x86: Add test cases for LAM
v3 cover x86: Add test cases for LAM
Intel Linear-address masking (LAM) [1], modifies the checking that is applied to *64-bit* linear addresses, allowing software to use of the untranslated address bits for metadata.
The patch series add test cases for KVM LAM:
Patch 1 makes change to allow setting of CR3 LAM bits in vmlaunch tests. Patch 2~4 add test cases for LAM supervisor mode and user mode, including:
- For supervisor mode
CR4.LAM_SUP toggle
Memory/MMIO access with tagged pointer
INVLPG
INVPCID
INVVPID (also used to cover VMX instruction VMExit path)
- For user mode
CR3 LAM bits toggle
Memory/MMIO access with tagged pointer
[1] Intel ISE https://cdrdv2.intel.com/v1/dl/getContent/671368
Chapter Linear Address Masking (LAM)
---
Changelog
v2 --> v3:
Simplify the implementation of set_metadata() (Suggested by Chao Gao) Move definition of X86_FEATURE_LAM to patch 1.
Remove strcpy case and unify memory & MMIO address access using MOV instruction. Some code cleanups.
Add reviewed-by from Chao Gao in patch 1.
v1 --> v2:
Add cases to test INVLPG, INVPCID, INVVPID with LAM_SUP Add cases to test LAM_{U48,U57}
Binbin Wu (3):
x86: Allow setting of CR3 LAM bits if LAM supported
x86: Add test cases for LAM_{U48,U57}
x86: Add test case for INVVPID with LAM
Robert Hoo (1):
x86: Add test case for LAM_SUP
lib/x86/processor.h | 7 + x86/Makefile.x86_64 | 1 + x86/lam.c | 315 ++++++++++++++++++++++++++++++++++++++++++++ x86/unittests.cfg | 10 ++ x86/vmx_tests.c | 66 +++++++++- 5 files changed, 398 insertions(+), 1 deletion(-) create mode 100644 x86/lam.c
Signature · 1 line
2.25.1
PATCH v3 1/4 x86: Allow setting of CR3 LAM bits if LAM supported
If LAM is supported, VM entry allows CR3.LAM_U48 (bit 62) and CR3.LAM_U57 (bit 61) to be set in CR3 field.
Change the test result expectations when setting CR3.LAM_U48 or CR3.LAM_U57 on vmlaunch tests when LAM is supported.
--- lib/x86/processor.h | 3 +++ x86/vmx_tests.c | 6 +++++- 2 files changed, 8 insertions(+), 1 deletion(-)
lib/x86/processor.h +3 -0
@@ -55,6 +55,8 @@#define X86_CR0_PG BIT(X86_CR0_PG_BIT)#define X86_CR3_PCID_MASK GENMASK(11, 0) +#define X86_CR3_LAM_U57_BIT (61) +#define X86_CR3_LAM_U48_BIT (62)#define X86_CR4_VME_BIT (0)#define X86_CR4_VME BIT(X86_CR4_VME_BIT)
@@ -248,6 +250,7 @@ static inline bool is_intel(void)#define X86_FEATURE_SPEC_CTRL (CPUID(0x7, 0, EDX, 26))#define X86_FEATURE_ARCH_CAPABILITIES (CPUID(0x7, 0, EDX, 29))#define X86_FEATURE_PKS (CPUID(0x7, 0, ECX, 31)) +#define X86_FEATURE_LAM (CPUID(0x7, 1, EAX, 26))/* * Extended Leafs, a.k.a. AMD defined
x86/vmx_tests.c +5 -1
@@ -7000,7 +7000,11 @@ static void test_host_ctl_regs(void) cr3 = cr3_saved | (1ul << i);vmcs_write (HOST_CR3 , cr3);report_prefix_pushf ("HOST_CR3 %lx" , cr3); -test_vmx_vmlaunch (VMXERR_ENTRY_INVALID_HOST_STATE_FIELD ); +if (this_cpu_has (X86_FEATURE_LAM ) && + ((i ==X86_CR3_LAM_U57_BIT ) || ( i ==X86_CR3_LAM_U48_BIT ))) +test_vmx_vmlaunch (0 ); +else +test_vmx_vmlaunch (VMXERR_ENTRY_INVALID_HOST_STATE_FIELD );report_prefix_pop (); }
Signature · 1 line
2.25.1
PATCH v3 2/4 x86: Add test case for LAM_SUP
In reply to Binbin WuFrom: Robert Hoo <[email protected]>
This unit test covers:
1. CR4.LAM_SUP toggles.
2. Memory & MMIO access with supervisor mode address with LAM metadata.
3. INVLPG memory operand doesn't contain LAM meta data, if the address is non-canonical form then the INVLPG is the same as a NOP (no #GP).
4. INVPCID memory operand (descriptor pointer) could contain LAM meta data, however, the address in the descriptor should be canonical.
