diff options
Diffstat (limited to 'arch/powerpc/kvm/book3s_hv_rmhandlers.S')
| -rw-r--r-- | arch/powerpc/kvm/book3s_hv_rmhandlers.S | 373 |
1 files changed, 337 insertions, 36 deletions
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S index 818dce344e82..974793435a2e 100644 --- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S +++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S @@ -28,6 +28,9 @@ #include <asm/exception-64s.h> #include <asm/kvm_book3s_asm.h> #include <asm/mmu-hash64.h> +#include <asm/tm.h> + +#define VCPU_GPRS_TM(reg) (((reg) * ULONG_SIZE) + VCPU_GPR_TM) #ifdef __LITTLE_ENDIAN__ #error Need to fix lppaca and SLB shadow accesses in little endian mode @@ -75,14 +78,20 @@ BEGIN_FTR_SECTION END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_207S) /* Restore SPRG3 */ - ld r3,PACA_SPRG3(r13) - mtspr SPRN_SPRG3,r3 + ld r3,PACA_SPRG_VDSO(r13) + mtspr SPRN_SPRG_VDSO_WRITE,r3 /* Reload the host's PMU registers */ ld r3, PACALPPACAPTR(r13) /* is the host using the PMU? */ lbz r4, LPPACA_PMCINUSE(r3) cmpwi r4, 0 beq 23f /* skip if not */ +BEGIN_FTR_SECTION + ld r3, HSTATE_MMCR(r13) + andi. r4, r3, MMCR0_PMAO_SYNC | MMCR0_PMAO + cmpwi r4, MMCR0_PMAO + beql kvmppc_fix_pmao +END_FTR_SECTION_IFSET(CPU_FTR_PMAO_BUG) lwz r3, HSTATE_PMC(r13) lwz r4, HSTATE_PMC + 4(r13) lwz r5, HSTATE_PMC + 8(r13) @@ -106,8 +115,18 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_201) ld r3, HSTATE_MMCR(r13) ld r4, HSTATE_MMCR + 8(r13) ld r5, HSTATE_MMCR + 16(r13) + ld r6, HSTATE_MMCR + 24(r13) + ld r7, HSTATE_MMCR + 32(r13) mtspr SPRN_MMCR1, r4 mtspr SPRN_MMCRA, r5 + mtspr SPRN_SIAR, r6 + mtspr SPRN_SDAR, r7 +BEGIN_FTR_SECTION + ld r8, HSTATE_MMCR + 40(r13) + ld r9, HSTATE_MMCR + 48(r13) + mtspr SPRN_MMCR2, r8 + mtspr SPRN_SIER, r9 +END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S) mtspr SPRN_MMCR0, r3 isync 23: @@ -229,6 +248,12 @@ kvm_novcpu_exit: */ .globl kvm_start_guest kvm_start_guest: + + /* Set runlatch bit the minute you wake up from nap */ + mfspr r1, SPRN_CTRLF + ori r1, r1, 1 + mtspr SPRN_CTRLT, r1 + ld r2,PACATOC(r13) li r0,KVM_HWTHREAD_IN_KVM @@ -296,6 +321,11 @@ kvm_no_guest: li r0, KVM_HWTHREAD_IN_NAP stb r0, HSTATE_HWTHREAD_STATE(r13) kvm_do_nap: + /* Clear the runlatch bit before napping */ + mfspr r2, SPRN_CTRLF + clrrdi r2, r2, 1 + mtspr SPRN_CTRLT, r2 + li r3, LPCR_PECE0 mfspr r4, SPRN_LPCR rlwimi r4, r3, 0, LPCR_PECE0 | LPCR_PECE1 @@ -597,12 +627,128 @@ BEGIN_FTR_SECTION END_FTR_SECTION_NESTED(CPU_FTR_ARCH_206, CPU_FTR_ARCH_206, 89) END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_207S) +#ifdef CONFIG_PPC_TRANSACTIONAL_MEM +BEGIN_FTR_SECTION + b skip_tm +END_FTR_SECTION_IFCLR(CPU_FTR_TM) + + /* Turn on TM/FP/VSX/VMX so we can restore them. */ + mfmsr r5 + li r6, MSR_TM >> 32 + sldi r6, r6, 32 + or r5, r5, r6 + ori r5, r5, MSR_FP + oris r5, r5, (MSR_VEC | MSR_VSX)@h + mtmsrd r5 + + /* + * The user may change these outside of a transaction, so they must + * always be context switched. + */ + ld r5, VCPU_TFHAR(r4) + ld r6, VCPU_TFIAR(r4) + ld r7, VCPU_TEXASR(r4) + mtspr SPRN_TFHAR, r5 + mtspr SPRN_TFIAR, r6 + mtspr SPRN_TEXASR, r7 + + ld r5, VCPU_MSR(r4) + rldicl. r5, r5, 64 - MSR_TS_S_LG, 62 + beq skip_tm /* TM not active in guest */ + + /* Make sure the failure summary is set, otherwise we'll program check + * when we trechkpt. It's possible that this might have been not set + * on a kvmppc_set_one_reg() call but we shouldn't let this crash the + * host. + */ + oris r7, r7, (TEXASR_FS)@h + mtspr SPRN_TEXASR, r7 + + /* + * We need to load up the checkpointed state for the guest. + * We need to do this early as it will blow away any GPRs, VSRs and + * some SPRs. + */ + + mr r31, r4 + addi r3, r31, VCPU_FPRS_TM + bl .load_fp_state + addi r3, r31, VCPU_VRS_TM + bl .load_vr_state + mr r4, r31 + lwz r7, VCPU_VRSAVE_TM(r4) + mtspr SPRN_VRSAVE, r7 + + ld r5, VCPU_LR_TM(r4) + lwz r6, VCPU_CR_TM(r4) + ld r7, VCPU_CTR_TM(r4) + ld r8, VCPU_AMR_TM(r4) + ld r9, VCPU_TAR_TM(r4) + mtlr r5 + mtcr r6 + mtctr r7 + mtspr SPRN_AMR, r8 + mtspr SPRN_TAR, r9 + + /* + * Load up PPR and DSCR values but don't put them in the actual SPRs + * till the last moment to avoid running with userspace PPR and DSCR for + * too long. + */ + ld r29, VCPU_DSCR_TM(r4) + ld r30, VCPU_PPR_TM(r4) + + std r2, PACATMSCRATCH(r13) /* Save TOC */ + + /* Clear the MSR RI since r1, r13 are all going to be foobar. */ + li r5, 0 + mtmsrd r5, 1 + + /* Load GPRs r0-r28 */ + reg = 0 + .rept 29 + ld reg, VCPU_GPRS_TM(reg)(r31) + reg = reg + 1 + .endr + + mtspr SPRN_DSCR, r29 + mtspr SPRN_PPR, r30 + + /* Load final GPRs */ + ld 29, VCPU_GPRS_TM(29)(r31) + ld 30, VCPU_GPRS_TM(30)(r31) + ld 31, VCPU_GPRS_TM(31)(r31) + + /* TM checkpointed state is now setup. All GPRs are now volatile. */ + TRECHKPT + + /* Now let's get back the state we need. */ + HMT_MEDIUM + GET_PACA(r13) + ld r29, HSTATE_DSCR(r13) + mtspr SPRN_DSCR, r29 + ld r4, HSTATE_KVM_VCPU(r13) + ld r1, HSTATE_HOST_R1(r13) + ld r2, PACATMSCRATCH(r13) + + /* Set the MSR RI since we have our registers back. */ + li r5, MSR_RI + mtmsrd r5, 1 +skip_tm: +#endif + /* Load guest PMU registers */ /* R4 is live here (vcpu pointer) */ li r3, 1 sldi r3, r3, 31 /* MMCR0_FC (freeze counters) bit */ mtspr SPRN_MMCR0, r3 /* freeze all counters, disable ints */ isync +BEGIN_FTR_SECTION + ld r3, VCPU_MMCR(r4) + andi. r5, r3, MMCR0_PMAO_SYNC | MMCR0_PMAO + cmpwi r5, MMCR0_PMAO + beql kvmppc_fix_pmao +END_FTR_SECTION_IFSET(CPU_FTR_PMAO_BUG) lwz r3, VCPU_PMC(r4) /* always load up guest PMU registers */ lwz r5, VCPU_PMC + 4(r4) /* to prevent information leak */ lwz r6, VCPU_PMC + 8(r4) @@ -704,14 +850,6 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_207S) ld r6, VCPU_VTB(r4) mtspr SPRN_IC, r5 mtspr SPRN_VTB, r6 -#ifdef CONFIG_PPC_TRANSACTIONAL_MEM - ld r5, VCPU_TFHAR(r4) - ld r6, VCPU_TFIAR(r4) - ld r7, VCPU_TEXASR(r4) - mtspr SPRN_TFHAR, r5 - mtspr SPRN_TFIAR, r6 - mtspr SPRN_TEXASR, r7 -#endif ld r8, VCPU_EBBHR(r4) mtspr SPRN_EBBHR, r8 ld r5, VCPU_EBBRR(r4) @@ -736,6 +874,10 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_207S) * Set the decrementer to the guest decrementer. */ ld r8,VCPU_DEC_EXPIRES(r4) + /* r8 