| Message ID | 1629961999-23407-1-git-send-email-weijiang.yang@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [kvm-unit-tests,v2] x86: vpmu: Add tests for Arch LBR support | expand |
Hi weijiang, would you help update the status of this test ? On 26/8/2021 3:13 pm, Yang Weijiang wrote: > The added code is to check whether basic Arch LBR function is supported > in guest, including read/write to LBR record MSRs, validation of control/ > depth MSRs, write to depth MSR resetting all record MSRs and xsaves/xrstors > support for guest Arch LBR. The purpose of the code is to do basic checks for > Arch LBR, not to make everything around Arch LBR be tested. > > Change in v2: > - Per Like's review feedback, changed below things: > 1. Combined Arch LBR tests with existing LBR tests. > 2. Added sanity check tests for control & depth MSRs. > 3. Added test for write depth MSR to reset all record MSRs. > - Added more test checkpoints in output report. > - Refactored part of the code. > > Signed-off-by: Yang Weijiang <weijiang.yang@intel.com> > --- > x86/pmu_lbr.c | 292 +++++++++++++++++++++++++++++++++++++++++++++++--- > 1 file changed, 275 insertions(+), 17 deletions(-) > > diff --git a/x86/pmu_lbr.c b/x86/pmu_lbr.c > index 3bd9e9f..6c22cad 100644 > --- a/x86/pmu_lbr.c > +++ b/x86/pmu_lbr.c > @@ -1,19 +1,88 @@ > +#include "asm-generic/page.h" > #include "x86/msr.h" > #include "x86/processor.h" > #include "x86/desc.h" > +#include "bitops.h" > > #define N 1000000 > -#define MAX_NUM_LBR_ENTRY 32 > -#define DEBUGCTLMSR_LBR (1UL << 0) > -#define PMU_CAP_LBR_FMT 0x3f > +#define MAX_NUM_LBR_ENTRY 32 > +#define DEBUGCTLMSR_LBR (1UL << 0) > +#define PMU_CAP_LBR_FMT 0x3f > > #define MSR_LBR_NHM_FROM 0x00000680 > #define MSR_LBR_NHM_TO 0x000006c0 > #define MSR_LBR_CORE_FROM 0x00000040 > -#define MSR_LBR_CORE_TO 0x00000060 > +#define MSR_LBR_CORE_TO 0x00000060 > #define MSR_LBR_TOS 0x000001c9 > #define MSR_LBR_SELECT 0x000001c8 > > +#define MSR_ARCH_LBR_CTL 0x000014ce > +#define MSR_ARCH_LBR_DEPTH 0x000014cf > +#define MSR_ARCH_LBR_FROM_0 0x00001500 > +#define MSR_ARCH_LBR_TO_0 0x00001600 > +#define MSR_ARCH_LBR_INFO_0 0x00001200 > + > +#define MSR_IA32_XSS 0x00000da0 > + > +#define IA32_XSS_ARCH_LBR (1UL << 15) > +#define CR4_OSXSAVE_BIT (1UL << 18) > +#define CPUID_EDX_ARCH_LBR (1UL << 19) > + > +#define ARCH_LBR_CTL_BITS 0x3f0003 > + > +#define XSAVES ".byte 0x48,0x0f,0xc7,0x2f\n\t" > +#define XRSTORS ".byte 0x48,0x0f,0xc7,0x1f\n\t" > + > +static int run_arch_lbr_test(void); > + > +struct xstate_header { > + u64 xfeatures; > + u64 xcomp_bv; > + u64 reserved[6]; > +} __attribute__((packed)); > + > +struct arch_lbr_entry { > + u64 lbr_from; > + u64 lbr_to; > + u64 lbr_info; > +}__attribute__((packed)); > + > +struct arch_lbr_struct { > + u64 lbr_ctl; > + u64 lbr_depth; > + u64 ler_from; > + u64 ler_to; > + u64 ler_info; > + struct arch_lbr_entry lbr_records[MAX_NUM_LBR_ENTRY]; > +}__attribute__((packed)); > + > +struct xsave_struct { > + u8 fpu_sse[512]; > + struct xstate_header xstate_hdr; > + struct arch_lbr_struct records; > +} __attribute__((packed)); > + > +u8 __attribute__((__aligned__(64))) xsave_buffer[PAGE_SIZE]; > + > +struct xsave_struct *test_buf = (struct xsave_struct *)xsave_buffer; > + > +union cpuid10_eax { > + struct { > + unsigned int version_id:8; > + unsigned int num_counters:8; > + unsigned int bit_width:8; > + unsigned int mask_length:8; > + } split; > + unsigned int full; > +} eax; > + > +u32 lbr_from, lbr_to; > +u64 lbr_ctl, lbr_depth; > + > +u64 