hyperv: Fixes for get_vtl(), hv_vtl_apicid_to_vp_id()
Subsystems (3)
v4 cover hyperv: Fixes for get_vtl(), hv_vtl_apicid_to_vp_id()
The get_vtl(void) function
* has got one bug when the code started using a wrong pointer type after refactoring, and also
* the both function in question don't adhere to the requirements of the Hypervisor Top-Level Funactional Specification[1, 2] as the code overlaps the input and output areas for a hypercall.
The first issue leads to a wrong 100% reproducible computation due to reading a byte worth of data at a wrong offset. That in turn leads to using a nonsensical value ("fortunately", could catch it easily!) for the current VTL when initiating VMBus communications. As a repercussion from that, the system wouldn't boot. The fix is straightforward: use the correct pointer type.
The second issue doesn't seem to lead to any reproducible breakage just yet. It is fixed with using the output hypercall pages allocated per-CPU, and that isn't the only or the most obvious choice so let me elaborate why that fix appears to be the best one in my opinion out of the options I could conceive of.
The approach chosen for fixing the second issue makes two things shine through:
* these functions just get a vCPU register, no special treatment needs to be involved,
* VTLs and dom0 can and should share code as both exist to provide services to a guest(s), be that from within the partition or from outside of it.
The projected benefits include replacing the functions in question with a future hv_get_vp_registers one shared between dom0 and VTLs to allow for a better test coverage.
I have validated the fixes by booting the fixed kernel in VTL2 up using OpenVMM and OpenHCL[3, 4].
[1] https://learn.microsoft.com/en-us/virtualization/hyper-v-on-windows/tlfs/hypercall-interface
[2] https://github.com/MicrosoftDocs/Virtualization-Documentation/tree/main/tlfs
[3] https://openvmm.dev/guide/user_guide/openhcl.html
[4] https://github.com/microsoft/OpenVMM
[v4]
- Wrapped DECLARE_FLEX_ARRAY into a struct and added one more member as the documentation requires,
- Removed superfluous type coercion,
- Fixed tags,
- Rebased onto the latest hyperv-next branch.
[v3]: https://lore.kernel.org/lkml/[email protected]/
- Added a fix for hv_vtl_apicid_to_vp_id(),
- Split out the patch for enabling the hypercall output page,
- Updated the title of the patch series,
[v2]: https://lore.kernel.org/lkml/[email protected]/
- Used the suggestions to define an additional structure to improve code readability,
- Split out the patch with that definition.
[v1]: https://lore.kernel.org/lkml/[email protected]/
Roman Kisel (5):
hyperv: Define struct hv_output_get_vp_registers
hyperv: Fix pointer type in get_vtl(void)
hyperv: Enable the hypercall output page for the VTL mode
hyperv: Do not overlap the hvcall IO areas in get_vtl()
hyperv: Do not overlap the hvcall IO areas in hv_vtl_apicid_to_vp_id()
arch/x86/hyperv/hv_init.c | 6 ++-- arch/x86/hyperv/hv_vtl.c | 2 +- drivers/hv/hv_common.c | 6 ++-- include/hyperv/hvgdk_mini.h | 65 +++++++++++++++++++++++++++++++++++-- 4 files changed, 70 insertions(+), 9 deletions(-)
base-commit: 26e1b813fcd02984b1cac5f3decdf4b0bb56fe02
Signature · 1 line
2.34.1
PATCH v4 1/5 hyperv: Define struct hv_output_get_vp_registers
There is no definition of the output structure for the GetVpRegisters hypercall. Hence, using the hypercall is not possible when the output value has some structure to it. Even getting a datum of a primitive type reads as ad-hoc without that definition.
Define struct hv_output_get_vp_registers to enable using the GetVpRegisters hypercall. Make provisions for all supported architectures. No functional changes.
