powerpc/mm: don't do tlbie for updatepp request with NO HPTE fault

upatepp can get called for a nohpte fault when we find from the linux
page table that the translation was hashed before. In that case
we are sure that there is no existing translation, hence we could
avoid doing tlbie.

We could possibly race with a parallel fault filling the TLB. But
that should be ok because updatepp is only ever relaxing permissions.
We also look at linux pte permission bits when filling hash pte
permission bits. We also hold the linux pte busy bits while
inserting/updating a hashpte entry, hence a paralle update of
linux pte is not possible. On the other hand mprotect involves
ptep_modify_prot_start which cause a hpte invalidate and not updatepp.

Performance number:
We use randbox_access_bench written by Anton.

Kernel with THP disabled and smaller hash page table size.

    86.60%  random_access_b  [kernel.kallsyms]                [k] .native_hpte_updatepp
     2.10%  random_access_b  random_access_bench              [.] doit
     1.99%  random_access_b  [kernel.kallsyms]                [k] .do_raw_spin_lock
     1.85%  random_access_b  [kernel.kallsyms]                [k] .native_hpte_insert
     1.26%  random_access_b  [kernel.kallsyms]                [k] .native_flush_hash_range
     1.18%  random_access_b  [kernel.kallsyms]                [k] .__delay
     0.69%  random_access_b  [kernel.kallsyms]                [k] .native_hpte_remove
     0.37%  random_access_b  [kernel.kallsyms]                [k] .clear_user_page
     0.34%  random_access_b  [kernel.kallsyms]                [k] .__hash_page_64K
     0.32%  random_access_b  [kernel.kallsyms]                [k] fast_exception_return
     0.30%  random_access_b  [kernel.kallsyms]                [k] .hash_page_mm

With Fix:

    27.54%  random_access_b  random_access_bench              [.] doit
    22.90%  random_access_b  [kernel.kallsyms]                [k] .native_hpte_insert
     5.76%  random_access_b  [kernel.kallsyms]                [k] .native_hpte_remove
     5.20%  random_access_b  [kernel.kallsyms]                [k] fast_exception_return
     5.12%  random_access_b  [kernel.kallsyms]                [k] .__hash_page_64K
     4.80%  random_access_b  [kernel.kallsyms]                [k] .hash_page_mm
     3.31%  random_access_b  [kernel.kallsyms]                [k] data_access_common
     1.84%  random_access_b  [kernel.kallsyms]                [k] .trace_hardirqs_on_caller

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

6 files changed, 58 insertions, 37 deletions

diff --git a/arch/powerpc/mm/hash_low_64.S b/arch/powerpc/mm/hash_low_64.S
index 5094f32b706e..463174a4a647 100644
--- a/arch/powerpc/mm/hash_low_64.S
+++ b/arch/powerpc/mm/hash_low_64.S
@@ -46,7 +46,8 @@
 
 /*
  * _hash_page_4K(unsigned long ea, unsigned long access, unsigned long vsid,
- *		 pte_t *ptep, unsigned long trap, int local, int ssize)
+ *		 pte_t *ptep, unsigned long trap, unsigned long flags,
+ *		 int ssize)
  *
  * Adds a 4K page to the hash table in a segment of 4K pages only
  */
@@ -298,7 +299,7 @@ htab_modify_pte:
 	li	r6,MMU_PAGE_4K		/* base page size */
 	li	r7,MMU_PAGE_4K		/* actual page size */
 	ld	r8,STK_PARAM(R9)(r1)	/* segment size */
-	ld	r9,STK_PARAM(R8)(r1)	/* get "local" param */
+	ld	r9,STK_PARAM(R8)(r1)	/* get "flags" param */
 .globl htab_call_hpte_updatepp
 htab_call_hpte_updatepp:
 	bl	.			/* Patched by htab_finish_init() */
@@ -338,8 +339,8 @@ htab_pte_insert_failure:
  *****************************************************************************/
 
