Skip to contentSkip to conversation

Off-load TLB invalidations to host for !GTSE

PATCH Applied Bharata B Rao 7 participants 13 messages Jul 3, 2020 – Jul 18, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v3 cover Off-load TLB invalidations to host for !GTSE

Hypervisor may choose not to enable Guest Translation Shootdown Enable (GTSE) option for the guest. When GTSE isn't ON, the guest OS isn't permitted to use instructions like tblie and tlbsync directly, but is expected to make hypervisor calls to get the TLB flushed.

This series enables the TLB flush routines in the radix code to off-load TLB flushing to hypervisor via the newly proposed hcall H_RPT_INVALIDATE.

To easily check the availability of GTSE, it is made an MMU feature. The OV5 handling and H_REGISTER_PROC_TBL hcall are changed to handle GTSE as an optionally available feature and to not assume GTSE when radix support is available.

The actual hcall implementation for KVM isn't included in this patchset and will be posted separately.

Changes in v3
=============
- Fixed a bug in the hcall wrapper code where we were missing setting H_RPTI_TYPE_NESTED while retrying the failed flush request with a full flush for the nested case.
- s/psize_to_h_rpti/psize_to_rpti_pgsize

v2: https://lore.kernel.org/linuxppc-dev/[email protected]/T/#t

Bharata B Rao (2):
powerpc/mm: Enable radix GTSE only if supported. powerpc/pseries: H_REGISTER_PROC_TBL should ask for GTSE only if
enabled

Nicholas Piggin (1):
powerpc/mm/book3s64/radix: Off-load TLB invalidations to host when
!GTSE

 .../include/asm/book3s/64/tlbflush-radix.h    | 15 ++++
 arch/powerpc/include/asm/hvcall.h             | 34 +++++++-
 arch/powerpc/include/asm/mmu.h                |  4 +
 arch/powerpc/include/asm/plpar_wrappers.h     | 52 ++++++++++++
 arch/powerpc/kernel/dt_cpu_ftrs.c             |  1 +
 arch/powerpc/kernel/prom_init.c               | 13 +--
 arch/powerpc/mm/book3s64/radix_tlb.c          | 82 +++++++++++++++++--
 arch/powerpc/mm/init_64.c                     |  5 +-
 arch/powerpc/platforms/pseries/lpar.c         |  8 +-
 9 files changed, 197 insertions(+), 17 deletions(-)
Signature · 1 line
2.21.3
#1 lore raw

PATCH v3 1/3 powerpc/mm: Enable radix GTSE only if supported.

Make GTSE an MMU feature and enable it by default for radix. However for guest, conditionally enable it if hypervisor supports it via OV5 vector. Let prom_init ask for radix GTSE only if the support exists.

Having GTSE as an MMU feature will make it easy to enable radix without GTSE. Currently radix assumes GTSE is enabled by default.

---
 arch/powerpc/include/asm/mmu.h    |  4 ++++
 arch/powerpc/kernel/dt_cpu_ftrs.c |  1 +
 arch/powerpc/kernel/prom_init.c   | 13 ++++++++-----
 arch/powerpc/mm/init_64.c         |  5 ++++-
 4 files changed, 17 insertions(+), 6 deletions(-)
arch/powerpc/include/asm/mmu.h +4 -0
index f4ac25d4df05..884d51995934 100644
@@ -28,6 +28,9 @@
  * Individual features below.
  */
 
+/* Guest Translation Shootdown Enable */
+#define MMU_FTR_GTSE			ASM_CONST(0x00001000)
+
 /*
  * Support for 68 bit VA space. We added that from ISA 2.05
  */
@@ -173,6 +176,7 @@ enum {
 #endif
 #ifdef CONFIG_PPC_RADIX_MMU
 		MMU_FTR_TYPE_RADIX |
+		MMU_FTR_GTSE |
 #ifdef CONFIG_PPC_KUAP
 		MMU_FTR_RADIX_KUAP |
 #endif /* CONFIG_PPC_KUAP */
arch/powerpc/kernel/dt_cpu_ftrs.c +1 -0
index a0edeb391e3e..ac650c233cd9 100644
@@ -336,6 +336,7 @@ static int __init feat_enable_mmu_radix(struct dt_cpu_feature *f)
 #ifdef CONFIG_PPC_RADIX_MMU
 	cur_cpu_spec->mmu_features |= MMU_FTR_TYPE_RADIX;
 	cur_cpu_spec->mmu_features |= MMU_FTRS_HASH_BASE;
+	cur_cpu_spec->mmu_features |= MMU_FTR_GTSE;
 	cur_cpu_spec->cpu_user_features |= PPC_FEATURE_HAS_MMU;
 
 	return 1;
arch/powerpc/kernel/prom_init.c +8 -5
index 90c604d00b7d..cbc605cfdec0 100644
@@ -1336,12 +1336,15 @@ static void __init prom_check_platform_support(void)
 		}
 	}
 
-	if (supported.radix_mmu && supported.radix_gtse &&
-	    IS_ENABLED(CONFIG_PPC_RADIX_MMU)) {
-		/* Radix preferred - but we require GTSE for now */
-		prom_debug("Asking for radix with GTSE\n");
+	if (supported.radix_mmu && IS_ENABLED(CONFIG_PPC_RADIX_MMU)) {
+		/* Radix preferred - Check if GTSE is also supported */
+		prom_debug("Asking for radix\n");
 		ibm_architecture_vec.vec5.mmu = OV5_FEAT(OV5_MMU_RADIX);
-		ibm_architecture_vec.vec5.radix_ext = OV5_FEAT(OV5_RADIX_GTSE);
+		if (supported.radix_gtse)
+			ibm_architecture_vec.vec5.radix_ext =
+					OV5_FEAT(OV5_RADIX_GTSE);
+		else
+			prom_debug("Radix GTSE isn't supported\n");
 	} else if (supported.hash_mmu) {
 		/* Default to hash mmu (if we can) */
 		prom_debug("Asking for hash\n");
arch/powerpc/mm/init_64.c +4 -1
index bc73abf0bc25..152aa0200cef 100644
@@ -407,12 +407,15 @@ static void __init early_check_vec5(void)
 		if (!(vec5[OV5_INDX(OV5_RADIX_GTSE)] &
 						OV5_FEAT(OV5_RADIX_GTSE))) {
 			pr_warn("WARNING: Hypervisor doesn't support RADIX with GTSE\n");
-		}
+			cur_cpu_spec->mmu_features &= ~MMU_FTR_GTSE;
+		} else
+			cur_cpu_spec->mmu_features |= MMU_FTR_GTSE;
 		/* Do radix anyway - the hypervisor said we had to */
 		cur_cpu_spec->mmu_features |= MMU_FTR_TYPE_RADIX;
 	} else if (mmu_supported == OV5_FEAT(OV5_MMU_HASH)) {
 		/* Hypervisor only supports hash - disable radix */
 		cur_cpu_spec->mmu_features &= ~MMU_FTR_TYPE_RADIX;
+		cur_cpu_spec->mmu_features &= ~MMU_FTR_GTSE;
 	}
 }
 
Signature · 1 line
2.21.3
#2 lore raw

Re: [PATCH v3 1/3] powerpc/mm: Enable radix GTSE only if supported.