arch_lbr_from[MAX_NUM_LBR_ENTRY]; > +u64 arch_lbr_to[MAX_NUM_LBR_ENTRY]; > +u64 arch_lbr_info[MAX_NUM_LBR_ENTRY]; > + > volatile int count; > > static __attribute__((noinline)) int compute_flag(int i) > @@ -38,36 +107,225 @@ static __attribute__((noinline)) int lbr_test(void) > return 0; > } > > -union cpuid10_eax { > - struct { > - unsigned int version_id:8; > - unsigned int num_counters:8; > - unsigned int bit_width:8; > - unsigned int mask_length:8; > - } split; > - unsigned int full; > -} eax; > - > -u32 lbr_from, lbr_to; > - > static void init_lbr(void *index) > { > wrmsr(lbr_from + *(int *) index, 0); > wrmsr(lbr_to + *(int *)index, 0); > } > > +static void test_lbr_depth(void *data) > +{ > + wrmsr(MSR_ARCH_LBR_DEPTH, *(int *)data); > +} > + > +static void test_lbr_control(void *data) > +{ > + wrmsr(MSR_ARCH_LBR_CTL, *(int *)data); > +} > + > static bool test_init_lbr_from_exception(u64 index) > { > return test_for_exception(GP_VECTOR, init_lbr, &index); > } > > +static bool test_lbr_depth_from_exception(u64 data) > +{ > + return test_for_exception(GP_VECTOR, test_lbr_depth, &data); > +} > + > +static bool test_lbr_control_from_exception(u64 data) > +{ > + return test_for_exception(GP_VECTOR, test_lbr_control, &data); > +} > + > +static inline void xrstors(struct xsave_struct *fx, unsigned long mask) > +{ > + u32 lmask = mask; > + u32 hmask = mask >> 32; > + > + asm volatile(XRSTORS > + : : "D" (fx), "m" (*fx), "a" (lmask), "d" (hmask) > + : "memory"); > +} > + > +static inline int xsaves(struct xsave_struct *fx, unsigned long mask) > +{ > + u32 lmask = mask; > + u32 hmask = mask >> 32; > + int err = 0; > + > + asm volatile(XSAVES > + : [err] "=r" (err) : "D" (fx), "m" (*fx), "a" (lmask), "d" (hmask) > + : "memory"); > + return err; > +} > + > +static int test_clear_lbr_records(void) > +{ > + int i; > + > + for (i = 0; i < lbr_depth; ++i) { > + if (rdmsr(MSR_ARCH_LBR_FROM_0 + i) != 0 || > + rdmsr(MSR_ARCH_LBR_TO_0 + i) != 0 || > + rdmsr(MSR_ARCH_LBR_INFO_0 + i) != 0) > + break; > + } > + > + return (i == lbr_depth) ? 0 : -1; > + > +} > + > +static bool check_xsaves_records(void) > +{ > + int i; > + struct arch_lbr_entry *records = test_buf->records.lbr_records; > + > + for (i = 0; i < lbr_depth; ++i) { > + if (arch_lbr_from[i] != (*(records + i)).lbr_from || > + arch_lbr_to[i] != (*(records + i)).lbr_to || > + arch_lbr_info[i] != (*(records + i)).lbr_info) > + break; > + } > + > + return i == lbr_depth; > +} > + > +static bool check_msrs_records(void) > +{ > + int i; > + > + for (i = 0; i < lbr_depth; ++i) { > + if (arch_lbr_from[i] != rdmsr(MSR_ARCH_LBR_FROM_0 + i) || > + arch_lbr_to[i] != rdmsr(MSR_ARCH_LBR_TO_0 + i) || > + arch_lbr_info[i] != rdmsr(MSR_ARCH_LBR_INFO_0 + i)) > + break; > + } > + > + return i == lbr_depth; > +} > + > +static void test_with_xsaves(void) > +{ > + u32 cr4; > + > + /* Only test Arch LBR save/restore, ignore other features.*/ > + wrmsr(MSR_IA32_XSS, IA32_XSS_ARCH_LBR); > + > + cr4 = read_cr4(); > + write_cr4(cr4 | CR4_OSXSAVE_BIT); > + > + xsaves(test_buf, IA32_XSS_ARCH_LBR | 0x3); > + > + report(check_xsaves_records(), > + "The saved(XSAVES) LBR records do match the original values."); > + > + wrmsr(MSR_ARCH_LBR_DEPTH, lbr_depth); > + > + xrstors(test_buf, IA32_XSS_ARCH_LBR | 0x3); > + > + report(check_msrs_records(), > + "The restored(XRSTORS) LBR records do match the original values."); > +} > + > +static int test_arch_lbr_entry(void) > +{ > + int max, ret = -1; > + > + lbr_from = MSR_ARCH_LBR_FROM_0; > + lbr_to = MSR_ARCH_LBR_TO_0; > + > + if (test_init_lbr_from_exception(0)) { > + printf("Arch LBR on this platform is not supported!