 /* _hash_page_4K(unsigned long ea, unsigned long access, unsigned long vsid,
- *		 pte_t *ptep, unsigned long trap, int local, int ssize,
- *		 int subpg_prot)
+ *		 pte_t *ptep, unsigned long trap, unsigned local flags,
+ *		 int ssize, int subpg_prot)
  */
 
 /*
@@ -594,7 +595,7 @@ htab_inval_old_hpte:
 	li	r5,0			/* PTE.hidx */
 	li	r6,MMU_PAGE_64K		/* psize */
 	ld	r7,STK_PARAM(R9)(r1)	/* ssize */
-	ld	r8,STK_PARAM(R8)(r1)	/* local */
+	ld	r8,STK_PARAM(R8)(r1)	/* flags */
 	bl	flush_hash_page
 	/* Clear out _PAGE_HPTE_SUB bits in the new linux PTE */
 	lis	r0,_PAGE_HPTE_SUB@h
@@ -666,7 +667,7 @@ htab_modify_pte:
 	li	r6,MMU_PAGE_4K		/* base page size */
 	li	r7,MMU_PAGE_4K		/* actual page size */
 	ld	r8,STK_PARAM(R9)(r1)	/* segment size */
-	ld	r9,STK_PARAM(R8)(r1)	/* get "local" param */
+	ld	r9,STK_PARAM(R8)(r1)	/* get "flags" param */
 .globl htab_call_hpte_updatepp
 htab_call_hpte_updatepp:
 	bl	.			/* patched by htab_finish_init() */
@@ -962,7 +963,7 @@ ht64_modify_pte:
 	li	r6,MMU_PAGE_64K		/* base page size */
 	li	r7,MMU_PAGE_64K		/* actual page size */
 	ld	r8,STK_PARAM(R9)(r1)	/* segment size */
-	ld	r9,STK_PARAM(R8)(r1)	/* get "local" param */
+	ld	r9,STK_PARAM(R8)(r1)	/* get "flags" param */
 .globl ht64_call_hpte_updatepp
 ht64_call_hpte_updatepp:
 	bl	.			/* patched by htab_finish_init() */
diff --git a/arch/powerpc/mm/hash_native_64.c b/arch/powerpc/mm/hash_native_64.c
index 13700911b522..9c4880ddecd6 100644
--- a/arch/powerpc/mm/hash_native_64.c
+++ b/arch/powerpc/mm/hash_native_64.c
@@ -283,11 +283,11 @@ static long native_hpte_remove(unsigned long hpte_group)
 
 static long native_hpte_updatepp(unsigned long slot, unsigned long newpp,
 				 unsigned long vpn, int bpsize,
-				 int apsize, int ssize, int local)
+				 int apsize, int ssize, unsigned long flags)
 {
 	struct hash_pte *hptep = htab_address + slot;
 	unsigned long hpte_v, want_v;
-	int ret = 0;
+	int ret = 0, local = 0;
 
 	want_v = hpte_encode_avpn(vpn, bpsize, ssize);
 
@@ -322,8 +322,15 @@ static long native_hpte_updatepp(unsigned long slot, unsigned long newpp,
 		}
 		native_unlock_hpte(hptep);
 	}
-	/* Ensure it is out of the tlb too. */
-	tlbie(vpn, bpsize, apsize, ssize, local);
+
+	if (flags & HPTE_LOCAL_UPDATE)
+		local = 1;
+	/*
+	 * Ensure it is out of the tlb too if it is not a nohpte fault
+	 */
+	if (!(flags & HPTE_NOHPTE_UPDATE))
+		tlbie(vpn, bpsize, apsize, ssize, local);
+
 	return ret;
 }
 
