Skip to contentSkip to conversation

powerpc/kvm: Move pmu code in kvm folder to separate file for power9 and later platforms

PATCH Applied Kajol Jain 4 participants 6 messages Jul 11, 2022 – Aug 2, 2022

Subsystems (2)
lore
Share & tools
lore

Kajol Jain

author
#0 lore raw

PATCH 1/2 powerpc/kvm: Move pmu code in kvm folder to separate file for power9 and later platforms

File book3s_hv_p9_entry.c in powerpc/kvm folder consists of functions like freeze_pmu, switch_pmu_to_guest and switch_pmu_to_host which are specific to Performance Monitoring Unit(PMU) for power9 and later platforms.

For better maintenance, moving pmu related code from book3s_hv_p9_entry.c to a new file called book3s_hv_p9_perf.c, without any logic change.
Also make corresponding changes in the Makefile to include book3s_hv_p9_perf.c during compilation.

---
 arch/powerpc/kvm/Makefile             |   1 +
 arch/powerpc/kvm/book3s_hv_p9_entry.c | 221 -------------------------
 arch/powerpc/kvm/book3s_hv_p9_perf.c  | 225 ++++++++++++++++++++++++++
 3 files changed, 226 insertions(+), 221 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_hv_p9_perf.c
arch/powerpc/kvm/Makefile +1 -0
index 0cd23ce07d68..5319d889b184 100644
@@ -86,6 +86,7 @@ kvm-book3s_64-builtin-objs-$(CONFIG_KVM_BOOK3S_64_HANDLER) += \
 	book3s_hv_rm_mmu.o \
 	book3s_hv_ras.o \
 	book3s_hv_builtin.o \
+	book3s_hv_p9_perf.o \
 	$(kvm-book3s_64-builtin-tm-objs-y) \
 	$(kvm-book3s_64-builtin-xics-objs-y)
 endif
arch/powerpc/kvm/book3s_hv_p9_entry.c +0 -221

231 lines not on this page · raw message

arch/powerpc/kvm/book3s_hv_p9_perf.c new file +225 -0

225 lines not on this page · raw message

Signature · 1 line
2.27.0
author
#1 lore raw

PATCH 2/2 powerpc/kvm: Remove comment related to moving PMU code to perf subsystem

Commit aabcaf6ae2a0 ("KVM: PPC: Book3S HV P9: Move host OS save/restore functions to built-in") added a comment in switch_pmu_to_guest function, indicating possibility of moving PMU handling code to perf subsystem. But perf subsystem code compilation depends upon the enablement of CONFIG_PERF_EVENTS whereas, kvm code don't have any dependency on this config.
Patch remove this comment as switch_pmu_to_guest functionality is needed even if perf subsystem is disabled.

---
 arch/powerpc/kvm/book3s_hv_p9_perf.c | 6 ------
 1 file changed, 6 deletions(-)
arch/powerpc/kvm/book3s_hv_p9_perf.c +0 -6
index da3135cab9ea..44d24cca3df1 100644
@@ -44,12 +44,6 @@ void switch_pmu_to_guest(struct kvm_vcpu *vcpu,
 
 	/* Save host */
 	if (ppc_get_pmu_inuse()) {
-		/*
-		 * It might be better to put PMU handling (at least for the
-		 * host) in the perf subsystem because it knows more about what
-		 * is being used.
-		 */
-
 		/* POWER9, POWER10 do not implement HPMC or SPMC */
 
 		host_os_sprs->mmcr0 = mfspr(SPRN_MMCR0);
Signature · 1 line
2.27.0
#2 lore raw

Re: [PATCH 2/2] powerpc/kvm: Remove comment related to moving PMU code to perf subsystem

Excerpts from Kajol Jain's message of July 11, 2022 1:49 pm:

Quoted 10 lines

Commit aabcaf6ae2a0 ("KVM: PPC: Book3S HV P9: Move host OS save/restore functions to built-in") added a comment in switch_pmu_to_guest function, indicating possibility of moving PMU handling code to perf subsystem. But perf subsystem code compilation depends upon the enablement of CONFIG_PERF_EVENTS whereas, kvm code don't have any dependency on this config.
Patch remove this comment as switch_pmu_to_guest functionality is needed even if perf subsystem is disabled.