In x86/unittests.cfg, add 2 test cases/guest conf, with and without LAM.
LAM feature spec: https://cdrdv2.intel.com/v1/dl/getContent/671368, Chap 7 LINEAR ADDRESS MASKING (LAM)
--- lib/x86/processor.h | 2 + x86/Makefile.x86_64 | 1 + x86/lam.c | 244 ++++++++++++++++++++++++++++++++++++++++++++ x86/unittests.cfg | 10 ++ 4 files changed, 257 insertions(+) create mode 100644 x86/lam.c
lib/x86/processor.h +2 -0
@@ -107,6 +107,8 @@#define X86_CR4_CET BIT(X86_CR4_CET_BIT)#define X86_CR4_PKS_BIT (24)#define X86_CR4_PKS BIT(X86_CR4_PKS_BIT) +#define X86_CR4_LAM_SUP_BIT (28) +#define X86_CR4_LAM_SUP BIT(X86_CR4_LAM_SUP_BIT)#define X86_EFLAGS_CF_BIT (0)#define X86_EFLAGS_CF BIT(X86_EFLAGS_CF_BIT)
x86/Makefile.x86_64 +1 -0
@@ -34,6 +34,7 @@ tests += $(TEST_DIR)/rdpru.$(exe) tests += $(TEST_DIR)/pks.$(exe) tests += $(TEST_DIR)/pmu_lbr.$(exe) tests += $(TEST_DIR)/pmu_pebs.$(exe) +tests += $(TEST_DIR)/lam.$(exe) ifeq ($(CONFIG_EFI),y) tests += $(TEST_DIR)/amd_sev.$(exe)
x86/lam.c new file +244 -0
244 lines not on this page · raw message
x86/unittests.cfg +10 -0
@@ -478,3 +478,13 @@ file = cet.flat arch = x86_64 smp = 2 extra_params = -enable-kvm -m 2048 -cpu host + +[intel-lam] +file = lam.flat +arch = x86_64 +extra_params = -enable-kvm -cpu host + +[intel-no-lam] +file = lam.flat +arch = x86_64 +extra_params = -enable-kvm -cpu host,-lam
Signature · 1 line
2.25.1
PATCH v3 3/4 x86: Add test cases for LAM_{U48,U57}
In reply to Binbin WuThis unit test covers:
1. CR3 LAM bits toggles.
2. Memory/MMIO access with user mode address containing LAM metadata.
--- lib/x86/processor.h | 2 ++ x86/lam.c | 71 +++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 73 insertions(+)
lib/x86/processor.h +2 -0
@@ -56,7 +56,9 @@#define X86_CR3_PCID_MASK GENMASK(11, 0)#define X86_CR3_LAM_U57_BIT (61) +#define X86_CR3_LAM_U57 BIT_ULL(X86_CR3_LAM_U57_BIT)#define X86_CR3_LAM_U48_BIT (62) +#define X86_CR3_LAM_U48 BIT_ULL(X86_CR3_LAM_U48_BIT)#define X86_CR4_VME_BIT (0)#define X86_CR4_VME BIT(X86_CR4_VME_BIT)
x86/lam.c +71 -0
110 lines not on this page · raw message
Signature · 1 line
2.25.1
PATCH v3 4/4 x86: Add test case for INVVPID with LAM
In reply to Binbin WuWhen LAM is on, the linear address of INVVPID operand can contain metadata, and the linear address in the INVVPID descriptor can contain metadata.
The added cases use tagged descriptor address or/and tagged target invalidation address to make sure the behaviors are expected when LAM is on.
Also, INVVPID cases can be used as the common test cases for VMX instruction VMExits.
--- x86/vmx_tests.c | 60 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 60 insertions(+)
x86/vmx_tests.c +60 -0
@@ -3225,6 +3225,65 @@ static void invvpid_test_not_in_vmx_operation(void)TEST_ASSERT (!vmx_on ()); } +#define LAM57_MASK GENMASK_ULL(62, 57) +#define LAM48_MASK GENMASK_ULL(62, 48) + +static inline u64 set_metadata (u64 src,u64 metadata_mask) +{ +return (src & ~metadata_mask) | (NONCANONICAL & metadata_mask); +} + +/* LAM applies to the target address inside the descriptor of invvpid */ +static void invvpid_test_lam (void ) +{ +void *vaddr; +struct invvpid_operand *operand; +u64 lam_mask =LAM48_MASK ; +bool fault; + +if (!this_cpu_has (X86_FEATURE_LAM )) { +report_skip ("LAM is not supported, skip INVVPID with LAM" ); +return ; + } + +if (this_cpu_has (X86_FEATURE_LA57 ) &&read_cr4 () &X86_CR4_LA57 ) + lam_mask =LAM57_MASK ; + + vaddr =alloc_vpage (); +install_page (current_page_table (),virt_to_phys (alloc_page ()), vaddr); +/* +* Since the stack memory address in KUT doesn't follow kernel address +* space partition rule, reuse the memory address for descriptor and +* the target address in the descriptor of invvpid. +*/ + operand = (struct invvpid_operand *)vaddr; + operand->vpid =0xffff ; + operand->gla = (u64 )vaddr; + +write_cr4_safe (read_cr4 () |X86_CR4_LAM_SUP ); +if (!(read_cr4 () &X86_CR4_LAM_SUP )) { +report_skip ("Failed to enable LAM_SUP" ); +return ; + } + + operand = (struct invvpid_operand *)vaddr; + operand->gla =set_metadata (operand->gla, lam_mask); + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): untagged pointer + tagged addr" ); + + operand = (struct invvpid_operand *)set_metadata ((u64 )operand, lam_mask); + operand->gla = (u64 )vaddr; + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): tagged pointer + untagged addr" ); + + operand = (struct invvpid_operand *)set_metadata ((u64 )operand, lam_mask); + operand->gla =set_metadata (operand->gla, lam_mask); + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): tagged pointer + tagged addr" ); + +write_cr4_safe (read_cr4 () & ~X86_CR4_LAM_SUP ); +} +/* * This does not test real-address mode, virtual-8086 mode, protected mode, * or CPL > 0.