--- include/hyperv/hvgdk_mini.h | 65 +++++++++++++++++++++++++++++++++++-- 1 file changed, 63 insertions(+), 2 deletions(-)
include/hyperv/hvgdk_mini.h +63 -2
@@ -781,6 +781,8 @@ struct hv_timer_message_payload {__u64 delivery_time;/* When the message was delivered */ } __packed; +#if defined(CONFIG_X86) + struct hv_x64_segment_register { u64 base; u32 limit;
@@ -807,6 +809,8 @@ struct hv_x64_table_register {u64 base; } __packed; +#endif +union hv_input_vtl {u8 as_uint8;struct {
@@ -1068,6 +1072,41 @@ union hv_dispatch_suspend_register { } __packed; }; +#if defined(CONFIG_ARM64 ) + +union hv_arm64_pending_interruption_register { +u64 as_uint64; +struct { +u64 interruption_pending :1 ; +u64 interruption_type :1 ; +u64 reserved :30 ; +u32 error_code; + } __packed; +}; + +union hv_arm64_interrupt_state_register { +u64 as_uint64; +struct { +u64 interrupt_shadow :1 ; +u64 reserved :63 ; + } __packed; +}; + +union hv_arm64_pending_synthetic_exception_event { +u64 as_uint64[2 ]; +struct { +u32 event_pending :1 ; +u32 event_type :3 ; +u32 reserved :4 ; +u32 exception_type; +u64 context; + } __packed; +}; + +#endif + +#if defined(CONFIG_X86) + union hv_x64_interrupt_state_register { u64 as_uint64; struct {
@@ -1091,6 +1130,8 @@ union hv_x64_pending_interruption_register { } __packed; }; +#endif +union hv_register_value {struct hv_u128 reg128;u64 reg64;
@@ -1098,13 +1139,33 @@ union hv_register_value {u16 reg16;u8 reg8; -struct hv_x64_segment_register segment; -struct hv_x64_table_register table;union hv_explicit_suspend_register explicit_suspend;union hv_intercept_suspend_register intercept_suspend;union hv_dispatch_suspend_register dispatch_suspend; +#if defined(CONFIG_X86 ) +struct hv_x64_segment_register segment; +struct hv_x64_table_register table;union hv_x64_interrupt_state_register interrupt_state;union hv_x64_pending_interruption_register pending_interruption; +#endif +#if defined(CONFIG_ARM64 ) +union hv_arm64_pending_interruption_register pending_interruption; +union hv_arm64_interrupt_state_register interrupt_state; +union hv_arm64_pending_synthetic_exception_event pending_synthetic_exception_event; +#endif +}; + +/* +* NOTE: Linux helper struct - NOT from Hyper-V code. +* DECLARE_FLEX_ARRAY() needs to be wrapped into +* a structure and have at least one more member besides +* DECLARE_FLEX_ARRAY. +*/ +struct hv_output_get_vp_registers { +struct { + DECLARE_FLEX_ARRAY(union hv_register_value, values); +struct {} values_end; + }; };#if defined(CONFIG_ARM64 )
Signature · 1 line
2.34.1
Re: [PATCH v4 1/5] hyperv: Define struct hv_output_get_vp_registers
Quoted 6 lines · Roman Kisel
There is no definition of the output structure for the GetVpRegisters hypercall. Hence, using the hypercall is not possible when the output value has some structure to it. Even getting a datum of a primitive type reads as ad-hoc without that definition.
Quoted 9 lines
Define struct
hv_output_get_vp_registersto enable using the GetVpRegisters hypercall. Make provisions for all supported architectures. No functional changes.Signed-off-by Roman Kisel <[email protected]>--- include/hyperv/hvgdk_mini.h | 65 +++++++++++++++++++++++++++++++++++-- 1 file changed, 63 insertions(+), 2 deletions(-)
Looks good to me.