diff --git a/arch/powerpc/mm/hash_utils_64.c b/arch/powerpc/mm/hash_utils_64.c
index 68211d398fdb..e56a307bc676 100644
--- a/arch/powerpc/mm/hash_utils_64.c
+++ b/arch/powerpc/mm/hash_utils_64.c
@@ -989,7 +989,9 @@ static void check_paca_psize(unsigned long ea, struct mm_struct *mm,
  * -1 - critical hash insertion error
  * -2 - access not permitted by subpage protection mechanism
  */
-int hash_page_mm(struct mm_struct *mm, unsigned long ea, unsigned long access, unsigned long trap)
+int hash_page_mm(struct mm_struct *mm, unsigned long ea,
+		 unsigned long access, unsigned long trap,
+		 unsigned long flags)
 {
 	enum ctx_state prev_state = exception_enter();
 	pgd_t *pgdir;
@@ -997,7 +999,7 @@ int hash_page_mm(struct mm_struct *mm, unsigned long ea, unsigned long access, u
 	pte_t *ptep;
 	unsigned hugeshift;
 	const struct cpumask *tmp;
-	int rc, user_region = 0, local = 0;
+	int rc, user_region = 0;
 	int psize, ssize;
 
 	DBG_LOW("hash_page(ea=%016lx, access=%lx, trap=%lx\n",
@@ -1049,7 +1051,7 @@ int hash_page_mm(struct mm_struct *mm, unsigned long ea, unsigned long access, u
 	/* Check CPU locality */
 	tmp = cpumask_of(smp_processor_id());
 	if (user_region && cpumask_equal(mm_cpumask(mm), tmp))
-		local = 1;
+		flags |= HPTE_LOCAL_UPDATE;
 
 #ifndef CONFIG_PPC_64K_PAGES
 	/* If we use 4K pages and our psize is not 4K, then we might
@@ -1086,11 +1088,11 @@ int hash_page_mm(struct mm_struct *mm, unsigned long ea, unsigned long access, u
 	if (hugeshift) {
 		if (pmd_trans_huge(*(pmd_t *)ptep))
 			rc = __hash_page_thp(ea, access, vsid, (pmd_t *)ptep,
-					     trap, local, ssize, psize);
+					     trap, flags, ssize, psize);
 #ifdef CONFIG_HUGETLB_PAGE
 		else
 			rc = __hash_page_huge(ea, access, vsid, ptep, trap,
-					      local, ssize, hugeshift, psize);
+					      flags, ssize, hugeshift, psize);
 #else
 		else {
 			/*
@@ -1149,7 +1151,8 @@ int hash_page_mm(struct mm_struct *mm, unsigned long ea, unsigned long access, u
 
 #ifdef CONFIG_PPC_HAS_HASH_64K
 	if (psize == MMU_PAGE_64K)
-		rc = __hash_page_64K(ea, access, vsid, ptep, trap, local, ssize);
+		rc = __hash_page_64K(ea, access, vsid, ptep, trap,
+				     flags, ssize);
 	else
 #endif /* CONFIG_PPC_HAS_HASH_64K */
 	{
@@ -1158,7 +1161,7 @@ int hash_page_mm(struct mm_struct *mm, unsigned long ea, unsigned long access, u
 			rc = -2;
 		else
 			rc = __hash_page_4K(ea, access, vsid, ptep, trap,
-					    local, ssize, spp);
+					    flags, ssize, spp);
 	}
 
 	/* Dump some info in case of hash insertion failure, they should
@@ -1181,14 +1184,19 @@ bail:
 }
 EXPORT_SYMBOL_GPL(hash_page_mm);
 
-int hash_page(unsigned long ea, unsigned long access, unsigned long trap)
+int hash_page(unsigned long ea, unsigned long access, unsigned long trap,
+	      unsigned long dsisr)
 {
+	unsigned long flags = 0;
 	struct mm_struct *mm = current->mm;
 
 	if (REGION_ID(ea) == VMALLOC_REGION_ID)
 		mm = &init_mm;
 