@@ -3282,6 +3341,7 @@ static void invvpid_test(void)invvpid_test_pf ();invvpid_test_compatibility_mode ();invvpid_test_not_in_vmx_operation (); +invvpid_test_lam (); }/*
Signature · 1 line
2.25.1
Re: [kvm-unit-tests v3 1/4] x86: Allow setting of CR3 LAM bits if LAM supported
In reply to Binbin WuQuoted 39 lines · Binbin Wu
If LAM is supported, VM entry allows CR3.LAM_U48 (bit 62) and CR3.LAM_U57 (bit 61) to be set in CR3 field.
Change the test result expectations when setting CR3.LAM_U48 or CR3.LAM_U57 on vmlaunch tests when LAM is supported.
Signed-off-by Binbin Wu <[email protected]>Reviewed-by Chao Gao <[email protected]>---
lib/x86/processor.h| 3 +++x86/vmx_tests.c| 6 +++++-2 files changed, 8 insertions(+), 1 deletion(-)lib/x86/processor.h +3
-0@@ -55,6 +55,8 @@#define X86_CR0_PG BIT(X86_CR0_PG_BIT)#define X86_CR3_PCID_MASK GENMASK(11, 0) +#define X86_CR3_LAM_U57_BIT (61) +#define X86_CR3_LAM_U48_BIT (62)#define X86_CR4_VME_BIT (0)#define X86_CR4_VME BIT(X86_CR4_VME_BIT)@@ -248,6 +250,7 @@ static inline bool is_intel(void)#define X86_FEATURE_SPEC_CTRL (CPUID(0x7, 0, EDX, 26))#define X86_FEATURE_ARCH_CAPABILITIES (CPUID(0x7, 0, EDX, 29))#define X86_FEATURE_PKS (CPUID(0x7, 0, ECX, 31)) +#define X86_FEATURE_LAM (CPUID(0x7, 1, EAX, 26))/* * Extended Leafs, a.k.a. AMD defined x86/vmx_tests.c +0
-0
x86/vmx_tests.c static void test_host_ctl_regs(void) · line 7004
Quoted 6 lines
x86/vmx_tests.c +2
-1@@ -7000,7 +7000,11 @@ static void test_host_ctl_regs(void) cr3 = cr3_saved | (1ul << i);vmcs_write (HOST_CR3 , cr3);report_prefix_pushf ("HOST_CR3 %lx" , cr3); -test_vmx_vmlaunch (VMXERR_ENTRY_INVALID_HOST_STATE_FIELD ); +if (this_cpu_has (X86_FEATURE_LAM ) && + ((i ==X86_CR3_LAM_U57_BIT ) || ( i ==X86_CR3_LAM_U48_BIT )))
^ stray space
Quoted 9 lines
x86/vmx_tests.c +3
-0+test_vmx_vmlaunch (0 ); + else +test_vmx_vmlaunch (VMXERR_ENTRY_INVALID_HOST_STATE_FIELD );report_prefix_pop (); }Signature · 1 line
2.25.1
Re: [kvm-unit-tests v3 2/4] x86: Add test case for LAM_SUP
In reply to Binbin WuQuoted 10 lines · Binbin Wu
From: Robert Hoo <[email protected]>
This unit test covers:
1. CR4.LAM_SUP toggles.
2. Memory & MMIO access with supervisor mode address with LAM metadata.
3. INVLPG memory operand doesn't contain LAM meta data, if the address is non-canonical form then the INVLPG is the same as a NOP (no #GP).
4. INVPCID memory operand (descriptor pointer) could contain LAM meta data, however, the address in the descriptor should be canonical.