Hi Bharata,

Quoted 81 lines · Bharata B Rao

Make GTSE an MMU feature and enable it by default for radix. However for guest, conditionally enable it if hypervisor supports it via OV5 vector. Let prom_init ask for radix GTSE only if the support exists.

Having GTSE as an MMU feature will make it easy to enable radix without GTSE. Currently radix assumes GTSE is enabled by default.

---
 arch/powerpc/include/asm/mmu.h    |  4 ++++
 arch/powerpc/kernel/dt_cpu_ftrs.c |  1 +
 arch/powerpc/kernel/prom_init.c   | 13 ++++++++-----
 arch/powerpc/mm/init_64.c         |  5 ++++-
 4 files changed, 17 insertions(+), 6 deletions(-)
arch/powerpc/include/asm/mmu.h +4 -0
index f4ac25d4df05..884d51995934 100644
@@ -28,6 +28,9 @@
  * Individual features below.
  */
 
+/* Guest Translation Shootdown Enable */
+#define MMU_FTR_GTSE			ASM_CONST(0x00001000)
+
 /*
  * Support for 68 bit VA space. We added that from ISA 2.05
  */
@@ -173,6 +176,7 @@ enum {
 #endif
 #ifdef CONFIG_PPC_RADIX_MMU
 		MMU_FTR_TYPE_RADIX |
+		MMU_FTR_GTSE |
 #ifdef CONFIG_PPC_KUAP
 		MMU_FTR_RADIX_KUAP |
 #endif /* CONFIG_PPC_KUAP */
arch/powerpc/kernel/dt_cpu_ftrs.c +1 -0
index a0edeb391e3e..ac650c233cd9 100644
@@ -336,6 +336,7 @@ static int __init feat_enable_mmu_radix(struct dt_cpu_feature *f)
 #ifdef CONFIG_PPC_RADIX_MMU
 	cur_cpu_spec->mmu_features |= MMU_FTR_TYPE_RADIX;
 	cur_cpu_spec->mmu_features |= MMU_FTRS_HASH_BASE;
+	cur_cpu_spec->mmu_features |= MMU_FTR_GTSE;
 	cur_cpu_spec->cpu_user_features |= PPC_FEATURE_HAS_MMU;
 
 	return 1;
arch/powerpc/kernel/prom_init.c +8 -5
index 90c604d00b7d..cbc605cfdec0 100644
@@ -1336,12 +1336,15 @@ static void __init prom_check_platform_support(void)
 		}
 	}
 
-	if (supported.radix_mmu && supported.radix_gtse &&
-	    IS_ENABLED(CONFIG_PPC_RADIX_MMU)) {
-		/* Radix preferred - but we require GTSE for now */
-		prom_debug("Asking for radix with GTSE\n");
+	if (supported.radix_mmu && IS_ENABLED(CONFIG_PPC_RADIX_MMU)) {
+		/* Radix preferred - Check if GTSE is also supported */
+		prom_debug("Asking for radix\n");
 		ibm_architecture_vec.vec5.mmu = OV5_FEAT(OV5_MMU_RADIX);
-		ibm_architecture_vec.vec5.radix_ext = OV5_FEAT(OV5_RADIX_GTSE);
+		if (supported.radix_gtse)
+			ibm_architecture_vec.vec5.radix_ext =
+					OV5_FEAT(OV5_RADIX_GTSE);
+		else
+			prom_debug("Radix GTSE isn't supported\n");
 	} else if (supported.hash_mmu) {
 		/* Default to hash mmu (if we can) */
 		prom_debug("Asking for hash\n");
arch/powerpc/mm/init_64.c +0 -0
index bc73abf0bc25..152aa0200cef 100644

arch/powerpc/mm/init_64.c static void __init early_check_vec5(void) · line 412

Quoted 7 lines
arch/powerpc/mm/init_64.c +3 -1
@@ -407,12 +407,15 @@ static void __init early_check_vec5(void)
 		if (!(vec5[OV5_INDX(OV5_RADIX_GTSE)] &
 						OV5_FEAT(OV5_RADIX_GTSE))) {
 			pr_warn("WARNING: Hypervisor doesn't support RADIX with GTSE\n");
-		}
+			cur_cpu_spec->mmu_features &= ~MMU_FTR_GTSE;
+		} else
+			cur_cpu_spec->mmu_features |= MMU_FTR_GTSE;

The GTSE flag is set by default in feat_enable_mmu_radix(), should it be set again here?

Thanks,
Santosh

Quoted 11 lines
arch/powerpc/mm/init_64.c +1 -0
 		/* Do radix anyway - the hypervisor said we had to */
 		cur_cpu_spec->mmu_features |= MMU_FTR_TYPE_RADIX;
 	} else if (mmu_supported == OV5_FEAT(OV5_MMU_HASH)) {
 		/* Hypervisor only supports hash - disable radix */
 		cur_cpu_spec->mmu_features &= ~MMU_FTR_TYPE_RADIX;
+		cur_cpu_spec->mmu_features &= ~MMU_FTR_GTSE;
 	}
 }
 
Signature · 1 line
2.21.3
#3 lore raw

Re: [PATCH v3 1/3] powerpc/mm: Enable radix GTSE only if supported.

Quoted 88 lines · Santosh Sivaraj

Hi Bharata,

Quoted 83 lines · Bharata B Rao

Make GTSE an MMU feature and enable it by default for radix. However for guest, conditionally enable it if hypervisor supports it via OV5 vector. Let prom_init ask for radix GTSE only if the support exists.

Having GTSE as an MMU feature will make it easy to enable radix without GTSE. Currently radix assumes GTSE is enabled by default.

---
 arch/powerpc/include/asm/mmu.h    |  4 ++++
 arch/powerpc/kernel/dt_cpu_ftrs.c |  1 +
 arch/powerpc/kernel/prom_init.c   | 13 ++++++++-----
 arch/powerpc/mm/init_64.c         |  5 ++++-
 4 files changed, 17 insertions(+), 6 deletions(-)
arch/powerpc/include/asm/mmu.h +4 -0
index f4ac25d4df05..884d51995934 100644
@@ -28,6 +28,9 @@
  * Individual features below.
  */
 
+/* Guest Translation Shootdown Enable */
+#define MMU_FTR_GTSE			ASM_CONST(0x00001000)
+
 /*
  * Support for 68 bit VA space. We added that from ISA 2.05
  */
@@ -173,6 +176,7 @@ enum {
 #endif
 #ifdef CONFIG_PPC_RADIX_MMU
 		MMU_FTR_TYPE_RADIX |
+		MMU_FTR_GTSE |
 #ifdef CONFIG_PPC_KUAP
 		MMU_FTR_RADIX_KUAP |
 #endif /* CONFIG_PPC_KUAP */
arch/powerpc/kernel/dt_cpu_ftrs.c +1 -0
index a0edeb391e3e..ac650c233cd9 100644
@@ -336,6 +336,7 @@ static int __init feat_enable_mmu_radix(struct dt_cpu_feature *f)
 #ifdef CONFIG_PPC_RADIX_MMU
 	cur_cpu_spec->mmu_features |= MMU_FTR_TYPE_RADIX;
 	cur_cpu_spec->mmu_features |= MMU_FTRS_HASH_BASE;
+	cur_cpu_spec->mmu_features |= MMU_FTR_GTSE;
 	cur_cpu_spec->cpu_user_features |= PPC_FEATURE_HAS_MMU;
 
 	return 1;
arch/powerpc/kernel/prom_init.c +8 -5
index 90c604d00b7d..cbc605cfdec0 100644
@@ -1336,12 +1336,15 @@ static void __init prom_check_platform_support(void)
 		}
 	}
 