Does the host PMU state need to be saved/restored if we don't have PERF_EVENTS enabled? Guest yes, but host maybe could become a no-op?

Thanks,
Nick

Quoted 25 lines
---
 arch/powerpc/kvm/book3s_hv_p9_perf.c | 6 ------
 1 file changed, 6 deletions(-)
arch/powerpc/kvm/book3s_hv_p9_perf.c +0 -6
index da3135cab9ea..44d24cca3df1 100644
@@ -44,12 +44,6 @@ void switch_pmu_to_guest(struct kvm_vcpu *vcpu,
 
 	/* Save host */
 	if (ppc_get_pmu_inuse()) {
-		/*
-		 * It might be better to put PMU handling (at least for the
-		 * host) in the perf subsystem because it knows more about what
-		 * is being used.
-		 */
-
 		/* POWER9, POWER10 do not implement HPMC or SPMC */
 
 		host_os_sprs->mmcr0 = mfspr(SPRN_MMCR0);
Signature · 1 line
2.27.0
reviewer Reviewed-by on v1 1/2
#3 lore raw
In reply to Kajol Jain

Excerpts from Kajol Jain's message of July 11, 2022 1:49 pm:

Quoted 30 lines

File book3s_hv_p9_entry.c in powerpc/kvm folder consists of functions like freeze_pmu, switch_pmu_to_guest and switch_pmu_to_host which are specific to Performance Monitoring Unit(PMU) for power9 and later platforms.

For better maintenance, moving pmu related code from book3s_hv_p9_entry.c to a new file called book3s_hv_p9_perf.c, without any logic change.
Also make corresponding changes in the Makefile to include book3s_hv_p9_perf.c during compilation.

arch/powerpc/kvm/book3s_hv_p9_perf.c new file +17 -0
+
+	if (ppc_get_pmu_inuse()) {
+		mtspr(SPRN_PMC1, host_os_sprs->pmc1);
+		mtspr(SPRN_PMC2, host_os_sprs->pmc2);
+		mtspr(SPRN_PMC3, host_os_sprs->pmc3);
+		mtspr(SPRN_PMC4, host_os_sprs->pmc4);
+		mtspr(SPRN_PMC5, host_os_sprs->pmc5);
+		mtspr(SPRN_PMC6, host_os_sprs->pmc6);
+		mtspr(SPRN_MMCR1, host_os_sprs->mmcr1);
+		mtspr(SPRN_MMCR2, host_os_sprs->mmcr2);
+		mtspr(SPRN_SDAR, host_os_sprs->sdar);
+		mtspr(SPRN_SIAR, host_os_sprs->siar);
+		mtspr(SPRN_SIER, host_os_sprs->sier1);
+
+		if (cpu_has_feature(CPU_FTR_ARCH_31)) {
+			mtspr(SPRN_MMCR3, host_os_sprs->mmcr3);
+			mtspr(SPRN_SIER2, host_os_sprs->sier2);

arch/powerpc/kvm/book3s_hv_p9_perf.c line 225

Quoted 10 lines
arch/powerpc/kvm/book3s_hv_p9_perf.c new file +10 -0
+			mtspr(SPRN_SIER3, host_os_sprs->sier3);
+		}
+
+		/* Set MMCRA then MMCR0 last */
+		mtspr(SPRN_MMCRA, host_os_sprs->mmcra);
+		mtspr(SPRN_MMCR0, host_os_sprs->mmcr0);
+		isync();
+	}
+}
+EXPORT_SYMBOL_GPL(switch_pmu_to_host);

I'm still thinking these parts of the code in particular that do the host PMU save/restore could be handled by calls into perf subsystem. In some cases it doesn't need to save SPRs because it can recreate them or is not using them. Maybe it's not so simple.

Either way, I'm fine with this move to stat with.