Quoted 8 lines
In x86/unittests.cfg, add 2 test cases/guest conf, with and without LAM.
LAM feature spec: https://cdrdv2.intel.com/v1/dl/getContent/671368, Chap 7 LINEAR ADDRESS MASKING (LAM)
Signed-off-by Robert Hoo <[email protected]>Co-developed-by Binbin Wu <[email protected]>Signed-off-by Binbin Wu <[email protected]>
one nit below
x86/lam.c line 138
Quoted 3 lines
x86/lam.c new file +3
-0+static void do_invpcid (void *desc) +{ +unsigned long type =0 ;
the local variable isn't needed ...
x86/lam.c line 142
Quoted 4 lines
x86/lam.c new file +4
-0+struct invpcid_desc *desc_ptr = (struct invpcid_desc *)desc; + + asmvolatile ("invpcid %0, %1" : + : "m" (*desc_ptr), "r" (type)
You can simply use
"r" (0)
.
Quoted 2 lines
(quoted diff) +2
-0+ : "memory"); +}
Re: [kvm-unit-tests v3 3/4] x86: Add test cases for LAM_{U48,U57}
In reply to Binbin WuQuoted 5 lines · Binbin Wu
This unit test covers:
1. CR3 LAM bits toggles.
2. Memory/MMIO access with user mode address containing LAM metadata.Signed-off-by Binbin Wu <[email protected]>
two nits below:
x86/lam.c static void test_lam_sup(bool has_lam, bool fep_available) · line 285
Quoted 11 lines
---
x86/lam.c +10
-0+static void test_lam_user (bool has_lam) +{ +phys_addr_t paddr; +unsigned long cr3 =read_cr3 (); + +/* +* The physical address width is within 36 bits, so that using identical +* mapping, the linear address will be considered as user mode address +* from the view of LAM. +*/
Why 36 bits (i.e., 64G)?
would you mind adding a comment in patch 2 to explain why the virtual addresses are kernel mode addresses?
x86/lam.c static void test_lam_sup(bool has_lam, bool fep_available) · line 288
Quoted 3 lines
x86/lam.c +3
-0+ paddr =virt_to_phys (alloc_page ()); +install_page ((void *)cr3, paddr, (void *)paddr); +install_page ((void *)cr3,IORAM_BASE_PHYS , (void *)IORAM_BASE_PHYS );
are the two lines necessary?
Quoted 19 lines
x86/lam.c +5
-0+ +test_lam_user_mode (has_lam,LAM48_MASK , paddr,IORAM_BASE_PHYS ); +test_lam_user_mode (has_lam,LAM57_MASK , paddr,IORAM_BASE_PHYS ); +} +int
main(int ac, char **av){ bool has_lam;x86/lam.c +1
-0@@ -239,6 +309,7 @@ int main(int ac, char **av) "use kvm.force_emulation_prefix=1 to enable\n");test_lam_sup (has_lam, fep_available); +test_lam_user (has_lam);return report_summary (); }Signature · 1 line
2.25.1
Re: [kvm-unit-tests v3 4/4] x86: Add test case for INVVPID with LAM
In reply to Binbin WuQuoted 11 lines · Binbin Wu
When LAM is on, the linear address of INVVPID operand can contain metadata, and the linear address in the INVVPID descriptor can contain metadata.
The added cases use tagged descriptor address or/and tagged target invalidation address to make sure the behaviors are expected when LAM is on.
Also, INVVPID cases can be used as the common test cases for VMX instruction VMExits.Signed-off-by Binbin Wu <[email protected]>
with a few cosmetic comments below:
Quoted 9 lines
---
x86/vmx_tests.c| 60 +++++++++++++++++++++++++++++++++++++++++++++++++1 file changed, 60 insertions(+)x86/vmx_tests.c +0
-0
x86/vmx_tests.c static void invvpid_test_not_in_vmx_operation(void) · line 3234
Quoted 10 lines
x86/vmx_tests.c +7
-0@@ -3225,6 +3225,65 @@ static void invvpid_test_not_in_vmx_operation(void)TEST_ASSERT (!vmx_on ()); } +#define LAM57_MASK GENMASK_ULL(62, 57) +#define LAM48_MASK GENMASK_ULL(62, 48) + +static inline u64 set_metadata (u64 src,u64 metadata_mask) +{ +return (src & ~metadata_mask) | (NONCANONICAL & metadata_mask); +}
Can you move the duplicate defintions and functions to a header file?
Quoted 13 lines
x86/vmx_tests.c +13
-0+ +/* LAM applies to the target address inside the descriptor of invvpid */ +static void invvpid_test_lam (void ) +{ +void *vaddr; +struct invvpid_operand *operand; +u64 lam_mask =LAM48_MASK ; +bool fault; + +if (!this_cpu_has (X86_FEATURE_LAM )) { +report_skip ("LAM is not supported, skip INVVPID with LAM" ); +return ; + }
...