-	if (supported.radix_mmu && supported.radix_gtse &&
-	    IS_ENABLED(CONFIG_PPC_RADIX_MMU)) {
-		/* Radix preferred - but we require GTSE for now */
-		prom_debug("Asking for radix with GTSE\n");
+	if (supported.radix_mmu && IS_ENABLED(CONFIG_PPC_RADIX_MMU)) {
+		/* Radix preferred - Check if GTSE is also supported */
+		prom_debug("Asking for radix\n");
 		ibm_architecture_vec.vec5.mmu = OV5_FEAT(OV5_MMU_RADIX);
-		ibm_architecture_vec.vec5.radix_ext = OV5_FEAT(OV5_RADIX_GTSE);
+		if (supported.radix_gtse)
+			ibm_architecture_vec.vec5.radix_ext =
+					OV5_FEAT(OV5_RADIX_GTSE);
+		else
+			prom_debug("Radix GTSE isn't supported\n");
 	} else if (supported.hash_mmu) {
 		/* Default to hash mmu (if we can) */
 		prom_debug("Asking for hash\n");
arch/powerpc/mm/init_64.c +0 -0
index bc73abf0bc25..152aa0200cef 100644
@@ -407,12 +407,15 @@ static void __init early_check_vec5(void)
 		if (!(vec5[OV5_INDX(OV5_RADIX_GTSE)] &
 						OV5_FEAT(OV5_RADIX_GTSE))) {
Quoted 8 lines

arch/powerpc/mm/init_64.c static void __init early_check_vec5(void) · line 412

Quoted 5 lines
arch/powerpc/mm/init_64.c +3 -1
 			pr_warn("WARNING: Hypervisor doesn't support RADIX with GTSE\n");
-		}
+			cur_cpu_spec->mmu_features &= ~MMU_FTR_GTSE;
+		} else
+			cur_cpu_spec->mmu_features |= MMU_FTR_GTSE;

The GTSE flag is set by default in feat_enable_mmu_radix(), should it be set again here?

Strictly speaking no, but makes it a bit explicit and also follows what the related feature does below:

Quoted 2 lines

arch/powerpc/mm/init_64.c static void __init early_check_vec5(void) · line 414

Quoted 2 lines
arch/powerpc/mm/init_64.c +0 -0
 		/* Do radix anyway - the hypervisor said we had to */
 		cur_cpu_spec->mmu_features |= MMU_FTR_TYPE_RADIX;

Regards,
Bharata.

#4 lore raw

PATCH v3 2/3 powerpc/pseries: H_REGISTER_PROC_TBL should ask for GTSE only if enabled

In reply to Bharata B Rao

H_REGISTER_PROC_TBL asks for GTSE by default. GTSE flag bit should be set only when GTSE is supported.

---
 arch/powerpc/platforms/pseries/lpar.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
arch/powerpc/platforms/pseries/lpar.c +5 -3
index fd26f3d21d7b..f82569a505f1 100644
@@ -1680,9 +1680,11 @@ static int pseries_lpar_register_process_table(unsigned long base,
 
 	if (table_size)
 		flags |= PROC_TABLE_NEW;
-	if (radix_enabled())
-		flags |= PROC_TABLE_RADIX | PROC_TABLE_GTSE;
-	else
+	if (radix_enabled()) {
+		flags |= PROC_TABLE_RADIX;
+		if (mmu_has_feature(MMU_FTR_GTSE))
+			flags |= PROC_TABLE_GTSE;
+	} else
 		flags |= PROC_TABLE_HPT_SLB;
 	for (;;) {
 		rc = plpar_hcall_norets(H_REGISTER_PROC_TBL, flags, base,
Signature · 1 line
2.21.3
#5 lore raw

PATCH v3 3/3 powerpc/mm/book3s64/radix: Off-load TLB invalidations to host when !GTSE

In reply to Bharata B Rao

From: Nicholas Piggin <[email protected]>

When platform doesn't support GTSE, let TLB invalidation requests for radix guests be off-loaded to the host using H_RPT_INVALIDATE hcall.

[hcall wrapper, error path handling and renames]
---
 .../include/asm/book3s/64/tlbflush-radix.h    | 15 ++++
 arch/powerpc/include/asm/hvcall.h             | 34 +++++++-
 arch/powerpc/include/asm/plpar_wrappers.h     | 52 ++++++++++++
 arch/powerpc/mm/book3s64/radix_tlb.c          | 82 +++++++++++++++++--
 4 files changed, 175 insertions(+), 8 deletions(-)
arch/powerpc/include/asm/book3s/64/tlbflush-radix.h +15 -0
index ca8db193ae38..94439e0cefc9 100644
@@ -2,10 +2,25 @@
 #ifndef _ASM_POWERPC_TLBFLUSH_RADIX_H
 #define _ASM_POWERPC_TLBFLUSH_RADIX_H
 
+#include <asm/hvcall.h>
+
 struct vm_area_struct;
 struct mm_struct;
 struct mmu_gather;
 
+static inline u64 psize_to_rpti_pgsize(unsigned long psize)
+{
+	if (psize == MMU_PAGE_4K)
+		return H_RPTI_PAGE_4K;
+	if (psize == MMU_PAGE_64K)
+		return H_RPTI_PAGE_64K;
+	if (psize == MMU_PAGE_2M)
+		return H_RPTI_PAGE_2M;
+	if (psize == MMU_PAGE_1G)
+		return H_RPTI_PAGE_1G;
+	return H_RPTI_PAGE_ALL;
+}
+
 static inline int mmu_get_ap(int psize)
 {
 	return mmu_psize_defs[psize].ap;
arch/powerpc/include/asm/hvcall.h +33 -1
index e90c073e437e..43486e773bd6 100644
@@ -305,7 +305,8 @@
 #define H_SCM_UNBIND_ALL        0x3FC
 #define H_SCM_HEALTH            0x400
 #define H_SCM_PERFORMANCE_STATS 0x418
-#define MAX_HCALL_OPCODE	H_SCM_PERFORMANCE_STATS
+#define H_RPT_INVALIDATE	0x448
+#define MAX_HCALL_OPCODE	H_RPT_INVALIDATE
 
 /* Scope args for H_SCM_UNBIND_ALL */
 #define H_UNBIND_SCOPE_ALL (0x1)
@@ -389,6 +390,37 @@
 #define PROC_TABLE_RADIX	0x04
 #define PROC_TABLE_GTSE		0x01
 