is a host timebase value here, convert to guest TB */ + ld r5,HSTATE_KVM_VCORE(r13) + ld r6,VCORE_TB_OFFSET(r5) + add r8,r8,r6 mftb r7 subf r3,r7,r8 mtspr SPRN_DEC,r3 @@ -817,7 +959,8 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_206) 12: mtspr SPRN_SRR0, r10 mr r10,r0 mtspr SPRN_SRR1, r11 - ld r11, VCPU_INTR_MSR(r4) + mr r9, r4 + bl kvmppc_msr_interrupt 5: /* @@ -1098,17 +1241,15 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_201) mftb r6 extsw r5,r5 add r5,r5,r6 + /* r5 is a guest timebase value here, convert to host TB */ + ld r3,HSTATE_KVM_VCORE(r13) + ld r4,VCORE_TB_OFFSET(r3) + subf r5,r4,r5 std r5,VCPU_DEC_EXPIRES(r9) BEGIN_FTR_SECTION b 8f END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_207S) - /* Turn on TM so we can access TFHAR/TFIAR/TEXASR */ - mfmsr r8 - li r0, 1 - rldimi r8, r0, MSR_TM_LG, 63-MSR_TM_LG - mtmsrd r8 - /* Save POWER8-specific registers */ mfspr r5, SPRN_IAMR mfspr r6, SPRN_PSPB @@ -1122,14 +1263,6 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_207S) std r5, VCPU_IC(r9) std r6, VCPU_VTB(r9) std r7, VCPU_TAR(r9) -#ifdef CONFIG_PPC_TRANSACTIONAL_MEM - mfspr r5, SPRN_TFHAR - mfspr r6, SPRN_TFIAR - mfspr r7, SPRN_TEXASR - std r5, VCPU_TFHAR(r9) - std r6, VCPU_TFIAR(r9) - std r7, VCPU_TEXASR(r9) -#endif mfspr r8, SPRN_EBBHR std r8, VCPU_EBBHR(r9) mfspr r5, SPRN_EBBRR @@ -1202,6 +1335,110 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_206) mr r3, r9 bl kvmppc_save_fp +#ifdef CONFIG_PPC_TRANSACTIONAL_MEM +BEGIN_FTR_SECTION + b 2f +END_FTR_SECTION_IFCLR(CPU_FTR_TM) + /* Turn on TM. */ + mfmsr r8 + li r0, 1 + rldimi r8, r0, MSR_TM_LG, 63-MSR_TM_LG + mtmsrd r8 + + ld r5, VCPU_MSR(r9) + rldicl. r5, r5, 64 - MSR_TS_S_LG, 62 + beq 1f /* TM not active in guest. */ + + li r3, TM_CAUSE_KVM_RESCHED + + /* Clear the MSR RI since r1, r13 are all going to be foobar. */ + li r5, 0 + mtmsrd r5, 1 + + /* All GPRs are volatile at this point. */ + TRECLAIM(R3) + + /* Temporarily store r13 and r9 so we have some regs to play with */ + SET_SCRATCH0(r13) + GET_PACA(r13) + std r9, PACATMSCRATCH(r13) + ld r9, HSTATE_KVM_VCPU(r13) + + /* Get a few more GPRs free. */ + std r29, VCPU_GPRS_TM(29)(r9) + std r30, VCPU_GPRS_TM(30)(r9) + std r31, VCPU_GPRS_TM(31)(r9) + + /* Save away PPR and DSCR soon so don't run with user values. */ + mfspr r31, SPRN_PPR + HMT_MEDIUM + mfspr r30, SPRN_DSCR + ld r29, HSTATE_DSCR(r13) + mtspr SPRN_DSCR, r29 + + /* Save all but r9, r13 & r29-r31 */ + reg = 0 + .rept 29 + .if (reg != 9) && (reg != 13) + std reg, VCPU_GPRS_TM(reg)(r9) + .endif + reg = reg + 1 + .endr + /* ... now save r13 */ + GET_SCRATCH0(r4) + std r4, VCPU_GPRS_TM(13)(r9) + /* ... and save r9 */ + ld r4, PACATMSCRATCH(r13) + std r4, VCPU_GPRS_TM(9)(r9) + + /* Reload stack pointer and TOC. */ + ld r1, HSTATE_HOST_R1(r13) + ld r2, PACATOC(r13) + + /* Set MSR RI now we have r1 and r13 back. */ + li r5, MSR_RI + mtmsrd r5, 1 + + /* Save away checkpinted SPRs. */ + std r31, VCPU_PPR_TM(r9) + std r30, VCPU_DSCR_TM(r9) + mflr r5 + mfcr r6 + mfctr r7 + mfspr r8, SPRN_AMR + mfspr r10, SPRN_TAR + std r5, VCPU_LR_TM(r9) + stw r6, VCPU_CR_TM(r9) + std r7, VCPU_CTR_TM(r9) + std r8, VCPU_AMR_TM(r9) + std r10, VCPU_TAR_TM(r9) + + /* Restore r12 as trap number. */ + lwz r12, VCPU_TRAP(r9) + + /* Save FP/VSX. */ + addi r3, r9, VCPU_FPRS_TM + bl .store_fp_state + addi r3, r9, VCPU_VRS_TM + bl .store_vr_state + mfspr r6, SPRN_VRSAVE + stw r6, VCPU_VRSAVE_TM(r9) +1: + /* + * We need to save these SPRs after the treclaim so that the software + * error code is recorded correctly in the TEXASR. Also the user may + * change these outside of a transaction, so they must always be + * context switched. + */ + mfspr r5, SPRN_TFHAR + mfspr r6, SPRN_TFIAR + mfspr r7, SPRN_TEXASR + std r5, VCPU_TFHAR(r9) + std r6, VCPU_TFIAR(r9) + std r7, VCPU_TEXASR(r9) +2: +#endif + /* Increment yield count if they have a VPA */ ld r8, VCPU_VPA(r9) /* do they have a VPA? */ cmpdi r8, 0 @@ -1214,6 +1451,30 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_206) 25: /* Save PMU registers if requested */ /* r8 and cr0.eq are live here */ +BEGIN_FTR_SECTION + /* + * POWER8 seems to have a hardware bug where setting + * MMCR0[PMAE] along with MMCR0[PMC1CE] and/or MMCR0[PMCjCE] + * when some counters are already negative doesn't seem + * to cause a performance monitor alert (and hence interrupt). + * The effect of this is that when saving the PMU state, + * if there is no PMU alert pending when we read MMCR0 + * before freezing the counters, but one becomes pending + * before we read the counters, we lose it. + * To work around this, we need a way to freeze the counters + * before reading MMCR0. Normally, freezing the counters + * is done by writing MMCR0 (to set MMCR0[FC]) which + * unavoidably writes MMCR0[PMA0] as well. On POWER8, + * we can also freeze the counters using MMCR2, by writing + * 1s to all the counter freeze condition bits (there are + * 9 bits each for 6 counters). + */ + li r3, -1 /* set all freeze bits */ + clrrdi r3, r3, 10 + mfspr r10, SPRN_MMCR2 + mtspr SPRN_MMCR2, r3 + isync +END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S) li r3, 1 sldi r3, r3, 31 /* MMCR0_FC (freeze counters) bit */ mfspr r4, SPRN_MMCR0 /* save MMCR0 */ @@ -1237,6 +1498,9 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_206) std r4, VCPU_MMCR(r9) std r5, VCPU_MMCR + 8(r9) std r6, VCPU_MMCR + 16(r9) +BEGIN_FTR_SECTION + std r10, VCPU_MMCR + 24(r9) +END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S) std r7, VCPU_SIAR(r9) std r8, VCPU_SDAR(r9) mfspr r3, SPRN_PMC1 @@ -1260,12 +1524,10 @@ BEGIN_FTR_SECTION stw r11, VCPU_PMC + 28(r9) END_FTR_SECTION_IFSET(CPU_FTR_ARCH_201) BEGIN_FTR_SECTION - mfspr r4, SPRN_MMCR2 mfspr r5, SPRN_SIER mfspr r6, SPRN_SPMC1 mfspr r7, SPRN_SPMC2 mfspr r8, SPRN_MMCRS - std r4, VCPU_MMCR + 24(r9) std r5, VCPU_SIER(r9) stw r6, VCPU_PMC + 24(r9) stw r7, VCPU_PMC + 28(r9) @@ -1387,7 +1649,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S) ld r8,VCORE_TB_OFFSET(r5) cmpdi r8,0 beq 17f - mftb r6 /* current host timebase */ + mftb r6 /* current guest timebase */ subf r8,r8,r6 mtspr SPRN_TBU40,r8 /* update upper 40 