kernel test robot
build botRe: [PATCH v4 1/5] hyperv: Define struct hv_output_get_vp_registers
In reply to Roman Kiselkernel test robot noticed the following build errors:
Hi Roman,
kernel test robot noticed the following build errors:
[auto build test ERROR on 26e1b813fcd02984b1cac5f3decdf4b0bb56fe02]
url: https://github.com/intel-lab-lkp/linux/commits/Roman-Kisel/hyperv-Define-struct-hv_output_get_vp_registers/20241228-023454 base: 26e1b813fcd02984b1cac5f3decdf4b0bb56fe02 patch link: https://lore.kernel.org/r/20241227183155.122827-2-romank%40linux.microsoft.com patch subject: [PATCH v4 1/5] hyperv: Define struct hv_output_get_vp_registers config: arm64-randconfig-001-20241228 (https://download.01.org/0day-ci/archive/20241228/[email protected]/config) compiler: aarch64-linux-gcc (GCC) 14.2.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20241228/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All errors (new ones prefixed by >>):
In file included from include/hyperv/hvhdk_mini.h:8,
from include/hyperv/hvhdk.h:10,
from arch/arm64/hyperv/hv_core.c:17:
>> include/hyperv/hvgdk_mini.h:828:40: error: field 'cs' has incomplete type
828 | struct hv_x64_segment_register cs;
| ^~
>> include/hyperv/hvgdk_mini.h:829:40: error: field 'ds' has incomplete type
829 | struct hv_x64_segment_register ds;
| ^~
>> include/hyperv/hvgdk_mini.h:830:40: error: field 'es' has incomplete type
830 | struct hv_x64_segment_register es;
| ^~
>> include/hyperv/hvgdk_mini.h:831:40: error: field 'fs' has incomplete type
831 | struct hv_x64_segment_register fs;
| ^~
>> include/hyperv/hvgdk_mini.h:832:40: error: field 'gs' has incomplete type
832 | struct hv_x64_segment_register gs;
| ^~
>> include/hyperv/hvgdk_mini.h:833:40: error: field 'ss' has incomplete type
833 | struct hv_x64_segment_register ss;
| ^~
>> include/hyperv/hvgdk_mini.h:834:40: error: field 'tr' has incomplete type
834 | struct hv_x64_segment_register tr;
| ^~
>> include/hyperv/hvgdk_mini.h:835:40: error: field 'ldtr' has incomplete type
835 | struct hv_x64_segment_register ldtr;
| ^~~~
>> include/hyperv/hvgdk_mini.h:837:38: error: field 'idtr' has incomplete type
837 | struct hv_x64_table_register idtr;
| ^~~~
>> include/hyperv/hvgdk_mini.h:838:38: error: field 'gdtr' has incomplete type
838 | struct hv_x64_table_register gdtr;
| ^~~~
>> include/hyperv/hvhdk.h:78:56: error: field 'es' has incomplete type
78 | struct hv_x64_segment_register es;
| ^~
>> include/hyperv/hvhdk.h:79:56: error: field 'cs' has incomplete type
79 | struct hv_x64_segment_register cs;
| ^~
>> include/hyperv/hvhdk.h:80:56: error: field 'ss' has incomplete type
80 | struct hv_x64_segment_register ss;
| ^~
>> include/hyperv/hvhdk.h:81:56: error: field 'ds' has incomplete type
81 | struct hv_x64_segment_register ds;
| ^~
>> include/hyperv/hvhdk.h:82:56: error: field 'fs' has incomplete type
82 | struct hv_x64_segment_register fs;
| ^~
>> include/hyperv/hvhdk.h:83:56: error: field 'gs' has incomplete type
83 | struct hv_x64_segment_register gs;
| ^~
>> include/hyperv/hvhdk.h:86:48: error: array type has incomplete element type 'struct hv_x64_segment_register'
86 | struct hv_x64_segment_register segment_registers[6];
| ^~~~~~~~~~~~~~~~~
>> include/hyperv/hvhdk.h:95:52: error: field 'pending_interruption' has incomplete type
95 | union hv_x64_pending_interruption_register pending_interruption;
| ^~~~~~~~~~~~~~~~~~~~
>> include/hyperv/hvhdk.h:96:47: error: field 'interrupt_state' has incomplete type
96 | union hv_x64_interrupt_state_register interrupt_state;
| ^~~~~~~~~~~~~~~vim +/cs +828 include/hyperv/hvgdk_mini.h
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 822
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 823 struct hv_init_vp_context {
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 824 u64 rip;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 825 u64 rsp;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 826 u64 rflags;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 827
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @828 struct hv_x64_segment_register cs;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @829 struct hv_x64_segment_register ds;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @830 struct hv_x64_segment_register es;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @831 struct hv_x64_segment_register fs;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @832 struct hv_x64_segment_register gs;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @833 struct hv_x64_segment_register ss;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @834 struct hv_x64_segment_register tr;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @835 struct hv_x64_segment_register ldtr;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 836
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @837 struct hv_x64_table_register idtr;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 @838 struct hv_x64_table_register gdtr;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 839
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 840 u64 efer;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 841 u64 cr0;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 842 u64 cr3;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 843 u64 cr4;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 844 u64 msr_cr_pat;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 845 } __packed;
dd4f3a2b21ac14 Nuno Das Neves 2024-11-25 846 Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
RE: [PATCH v4 1/5] hyperv: Define struct hv_output_get_vp_registers
In reply to Roman KiselFrom: Roman Kisel <[email protected]> Sent: Friday, December 27, 2024 10:32 AM
Quoted 21 lines
There is no definition of the output structure for the GetVpRegisters hypercall. Hence, using the hypercall is not possible when the output value has some structure to it. Even getting a datum of a primitive type reads as ad-hoc without that definition.