Quoted 10 lines
x86/vmx_tests.c +10
-0+ +if (this_cpu_has (X86_FEATURE_LA57 ) &&read_cr4 () &X86_CR4_LA57 ) + lam_mask =LAM57_MASK ; + + vaddr =alloc_vpage (); +install_page (current_page_table (),virt_to_phys (alloc_page ()), vaddr); +/* +* Since the stack memory address in KUT doesn't follow kernel address +* space partition rule, reuse the memory address for descriptor and +* the target address in the descriptor of invvpid.
x86/vmx_tests.c static void invvpid_test_not_in_vmx_operation(void) · line 3267
Quoted 10 lines
x86/vmx_tests.c +10
-0+*/ + operand = (struct invvpid_operand *)vaddr; + operand->vpid =0xffff ; + operand->gla = (u64 )vaddr; + +write_cr4_safe (read_cr4 () |X86_CR4_LAM_SUP ); +if (!(read_cr4 () &X86_CR4_LAM_SUP )) { +report_skip ("Failed to enable LAM_SUP" ); +return ; + }
It might be better to enable LAM_SUP right after above check for the LAM CPUID bit. And no need to verify the result because there is a dedicated test case already in patch 2.
Quoted 23 lines
x86/vmx_tests.c +19
-0+ + operand = (struct invvpid_operand *)vaddr; + operand->gla =set_metadata (operand->gla, lam_mask); + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): untagged pointer + tagged addr" ); + + operand = (struct invvpid_operand *)set_metadata ((u64 )operand, lam_mask); + operand->gla = (u64 )vaddr; + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): tagged pointer + untagged addr" ); + + operand = (struct invvpid_operand *)set_metadata ((u64 )operand, lam_mask); + operand->gla =set_metadata (operand->gla, lam_mask); + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): tagged pointer + tagged addr" ); + +write_cr4_safe (read_cr4 () & ~X86_CR4_LAM_SUP ); +} +/*
* This does not test real-address mode, virtual-8086 mode, protected mode,
* or CPL > 0.
x86/vmx_tests.c static void invvpid_test(void) · line 3344
Quoted 4 lines
x86/vmx_tests.c +1
-0@@ -3282,6 +3341,7 @@ static void invvpid_test(void)invvpid_test_pf ();invvpid_test_compatibility_mode ();invvpid_test_not_in_vmx_operation (); +invvpid_test_lam ();
operand->gla is checked only in INVVPID_ADDR mode. So, the lam test should be moved under "if (types & (1u << INVVPID_ADDR))" a few lines above.
Quoted 6 lines
x86/vmx_tests.c +0
-0}/* Signature · 1 line
2.25.1
Re: [kvm-unit-tests v3 1/4] x86: Allow setting of CR3 LAM bits if LAM supported
In reply to Chao GaoQuoted 33 lines · Chao Gao
Quoted 32 lines · Binbin Wu
If LAM is supported, VM entry allows CR3.LAM_U48 (bit 62) and CR3.LAM_U57 (bit 61) to be set in CR3 field.
Change the test result expectations when setting CR3.LAM_U48 or CR3.LAM_U57 on vmlaunch tests when LAM is supported.
Signed-off-by Binbin Wu <[email protected]>Reviewed-by Chao Gao <[email protected]>---
lib/x86/processor.h| 3 +++x86/vmx_tests.c| 6 +++++-2 files changed, 8 insertions(+), 1 deletion(-)lib/x86/processor.h +3
-0@@ -55,6 +55,8 @@#define X86_CR0_PG BIT(X86_CR0_PG_BIT)#define X86_CR3_PCID_MASK GENMASK(11, 0) +#define X86_CR3_LAM_U57_BIT (61) +#define X86_CR3_LAM_U48_BIT (62)#define X86_CR4_VME_BIT (0)#define X86_CR4_VME BIT(X86_CR4_VME_BIT)@@ -248,6 +250,7 @@ static inline bool is_intel(void)#define X86_FEATURE_SPEC_CTRL (CPUID(0x7, 0, EDX, 26))#define X86_FEATURE_ARCH_CAPABILITIES (CPUID(0x7, 0, EDX, 29))#define X86_FEATURE_PKS (CPUID(0x7, 0, ECX, 31)) +#define X86_FEATURE_LAM (CPUID(0x7, 1, EAX, 26))
Quoted 14 lines
x86/vmx_tests.c static void test_host_ctl_regs(void) · line 7004
Quoted 13 lines
lib/x86/processor.h +0
-0/* * Extended Leafs, a.k.a. AMD defined x86/vmx_tests.c +2
-1@@ -7000,7 +7000,11 @@ static void test_host_ctl_regs(void) cr3 = cr3_saved | (1ul << i);vmcs_write (HOST_CR3 , cr3);report_prefix_pushf ("HOST_CR3 %lx" , cr3); -test_vmx_vmlaunch (VMXERR_ENTRY_INVALID_HOST_STATE_FIELD ); +if (this_cpu_has (X86_FEATURE_LAM ) && + ((i ==X86_CR3_LAM_U57_BIT ) || ( i ==X86_CR3_LAM_U48_BIT )))^ stray space
Thanks, will remove it.