bits */ mftb r7 /* check if lower 24 bits overflowed */ @@ -1557,7 +1819,7 @@ kvmppc_hdsi: mtspr SPRN_SRR0, r10 mtspr SPRN_SRR1, r11 li r10, BOOK3S_INTERRUPT_DATA_STORAGE - ld r11, VCPU_INTR_MSR(r9) + bl kvmppc_msr_interrupt fast_interrupt_c_return: 6: ld r7, VCPU_CTR(r9) lwz r8, VCPU_XER(r9) @@ -1626,7 +1888,7 @@ kvmppc_hisi: 1: mtspr SPRN_SRR0, r10 mtspr SPRN_SRR1, r11 li r10, BOOK3S_INTERRUPT_INST_STORAGE - ld r11, VCPU_INTR_MSR(r9) + bl kvmppc_msr_interrupt b fast_interrupt_c_return 3: ld r6, VCPU_KVM(r9) /* not relocated, use VRMA */ @@ -1669,7 +1931,7 @@ sc_1_fast_return: mtspr SPRN_SRR0,r10 mtspr SPRN_SRR1,r11 li r10, BOOK3S_INTERRUPT_SYSCALL - ld r11, VCPU_INTR_MSR(r9) + bl kvmppc_msr_interrupt mr r4,r9 b fast_guest_return @@ -1691,7 +1953,7 @@ hcall_real_table: .long 0 /* 0x10 - H_CLEAR_MOD */ .long 0 /* 0x14 - H_CLEAR_REF */ .long .kvmppc_h_protect - hcall_real_table - .long 0 /* 0x1c - H_GET_TCE */ + .long .kvmppc_h_get_tce - hcall_real_table .long .kvmppc_h_put_tce - hcall_real_table .long 0 /* 0x24 - H_SET_SPRG0 */ .long .kvmppc_h_set_dabr - hcall_real_table @@ -1889,8 +2151,13 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_206) /* * Take a nap until a decrementer or external or doobell interrupt - * occurs, with PECE1, PECE0 and PECEDP set in LPCR + * occurs, with PECE1, PECE0 and PECEDP set in LPCR. Also clear the + * runlatch bit before napping. */ + mfspr r2, SPRN_CTRLF + clrrdi r2, r2, 1 + mtspr SPRN_CTRLT, r2 + li r0,1 stb r0,HSTATE_HWTHREAD_REQ(r13) mfspr r5,SPRN_LPCR @@ -1997,7 +2264,8 @@ machine_check_realmode: beq mc_cont /* If not, deliver a machine check. SRR0/1 are already set */ li r10, BOOK3S_INTERRUPT_MACHINE_CHECK - ld r11, VCPU_INTR_MSR(r9) + ld r11, VCPU_MSR(r9) + bl kvmppc_msr_interrupt b fast_interrupt_c_return /* @@ -2138,8 +2406,6 @@ END_FTR_SECTION_IFSET(CPU_FTR_ALTIVEC) mfspr r6,SPRN_VRSAVE stw r6,VCPU_VRSAVE(r31) mtlr r30 - mtmsrd r5 - isync blr /* @@ -2186,3 +2452,38 @@ END_FTR_SECTION_IFSET(CPU_FTR_ALTIVEC) */ kvmppc_bad_host_intr: b . + +/* + * This mimics the MSR transition on IRQ delivery. The new guest MSR is taken + * from VCPU_INTR_MSR and is modified based on the required TM state changes. + * r11 has the guest MSR value (in/out) + * r9 has a vcpu pointer (in) + * r0 is used as a scratch register + */ +kvmppc_msr_interrupt: + rldicl r0, r11, 64 - MSR_TS_S_LG, 62 + cmpwi r0, 2 /* Check if we are in transactional state.. */ + ld r11, VCPU_INTR_MSR(r9) + bne 1f + /* ... if transactional, change to suspended */ + li r0, 1 +1: rldimi r11, r0, MSR_TS_S_LG, 63 - MSR_TS_T_LG + blr + +/* + * This works around a hardware bug on POWER8E processors, where + * writing a 1 to the MMCR0[PMAO] bit doesn't generate a + * performance monitor interrupt. Instead, when we need to have + * an interrupt pending, we have to arrange for a counter to overflow. + */ +kvmppc_fix_pmao: + li r3, 0 + mtspr SPRN_MMCR2, r3 + lis r3, (MMCR0_PMXE | MMCR0_FCECE)@h + ori r3, r3, MMCR0_PMCjCE | MMCR0_C56RUN + mtspr SPRN_MMCR0, r3 + lis r3, 0x7fff + ori r3, r3, 0xffff + mtspr SPRN_PMC6, r3 + isync + blr |