Define struct
hv_output_get_vp_registersto enable using the GetVpRegisters hypercall. Make provisions for all supported architectures. No functional changes.Signed-off-by Roman Kisel <[email protected]>--- include/hyperv/hvgdk_mini.h | 65 +++++++++++++++++++++++++++++++++++-- 1 file changed, 63 insertions(+), 2 deletions(-)include/hyperv/hvgdk_mini.h +0
-0
include/hyperv/hvgdk_mini.h struct hv_timer_message_payload { · line 785
Quoted 5 lines
include/hyperv/hvgdk_mini.h +2
-0@@ -781,6 +781,8 @@ struct hv_timer_message_payload {__u64 delivery_time;/* When the message was delivered */ } __packed; +#if defined(CONFIG_X86 ) +
From a thread [1] with Nuno, my understanding is that the include/hyperv/* files should *not* use #ifdef <arch> unless strictly necessary because a structure or symbol is used in arch neutral code and has a different definition for each arch.
When the structure or symbol definition name contains "x64" or "arm64", such conflicts won't occur, and the #ifdef isn't needed. This does means that when compiling for either x86 or arm64, the symbols from both architectures will be visible in the symbol namespace, but that shouldn't cause any problems.
So it looks to me like none of the #ifdef's in this patch are needed. Nuno -- please jump in if I have misunderstood.
Michael
Quoted 102 lines
include/hyperv/hvgdk_mini.h +61
-2struct hv_x64_segment_register {u64 base;u32 limit;@@ -807,6 +809,8 @@ struct hv_x64_table_register {u64 base; } __packed; +#endif +union hv_input_vtl {u8 as_uint8;struct {@@ -1068,6 +1072,41 @@ union hv_dispatch_suspend_register { } __packed; }; +#if defined(CONFIG_ARM64 ) + +union hv_arm64_pending_interruption_register { +u64 as_uint64; +struct { +u64 interruption_pending :1 ; +u64 interruption_type :1 ; +u64 reserved :30 ; +u32 error_code; + } __packed; +}; + +union hv_arm64_interrupt_state_register { +u64 as_uint64; +struct { +u64 interrupt_shadow :1 ; +u64 reserved :63 ; + } __packed; +}; + +union hv_arm64_pending_synthetic_exception_event { +u64 as_uint64[2 ]; +struct { +u32 event_pending :1 ; +u32 event_type :3 ; +u32 reserved :4 ; +u32 exception_type; +u64 context; + } __packed; +}; + +#endif + +#if defined(CONFIG_X86) + union hv_x64_interrupt_state_register { u64 as_uint64; struct {@@ -1091,6 +1130,8 @@ union hv_x64_pending_interruption_register { } __packed; }; +#endif +union hv_register_value {struct hv_u128 reg128;u64 reg64;@@ -1098,13 +1139,33 @@ union hv_register_value {u16 reg16;u8 reg8; -struct hv_x64_segment_register segment; -struct hv_x64_table_register table;union hv_explicit_suspend_register explicit_suspend;union hv_intercept_suspend_register intercept_suspend;union hv_dispatch_suspend_register dispatch_suspend; +#if defined(CONFIG_X86 ) +struct hv_x64_segment_register segment; +struct hv_x64_table_register table;union hv_x64_interrupt_state_register interrupt_state;union hv_x64_pending_interruption_register pending_interruption; +#endif +#if defined(CONFIG_ARM64 ) +union hv_arm64_pending_interruption_register pending_interruption; +union hv_arm64_interrupt_state_register interrupt_state; +union hv_arm64_pending_synthetic_exception_event pending_synthetic_exception_event; +#endif +}; + +/* +* NOTE: Linux helper struct - NOT from Hyper-V code. +* DECLARE_FLEX_ARRAY() needs to be wrapped into +* a structure and have at least one more member besides +* DECLARE_FLEX_ARRAY. +*/ +struct hv_output_get_vp_registers { +struct { + DECLARE_FLEX_ARRAY(union hv_register_value, values); +struct {} values_end; + }; };#if
defined(CONFIG_ARM64)
--
2.34.1
Re: [PATCH v4 1/5] hyperv: Define struct hv_output_get_vp_registers
Quoted 1 line · Michael Kelley
From: Roman Kisel <[email protected]> Sent: Friday, December 27, 2024 10:32 AM
[...]