\n"); > + return ret; > + } > + > + for (max = 0; max < lbr_depth; max++) { > + if (test_init_lbr_from_exception(max)) > + break; > + } > + > + if (max == lbr_depth) { > + report(true, "The number of guest LBR entries is good."); > + ret = 0; > + } > + > + return ret; > +} > + > +static int run_arch_lbr_test(void) > +{ > + struct cpuid id; > + int i; > + > + printf("Start Arch LBR testing...\n"); > + id = raw_cpuid(0xd, 1); > + if (!(id.a & 0x8)) { > + printf("XSAVES is not supported!.\n"); > + return report_summary(); > + } > + > + id = cpuid(0x1c); > + lbr_depth = (fls(id.a & 0xff) + 1)*8; > + > + if (test_arch_lbr_entry()) { > + printf("Arch LBR entry test failed!.\n"); > + return report_summary(); > + } > + > + report(test_lbr_control_from_exception(ARCH_LBR_CTL_BITS | 0xfff0), > + "Invalid Arch LBR control can trigger #GP."); > + report(test_lbr_depth_from_exception(lbr_depth + 0x10), > + "Unsupported Arch LBR depth can trigger #GP."); > + > + wrmsr(MSR_ARCH_LBR_DEPTH, lbr_depth); > + report(!test_clear_lbr_records(), > + "Write to depth MSR can reset record MSRs to 0s."); > + > + lbr_ctl = ARCH_LBR_CTL_BITS; > + wrmsr(MSR_ARCH_LBR_CTL, lbr_ctl); > + > + lbr_test(); > + > + /* Disable Arch LBR sampling before run sanity checks. */ > + lbr_ctl &= ~0x1; > + wrmsr(MSR_ARCH_LBR_CTL, lbr_ctl); > + > + /* > + * Save Arch LBR records from MSRs for following comparison with the > + * values from xsaves/xrstors. > + */ > + for (i = 0; i < lbr_depth; ++i) { > + if (!(arch_lbr_from[i] = rdmsr(MSR_ARCH_LBR_FROM_0 + i)) || > + !(arch_lbr_to[i] = rdmsr(MSR_ARCH_LBR_TO_0 + i)) || > + !(arch_lbr_info[i] = rdmsr(MSR_ARCH_LBR_INFO_0 + i))) > + break; > + } > + > + if (i != lbr_depth) { > + printf("Invalid Arch LBR records.\n"); > + return report_summary(); > + } > + > + test_with_xsaves(); > + > + return 0; > +} > + > int main(int ac, char **av) > { > - struct cpuid id = cpuid(10); > + struct cpuid id; > u64 perf_cap; > int max, i; > > setup_vm(); > + > + id = cpuid(0x7); > + if ((id.d & CPUID_EDX_ARCH_LBR)) { > + run_arch_lbr_test(); > + return report_summary(); > + } > + > + id = cpuid(10); > perf_cap = rdmsr(MSR_IA32_PERF_CAPABILITIES); > eax.full = id.a; >
On 5/26/2022 5:48 PM, Like Xu wrote: > Hi weijiang, would you help update the status of this test ? Hi, Like, This unit tests app can pass both on Arch LBR capable platform or legacy platform with v12 arch-lbr KVM series + qemu series, see below results: ========================================================= Arch LBR capable platform(SPR): Start Arch LBR testing... PASS: The number of guest LBR entries is good. PASS: Invalid Arch LBR control can trigger #GP. PASS: Unsupported Arch LBR depth can trigger #GP. PASS: Write to depth MSR can reset record MSRs to 0s. PASS: The saved(XSAVES) LBR records do match the original values. PASS: The restored(XRSTORS) LBR records do match the original values. SUMMARY: 6 tests PASS pmu_lbr (6 tests) =========================================================== Legacy platform: PMU version: 2 LBR version: 5 PASS: The number of guest LBR entries is good. PASS: The guest LBR MSR_LBR_TOS value is good. PASS: The guest LBR FROM_IP/TO_IP values are good. SUMMARY: 3 tests PASS pmu_lbr (3 tests) > > On 26/8/2021 3:13 pm, Yang Weijiang wrote: >> The added code is to check whether basic Arch LBR function is supported >> in guest, including read/write to LBR record MSRs, validation