Quoted 10 lines
Quoted 10 lines
x86/vmx_tests.c +3
-0+test_vmx_vmlaunch (0 ); + else +test_vmx_vmlaunch (VMXERR_ENTRY_INVALID_HOST_STATE_FIELD );report_prefix_pop (); }Signature · 1 line
2.25.1
Re: [kvm-unit-tests v3 3/4] x86: Add test cases for LAM_{U48,U57}
In reply to Chao GaoQuoted 10 lines · Chao Gao
Quoted 5 lines · Binbin Wu
This unit test covers:
1. CR3 LAM bits toggles.
2. Memory/MMIO access with user mode address containing LAM metadata.Signed-off-by Binbin Wu <[email protected]>Reviewed-by Chao Gao <[email protected]>two nits below:
Quoted 12 lines
x86/lam.c static void test_lam_sup(bool has_lam, bool fep_available) · line 285
Quoted 11 lines
---
x86/lam.c +10
-0+static void test_lam_user (bool has_lam) +{ +phys_addr_t paddr; +unsigned long cr3 =read_cr3 (); + +/* +* The physical address width is within 36 bits, so that using identical +* mapping, the linear address will be considered as user mode address +* from the view of LAM. +*/ Why 36 bits (i.e., 64G)?
KUT memory allocator supports 4 Memory Areas:AREA_NORMAL, AREA_HIGH, AREA_LOW and AREA_LOWEST. AREA_NORMAL has the maximum address width, which is 36 bits.
How about change the comment to this:
+ /*
+ * The physical address width supported by KUT memory allocator is within 36 bits,
+ * so that using identical mapping, the linear address will be considered as user
+ * mode address from the view of LAM.
+ */
Quoted 3 lines
would you mind adding a comment in patch 2 to explain why the virtual addresses are kernel mode addresses?
OK. Will add the following comments:
/*
* KUT initializes vfree_top to 0 for X86_64, and each virtual address
* allocation decreases the size from vfree_top. It's guaranteed that
* the return value of alloc_vpage() is considered as kernel mode
* address and canonical since only a small mount virtual address range
* is allocated in this test.
*/
Quoted 5 lines
x86/lam.c static void test_lam_sup(bool has_lam, bool fep_available) · line 288
Quoted 4 lines
x86/lam.c +3
-0+ paddr =virt_to_phys (alloc_page ()); +install_page ((void *)cr3, paddr, (void *)paddr); +install_page ((void *)cr3,IORAM_BASE_PHYS , (void *)IORAM_BASE_PHYS );are the two lines necessary?
The two lines are not necessary.
Check setup_mmu(), it already sets up the identical mapping for avialble physcial memory/MMIO. Will remove them.
Quoted 20 lines
Quoted 20 lines
x86/lam.c +5
-0+ +test_lam_user_mode (has_lam,LAM48_MASK , paddr,IORAM_BASE_PHYS ); +test_lam_user_mode (has_lam,LAM57_MASK , paddr,IORAM_BASE_PHYS ); +} +int
main(int ac, char **av){ bool has_lam;x86/lam.c +1
-0@@ -239,6 +309,7 @@ int main(int ac, char **av) "use kvm.force_emulation_prefix=1 to enable\n");test_lam_sup (has_lam, fep_available); +test_lam_user (has_lam);return report_summary (); }Signature · 1 line
2.25.1
Re: [kvm-unit-tests v3 4/4] x86: Add test case for INVVPID with LAM
In reply to Chao GaoQuoted 28 lines · Chao Gao
Quoted 11 lines · Binbin Wu
When LAM is on, the linear address of INVVPID operand can contain metadata, and the linear address in the INVVPID descriptor can contain metadata.
The added cases use tagged descriptor address or/and tagged target invalidation address to make sure the behaviors are expected when LAM is on.
Also, INVVPID cases can be used as the common test cases for VMX instruction VMExits.Signed-off-by Binbin Wu <[email protected]>Reviewed-by Chao Gao <[email protected]>with a few cosmetic comments below:
Quoted 12 lines
---
x86/vmx_tests.c| 60 +++++++++++++++++++++++++++++++++++++++++++++++++1 file changed, 60 insertions(+)x86/vmx_tests.c +0
-0@@ -3225,6 +3225,65 @@ static void invvpid_test_not_in_vmx_operation(void)TEST_ASSERT (!vmx_on ()); }
Quoted 8 lines
x86/vmx_tests.c static void invvpid_test_not_in_vmx_operation(void) · line 3234
Quoted 7 lines
x86/vmx_tests.c +7
-0+#define LAM57_MASK GENMASK_ULL(62, 57) +#define LAM48_MASK GENMASK_ULL(62, 48) + +static inline u64 set_metadata (u64 src,u64 metadata_mask) +{ +return (src & ~metadata_mask) | (NONCANONICAL & metadata_mask); +}Can you move the duplicate defintions and functions to a header file?