Quoted 12 lines
Quoted 1 line
From a thread [1] with Nuno, my understanding is that the
include/hyperv/*files should *not* use #ifdef <arch> unless strictly necessary because a structure or symbol is used in arch neutral code and has a different definition for each arch.When the structure or symbol definition name contains "x64" or "arm64", such conflicts won't occur, and the #ifdef isn't needed. This does means that when compiling for either x86 or arm64, the symbols from both architectures will be visible in the symbol namespace, but that shouldn't cause any problems.
Quoted 3 lines
So it looks to me like none of the #ifdef's in this patch are needed. Nuno -- please jump in if I have misunderstood.
Understood, thank you, Michael! From the thread you linked, it appears that there were other approaches with other trade-offs to be had, and this route was considered to be the most beneficial one.
Let me spin up another version with these guards removed. Thanks you all for bearing with me so far!
Quoted 106 lines
Michael
Quoted 102 lines
struct hv_x64_segment_register { u64 base; u32 limit;include/hyperv/hvgdk_mini.h +61
-2@@ -807,6 +809,8 @@ struct hv_x64_table_register {u64 base; } __packed; +#endif +union hv_input_vtl {u8 as_uint8;struct {@@ -1068,6 +1072,41 @@ union hv_dispatch_suspend_register { } __packed; }; +#if defined(CONFIG_ARM64 ) + +union hv_arm64_pending_interruption_register { +u64 as_uint64; +struct { +u64 interruption_pending :1 ; +u64 interruption_type :1 ; +u64 reserved :30 ; +u32 error_code; + } __packed; +}; + +union hv_arm64_interrupt_state_register { +u64 as_uint64; +struct { +u64 interrupt_shadow :1 ; +u64 reserved :63 ; + } __packed; +}; + +union hv_arm64_pending_synthetic_exception_event { +u64 as_uint64[2 ]; +struct { +u32 event_pending :1 ; +u32 event_type :3 ; +u32 reserved :4 ; +u32 exception_type; +u64 context; + } __packed; +}; + +#endif + +#if defined(CONFIG_X86) + union hv_x64_interrupt_state_register { u64 as_uint64; struct {@@ -1091,6 +1130,8 @@ union hv_x64_pending_interruption_register { } __packed; }; +#endif +union hv_register_value {struct hv_u128 reg128;u64 reg64;@@ -1098,13 +1139,33 @@ union hv_register_value {u16 reg16;u8 reg8; -struct hv_x64_segment_register segment; -struct hv_x64_table_register table;union hv_explicit_suspend_register explicit_suspend;union hv_intercept_suspend_register intercept_suspend;union hv_dispatch_suspend_register dispatch_suspend; +#if defined(CONFIG_X86 ) +struct hv_x64_segment_register segment; +struct hv_x64_table_register table;union hv_x64_interrupt_state_register interrupt_state;union hv_x64_pending_interruption_register pending_interruption; +#endif +#if defined(CONFIG_ARM64 ) +union hv_arm64_pending_interruption_register pending_interruption; +union hv_arm64_interrupt_state_register interrupt_state; +union hv_arm64_pending_synthetic_exception_event pending_synthetic_exception_event; +#endif +}; + +/* +* NOTE: Linux helper struct - NOT from Hyper-V code. +* DECLARE_FLEX_ARRAY() needs to be wrapped into +* a structure and have at least one more member besides +* DECLARE_FLEX_ARRAY. +*/ +struct hv_output_get_vp_registers { +struct { + DECLARE_FLEX_ARRAY(union hv_register_value, values); +struct {} values_end; + }; };#if
defined(CONFIG_ARM64)
--
2.34.1
Signature · 2 lines
Thank you, Roman
PATCH v4 2/5 hyperv: Fix pointer type in get_vtl(void)
In reply to Roman KiselCommit bc905fa8b633 ("hyperv: Switch from hyperv-tlfs.h to hyperv/hvhdk.h") changed the type of the output pointer to struct hv_register_assoc from struct hv_get_vp_registers_output. That leads to an incorrect computation, and leaves the system broken.
Use the correct pointer type for the output of the GetVpRegisters hypercall.