#4 lore raw
Quoted 36 lines · Nicholas Piggin

Excerpts from Kajol Jain's message of July 11, 2022 1:49 pm:

Quoted 35 lines

File book3s_hv_p9_entry.c in powerpc/kvm folder consists of functions like freeze_pmu, switch_pmu_to_guest and switch_pmu_to_host which are specific to Performance Monitoring Unit(PMU) for power9 and later platforms.

For better maintenance, moving pmu related code from book3s_hv_p9_entry.c to a new file called book3s_hv_p9_perf.c, without any logic change.
Also make corresponding changes in the Makefile to include book3s_hv_p9_perf.c during compilation.

arch/powerpc/kvm/book3s_hv_p9_perf.c new file +24 -0
+
+	if (ppc_get_pmu_inuse()) {
+		mtspr(SPRN_PMC1, host_os_sprs->pmc1);
+		mtspr(SPRN_PMC2, host_os_sprs->pmc2);
+		mtspr(SPRN_PMC3, host_os_sprs->pmc3);
+		mtspr(SPRN_PMC4, host_os_sprs->pmc4);
+		mtspr(SPRN_PMC5, host_os_sprs->pmc5);
+		mtspr(SPRN_PMC6, host_os_sprs->pmc6);
+		mtspr(SPRN_MMCR1, host_os_sprs->mmcr1);
+		mtspr(SPRN_MMCR2, host_os_sprs->mmcr2);
+		mtspr(SPRN_SDAR, host_os_sprs->sdar);
+		mtspr(SPRN_SIAR, host_os_sprs->siar);
+		mtspr(SPRN_SIER, host_os_sprs->sier1);
+
+		if (cpu_has_feature(CPU_FTR_ARCH_31)) {
+			mtspr(SPRN_MMCR3, host_os_sprs->mmcr3);
+			mtspr(SPRN_SIER2, host_os_sprs->sier2);
+			mtspr(SPRN_SIER3, host_os_sprs->sier3);
+		}
+
+		/* Set MMCRA then MMCR0 last */
+		mtspr(SPRN_MMCRA, host_os_sprs->mmcra);
+		mtspr(SPRN_MMCR0, host_os_sprs->mmcr0);
+		isync();
Quoted 8 lines

arch/powerpc/kvm/book3s_hv_p9_perf.c line 225

Quoted 4 lines
arch/powerpc/kvm/book3s_hv_p9_perf.c new file +3 -0
+	}
+}
+EXPORT_SYMBOL_GPL(switch_pmu_to_host);

I'm still thinking these parts of the code in particular that do the host PMU save/restore could be handled by calls into perf subsystem. In some cases it doesn't need to save SPRs because it can recreate them or is not using them. Maybe it's not so simple.

Yes, we looked at this. Concern for me is counter data leak.
Host application will still have read access to these SPRs
in power9 and before.So I would recommend to save/restore host values as part of it.

Maddy

Quoted 4 lines

Either way, I'm fine with this move to stat with.

#5 lore raw
In reply to Kajol Jain
Quoted 5 lines · Kajol Jain

File book3s_hv_p9_entry.c in powerpc/kvm folder consists of functions like freeze_pmu, switch_pmu_to_guest and switch_pmu_to_host which are specific to Performance Monitoring Unit(PMU) for power9 and later platforms.

Quoted 7 lines

For better maintenance, moving pmu related code from book3s_hv_p9_entry.c to a new file called book3s_hv_p9_perf.c, without any logic change.
Also make corresponding changes in the Makefile to include book3s_hv_p9_perf.c during compilation.

[...]

Applied to powerpc/topic/ppc-kvm.

[1/2] powerpc/kvm: Move pmu code in kvm folder to separate file for power9 and later platforms

      https://git.kernel.org/powerpc/c/db5360840f09eded71009a981084ab10a93a08d8
[2/2] powerpc/kvm: Remove comment related to moving PMU code to perf subsystem
      https://git.kernel.org/powerpc/c/4008d54e29531813e800580f8309133b9b14a921

cheers