of >> control/ >> depth MSRs, write to depth MSR resetting all record MSRs and >> xsaves/xrstors >> support for guest Arch LBR. The purpose of the code is to do basic >> checks for >> Arch LBR, not to make everything around Arch LBR be tested. >> >> Change in v2: >> - Per Like's review feedback, changed below things: >> 1. Combined Arch LBR tests with existing LBR tests. >> 2. Added sanity check tests for control & depth MSRs. >> 3. Added test for write depth MSR to reset all record MSRs. >> - Added more test checkpoints in output report. >> - Refactored part of the code. >> >> Signed-off-by: Yang Weijiang <weijiang.yang@intel.com> >> --- >> x86/pmu_lbr.c | 292 +++++++++++++++++++++++++++++++++++++++++++++++--- >> 1 file changed, 275 insertions(+), 17 deletions(-) >> >> diff --git a/x86/pmu_lbr.c b/x86/pmu_lbr.c >> index 3bd9e9f..6c22cad 100644 >> --- a/x86/pmu_lbr.c >> +++ b/x86/pmu_lbr.c >> @@ -1,19 +1,88 @@ >> +#include "asm-generic/page.h" >> #include "x86/msr.h" >> #include "x86/processor.h" >> #include "x86/desc.h" >> +#include "bitops.h" >> #define N 1000000 >> -#define MAX_NUM_LBR_ENTRY 32 >> -#define DEBUGCTLMSR_LBR (1UL << 0) >> -#define PMU_CAP_LBR_FMT 0x3f >> +#define MAX_NUM_LBR_ENTRY 32 >> +#define DEBUGCTLMSR_LBR (1UL << 0) >> +#define PMU_CAP_LBR_FMT 0x3f >> #define MSR_LBR_NHM_FROM 0x00000680 >> #define MSR_LBR_NHM_TO 0x000006c0 >> #define MSR_LBR_CORE_FROM 0x00000040 >> -#define MSR_LBR_CORE_TO 0x00000060 >> +#define MSR_LBR_CORE_TO 0x00000060 >> #define MSR_LBR_TOS 0x000001c9 >> #define MSR_LBR_SELECT 0x000001c8 >> +#define MSR_ARCH_LBR_CTL 0x000014ce >> +#define MSR_ARCH_LBR_DEPTH 0x000014cf >> +#define MSR_ARCH_LBR_FROM_0 0x00001500 >> +#define MSR_ARCH_LBR_TO_0 0x00001600 >> +#define MSR_ARCH_LBR_INFO_0 0x00001200 >> + >> +#define MSR_IA32_XSS 0x00000da0 >> + >> +#define IA32_XSS_ARCH_LBR (1UL << 15) >> +#define CR4_OSXSAVE_BIT (1UL << 18) >> +#define CPUID_EDX_ARCH_LBR (1UL << 19) >> + >> +#define ARCH_LBR_CTL_BITS 0x3f0003 >> + >> +#define XSAVES ".byte 0x48,0x0f,0xc7,0x2f\n\t" >> +#define XRSTORS ".byte 0x48,0x0f,0xc7,0x1f\n\t" >> + >> +static int run_arch_lbr_test(void); >> + >> +struct xstate_header { >> + u64 xfeatures; >> + u64 xcomp_bv; >> + u64 reserved[6]; >> +} __attribute__((packed)); >> + >> +struct arch_lbr_entry { >> + u64 lbr_from; >> + u64 lbr_to; >> + u64 lbr_info; >> +}__attribute__((packed)); >> + >> +struct arch_lbr_struct { >> + u64 lbr_ctl; >> + u64 lbr_depth; >> + u64 ler_from; >> + u64 ler_to; >> + u64 ler_info; >> + struct arch_lbr_entry lbr_records[MAX_NUM_LBR_ENTRY]; >> +}__attribute__((packed)); >> + >> +struct xsave_struct { >> + u8 fpu_sse[512]; >> + struct xstate_header xstate_hdr; >> + struct arch_lbr_struct records; >> +} __attribute__((packed)); >> + >> +u8 __attribute__((__aligned__(64))) xsave_buffer[PAGE_SIZE]; >> + >> +struct xsave_struct *test_buf = (struct xsave_struct *)xsave_buffer; >> + >> +union cpuid10_eax { >> + struct { >> + unsigned int version_id:8; >> + unsigned int num_counters:8; >> + unsigned int bit_width:8; >> + unsigned int mask_length:8; >> + } split; >> + unsigned int full; >> +} eax; >> + >> +u32 lbr_from, lbr_to; >> +u64 lbr_ctl, lbr_depth; >> + >> +u64 arch_lbr_from[MAX_NUM_LBR_ENTRY]; >> +u64 arch_lbr_to[MAX_NUM_LBR_ENTRY]; >> +u64 arch_lbr_info[MAX_NUM_LBR_ENTRY]; >> + >> volatile int count; >> static __attribute__((noinline)) int compute_flag(int i) >> @@ -38,36 +107,225 @@ static __attribute__((noinline)) int >> lbr_test(void) >> return 0; >> } >> -union cpuid10_eax { >> - struct { >> - unsigned int version_id:8; >> - unsigned int num_counters:8; >> - unsigned int bit_width:8; >> - unsigned int mask_length:8; >> - } split; >> - unsigned int full; >> -} eax; >> - >> -u32 lbr_from, lbr_to; >> - >> static void init_lbr(void *index) >> { >> wrmsr(lbr_from + *(int *) index, 0); >> wrmsr(lbr_to + *(int *)index, 0); >> } >> +static void test_lbr_depth(void *data) >> +{ >> + wrmsr(MSR_ARCH_LBR_DEPTH, *(int *)data); >> +} >> + >> +static void test_lbr_control(void *data) >> +{ >> + wrmsr(MSR_ARCH_LBR_CTL, *(int *)data); >> +} >> + >> static bool test_init_lbr_from_exception(u64 index) >> { >> return test_for_exception(GP_VECTOR, init_lbr, &index); >> } >> +static bool test_lbr_depth_from_exception(u64 data) >> +{ >> + return test_for_exception(GP_VECTOR, test_lbr_depth, &data); >> +} >> + >> +static bool test_lbr_control_from_exception(u64 data) >> +{ >> + return test_for_exception(GP_VECTOR, test_lbr_control, &data); >> +} >> + >> +static inline void xrstors(struct xsave_struct *fx, unsigned long >> mask) >> +{ >> + u32 lmask = mask; >> + u32 hmask = mask >> 32; >> + >> + asm volatile(XRSTORS >> + : : "D" (fx), "m" (*fx), "a" (lmask), "d" (hmask) >> + : "memory"); >> +} >> + >> +static inline int xsaves(struct xsave_struct *fx, unsigned long mask) >> +{ >> + u32 lmask = mask; >> + u32 hmask = mask >> 32; >> + int err = 0; >> + >> + asm volatile(XSAVES >> + : [err] "=r" (err) : "D" (fx), "m" (*fx), "a" >> (lmask), "d" (hmask) >> + : "memory"); >> + return err; >> +} >> + >> +static int test_clear_lbr_records(void) >> +{ >> + int i; >> + >> + for (i = 0; i < lbr_depth; ++i) { >> + if (rdmsr(MSR_ARCH_LBR_FROM_0 + i) != 0 || >> + rdmsr(MSR_ARCH_LBR_TO_0 + i) != 0 || >> + rdmsr(MSR_ARCH_LBR_INFO_0 + i) != 0) >> + break; >> + } >> + >> + return (i == lbr_depth) ? 0 : -1; >> + >> +} >> + >> +static bool check_xsaves_records(void) >> +{ >> + int i; >> + struct arch_lbr_entry *records = test_buf->records.lbr_records; >> + >> + for (i = 0; i < lbr_depth; ++i) { >> + if (arch_lbr_from[i] != (*(records + i)).lbr_from || >> + arch_lbr_to[i] != (*(records + i)).lbr_to || >> + arch_lbr_info[i] != (*(records + i)).lbr_info) >> + break; >> + } >> + >> + return i == lbr_depth; >> +} >> + >> +static bool check_msrs_records(void) >> +{ >> + int i; >> + >> + for (i = 0; i < lbr_depth; ++i) { >> + if (arch_lbr_from[i] != rdmsr(MSR_ARCH_LBR_FROM_0 + i) || >> + arch_lbr_to[i] != rdmsr(MSR_ARCH_LBR_TO_0 + i) || >> + arch_lbr_info[i] != rdmsr(MSR_ARCH_LBR_INFO_0 + i)) >> + break; >> + } >> + >> + return i == lbr_depth; >> +} >> + >> +static void test_with_xsaves(void) >> +{ >> + u32 cr4; >> + >> + /* Only test Arch LBR save/restore, ignore other features.*/ >> + wrmsr(MSR_IA32_XSS, IA32_XSS_ARCH_LBR); >> + >> + cr4 = read_cr4(); >> + write_cr4(cr4 | CR4_OSXSAVE_BIT); >> + >> + xsaves(test_buf, IA32_XSS_ARCH_LBR | 0x3); >> + >> + report(check_xsaves_records(), >> + "The saved(XSAVES) LBR records do match the original >> values."); >> + >> + wrmsr(MSR_ARCH_LBR_DEPTH, lbr_depth); >> + >> + xrstors(test_buf, IA32_XSS_ARCH_LBR | 0x3); >> + >> + report(check_msrs_records(), >> + "The restored(XRSTORS) LBR records do match the original >> values."); >> +} >> + >> +static int test_arch_lbr_entry(void) >> +{ >> + int max, ret = -1; >> + >> + lbr_from = MSR_ARCH_LBR_FROM_0; >> + lbr_to = MSR_ARCH_LBR_TO_0; >> + >> + if (test_init_lbr_from_exception(0)) { >> + printf("Arch LBR on this platform is not supported!