--- arch/x86/hyperv/hv_init.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
arch/x86/hyperv/hv_init.c +3 -3
@@ -416,13 +416,13 @@ static u8 __init get_vtl(void) {u64 control =HV_HYPERCALL_REP_COMP_1 |HVCALL_GET_VP_REGISTERS ;struct hv_input_get_vp_registers *input; -struct hv_register_assoc *output; +struct hv_output_get_vp_registers *output;unsigned long flags;u64 ret;local_irq_save (flags); input = *this_cpu_ptr (hyperv_pcpu_input_arg); - output = (struct hv_register_assoc *)input; + output = (struct hv_output_get_vp_registers *)input;memset (input,0 ,struct_size (input, names,1 )); input->partition_id =HV_PARTITION_ID_SELF ;
@@ -432,7 +432,7 @@ static u8 __init get_vtl(void) ret =hv_do_hypercall (control, input, output);if (hv_result_success (ret)) { - ret = output->value.reg8 &HV_X64_VTL_MASK ; + ret = output->values[0 ].reg8 &HV_X64_VTL_MASK ; }else {pr_err ("Failed to get VTL(error: %lld) exiting...\n" , ret);BUG ();
Signature · 1 line
2.34.1
PATCH v4 3/5 hyperv: Enable the hypercall output page for the VTL mode
In reply to Roman KiselDue to the hypercall page not being allocated in the VTL mode, the code resorts to using a part of the input page.
Allocate the hypercall output page in the VTL mode thus enabling it to use it for output and share code with dom0.
--- drivers/hv/hv_common.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/hv/hv_common.c +3 -3
@@ -340,7 +340,7 @@ int __init hv_common_init(void)BUG_ON (!hyperv_pcpu_input_arg);/* Allocate the per-CPU state for output arg for root */ -if (hv_root_partition) { +if (hv_root_partition ||IS_ENABLED (CONFIG_HYPERV_VTL_MODE )) { hyperv_pcpu_output_arg = alloc_percpu(void *);BUG_ON (!hyperv_pcpu_output_arg); }
@@ -435,7 +435,7 @@ int hv_common_cpu_init(unsigned int cpu)void **inputarg, **outputarg;u64 msr_vp_index;gfp_t flags; -int pgcount = hv_root_partition ?2 :1 ; +const int pgcount = (hv_root_partition ||IS_ENABLED (CONFIG_HYPERV_VTL_MODE )) ?2 :1 ;void *mem;int ret;
@@ -453,7 +453,7 @@ int hv_common_cpu_init(unsigned int cpu)if (!mem)return -ENOMEM ; -if (hv_root_partition) { +if (hv_root_partition ||IS_ENABLED (CONFIG_HYPERV_VTL_MODE )) { outputarg = (void **)this_cpu_ptr (hyperv_pcpu_output_arg); *outputarg = (char *)mem +HV_HYP_PAGE_SIZE ; }
Signature · 1 line
2.34.1
PATCH v4 4/5 hyperv: Do not overlap the hvcall IO areas in get_vtl()
In reply to Roman KiselThe Top-Level Functional Specification for Hyper-V, Section 3.6 [1, 2], disallows overlapping of the input and output hypercall areas, and get_vtl(void) does overlap them.
Use the output hypercall page of the current vCPU for the hypercall.
[1] https://learn.microsoft.com/en-us/virtualization/hyper-v-on-windows/tlfs/hypercall-interface
[2] https://github.com/MicrosoftDocs/Virtualization-Documentation/tree/main/tlfs
--- arch/x86/hyperv/hv_init.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/hyperv/hv_init.c +1 -1
@@ -422,7 +422,7 @@ static u8 __init get_vtl(void)local_irq_save (flags); input = *this_cpu_ptr (hyperv_pcpu_input_arg); - output = (struct hv_output_get_vp_registers *)input; + output = *this_cpu_ptr (hyperv_pcpu_output_arg);memset (input,0 ,struct_size (input, names,1 )); input->partition_id =HV_PARTITION_ID_SELF ;
Signature · 1 line
2.34.1
PATCH v4 5/5 hyperv: Do not overlap the hvcall IO areas in hv_vtl_apicid_to_vp_id()
In reply to Roman KiselThe Top-Level Functional Specification for Hyper-V, Section 3.6 [1, 2], disallows overlapping of the input and output hypercall areas, and hv_vtl_apicid_to_vp_id() overlaps them.
Use the output hypercall page of the current vCPU for the hypercall.