+/*
+ * Defines for
+ * H_RPT_INVALIDATE - Invalidate RPT translation lookaside information.
+ */
+
+/* Type of translation to invalidate (type) */
+#define H_RPTI_TYPE_NESTED	0x0001	/* Invalidate nested guest partition-scope */
+#define H_RPTI_TYPE_TLB		0x0002	/* Invalidate TLB */
+#define H_RPTI_TYPE_PWC		0x0004	/* Invalidate Page Walk Cache */
+/* Invalidate Process Table Entries if H_RPTI_TYPE_NESTED is clear */
+#define H_RPTI_TYPE_PRT		0x0008
+/* Invalidate Partition Table Entries if H_RPTI_TYPE_NESTED is set */
+#define H_RPTI_TYPE_PAT		0x0008
+#define H_RPTI_TYPE_ALL		(H_RPTI_TYPE_TLB | H_RPTI_TYPE_PWC | \
+				 H_RPTI_TYPE_PRT)
+#define H_RPTI_TYPE_NESTED_ALL	(H_RPTI_TYPE_TLB | H_RPTI_TYPE_PWC | \
+				 H_RPTI_TYPE_PAT)
+
+/* Invalidation targets (target) */
+#define H_RPTI_TARGET_CMMU		0x01 /* All virtual processors in the partition */
+#define H_RPTI_TARGET_CMMU_LOCAL	0x02 /* Current virtual processor */
+/* All nest/accelerator agents in use by the partition */
+#define H_RPTI_TARGET_NMMU		0x04
+
+/* Page size mask (page sizes) */
+#define H_RPTI_PAGE_4K	0x01
+#define H_RPTI_PAGE_64K	0x02
+#define H_RPTI_PAGE_2M	0x04
+#define H_RPTI_PAGE_1G	0x08
+#define H_RPTI_PAGE_ALL (-1UL)
+
 #ifndef __ASSEMBLY__
 #include <linux/types.h>
 
arch/powerpc/include/asm/plpar_wrappers.h +52 -0

64 lines not on this page · raw message

arch/powerpc/mm/book3s64/radix_tlb.c +75 -7

130 lines not on this page · raw message

Signature · 1 line
2.21.3
#6 lore raw
In reply to Bharata B Rao
Quoted 10 lines · Bharata B Rao

Hypervisor may choose not to enable Guest Translation Shootdown Enable (GTSE) option for the guest. When GTSE isn't ON, the guest OS isn't permitted to use instructions like tblie and tlbsync directly, but is expected to make hypervisor calls to get the TLB flushed.

This series enables the TLB flush routines in the radix code to off-load TLB flushing to hypervisor via the newly proposed hcall H_RPT_INVALIDATE.

[...]

Applied to powerpc/next.

[1/3] powerpc/mm: Enable radix GTSE only if supported.

      https://git.kernel.org/powerpc/c/029ab30b4c0a7ec587eece1ec07c3981fdff2bed
[2/3] powerpc/pseries: H_REGISTER_PROC_TBL should ask for GTSE only if enabled
      https://git.kernel.org/powerpc/c/b6c84175078ff022b343b7b0737aeb33001ca90c
[3/3] powerpc/mm/book3s64/radix: Off-load TLB invalidations to host when !GTSE
      https://git.kernel.org/powerpc/c/dd3d9aa5589c52efaec12ffeb84f0f5f8208fbc3

cheers

#7 lore raw
In reply to Bharata B Rao
Quoted 26 lines · Bharata B Rao

Hypervisor may choose not to enable Guest Translation Shootdown Enable (GTSE) option for the guest. When GTSE isn't ON, the guest OS isn't permitted to use instructions like tblie and tlbsync directly, but is expected to make hypervisor calls to get the TLB flushed.

This series enables the TLB flush routines in the radix code to off-load TLB flushing to hypervisor via the newly proposed hcall H_RPT_INVALIDATE.

To easily check the availability of GTSE, it is made an MMU feature. The OV5 handling and H_REGISTER_PROC_TBL hcall are changed to handle GTSE as an optionally available feature and to not assume GTSE when radix support is available.

The actual hcall implementation for KVM isn't included in this patchset and will be posted separately.

Changes in v3
=============
- Fixed a bug in the hcall wrapper code where we were missing setting H_RPTI_TYPE_NESTED while retrying the failed flush request with a full flush for the nested case.
- s/psize_to_h_rpti/psize_to_rpti_pgsize

v2: https://lore.kernel.org/linuxppc-dev/[email protected]/T/#t

Quoted 8 lines

Bharata B Rao (2):
powerpc/mm: Enable radix GTSE only if supported. powerpc/pseries: H_REGISTER_PROC_TBL should ask for GTSE only if
enabled

Nicholas Piggin (1):
powerpc/mm/book3s64/radix: Off-load TLB invalidations to host when
!GTSE

Reverting the whole series fixed random memory corruptions during boot on POWER9 PowerNV systems below.

IBM 8335-GTH (ibm,witherspoon)
POWER9, altivec supported
262144 MB memory, 2000 GB disk space

.config:
https://gitlab.com/cailca/linux-mm/-/blob/master/powerpc.config

[    9.338996][  T925] BUG: Unable to handle kernel instruction fetch (NULL pointer?)
[    9.339026][  T925] Faulting instruction address: 0x00000000
[    9.339051][  T925] Oops: Kernel access of bad area, sig: 11 [#1]
[    9.339064][  T925] LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=256 NUMA PowerNV
[    9.339098][  T925] Modules linked in: dm_mirror dm_region_hash dm_log dm_mod
[    9.339150][  T925] CPU: 92 PID: 925 Comm: (md-udevd) Not tainted 5.8.0-rc5-next-20200716 #3
[    9.339186][  T925] NIP:  0000000000000000 LR: c00000000021f2cc CTR: 0000000000000000
[    9.339210][  T925] REGS: c000201cb52d79b0 TRAP: 0400   Not tainted  (5.8.0-rc5-next-20200716)
[    9.339244][  T925] MSR:  9000000040009033 <SF,HV,EE,ME,IR,DR,RI,LE>  CR: 24222292  XER: 00000000
[    9.339278][  T925] CFAR: c00000000021f2c8 IRQMASK: 0 
[    9.339278][  T925] GPR00: c00000000021f2cc c000201cb52d7c40 c000000005901000 c000201cb52d7ca8 
[    9.339278][  T925] GPR04: c00800000ea60038 0000000000000000 000000007fff0000 000000007fff0000 
[    9.339278][  T925] GPR08: 0000000000000000 0000000000000000 c000201cb50bd500 0000000000000003 
[    9.339278][  T925] GPR12: 0000000000000000 c000201fff694500 00007fffa4a8a940 00007fffa4a8a6c8 
[    9.339278][  T925] GPR16: 00007fffa4a8a8f8 00007fffa4a8a650 00007fffa4a8a488 0000000000000000 
[    9.339278][  T925] GPR20: 0000000000050001 00007fffa4a8a984 000000007fff0000 c00000000a4545cc 
[    9.339278][  T925] GPR24: c000000000affe28 0000000000000000 0000000000000000 0000000000000166 
[    9.339278][  T925] GPR28: c000201cb52d7ca8 c00800000ea60000 c000201cc3b72600 000000007fff0000 
[    9.339493][  T925] NIP [0000000000000000] 0x0
[    9.339516][  T925] LR [c00000000021f2cc] __seccomp_filter+0xec/0x530

bpf_dispatcher_nop_func at include/linux/bpf.h:567 (inlined by) bpf_prog_run_pin_on_cpu at include/linux/filter.h:597 (inlined by) seccomp_run_filters at kernel/seccomp.c:324 (inlined by) __seccomp_filter at kernel/seccomp.c:937