\n"); >> + return ret; >> + } >> + >> + for (max = 0; max < lbr_depth; max++) { >> + if (test_init_lbr_from_exception(max)) >> + break; >> + } >> + >> + if (max == lbr_depth) { >> + report(true, "The number of guest LBR entries is good."); >> + ret = 0; >> + } >> + >> + return ret; >> +} >> + >> +static int run_arch_lbr_test(void) >> +{ >> + struct cpuid id; >> + int i; >> + >> + printf("Start Arch LBR testing...\n"); >> + id = raw_cpuid(0xd, 1); >> + if (!(id.a & 0x8)) { >> + printf("XSAVES is not supported!.\n"); >> + return report_summary(); >> + } >> + >> + id = cpuid(0x1c); >> + lbr_depth = (fls(id.a & 0xff) + 1)*8; >> + >> + if (test_arch_lbr_entry()) { >> + printf("Arch LBR entry test failed!.\n"); >> + return report_summary(); >> + } >> + >> + report(test_lbr_control_from_exception(ARCH_LBR_CTL_BITS | 0xfff0), >> + "Invalid Arch LBR control can trigger #GP."); >> + report(test_lbr_depth_from_exception(lbr_depth + 0x10), >> + "Unsupported Arch LBR depth can trigger #GP."); >> + >> + wrmsr(MSR_ARCH_LBR_DEPTH, lbr_depth); >> + report(!test_clear_lbr_records(), >> + "Write to depth MSR can reset record MSRs to 0s."); >> + >> + lbr_ctl = ARCH_LBR_CTL_BITS; >> + wrmsr(MSR_ARCH_LBR_CTL, lbr_ctl); >> + >> + lbr_test(); >> + >> + /* Disable Arch LBR sampling before run sanity checks. */ >> + lbr_ctl &= ~0x1; >> + wrmsr(MSR_ARCH_LBR_CTL, lbr_ctl); >> + >> + /* >> + * Save Arch LBR records from MSRs for following comparison with >> the >> + * values from xsaves/xrstors. >> + */ >> + for (i = 0; i < lbr_depth; ++i) { >> + if (!(arch_lbr_from[i] = rdmsr(MSR_ARCH_LBR_FROM_0 + i)) || >> + !(arch_lbr_to[i] = rdmsr(MSR_ARCH_LBR_TO_0 + i)) || >> + !(arch_lbr_info[i] = rdmsr(MSR_ARCH_LBR_INFO_0 + i))) >> + break; >> + } >> + >> + if (i != lbr_depth) { >> + printf("Invalid Arch LBR records.\n"); >> + return report_summary(); >> + } >> + >> + test_with_xsaves(); >> + >> + return 0; >> +} >> + >> int main(int ac, char **av) >> { >> - struct cpuid id = cpuid(10); >> + struct cpuid id; >> u64 perf_cap; >> int max, i; >> setup_vm(); >> + >> + id = cpuid(0x7); >> + if ((id.d & CPUID_EDX_ARCH_LBR)) { >> + run_arch_lbr_test(); >> + return report_summary(); >> + } >> + >> + id = cpuid(10); >> perf_cap = rdmsr(MSR_IA32_PERF_CAPABILITIES); >> eax.full = id.a;
diff --git a/x86/pmu_lbr.c b/x86/pmu_lbr.c index 3bd9e9f..6c22cad 100644 --- a/x86/pmu_lbr.c +++ b/x86/pmu_lbr.c @@ -1,19 +1,88 @@ +#include "asm-generic/page.h" #include "x86/msr.h" #include "x86/processor.h" #include "x86/desc.h" +#include "bitops.h" #define N 1000000 -#define MAX_NUM_LBR_ENTRY 32 -#define DEBUGCTLMSR_LBR (1UL << 0) -#define PMU_CAP_LBR_FMT 0x3f +#define MAX_NUM_LBR_ENTRY 32 +#define DEBUGCTLMSR_LBR (1UL << 0) +#define PMU_CAP_LBR_FMT 0x3f #define MSR_LBR_NHM_FROM 0x00000680 #define MSR_LBR_NHM_TO 0x000006c0 #define MSR_LBR_CORE_FROM 0x00000040 -#define MSR_LBR_CORE_TO 0x00000060 +#define MSR_LBR_CORE_TO 0x00000060 #define MSR_LBR_TOS 0x000001c9 #define MSR_LBR_SELECT 0x000001c8 +#define MSR_ARCH_LBR_CTL 0x000014ce +#define MSR_ARCH_LBR_DEPTH 0x000014cf +#define MSR_ARCH_LBR_FROM_0 0x00001500 +#define MSR_ARCH_LBR_TO_0 0x00001600 +#define MSR_ARCH_LBR_INFO_0 0x00001200 + +#define MSR_IA32_XSS 0x00000da0 + +#define IA32_XSS_ARCH_LBR (1UL << 15) +#define CR4_OSXSAVE_BIT (1UL << 18) +#define CPUID_EDX_ARCH_LBR (1UL << 19) + +#define ARCH_LBR_CTL_BITS 0x3f0003 + +#define XSAVES ".byte 0x48,0x0f,0xc7,0x2f\n\t" +#define XRSTORS ".byte 0x48,0x0f,0xc7,0x1f\n\t" + +static int run_arch_lbr_test(void); + +struct xstate_header { + u64 xfeatures; + u64 xcomp_bv; + u64 