[1] https://learn.microsoft.com/en-us/virtualization/hyper-v-on-windows/tlfs/hypercall-interface
[2] https://github.com/MicrosoftDocs/Virtualization-Documentation/tree/main/tlfs
--- arch/x86/hyperv/hv_vtl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/hyperv/hv_vtl.c +1 -1
@@ -189,7 +189,7 @@ static int hv_vtl_apicid_to_vp_id(u32 apic_id) input->partition_id =HV_PARTITION_ID_SELF ; input->apic_ids[0 ] = apic_id; - output = (u32 *)input; + output = *this_cpu_ptr (hyperv_pcpu_output_arg); control =HV_HYPERCALL_REP_COMP_1 |HVCALL_GET_VP_ID_FROM_APIC_ID ; status =hv_do_hypercall (control, input, output);
Signature · 1 line
2.34.1
Quoted 63 lines · Roman Kisel
The
get_vtl(void)function* has got one bug when the code started using a wrong pointer type after refactoring, and also
* the both function in question don't adhere to the requirements of the Hypervisor Top-Level Funactional Specification[1, 2] as the code overlaps the input and output areas for a hypercall.The first issue leads to a wrong 100% reproducible computation due to reading a byte worth of data at a wrong offset. That in turn leads to using a nonsensical value ("fortunately", could catch it easily!) for the current VTL when initiating VMBus communications. As a repercussion from that, the system wouldn't boot. The fix is straightforward: use the correct pointer type.
The second issue doesn't seem to lead to any reproducible breakage just yet. It is fixed with using the output hypercall pages allocated per-CPU, and that isn't the only or the most obvious choice so let me elaborate why that fix appears to be the best one in my opinion out of the options I could conceive of.
The approach chosen for fixing the second issue makes two things shine through:
* these functions just get a vCPU register, no special treatment needs to be involved,
* VTLs and dom0 can and should share code as both exist to provide services to a guest(s), be that from within the partition or from outside of it.The projected benefits include replacing the functions in question with a future
hv_get_vp_registersone shared between dom0 and VTLs to allow for a better test coverage.I have validated the fixes by booting the fixed kernel in VTL2 up using OpenVMM and OpenHCL[3, 4].
[1] https://learn.microsoft.com/en-us/virtualization/hyper-v-on-windows/tlfs/hypercall-interface
[2] https://github.com/MicrosoftDocs/Virtualization-Documentation/tree/main/tlfs
[3] https://openvmm.dev/guide/user_guide/openhcl.html
[4] https://github.com/microsoft/OpenVMM[v4]
- WrappedDECLARE_FLEX_ARRAYinto a struct and added one more member as the documentation requires,
- Removed superfluous type coercion,
- Fixed tags,
- Rebased onto the latest hyperv-next branch.[v3]: https://lore.kernel.org/lkml/[email protected]/
- Added a fix forhv_vtl_apicid_to_vp_id(),
- Split out the patch for enabling the hypercall output page,
- Updated the title of the patch series,[v2]: https://lore.kernel.org/lkml/[email protected]/
- Used the suggestions to define an additional structure to improve code readability,
- Split out the patch with that definition.[v1]: https://lore.kernel.org/lkml/[email protected]/
Roman Kisel (5):
hyperv: Define structhv_output_get_vp_registers
hyperv: Fix pointer type inget_vtl(void)
hyperv: Enable the hypercall output page for the VTL mode
hyperv: Do not overlap the hvcall IO areas inget_vtl()
hyperv: Do not overlap the hvcall IO areas inhv_vtl_apicid_to_vp_id()
Quoted 8 lines
arch/x86/hyperv/hv_init.c | 6 ++-- arch/x86/hyperv/hv_vtl.c | 2 +- drivers/hv/hv_common.c | 6 ++-- include/hyperv/hvgdk_mini.h | 65 +++++++++++++++++++++++++++++++++++-- 4 files changed, 70 insertions(+), 9 deletions(-)base-commit:
26e1b813fcd02984b1cac5f3decdf4b0bb56fe02
Thank you for the persistence!
For the series,
Quoted 1 line · Easwar Hariharan
On 12/27/2024 10:31 AM, Roman Kisel wrote:
[...]
Quoted 2 lines
Thank you for the persistence!
I feel most fortunate learning from you, Michael, and Nuno :) Thank you!
Quoted 4 lines
For the series,
Reviewed-by Easwar Hariharan <[email protected]>
Signature · 2 lines
Thank you, Roman