[    9.339538][  T925] Call Trace:
[    9.339548][  T925] [c000201cb52d7c40] [c00000000021f2cc] __seccomp_filter+0xec/0x530 (unreliable)
[    9.339566][  T925] [c000201cb52d7d50] [c000000000025af8] do_syscall_trace_enter+0xb8/0x470
do_seccomp at arch/powerpc/kernel/ptrace/ptrace.c:252
(inlined by) do_syscall_trace_enter at arch/powerpc/kernel/ptrace/ptrace.c:327
[    9.339600][  T925] [c000201cb52d7dc0] [c00000000002c8f8] system_call_exception+0x138/0x180
[    9.339625][  T925] [c000201cb52d7e20] [c00000000000c9e8] system_call_common+0xe8/0x214
[    9.339648][  T925] Instruction dump:
[    9.339667][  T925] XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX 
[    9.339706][  T925] XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX 
[    9.339748][  T925] ---[ end trace d89eb80f9a6bc141 ]---
[  OK  ] Started Journal Service.
[   10.452364][  T925] Kernel panic - not syncing: Fatal exception
[   11.876655][  T925] ---[ end Kernel panic - not syncing: Fatal exception ]---

There could also be lots of random userspace segfault like,

[   16.463545][  T771] rngd[771]: segfault (11) at 0 nip 0 lr 0 code 1 in rngd[106d60000+20000]
[   16.463620][  T771] rngd[771]: code: XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX 
[   16.463656][  T771] rngd[771]: code: XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX

Occasionally, there are many soft-lockups,

[  396.920702][   C99] watchdog: BUG: soft lockup - CPU#99 stuck for 22s! [(spawn):2692]
[  396.920754][   C99] Modules linked in: kvm_hv kvm ip_tables x_tables sd_mod bnx2x tg3 ahci mdio libahci libphy firmware_class libata dm_mirror dm_region_hash dm_log dm_mod
[  396.920843][   C99] irq event stamp: 1731717220
[  396.920860][   C99] hardirqs last  enabled at (1731717219): [<c00000000004d6f4>] do_page_fault+0x324/0xd90
[  396.920889][   C99] hardirqs last disabled at (1731717220): [<c000000000015638>] arch_local_irq_restore+0x48/0xd0
[  396.920919][   C99] softirqs last  enabled at (41260): [<c0000000009abbe8>] __do_softirq+0x648/0x8c8
[  396.920948][   C99] softirqs last disabled at (41125): [<c0000000000d717c>] irq_exit+0x15c/0x1c0
[  396.920976][   C99] CPU: 99 PID: 2692 Comm: (spawn) Tainted: G             L    5.8.0-rc5-next-20200716 #3
[  396.921001][   C99] NIP:  c0000000000152b4 LR: c000000000015640 CTR: 0000000000000000
[  396.921037][   C99] REGS: c000201cbc3d7178 TRAP: 0900   Tainted: G             L     (5.8.0-rc5-next-20200716)
[  396.921074][   C99] MSR:  9000000000001033 <SF,HV,ME,IR,DR,RI,LE>  CR: 28022482  XER: 20040000
[  396.921122][   C99] CFAR: 0000000000000000 IRQMASK: 3 
[  396.921122][   C99] GPR00: c000000000015640 c000201cbc3d7340 c000000005901000 c000201cbc3d7178 
[  396.921122][   C99] GPR04: c0000000057d7280 0000000000000000 000000000002000a 0000000000000003 
[  396.921122][   C99] GPR08: 0000201cc61c0000 0000000000000000 0000000000000001 c00000000593f868 
[  396.921122][   C99] GPR12: 0000000000002000 c000201fff67e700 00007fffdcda3918 0000000139eeba60 
[  396.921122][   C99] GPR16: 0000000139f30130 00007fffdcda39c8 0000000139eea708 0000000000000000 
[  396.921122][   C99] GPR20: 0000000000000000 0000000000000000 0000000000000000 0000000000000008 
[  396.921122][   C99] GPR24: 0000000000000e60 0000000000000900 0000000000000500 0000000000000a00 
[  396.921122][   C99] GPR28: 0000000000000f00 0000000000000002 0000000000000003 c000201ca4212400 
[  396.921468][   C99] NIP [c0000000000152b4] replay_soft_interrupts+0x74/0x3b0
replay_soft_interrupts at arch/powerpc/kernel/irq.c:216
[  396.921504][   C99] LR [c000000000015640] arch_local_irq_restore+0x50/0xd0
arch_local_irq_restore at arch/powerpc/kernel/irq.c:375
[  396.921539][   C99] Call Trace:
[  396.921560][   C99] [c000201cbc3d7340] [c000000000015640] arch_local_irq_restore+0x50/0xd0 (unreliable)
[  396.921602][   C99] [c000201cbc3d7360] [c0000000009a0c68] lock_is_held_type+0xf8/0x180
[  396.921641][   C99] [c000201cbc3d73c0] [c0000000003e8cf0] mem_cgroup_from_task+0xa0/0x130
[  396.921666][   C99] [c000201cbc3d7400] [c000000000337950] handle_mm_fault+0x140/0x1d20
[  396.921703][   C99] [c000201cbc3d7500] [c00000000004d5ac] do_page_fault+0x1dc/0xd90
[  396.921763][   C99] [c000201cbc3d7600] [c00000000000c028] handle_page_fault+0x10/0x2c
[  396.921804][   C99] --- interrupt: 300 at futex_cleanup+0x3c0/0x740
[  396.921804][   C99]     LR = futex_cleanup+0x35c/0x740
[  396.921879][   C99] [c000201cbc3d79c0] [c0000000001df2e8] futex_exec_release+0x28/0x50
[  396.921929][   C99] [c000201cbc3d79f0] [c0000000000c5e54] exec_mm_release+0x24/0x50
[  396.921968][   C99] [c000201cbc3d7a30] [c000000000421e84] begin_new_exec+0x324/0xea0
[  396.922005][   C99] [c000201cbc3d7af0] [c0000000004d8f1c] load_elf_binary+0x7fc/0x1110
[  396.922042][   C99] [c000201cbc3d7bf0] [c000000000420824] exec_binprm+0x1c4/0x7d0
[  396.922079][   C99] [c000201cbc3d7cb0] [c000000000421540] do_execveat_common+0x710/0x960
[  396.922117][   C99] [c000201cbc3d7d90] [c000000000422a44] sys_execve+0x44/0x60
[  396.922156][   C99] [c000201cbc3d7dc0] [c00000000002c8b8] system_call_exception+0xf8/0x180
[  396.922205][   C99] [c000201cbc3d7e20] [c00000000000c9e8] system_call_common+0xe8/0x214
[  396.922253][   C99] Instruction dump:
[  396.922286][   C99] 3b000e60 3b400500 3b600a00 3b800f00 f8010010 f821fe11 38610028 e92d0c70 
[  396.922316][   C99] f9210198 39200000 8aed0989 48037df9 <60000000> 39200003 f9210160 56e90738
[  248.821138][  T676] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. 
[  248.821170][  T676] khugepaged      D28416   682      2 0x00000800 
[  248.821212][  T676] Call Trace: 
[  248.821241][  T676] [c000001ff310f4e0] [c000000000caeb80] lru_add_drain_work+0x0/0x48 (unreliable) 
[  248.821275][  T676] [c000001ff310f6c0] [c00000000001a2d0] __switch_to+0x260/0x380 
[  248.821308][  T676] [c000001ff310f720] [c0000000009a18b8] __schedule+0x398/0x9f0 
[  248.821352][  T676] [c000001ff310f7f0] [c0000000009a1fa8] schedule+0x98/0x160 
[  248.821387][  T676] [c000001ff310f820] [c0000000009a9814] schedule_timeout+0x304/0x520 
[  248.821432][  T676] [c000001ff310f960] [c0000000009a3c84] wait_for_completion+0xc4/0x1b0 
[  248.821460][  T676] [c000001ff310f9d0] [c0000000000fd0c8] __flush_work+0x3b8/0x770 
[  248.821491][  T676] [c000001ff310faf0] [c0000000002e0ac4] lru_add_drain_all+0x3e4/0x760 
[  248.821521][  T676] [c000001ff310fbf0] [c0000000003e0f18] khugepaged+0xd8/0x1770 
[  248.821560][  T676] [c000001ff310fdb0] [c0000000001095fc] kthread+0x1bc/0x1d0 
[  248.821611][  T676] [c000001ff310fe20] [c00000000000cbc4] ret_from_kernel_thread+0x5c/0x78 
[  248.821655][  T676] INFO: task kworker/56:1:719 blocked for more than 122 seconds. 
[  248.821689][  T676]       Tainted: G             L    5.8.0-rc5-next-20200716 #3 
[  248.821729][  T676] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. 
[  248.821779][  T676] kworker/56:1    D27584   719      2 0x00000800 
[  248.821839][  T676] Workqueue: rcu_gp wait_rcu_exp_gp 
[  248.821862][  T676] Call Trace: 
[  248.821888][  T676] [c000001ff2b57660] [c0000000057c9fb0] rcu_state+0x4fb0/0x5100 (unreliable) 
[  248.821934][  T676] [c000001ff2b57840] [c00000000001a2d0] __switch_to+0x260/0x380 
[  248.821977][  T676] [c000001ff2b578a0] [c0000000009a18b8] __schedule+0x398/0x9f0 
[  248.822021][  T676] [c000001ff2b57970] [c0000000009a1fa8] schedule+0x98/0x160 
[  248.822066][  T676] [c000001ff2b579a0] [c0000000009a970c] schedule_timeout+0x1fc/0x520 
[  248.822110][  T676] [c000001ff2b57ae0] [c0000000001a86d0] rcu_exp_wait_wake+0x1b0/0x950 
[  248.822153][  T676] [c000001ff2b57c30] [c0000000000fb754] process_one_work+0x304/0x900 
[  248.822197][  T676] [c000001ff2b57d20] [c0000000000fbdc8] worker_thread+0x78/0x520 
[  248.822242][  T676] [c000001ff2b57db0] [c0000000001095fc] kthread+0x1bc/0x1d0 
[  248.822279][  T676] [c000001ff2b57e20] [c00000000000cbc4] ret_from_kernel_thread+0x5c/0x78 
[  248.822385][  T676] INFO: task lvm:3123 blocked for more than 122 seconds. 
[  248.822413][  T676]       Tainted: G             L    5.8.0-rc5-next-20200716 #3 
[  248.822462][  T676] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. 
[  248.822503][  T676] lvm             D26608  3123      1 0x00040000 
[  248.822552][  T676] Call Trace: 
[  248.822576][  T676] [c000001feee27a00] [c00000000001a2d0] __switch_to+0x260/0x380 
[  248.822620][  T676] [c000001feee27a60] [c0000000009a18b8] __schedule+0x398/0x9f0 
[  248.822648][  T676] [c000001feee27b30] [c0000000009a1fa8] schedule+0x98/0x160 
[  248.822680][  T676] [c000001feee27b60] [c0000000009a9814] schedule_timeout+0x304/0x520 
[  248.822724][  T676] [c000001feee27ca0] [c0000000009a3c84] wait_for_completion+0xc4/0x1b0 
[  248.822768][  T676] [c000001feee27d10] [c0000000004b0e88] sys_io_destroy+0x238/0x2f0 
[  248.822808][  T676] [c000001feee27dc0] [c00000000002c8b8] system_call_exception+0xf8/0x180 
[  248.822840][  T676] [c000001feee27e20] [c00000000000c9e8] system_call_common+0xe8/0x214 
[  248.822873][  T676] INFO: task lvm:3126 blocked for more than 122 seconds. 
[  248.822901][  T676]       Tainted: G             L    5.8.0-rc5-next-20200716 #3 
[  248.822938][  T676] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. 
[  248.822987][  T676] lvm             D26608  3126      1 0x00040000 
[  248.823017][  T676] Call Trace: 
[  248.823031][  T676] [c000001fc0b27a00] [c00000000001a2d0] __switch_to+0x260/0x380 
[  248.823075][  T676] [c000001fc0b27a60] [c0000000009a18b8] __schedule+0x398/0x9f0 
[  248.823113][  T676] [c000001fc0b27b30] [c0000000009a1fa8] schedule+0x98/0x160 
[  248.823158][  T676] [c000001fc0b27b60] [c0000000009a9814] schedule_timeout+0x304/0x520 
[  248.823199][  T676] [c000001fc0b27ca0] [c0000000009a3c84] wait_for_completion+0xc4/0x1b0 
[  248.823250][  T676] [c000001fc0b27d10] [c0000000004b0e88] sys_io_destroy+0x238/0x2f0 
[  248.823294][  T676] [c000001fc0b27dc0] [c00000000002c8b8] system_call_exception+0xf8/0x180 
[  248.823332][  T676] [c000001fc0b27e20] [c00000000000c9e8] system_call_common+0xe8/0x214 
[  248.823374][  T676] INFO: task auditd:3163 blocked for more than 122 seconds. 
[  248.823424][  T676]       Tainted: G             L    5.8.0-rc5-next-20200716 #3 
[  248.823471][  T676] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. 
[  248.823512][  T676] auditd          D27088  3163      1 0x00042408 
[  248.823551][  T676] Call Trace: 
[  248.823583][  T676] [c000001fc080f760] [0000000200000001] 0x200000001 (unreliable) 
[  248.823648][  T676] [c000001fc080f940] [c00000000001a2d0] __switch_to+0x260/0x380 
[  248.823689][  T676] [c000001fc080f9a0] [c0000000009a18b8] __schedule+0x398/0x9f0 
[  248.823742][  T676] [c000001fc080fa70] [c0000000009a1fa8] schedule+0x98/0x160 
[  248.823784][  T676] [c000001fc080faa0] [c0000000001a9244] synchronize_rcu_expedited+0x394/0x600 
[  248.823837][  T676] [c000001fc080fba0] [c0000000004504c4] namespace_unlock+0xf4/0x230 
[  248.823881][  T676] [c000001fc080fc00] [c000000000456dec] put_mnt_ns+0x5c/0x80 
[  248.823926][  T676] [c000001fc080fc30] [c00000000010ba6c] free_nsproxy+0x2c/0x1e0 
[  248.823966][  T676] [c000001fc080fc60] [c0000000000d5130] do_exit+0x4e0/0xee0 
[  248.823997][  T676] [c000001fc080fd60] [c0000000000d5bec] do_group_exit+0x5c/0xd0 
[  248.824019][  T676] [c000001fc080fda0] [c0000000000d5c7c] sys_exit_group+0x1c/0x20 
[  248.824060][  T676] [c000001fc080fdc0] [c00000000002c8b8] system_call_exception+0xf8/0x180 
[  248.824103][  T676] [c000001fc080fe20] [c00000000000c9e8] system_call_common+0xe8/0x214 
[  248.824192][  T676]  
[  248.824192][  T676] Showing all locks held in the system: 
[  248.824419][  T676] 1 lock held by khungtaskd/676: 
[  248.824455][  T676]  #0: c0000000057c44c0 (rcu_read_lock){....}-{1:2}, at: rcu_lock_acquire.constprop.29+0x8/0x30 
[  248.824531][  T676] 1 lock held by khugepaged/682: 
[  248.824565][  T676]  #0: c0000000057f42c8 (lock#4){+.+.}-{3:3}, at: lru_add_drain_all+0x68/0x760 
[  248.824679][  T676] 2 locks held by kworker/56:1/719: 
[  248.824742][  T676]  #0: c00000000bcc8938 ((wq_completion)rcu_gp){+.+.}-{0:0}, at: process_one_work+0x21c/0x900 
[  248.824857][  T676]  #1: c000001ff2b57c90 ((work_completion)(&rew.rew_work)){+.+.}-{0:0}, at: process_one_work+0x21c/0x900 
[  248.825026][  T676] 3 locks held by (spawn)/2692: 
[  248.825077][  T676] 1 lock held by auditd/3163: 
[  248.825135][  T676]  #0: c0000000057c9ee8 (rcu_state.exp_mutex){+.+.}-{3:3}, at: synchronize_rcu_expedited+0x254/0x600 
[  248.825296][  T676] ============================================= 
[  248.825296][  T676]  
Quoted 15 lines
 .../include/asm/book3s/64/tlbflush-radix.h    | 15 ++++
 arch/powerpc/include/asm/hvcall.h             | 34 +++++++-
 arch/powerpc/include/asm/mmu.h                |  4 +
 arch/powerpc/include/asm/plpar_wrappers.h     | 52 ++++++++++++
 arch/powerpc/kernel/dt_cpu_ftrs.c             |  1 +
 arch/powerpc/kernel/prom_init.c               | 13 +--
 arch/powerpc/mm/book3s64/radix_tlb.c          | 82 +++++++++++++++++--
 arch/powerpc/mm/init_64.c                     |  5 +-
 arch/powerpc/platforms/pseries/lpar.c         |  8 +-
 9 files changed, 197 insertions(+), 17 deletions(-)
Signature · 1 line
2.21.3
#8 lore raw