Then add a new header file lam.h?
Didn't find a suitable existant header file to add these definitions.
Quoted 31 lines
Quoted 14 lines
x86/vmx_tests.c +13
-0+ +/* LAM applies to the target address inside the descriptor of invvpid */ +static void invvpid_test_lam (void ) +{ +void *vaddr; +struct invvpid_operand *operand; +u64 lam_mask =LAM48_MASK ; +bool fault; + +if (!this_cpu_has (X86_FEATURE_LAM )) { +report_skip ("LAM is not supported, skip INVVPID with LAM" ); +return ; + }...
Quoted 15 lines
x86/vmx_tests.c +15
-0+ +if (this_cpu_has (X86_FEATURE_LA57 ) &&read_cr4 () &X86_CR4_LA57 ) + lam_mask =LAM57_MASK ; + + vaddr =alloc_vpage (); +install_page (current_page_table (),virt_to_phys (alloc_page ()), vaddr); +/* +* Since the stack memory address in KUT doesn't follow kernel address +* space partition rule, reuse the memory address for descriptor and +* the target address in the descriptor of invvpid. +*/ + operand = (struct invvpid_operand *)vaddr; + operand->vpid =0xffff ; + operand->gla = (u64 )vaddr; +
Quoted 8 lines
x86/vmx_tests.c static void invvpid_test_not_in_vmx_operation(void) · line 3267
Quoted 5 lines
x86/vmx_tests.c +5
-0+write_cr4_safe (read_cr4 () |X86_CR4_LAM_SUP ); +if (!(read_cr4 () &X86_CR4_LAM_SUP )) { +report_skip ("Failed to enable LAM_SUP" ); +return ; + }It might be better to enable
LAM_SUPright after above check for the LAM CPUID bit. And no need to verify the result because there is a dedicated test case already in patch 2.
OK.
Quoted 21 lines
Quoted 21 lines
x86/vmx_tests.c +19
-0+ + operand = (struct invvpid_operand *)vaddr; + operand->gla =set_metadata (operand->gla, lam_mask); + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): untagged pointer + tagged addr" ); + + operand = (struct invvpid_operand *)set_metadata ((u64 )operand, lam_mask); + operand->gla = (u64 )vaddr; + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): tagged pointer + untagged addr" ); + + operand = (struct invvpid_operand *)set_metadata ((u64 )operand, lam_mask); + operand->gla =set_metadata (operand->gla, lam_mask); + fault =test_for_exception (GP_VECTOR , ds_invvpid, operand); +report (!fault,"INVVPID (LAM on): tagged pointer + tagged addr" ); + +write_cr4_safe (read_cr4 () & ~X86_CR4_LAM_SUP ); +} +/*
* This does not test real-address mode, virtual-8086 mode, protected mode,
Quoted 8 lines
x86/vmx_tests.c static void invvpid_test(void) · line 3344
Quoted 6 lines
* or CPL > 0.
x86/vmx_tests.c +1
-0@@ -3282,6 +3341,7 @@ static void invvpid_test(void)invvpid_test_pf ();invvpid_test_compatibility_mode ();invvpid_test_not_in_vmx_operation (); +invvpid_test_lam ();
operand->glais checked only inINVVPID_ADDRmode. So, the lam test should be moved under "if (types & (1u <<INVVPID_ADDR))" a few lines above.
Yes, will update it.
Quoted 7 lines
Quoted 7 lines
x86/vmx_tests.c +0
-0}/* Signature · 1 line
2.25.1
Re: [kvm-unit-tests v3 4/4] x86: Add test case for INVVPID with LAM
Quoted 40 lines · Binbin Wu
Quoted 29 lines · Chao Gao
Quoted 11 lines · Binbin Wu
When LAM is on, the linear address of INVVPID operand can contain metadata, and the linear address in the INVVPID descriptor can contain metadata.
The added cases use tagged descriptor address or/and tagged target invalidation address to make sure the behaviors are expected when LAM is on.
Also, INVVPID cases can be used as the common test cases for VMX instruction VMExits.Signed-off-by Binbin Wu <[email protected]>Reviewed-by Chao Gao <[email protected]>with a few cosmetic comments below:
Quoted 13 lines
---
x86/vmx_tests.c| 60 +++++++++++++++++++++++++++++++++++++++++++++++++1 file changed, 60 insertions(+)x86/vmx_tests.c +0
-0@@ -3225,6 +3225,65 @@ static voidinvvpid_test_not_in_vmx_operation (void ) TEST_ASSERT(!vmx_on()); }Quoted 8 lines
x86/vmx_tests.c static void · line 3235
Quoted 7 lines
x86/vmx_tests.c +7
-0+#define LAM57_MASK GENMASK_ULL(62, 57) +#define LAM48_MASK GENMASK_ULL(62, 48) + +static inline u64 set_metadata (u64 src,u64 metadata_mask) +{ + return (src & ~metadata_mask) | (NONCANONICAL & metadata_mask); +}Can you move the duplicate defintions and functions to a header file?
