as "int" or "uint32_t"
in the HV. cpu_id type and logical_id type clean up will be done in
other patchs.
V1-->V2:
More clean up the type of vcpu id;
"%hu" is for vcpu id in the print function.
Signed-off-by: Xiangyang Wu <xiangyang.wu@intel.com>
There are two prefix (aka TRC and TRACE) for trace event. This patch make all
the trace event prefix consist with TRACE.
No functional change.
Signed-off-by: Kaige Fu <kaige.fu@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>
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>
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 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>
set trusty eptp to 0 after destroy, and only invalidate trusty eptp when it
exist.
Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Acked-by: Xu, Anthony <anthony.xu@intel.com>
use copy_from_gva in vie_init, if copy_from_gva meet -EFAULT, inject #PF.
And for decode_instruction, if return -EFAULT, the caller should keep return
path with successful status.
v2:
- remove vm_restart_instruction when inject #PF
Signed-off-by: Jason Chen CJ <jason.cj.chen@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>
We need know which tlb to flush: ept or vpid.
1. error handle for invept.
it's the same with invvpid error handle.
change its name to compatible with vpid.
2. the macro name for flush ept tlb request.
Signed-off-by: Li, Fei1 <fei1.li@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The following massages is not fatal error but should be print to serial and
sbuf at the same time. pr_fatal is not a good choice. pr_acrnlog is designed
to deal with the situation. So replace the following misused pr_fatal with
pr_acrnlog.
Signed-off-by: Kaige Fu <kaige.fu@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
According to the syntax defined in C99, each struct/union field must have an
identifier. This patch removes unnamed struct/union fields that can be easily
expressed in a C99-compatible way.
Here is a summary of structs/unions removed.
struct vhm_request:
union {
uint32_t type; uint32_t type;
int32_t reserved0[16]; => int32_t reserved0[15];
};
struct vhm_request_buffer:
struct vhm_request_buffer {
union { union vhm_request_buffer {
struct vhm_request ...; => struct vhm_request ...;
int8_t reserved[4096]; int8_t reserved[4096];
} }
}
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