Hi all,

Quoted 3 lines · On Thu, 16 Jul 2020 13:27:14 -0400 Qian Cai <[email protected]> wrote:

Reverting the whole series fixed random memory corruptions during boot on POWER9 PowerNV systems below.

I will revert those commits from linux-next today as well (they revert cleanly).

Signature · 2 lines
Cheers,
Stephen Rothwell
#9 lore raw
In reply to Qian Cai

Excerpts from Qian Cai's message of July 17, 2020 3:27 am:

Quoted 32 lines · Bharata B Rao
Quoted 31 lines

Hypervisor may choose not to enable Guest Translation Shootdown Enable (GTSE) option for the guest. When GTSE isn't ON, the guest OS isn't permitted to use instructions like tblie and tlbsync directly, but is expected to make hypervisor calls to get the TLB flushed.

This series enables the TLB flush routines in the radix code to off-load TLB flushing to hypervisor via the newly proposed hcall H_RPT_INVALIDATE.

To easily check the availability of GTSE, it is made an MMU feature. The OV5 handling and H_REGISTER_PROC_TBL hcall are changed to handle GTSE as an optionally available feature and to not assume GTSE when radix support is available.

The actual hcall implementation for KVM isn't included in this patchset and will be posted separately.

Changes in v3
=============
- Fixed a bug in the hcall wrapper code where we were missing setting H_RPTI_TYPE_NESTED while retrying the failed flush request with a full flush for the nested case.
- s/psize_to_h_rpti/psize_to_rpti_pgsize

v2: https://lore.kernel.org/linuxppc-dev/[email protected]/T/#t

Bharata B Rao (2):
powerpc/mm: Enable radix GTSE only if supported. powerpc/pseries: H_REGISTER_PROC_TBL should ask for GTSE only if
enabled

Quoted 6 lines
Quoted 3 lines

Nicholas Piggin (1):
powerpc/mm/book3s64/radix: Off-load TLB invalidations to host when
!GTSE

Reverting the whole series fixed random memory corruptions during boot on POWER9 PowerNV systems below.

If I s/mmu_has_feature(MMU_FTR_GTSE)/(1)/g in radix_tlb.c, then the .o disasm is the same as reverting my patch.

Feature bits not being set right? PowerNV should be pretty simple, seems to do the same as FTR_TYPE_RADIX.

So... test being done before static keys are set up? Shouldn't be. Must be something obvious I just can't see it.

Thanks,
Nick

#10 lore raw

Excerpts from Nicholas Piggin's message of July 17, 2020 12:08 pm:

Quoted 37 lines

Excerpts from Qian Cai's message of July 17, 2020 3:27 am:

Quoted 36 lines · Bharata B Rao
Quoted 34 lines

Hypervisor may choose not to enable Guest Translation Shootdown Enable (GTSE) option for the guest. When GTSE isn't ON, the guest OS isn't permitted to use instructions like tblie and tlbsync directly, but is expected to make hypervisor calls to get the TLB flushed.

This series enables the TLB flush routines in the radix code to off-load TLB flushing to hypervisor via the newly proposed hcall H_RPT_INVALIDATE.

To easily check the availability of GTSE, it is made an MMU feature. The OV5 handling and H_REGISTER_PROC_TBL hcall are changed to handle GTSE as an optionally available feature and to not assume GTSE when radix support is available.

The actual hcall implementation for KVM isn't included in this patchset and will be posted separately.

Changes in v3
=============
- Fixed a bug in the hcall wrapper code where we were missing setting H_RPTI_TYPE_NESTED while retrying the failed flush request with a full flush for the nested case.
- s/psize_to_h_rpti/psize_to_rpti_pgsize

v2: https://lore.kernel.org/linuxppc-dev/[email protected]/T/#t

Bharata B Rao (2):
powerpc/mm: Enable radix GTSE only if supported. powerpc/pseries: H_REGISTER_PROC_TBL should ask for GTSE only if
enabled

Nicholas Piggin (1):
powerpc/mm/book3s64/radix: Off-load TLB invalidations to host when
!GTSE

Quoted 8 lines
Quoted 2 lines

Reverting the whole series fixed random memory corruptions during boot on POWER9 PowerNV systems below.

If I s/mmu_has_feature(MMU_FTR_GTSE)/(1)/g in radix_tlb.c, then the .o disasm is the same as reverting my patch.

Feature bits not being set right? PowerNV should be pretty simple, seems to do the same as FTR_TYPE_RADIX.