-	return hash_page_mm(mm, ea, access, trap);
+	if (dsisr & DSISR_NOHPTE)
+		flags |= HPTE_NOHPTE_UPDATE;
+
+	return hash_page_mm(mm, ea, access, trap, flags);
 }
 EXPORT_SYMBOL_GPL(hash_page);
 
@@ -1200,7 +1208,7 @@ void hash_preload(struct mm_struct *mm, unsigned long ea,
 	pgd_t *pgdir;
 	pte_t *ptep;
 	unsigned long flags;
-	int rc, ssize, local = 0;
+	int rc, ssize, update_flags = 0;
 
 	BUG_ON(REGION_ID(ea) != USER_REGION_ID);
 
@@ -1251,16 +1259,17 @@ void hash_preload(struct mm_struct *mm, unsigned long ea,
 
 	/* Is that local to this CPU ? */
 	if (cpumask_equal(mm_cpumask(mm), cpumask_of(smp_processor_id())))
-		local = 1;
+		update_flags |= HPTE_LOCAL_UPDATE;
 
 	/* Hash it in */
 #ifdef CONFIG_PPC_HAS_HASH_64K
 	if (mm->context.user_psize == MMU_PAGE_64K)
-		rc = __hash_page_64K(ea, access, vsid, ptep, trap, local, ssize);
+		rc = __hash_page_64K(ea, access, vsid, ptep, trap,
+				     update_flags, ssize);
 	else
 #endif /* CONFIG_PPC_HAS_HASH_64K */
-		rc = __hash_page_4K(ea, access, vsid, ptep, trap, local, ssize,
-				    subpage_protection(mm, ea));
+		rc = __hash_page_4K(ea, access, vsid, ptep, trap, update_flags,
+				    ssize, subpage_protection(mm, ea));
 
 	/* Dump some info in case of hash insertion failure, they should
 	 * never happen so it is really useful to know if/when they do
@@ -1278,9 +1287,10 @@ out_exit:
  *          do not forget to update the assembly call site !
  */
 void flush_hash_page(unsigned long vpn, real_pte_t pte, int psize, int ssize,
-		     int local)
+		     unsigned long flags)
 {
 	unsigned long hash, index, shift, hidx, slot;
+	int local = flags & HPTE_LOCAL_UPDATE;
 
 	DBG_LOW("flush_hash_page(vpn=%016lx)\n", vpn);
 	pte_iterate_hashed_subpages(pte, psize, vpn, index, shift) {
@@ -1317,12 +1327,14 @@ void flush_hash_page(unsigned long vpn, real_pte_t pte, int psize, int ssize,
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
 void flush_hash_hugepage(unsigned long vsid, unsigned long addr,
-			 pmd_t *pmdp, unsigned int psize, int ssize, int local)
+			 pmd_t *pmdp, unsigned int psize, int ssize,
+			 unsigned long flags)
 {
 	int i, max_hpte_count, valid;
 	unsigned long s_addr;
 	unsigned char *hpte_slot_array;
 	unsigned long hidx, shift, vpn, hash, slot;
+	int local = flags & HPTE_LOCAL_UPDATE;
 
 	s_addr = addr & HPAGE_PMD_MASK;
 	hpte_slot_array = get_hpte_slot_array(pmdp);
diff --git a/arch/powerpc/mm/hugepage-hash64.c b/arch/powerpc/mm/hugepage-hash64.c
index 3a648cd363ae..86686514ae13 100644
--- a/arch/powerpc/mm/hugepage-hash64.c
+++ b/arch/powerpc/mm/hugepage-hash64.c
@@ -19,8 +19,8 @@
 #include <asm/machdep.h>
 