Quoted 2 lines
Then add a new header file
lam.h?
Didn't find a suitable existant header file to add these definitions.
I am about to move the definitions to lib/x86/processor.h and rename set_metadata() to set_la_non_canonical() to be more generic.
index e00a32b..236b537 100644
lib/x86/processor.h +0 -0
@@ -8,6 +8,14 @@
#include <stdint.h>
#define NONCANONICAL 0xaaaaaaaaaaaaaaaaull
+#define LAM57_MASK GENMASK_ULL(62, 57)
+#define LAM48_MASK GENMASK_ULL(62, 48)
+
+/* Set metadata with non-canonical pattern in mask bits of a linear address */ +static inline u64 set_la_non_canonical(u64 src, u64 mask)
+{
+ return (src & ~mask) | (NONCANONICAL & mask);
+}
Quoted 93 lines
Quoted 37 lines
Quoted 17 lines
x86/vmx_tests.c +2
-0+ +/* LAM applies to the target address inside the descriptor of invvpid */
x86/vmx_tests.c +11
-0+static void invvpid_test_lam (void ) +{ + void *vaddr; + struct invvpid_operand *operand; + u64 lam_mask = LAM48_MASK; + bool fault; + + if (!this_cpu_has(X86_FEATURE_LAM)) { + report_skip("LAM is not supported, skip INVVPID with LAM"); + return; + }...
Quoted 18 lines
x86/vmx_tests.c +6
-0+ + if (this_cpu_has(X86_FEATURE_LA57) && read_cr4() & X86_CR4_LA57) + lam_mask = LAM57_MASK; + + vaddr = alloc_vpage(); + install_page(current_page_table(), virt_to_phys(alloc_page()),vaddr);
+ /*
+ * Since the stack memory address in KUT doesn't follow kernel address
+ * space partition rule, reuse the memory address for descriptor and
+ * the target address in the descriptor of invvpid. + */
+ operand = (structinvvpid_operand*)vaddr;
+operand->vpid=0xffff;
+operand->gla= (u64)vaddr;
+Quoted 8 lines
Quoted 5 lines
+
write_cr4_safe(read_cr4() | X86_CR4_LAM_SUP); + if (!(read_cr4() &X86_CR4_LAM_SUP)) { +report_skip("Failed to enable LAM_SUP"); + return;
+ }It might be better to enable
LAM_SUPright after above check for the LAM CPUID bit. And no need to verify the result because there is a dedicated test case already in patch 2.OK.
Quoted 25 lines
Quoted 25 lines
+
+ operand = (structinvvpid_operand*)vaddr;
+operand->gla=set_metadata(operand->gla, lam_mask);
+ fault =test_for_exception(GP_VECTOR, ds_invvpid, operand); +report(!fault, "INVVPID (LAM on): untagged pointer + tagged addr");
+
+ operand = (structinvvpid_operand*)set_metadata((u64)operand,lam_mask);
+operand->gla= (u64)vaddr;
+ fault =test_for_exception(GP_VECTOR, ds_invvpid, operand); +report(!fault, "INVVPID (LAM on): tagged pointer + untagged addr");
+
+ operand = (structinvvpid_operand*)set_metadata((u64)operand,lam_mask);
+operand->gla=set_metadata(operand->gla, lam_mask);
+ fault =test_for_exception(GP_VECTOR, ds_invvpid, operand); +report(!fault, "INVVPID (LAM on): tagged pointer + tagged addr");x86/vmx_tests.c +4
-0+ + write_cr4_safe(read_cr4() & ~X86_CR4_LAM_SUP); +} +/*
* This does not test real-address mode, virtual-8086 mode, protected mode,Quoted 8 lines
x86/vmx_tests.c static void invvpid_test(void) · line 3344
Quoted 6 lines
* or CPL > 0.
x86/vmx_tests.c +1
-0@@ -3282,6 +3341,7 @@ static void invvpid_test(void) invvpid_test_pf(); invvpid_test_compatibility_mode(); invvpid_test_not_in_vmx_operation(); + invvpid_test_lam();
operand->glais checked only inINVVPID_ADDRmode. So, the lam test should be moved under "if (types & (1u <<INVVPID_ADDR))" a few lines above.Yes, will update it.
Quoted 7 lines
Quoted 7 lines
x86/vmx_tests.c +0
-0}/* Signature · 1 line
2.25.1