reserved[6]; +} __attribute__((packed)); + +struct arch_lbr_entry { + u64 lbr_from; + u64 lbr_to; + u64 lbr_info; +}__attribute__((packed)); + +struct arch_lbr_struct { + u64 lbr_ctl; + u64 lbr_depth; + u64 ler_from; + u64 ler_to; + u64 ler_info; + struct arch_lbr_entry lbr_records[MAX_NUM_LBR_ENTRY]; +}__attribute__((packed)); + +struct xsave_struct { + u8 fpu_sse[512]; + struct xstate_header xstate_hdr; + struct arch_lbr_struct records; +} __attribute__((packed)); + +u8 __attribute__((__aligned__(64))) xsave_buffer[PAGE_SIZE]; + +struct xsave_struct *test_buf = (struct xsave_struct *)xsave_buffer; + +union cpuid10_eax { + struct { + unsigned int version_id:8; + unsigned int num_counters:8; + unsigned int bit_width:8; + unsigned int mask_length:8; + } split; + unsigned int full; +} eax; + +u32 lbr_from, lbr_to; +u64 lbr_ctl, lbr_depth; + +u64 arch_lbr_from[MAX_NUM_LBR_ENTRY]; +u64 arch_lbr_to[MAX_NUM_LBR_ENTRY]; +u64 arch_lbr_info[MAX_NUM_LBR_ENTRY]; + volatile int count; static __attribute__((noinline)) int compute_flag(int i) @@ -38,36 +107,225 @@ static __attribute__((noinline)) int lbr_test(void) return 0; } -union cpuid10_eax { - struct { - unsigned int version_id:8; - unsigned int num_counters:8; - unsigned int bit_width:8; - unsigned int mask_length:8; - } split; - unsigned int full; -} eax; - -u32 lbr_from, lbr_to; - static void init_lbr(void *index) { wrmsr(lbr_from + *(int *) index, 0); wrmsr(lbr_to + *(int *)index, 0); } +static void test_lbr_depth(void *data) +{ + wrmsr(MSR_ARCH_LBR_DEPTH, *(int *)data); +} + +static void test_lbr_control(void *data) +{ + wrmsr(MSR_ARCH_LBR_CTL, *(int *)data); +} + static bool test_init_lbr_from_exception(u64 index) { return test_for_exception(GP_VECTOR, init_lbr, &index); } +static bool test_lbr_depth_from_exception(u64 data) +{ + return test_for_exception(GP_VECTOR, test_lbr_depth, &data); +} + +static bool test_lbr_control_from_exception(u64 data) +{ + return test_for_exception(GP_VECTOR, test_lbr_control, &data); +} + +static inline void xrstors(struct xsave_struct *fx, unsigned long mask) +{ + u32 lmask = mask; + u32 hmask = mask >> 32; + + asm volatile(XRSTORS + : : "D" (fx), "m" (*fx), "a" (lmask), "d" (hmask) + : "memory"); +} + +static inline int xsaves(struct xsave_struct *fx, unsigned long mask) +{ + u32 lmask = mask; + u32 hmask = mask >> 32; + int err = 0; + + asm volatile(XSAVES + : [err] "=r" (err) : "D" (fx), "m" (*fx), "a" (lmask), "d" (hmask) + : "memory"); + return err; +} + +static int test_clear_lbr_records(void) +{ + int i; + + for (i = 0; i < lbr_depth; ++i) { + if (rdmsr(MSR_ARCH_LBR_FROM_0 + i) != 0 || + rdmsr(MSR_ARCH_LBR_TO_0 + i) != 0 || + rdmsr(MSR_ARCH_LBR_INFO_0 + i) != 0) + break; + } + + return (i == lbr_depth) ? 0 : -1; + +} + +static bool check_xsaves_records(void) +{ + int i; + struct arch_lbr_entry *records = test_buf->records.lbr_records; + + for (i = 0; i < lbr_depth; ++i) { + if (arch_lbr_from[i] != (*(records + i)).lbr_from || + arch_lbr_to[i] != (*(records + i)).lbr_to || + arch_lbr_info[i] != (*(records + i)).lbr_info) + break; + } + + return i == lbr_depth; +} + +static bool check_msrs_records(void) +{ + int i; + + for (i = 0; i < lbr_depth; ++i) { + if (arch_lbr_from[i] != rdmsr(MSR_ARCH_LBR_FROM_0 + i) || + arch_lbr_to[i] != rdmsr(MSR_ARCH_LBR_TO_0 + i) || + arch_lbr_info[i] != rdmsr(MSR_ARCH_LBR_INFO_0 + i)) + break; + } + + return i == lbr_depth; +} + +static void test_with_xsaves(void) +{ + u32 cr4; + + /* Only test Arch LBR save/restore, ignore other features.