Might need this fix

---

arch/powerpc/kernel/prom.c +1 -1
index 9cc49f265c86..54c9bcea9d4e 100644
@@ -163,7 +163,7 @@ static struct ibm_pa_feature {
 	{ .pabyte = 0,  .pabit = 6, .cpu_features  = CPU_FTR_NOEXECUTE },
 	{ .pabyte = 1,  .pabit = 2, .mmu_features  = MMU_FTR_CI_LARGE_PAGE },
 #ifdef CONFIG_PPC_RADIX_MMU
-	{ .pabyte = 40, .pabit = 0, .mmu_features  = MMU_FTR_TYPE_RADIX },
+	{ .pabyte = 40, .pabit = 0, .mmu_features  = (MMU_FTR_TYPE_RADIX | MMU_FTR_GTSE) },
 #endif
 	{ .pabyte = 1,  .pabit = 1, .invert = 1, .cpu_features = CPU_FTR_NODSISRALIGN },
 	{ .pabyte = 5,  .pabit = 0, .cpu_features  = CPU_FTR_REAL_LE,
#11 lore raw
Quoted 56 lines · Nicholas Piggin

Excerpts from Nicholas Piggin's message of July 17, 2020 12:08 pm:

Quoted 37 lines

Excerpts from Qian Cai's message of July 17, 2020 3:27 am:

Quoted 36 lines · Bharata B Rao
Quoted 34 lines

Hypervisor may choose not to enable Guest Translation Shootdown Enable (GTSE) option for the guest. When GTSE isn't ON, the guest OS isn't permitted to use instructions like tblie and tlbsync directly, but is expected to make hypervisor calls to get the TLB flushed.

This series enables the TLB flush routines in the radix code to off-load TLB flushing to hypervisor via the newly proposed hcall H_RPT_INVALIDATE.

To easily check the availability of GTSE, it is made an MMU feature. The OV5 handling and H_REGISTER_PROC_TBL hcall are changed to handle GTSE as an optionally available feature and to not assume GTSE when radix support is available.

The actual hcall implementation for KVM isn't included in this patchset and will be posted separately.

Changes in v3
=============
- Fixed a bug in the hcall wrapper code where we were missing setting H_RPTI_TYPE_NESTED while retrying the failed flush request with a full flush for the nested case.
- s/psize_to_h_rpti/psize_to_rpti_pgsize

v2: https://lore.kernel.org/linuxppc-dev/[email protected]/T/#t

Bharata B Rao (2):
powerpc/mm: Enable radix GTSE only if supported. powerpc/pseries: H_REGISTER_PROC_TBL should ask for GTSE only if
enabled

Nicholas Piggin (1):
powerpc/mm/book3s64/radix: Off-load TLB invalidations to host when
!GTSE

Quoted 8 lines
Quoted 2 lines

Reverting the whole series fixed random memory corruptions during boot on POWER9 PowerNV systems below.

If I s/mmu_has_feature(MMU_FTR_GTSE)/(1)/g in radix_tlb.c, then the .o disasm is the same as reverting my patch.

Feature bits not being set right? PowerNV should be pretty simple, seems to do the same as FTR_TYPE_RADIX.

Might need this fix

---

arch/powerpc/kernel/prom.c +0 -0
index 9cc49f265c86..54c9bcea9d4e 100644

arch/powerpc/kernel/prom.c static struct ibm_pa_feature { · line 169

Quoted 8 lines
arch/powerpc/kernel/prom.c +1 -1
@@ -163,7 +163,7 @@ static struct ibm_pa_feature {
 	{ .pabyte = 0,  .pabit = 6, .cpu_features  = CPU_FTR_NOEXECUTE },
 	{ .pabyte = 1,  .pabit = 2, .mmu_features  = MMU_FTR_CI_LARGE_PAGE },
 #ifdef CONFIG_PPC_RADIX_MMU
-	{ .pabyte = 40, .pabit = 0, .mmu_features  = MMU_FTR_TYPE_RADIX },
+	{ .pabyte = 40, .pabit = 0, .mmu_features  = (MMU_FTR_TYPE_RADIX | MMU_FTR_GTSE) },
 #endif
 	{ .pabyte = 1,  .pabit = 1, .invert = 1, .cpu_features = CPU_FTR_NODSISRALIGN },
 	{ .pabyte = 5,  .pabit = 0, .cpu_features  = CPU_FTR_REAL_LE,

Michael - Let me know if this should be folded into 1/3 and the complete series resent.

Regards,
Bharata.

#12 lore raw
Quoted 52 lines · Bharata B Rao
Quoted 51 lines · Nicholas Piggin

Excerpts from Nicholas Piggin's message of July 17, 2020 12:08 pm:

Quoted 37 lines

Excerpts from Qian Cai's message of July 17, 2020 3:27 am:

Quoted 36 lines · Bharata B Rao
Quoted 34 lines

Hypervisor may choose not to enable Guest Translation Shootdown Enable (GTSE) option for the guest. When GTSE isn't ON, the guest OS isn't permitted to use instructions like tblie and tlbsync directly, but is expected to make hypervisor calls to get the TLB flushed.

This series enables the TLB flush routines in the radix code to off-load TLB flushing to hypervisor via the newly proposed hcall H_RPT_INVALIDATE.

To easily check the availability of GTSE, it is made an MMU feature. The OV5 handling and H_REGISTER_PROC_TBL hcall are changed to handle GTSE as an optionally available feature and to not assume GTSE when radix support is available.

The actual hcall implementation for KVM isn't included in this patchset and will be posted separately.

Changes in v3
=============
- Fixed a bug in the hcall wrapper code where we were missing setting H_RPTI_TYPE_NESTED while retrying the failed flush request with a full flush for the nested case.
- s/psize_to_h_rpti/psize_to_rpti_pgsize

v2: https://lore.kernel.org/linuxppc-dev/[email protected]/T/#t

Bharata B Rao (2):
powerpc/mm: Enable radix GTSE only if supported. powerpc/pseries: H_REGISTER_PROC_TBL should ask for GTSE only if
enabled

Nicholas Piggin (1):
powerpc/mm/book3s64/radix: Off-load TLB invalidations to host when
!GTSE

Quoted 8 lines
Quoted 2 lines

Reverting the whole series fixed random memory corruptions during boot on POWER9 PowerNV systems below.

If I s/mmu_has_feature(MMU_FTR_GTSE)/(1)/g in radix_tlb.c, then the .o disasm is the same as reverting my patch.

Feature bits not being set right? PowerNV should be pretty simple, seems to do the same as FTR_TYPE_RADIX.

Might need this fix

---

Quoted 16 lines

arch/powerpc/kernel/prom.c static struct ibm_pa_feature { · line 169

Quoted 13 lines
arch/powerpc/kernel/prom.c +1 -1
index 9cc49f265c86..54c9bcea9d4e 100644
@@ -163,7 +163,7 @@ static struct ibm_pa_feature {
 	{ .pabyte = 0,  .pabit = 6, .cpu_features  = CPU_FTR_NOEXECUTE },
 	{ .pabyte = 1,  .pabit = 2, .mmu_features  = MMU_FTR_CI_LARGE_PAGE },
 #ifdef CONFIG_PPC_RADIX_MMU
-	{ .pabyte = 40, .pabit = 0, .mmu_features  = MMU_FTR_TYPE_RADIX },
+	{ .pabyte = 40, .pabit = 0, .mmu_features  = (MMU_FTR_TYPE_RADIX | MMU_FTR_GTSE) },
 #endif
 	{ .pabyte = 1,  .pabit = 1, .invert = 1, .cpu_features = CPU_FTR_NODSISRALIGN },
 	{ .pabyte = 5,  .pabit = 0, .cpu_features  = CPU_FTR_REAL_LE,

Michael - Let me know if this should be folded into 1/3 and the complete series resent.

No it's already merged.

Please send a proper patch with a Fixes: tag and a change log that explains the details.

cheers