 int __hash_page_thp(unsigned long ea, unsigned long access, unsigned long vsid,
-		    pmd_t *pmdp, unsigned long trap, int local, int ssize,
-		    unsigned int psize)
+		    pmd_t *pmdp, unsigned long trap, unsigned long flags,
+		    int ssize, unsigned int psize)
 {
 	unsigned int index, valid;
 	unsigned char *hpte_slot_array;
@@ -95,7 +95,7 @@ int __hash_page_thp(unsigned long ea, unsigned long access, unsigned long vsid,
 		 */
 		if ((old_pmd & _PAGE_HASHPTE) && !(old_pmd & _PAGE_COMBO))
 			flush_hash_hugepage(vsid, ea, pmdp, MMU_PAGE_64K,
-					    ssize, local);
+					    ssize, flags);
 	}
 
 	valid = hpte_valid(hpte_slot_array, index);
@@ -108,7 +108,7 @@ int __hash_page_thp(unsigned long ea, unsigned long access, unsigned long vsid,
 		slot += hidx & _PTEIDX_GROUP_IX;
 
 		ret = ppc_md.hpte_updatepp(slot, rflags, vpn,
-					   psize, lpsize, ssize, local);
+					   psize, lpsize, ssize, flags);
 		/*
 		 * We failed to update, try to insert a new entry.
 		 */
diff --git a/arch/powerpc/mm/hugetlbpage-hash64.c b/arch/powerpc/mm/hugetlbpage-hash64.c
index a5bcf9301196..d94b1af53a93 100644
--- a/arch/powerpc/mm/hugetlbpage-hash64.c
+++ b/arch/powerpc/mm/hugetlbpage-hash64.c
@@ -19,8 +19,8 @@ extern long hpte_insert_repeating(unsigned long hash, unsigned long vpn,
 				  unsigned long vflags, int psize, int ssize);
 
 int __hash_page_huge(unsigned long ea, unsigned long access, unsigned long vsid,
-		     pte_t *ptep, unsigned long trap, int local, int ssize,
-		     unsigned int shift, unsigned int mmu_psize)
+		     pte_t *ptep, unsigned long trap, unsigned long flags,
+		     int ssize, unsigned int shift, unsigned int mmu_psize)
 {
 	unsigned long vpn;
 	unsigned long old_pte, new_pte;
@@ -81,7 +81,7 @@ int __hash_page_huge(unsigned long ea, unsigned long access, unsigned long vsid,
 		slot += (old_pte & _PAGE_F_GIX) >> 12;
 
 		if (ppc_md.hpte_updatepp(slot, rflags, vpn, mmu_psize,
-					 mmu_psize, ssize, local) == -1)
+					 mmu_psize, ssize, flags) == -1)
 			old_pte &= ~_PAGE_HPTEFLAGS;
 	}
 
diff --git a/arch/powerpc/mm/pgtable_64.c b/arch/powerpc/mm/pgtable_64.c
index eea9fa1f8ae7..4fe5f64cc179 100644
--- a/arch/powerpc/mm/pgtable_64.c
+++ b/arch/powerpc/mm/pgtable_64.c
@@ -739,9 +739,10 @@ void pmdp_invalidate(struct vm_area_struct *vma, unsigned long address,
 void hpte_do_hugepage_flush(struct mm_struct *mm, unsigned long addr,
 			    pmd_t *pmdp, unsigned long old_pmd)
 {
-	int ssize, local = 0;
+	int ssize;
 	unsigned int psize;
 	unsigned long vsid;
+	unsigned long flags = 0;
 	const struct cpumask *tmp;
 
 	/* get the base page size,vsid and segment size */
@@ -765,9 +766,9 @@ void hpte_do_hugepage_flush(struct mm_struct *mm, unsigned long addr,
 
 	tmp = cpumask_of(smp_processor_id());
 	if (cpumask_equal(mm_cpumask(mm), tmp))
-		local = 1;
+		flags |= HPTE_LOCAL_UPDATE;
 
-	return flush_hash_hugepage(vsid, addr, pmdp, psize, ssize, local);
+	return flush_hash_hugepage(vsid, addr, pmdp, psize, ssize, flags);
 }
 
 static pmd_t pmd_set_protbits(pmd_t pmd, pgprot_t pgprot)