*/ + wrmsr(MSR_IA32_XSS, IA32_XSS_ARCH_LBR); + + cr4 = read_cr4(); + write_cr4(cr4 | CR4_OSXSAVE_BIT); + + xsaves(test_buf, IA32_XSS_ARCH_LBR | 0x3); + + report(check_xsaves_records(), + "The saved(XSAVES) LBR records do match the original values."); + + wrmsr(MSR_ARCH_LBR_DEPTH, lbr_depth); + + xrstors(test_buf, IA32_XSS_ARCH_LBR | 0x3); + + report(check_msrs_records(), + "The restored(XRSTORS) LBR records do match the original values."); +} + +static int test_arch_lbr_entry(void) +{ + int max, ret = -1; + + lbr_from = MSR_ARCH_LBR_FROM_0; + lbr_to = MSR_ARCH_LBR_TO_0; + + if (test_init_lbr_from_exception(0)) { + printf("Arch LBR on this platform is not supported!\n"); + return ret; + } + + for (max = 0; max < lbr_depth; max++) { + if (test_init_lbr_from_exception(max)) + break; + } + + if (max == lbr_depth) { + report(true, "The number of guest LBR entries is good."); + ret = 0; + } + + return ret; +} + +static int run_arch_lbr_test(void) +{ + struct cpuid id; + int i; + + printf("Start Arch LBR testing...\n"); + id = raw_cpuid(0xd, 1); + if (!(id.a & 0x8)) { + printf("XSAVES is not supported!.\n"); + return report_summary(); + } + + id = cpuid(0x1c); + lbr_depth = (fls(id.a & 0xff) + 1)*8; + + if (test_arch_lbr_entry()) { + printf("Arch LBR entry test failed!.\n"); + return report_summary(); + } + + report(test_lbr_control_from_exception(ARCH_LBR_CTL_BITS | 0xfff0), + "Invalid Arch LBR control can trigger #GP."); + report(test_lbr_depth_from_exception(lbr_depth + 0x10), + "Unsupported Arch LBR depth can trigger #GP."); + + wrmsr(MSR_ARCH_LBR_DEPTH, lbr_depth); + report(!test_clear_lbr_records(), + "Write to depth MSR can reset record MSRs to 0s."); + + lbr_ctl = ARCH_LBR_CTL_BITS; + wrmsr(MSR_ARCH_LBR_CTL, lbr_ctl); + + lbr_test(); + + /* Disable Arch LBR sampling before run sanity checks. */ + lbr_ctl &= ~0x1; + wrmsr(MSR_ARCH_LBR_CTL, lbr_ctl); + + /* + * Save Arch LBR records from MSRs for following comparison with the + * values from xsaves/xrstors. + */ + for (i = 0; i < lbr_depth; ++i) { + if (!(arch_lbr_from[i] = rdmsr(MSR_ARCH_LBR_FROM_0 + i)) || + !(arch_lbr_to[i] = rdmsr(MSR_ARCH_LBR_TO_0 + i)) || + !(arch_lbr_info[i] = rdmsr(MSR_ARCH_LBR_INFO_0 + i))) + break; + } + + if (i != lbr_depth) { + printf("Invalid Arch LBR records.\n"); + return report_summary(); + } + + test_with_xsaves(); + + return 0; +} + int main(int ac, char **av) { - struct cpuid id = cpuid(10); + struct cpuid id; u64 perf_cap; int max, i; setup_vm(); + + id = cpuid(0x7); + if ((id.d & CPUID_EDX_ARCH_LBR)) { + run_arch_lbr_test(); + return report_summary(); + } + + id = cpuid(10); perf_cap = rdmsr(MSR_IA32_PERF_CAPABILITIES); eax.full = id.a;
The added code is to check whether basic Arch LBR function is supported in guest, including read/write to LBR record MSRs, validation of control/ depth MSRs, write to depth MSR resetting all record MSRs and xsaves/xrstors support for guest Arch LBR. The purpose of the code is to do basic checks for Arch LBR, not to make everything around Arch LBR be tested. Change in v2: - Per Like's review feedback, changed below things: 1. Combined Arch LBR tests with existing LBR tests. 2. Added sanity check tests for control & depth MSRs. 3. Added test for write depth MSR to reset all record MSRs. - Added more test checkpoints in output report. - Refactored part of the code. Signed-off-by: Yang Weijiang <weijiang.yang@intel.com> --- x86/pmu_lbr.c | 292 +++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 275 insertions(+), 17 deletions(-)