* [PATCH v1 1/2] target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert @ 2023-05-30 13:43 Nicholas Piggin 2023-05-30 13:43 ` [PATCH v1 2/2] target/ppc: PMU implement PERFM interrupts Nicholas Piggin 2023-06-05 11:10 ` [PATCH v1 1/2] target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert Daniel Henrique Barboza 0 siblings, 2 replies; 4+ messages in thread From: Nicholas Piggin @ 2023-05-30 13:43 UTC (permalink / raw) To: qemu-ppc Cc: Nicholas Piggin, qemu-devel, Daniel Henrique Barboza, Athira Rajeev FCECE does not get cleared according to the ISA v3.1B. Signed-off-by: Nicholas Piggin <npiggin@gmail.com> --- target/ppc/power8-pmu.c | 1 - 1 file changed, 1 deletion(-) diff --git a/target/ppc/power8-pmu.c b/target/ppc/power8-pmu.c index c4c331c6b5..af065115f2 100644 --- a/target/ppc/power8-pmu.c +++ b/target/ppc/power8-pmu.c @@ -289,7 +289,6 @@ static void fire_PMC_interrupt(PowerPCCPU *cpu) pmu_update_cycles(env); if (env->spr[SPR_POWER_MMCR0] & MMCR0_FCECE) { - env->spr[SPR_POWER_MMCR0] &= ~MMCR0_FCECE; env->spr[SPR_POWER_MMCR0] |= MMCR0_FC; /* Changing MMCR0_FC requires summaries and hflags update */ -- 2.40.1 ^ permalink raw reply related [flat|nested] 4+ messages in thread
* [PATCH v1 2/2] target/ppc: PMU implement PERFM interrupts 2023-05-30 13:43 [PATCH v1 1/2] target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert Nicholas Piggin @ 2023-05-30 13:43 ` Nicholas Piggin 2023-06-05 11:10 ` Daniel Henrique Barboza 2023-06-05 11:10 ` [PATCH v1 1/2] target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert Daniel Henrique Barboza 1 sibling, 1 reply; 4+ messages in thread From: Nicholas Piggin @ 2023-05-30 13:43 UTC (permalink / raw) To: qemu-ppc Cc: Nicholas Piggin, qemu-devel, Daniel Henrique Barboza, Athira Rajeev The PMU raises a performance monitor exception (causing an interrupt when MSR[EE]=1) when MMCR0[PMAO] is set, and lowers it when clear. Wire this up and implement the interrupt delivery for books. Linux perf record can now collect PMI-driven samples. fire_PMC_interrupt is renamed to perfm_alert, which matches a bit closer to the new terminology used in the ISA and distinguishes the alert condition (e.g., counter overflow) from the PERFM (or EBB) interrupts. Signed-off-by: Nicholas Piggin <npiggin@gmail.com> --- target/ppc/excp_helper.c | 2 +- target/ppc/power8-pmu.c | 20 +++++++++++++------- 2 files changed, 14 insertions(+), 8 deletions(-) diff --git a/target/ppc/excp_helper.c b/target/ppc/excp_helper.c index 0f7ed58673..4925996cf3 100644 --- a/target/ppc/excp_helper.c +++ b/target/ppc/excp_helper.c @@ -1611,6 +1611,7 @@ static void powerpc_excp_books(PowerPCCPU *cpu, int excp) case POWERPC_EXCP_ISEG: /* Instruction segment exception */ case POWERPC_EXCP_TRACE: /* Trace exception */ case POWERPC_EXCP_SDOOR: /* Doorbell interrupt */ + case POWERPC_EXCP_PERFM: /* Performance monitor interrupt */ break; case POWERPC_EXCP_HISI: /* Hypervisor instruction storage exception */ msr |= env->error_code; @@ -1664,7 +1665,6 @@ static void powerpc_excp_books(PowerPCCPU *cpu, int excp) */ return; case POWERPC_EXCP_THERM: /* Thermal interrupt */ - case POWERPC_EXCP_PERFM: /* Embedded performance monitor interrupt */ case POWERPC_EXCP_VPUA: /* Vector assist exception */ case POWERPC_EXCP_MAINT: /* Maintenance exception */ case POWERPC_EXCP_HV_MAINT: /* Hypervisor Maintenance exception */ diff --git a/target/ppc/power8-pmu.c b/target/ppc/power8-pmu.c index af065115f2..48f2868f8b 100644 --- a/target/ppc/power8-pmu.c +++ b/target/ppc/power8-pmu.c @@ -84,8 +84,16 @@ static void pmu_update_summaries(CPUPPCState *env) void pmu_mmcr01_updated(CPUPPCState *env) { + PowerPCCPU *cpu = env_archcpu(env); + pmu_update_summaries(env); hreg_update_pmu_hflags(env); + + if (env->spr[SPR_POWER_MMCR0] & MMCR0_PMAO) + ppc_set_irq(cpu, PPC_INTERRUPT_PERFM, 1); + else + ppc_set_irq(cpu, PPC_INTERRUPT_PERFM, 0); + /* * Should this update overflow timers (if mmcr0 is updated) so they * get set in cpu_post_load? @@ -282,7 +290,7 @@ void helper_store_pmc(CPUPPCState *env, uint32_t sprn, uint64_t value) pmc_update_overflow_timer(env, sprn); } -static void fire_PMC_interrupt(PowerPCCPU *cpu) +static void perfm_alert(PowerPCCPU *cpu) { CPUPPCState *env = &cpu->env; @@ -306,6 +314,7 @@ static void fire_PMC_interrupt(PowerPCCPU *cpu) /* These MMCR0 bits do not require summaries or hflags update. */ env->spr[SPR_POWER_MMCR0] &= ~MMCR0_PMAE; env->spr[SPR_POWER_MMCR0] |= MMCR0_PMAO; + ppc_set_irq(cpu, PPC_INTERRUPT_PERFM, 1); } raise_ebb_perfm_exception(env); @@ -314,20 +323,17 @@ static void fire_PMC_interrupt(PowerPCCPU *cpu) void helper_handle_pmc5_overflow(CPUPPCState *env) { env->spr[SPR_POWER_PMC5] = PMC_COUNTER_NEGATIVE_VAL; - fire_PMC_interrupt(env_archcpu(env)); + perfm_alert(env_archcpu(env)); } /* This helper assumes that the PMC is running. */ void helper_insns_inc(CPUPPCState *env, uint32_t num_insns) { bool overflow_triggered; - PowerPCCPU *cpu; overflow_triggered = pmu_increment_insns(env, num_insns); - if (overflow_triggered) { - cpu = env_archcpu(env); - fire_PMC_interrupt(cpu); + perfm_alert(env_archcpu(env)); } } @@ -335,7 +341,7 @@ static void cpu_ppc_pmu_timer_cb(void *opaque) { PowerPCCPU *cpu = opaque; - fire_PMC_interrupt(cpu); + perfm_alert(cpu); } void cpu_ppc_pmu_init(CPUPPCState *env) -- 2.40.1 ^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH v1 2/2] target/ppc: PMU implement PERFM interrupts 2023-05-30 13:43 ` [PATCH v1 2/2] target/ppc: PMU implement PERFM interrupts Nicholas Piggin @ 2023-06-05 11:10 ` Daniel Henrique Barboza 0 siblings, 0 replies; 4+ messages in thread From: Daniel Henrique Barboza @ 2023-06-05 11:10 UTC (permalink / raw) To: Nicholas Piggin, qemu-ppc Cc: qemu-devel, Daniel Henrique Barboza, Athira Rajeev On 5/30/23 10:43, Nicholas Piggin wrote: > The PMU raises a performance monitor exception (causing an interrupt > when MSR[EE]=1) when MMCR0[PMAO] is set, and lowers it when clear. > > Wire this up and implement the interrupt delivery for books. Linux perf > record can now collect PMI-driven samples. > > fire_PMC_interrupt is renamed to perfm_alert, which matches a bit closer > to the new terminology used in the ISA and distinguishes the alert > condition (e.g., counter overflow) from the PERFM (or EBB) interrupts. > > Signed-off-by: Nicholas Piggin <npiggin@gmail.com> > --- Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> And queued. Daniel > target/ppc/excp_helper.c | 2 +- > target/ppc/power8-pmu.c | 20 +++++++++++++------- > 2 files changed, 14 insertions(+), 8 deletions(-) > > diff --git a/target/ppc/excp_helper.c b/target/ppc/excp_helper.c > index 0f7ed58673..4925996cf3 100644 > --- a/target/ppc/excp_helper.c > +++ b/target/ppc/excp_helper.c > @@ -1611,6 +1611,7 @@ static void powerpc_excp_books(PowerPCCPU *cpu, int excp) > case POWERPC_EXCP_ISEG: /* Instruction segment exception */ > case POWERPC_EXCP_TRACE: /* Trace exception */ > case POWERPC_EXCP_SDOOR: /* Doorbell interrupt */ > + case POWERPC_EXCP_PERFM: /* Performance monitor interrupt */ > break; > case POWERPC_EXCP_HISI: /* Hypervisor instruction storage exception */ > msr |= env->error_code; > @@ -1664,7 +1665,6 @@ static void powerpc_excp_books(PowerPCCPU *cpu, int excp) > */ > return; > case POWERPC_EXCP_THERM: /* Thermal interrupt */ > - case POWERPC_EXCP_PERFM: /* Embedded performance monitor interrupt */ > case POWERPC_EXCP_VPUA: /* Vector assist exception */ > case POWERPC_EXCP_MAINT: /* Maintenance exception */ > case POWERPC_EXCP_HV_MAINT: /* Hypervisor Maintenance exception */ > diff --git a/target/ppc/power8-pmu.c b/target/ppc/power8-pmu.c > index af065115f2..48f2868f8b 100644 > --- a/target/ppc/power8-pmu.c > +++ b/target/ppc/power8-pmu.c > @@ -84,8 +84,16 @@ static void pmu_update_summaries(CPUPPCState *env) > > void pmu_mmcr01_updated(CPUPPCState *env) > { > + PowerPCCPU *cpu = env_archcpu(env); > + > pmu_update_summaries(env); > hreg_update_pmu_hflags(env); > + > + if (env->spr[SPR_POWER_MMCR0] & MMCR0_PMAO) > + ppc_set_irq(cpu, PPC_INTERRUPT_PERFM, 1); > + else > + ppc_set_irq(cpu, PPC_INTERRUPT_PERFM, 0); > + > /* > * Should this update overflow timers (if mmcr0 is updated) so they > * get set in cpu_post_load? > @@ -282,7 +290,7 @@ void helper_store_pmc(CPUPPCState *env, uint32_t sprn, uint64_t value) > pmc_update_overflow_timer(env, sprn); > } > > -static void fire_PMC_interrupt(PowerPCCPU *cpu) > +static void perfm_alert(PowerPCCPU *cpu) > { > CPUPPCState *env = &cpu->env; > > @@ -306,6 +314,7 @@ static void fire_PMC_interrupt(PowerPCCPU *cpu) > /* These MMCR0 bits do not require summaries or hflags update. */ > env->spr[SPR_POWER_MMCR0] &= ~MMCR0_PMAE; > env->spr[SPR_POWER_MMCR0] |= MMCR0_PMAO; > + ppc_set_irq(cpu, PPC_INTERRUPT_PERFM, 1); > } > > raise_ebb_perfm_exception(env); > @@ -314,20 +323,17 @@ static void fire_PMC_interrupt(PowerPCCPU *cpu) > void helper_handle_pmc5_overflow(CPUPPCState *env) > { > env->spr[SPR_POWER_PMC5] = PMC_COUNTER_NEGATIVE_VAL; > - fire_PMC_interrupt(env_archcpu(env)); > + perfm_alert(env_archcpu(env)); > } > > /* This helper assumes that the PMC is running. */ > void helper_insns_inc(CPUPPCState *env, uint32_t num_insns) > { > bool overflow_triggered; > - PowerPCCPU *cpu; > > overflow_triggered = pmu_increment_insns(env, num_insns); > - > if (overflow_triggered) { > - cpu = env_archcpu(env); > - fire_PMC_interrupt(cpu); > + perfm_alert(env_archcpu(env)); > } > } > > @@ -335,7 +341,7 @@ static void cpu_ppc_pmu_timer_cb(void *opaque) > { > PowerPCCPU *cpu = opaque; > > - fire_PMC_interrupt(cpu); > + perfm_alert(cpu); > } > > void cpu_ppc_pmu_init(CPUPPCState *env) ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v1 1/2] target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert 2023-05-30 13:43 [PATCH v1 1/2] target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert Nicholas Piggin 2023-05-30 13:43 ` [PATCH v1 2/2] target/ppc: PMU implement PERFM interrupts Nicholas Piggin @ 2023-06-05 11:10 ` Daniel Henrique Barboza 1 sibling, 0 replies; 4+ messages in thread From: Daniel Henrique Barboza @ 2023-06-05 11:10 UTC (permalink / raw) To: Nicholas Piggin, qemu-ppc Cc: qemu-devel, Daniel Henrique Barboza, Athira Rajeev On 5/30/23 10:43, Nicholas Piggin wrote: > FCECE does not get cleared according to the ISA v3.1B. > > Signed-off-by: Nicholas Piggin <npiggin@gmail.com> > --- Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> And queued. Daniel > target/ppc/power8-pmu.c | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/target/ppc/power8-pmu.c b/target/ppc/power8-pmu.c > index c4c331c6b5..af065115f2 100644 > --- a/target/ppc/power8-pmu.c > +++ b/target/ppc/power8-pmu.c > @@ -289,7 +289,6 @@ static void fire_PMC_interrupt(PowerPCCPU *cpu) > pmu_update_cycles(env); > > if (env->spr[SPR_POWER_MMCR0] & MMCR0_FCECE) { > - env->spr[SPR_POWER_MMCR0] &= ~MMCR0_FCECE; > env->spr[SPR_POWER_MMCR0] |= MMCR0_FC; > > /* Changing MMCR0_FC requires summaries and hflags update */ ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2023-06-05 11:11 UTC | newest] Thread overview: 4+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2023-05-30 13:43 [PATCH v1 1/2] target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert Nicholas Piggin 2023-05-30 13:43 ` [PATCH v1 2/2] target/ppc: PMU implement PERFM interrupts Nicholas Piggin 2023-06-05 11:10 ` Daniel Henrique Barboza 2023-06-05 11:10 ` [PATCH v1 1/2] target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert Daniel Henrique Barboza
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).