parisc: Optimize per-pagetable spinlocks

On parisc a spinlock is stored in the next page behind the pgd which
protects against parallel accesses to the pgd. That's why one additional
page (PGD_ALLOC_ORDER) is allocated for the pgd.

Matthew Wilcox suggested that we instead should use a pointer in the
struct page table for this spinlock and noted, that the comments for the
PGD_ORDER and PMD_ORDER defines were wrong.

Both suggestions are addressed with this patch. Instead of having an own
spinlock to protect the pgd, we now switch to use the existing
page_table_lock.  Additionally, beside loading the pgd into cr25 in
switch_mm_irqs_off(), the physical address of this lock is loaded into
cr28 (tr4), so that we can avoid implementing a complicated lookup in
assembly for this lock in the TLB fault handlers.

The existing Hybrid L2/L3 page table scheme (where the pmd is adjacent
to the pgd) has been dropped with this patch.

Remove the locking in set_pte() and the huge-page pte functions too.
They trigger a spinlock recursion on 32bit machines and seem unnecessary.

Suggested-by: Matthew Wilcox <willy@infradead.org>
Fixes: b37d1c1898b2 ("parisc: Use per-pagetable spinlock")
Signed-off-by: John David Anglin <dave.anglin@bell.net>
Signed-off-by: Helge Deller <deller@gmx.de>

2 files changed, 51 insertions, 66 deletions

diff --git a/arch/parisc/kernel/asm-offsets.c b/arch/parisc/kernel/asm-offsets.c
index 305768a40773..cd2cc1b1648c 100644
--- a/arch/parisc/kernel/asm-offsets.c
+++ b/arch/parisc/kernel/asm-offsets.c
@@ -268,7 +268,6 @@ int main(void)
 	DEFINE(ASM_BITS_PER_PGD, BITS_PER_PGD);
 	DEFINE(ASM_BITS_PER_PMD, BITS_PER_PMD);
 	DEFINE(ASM_BITS_PER_PTE, BITS_PER_PTE);
-	DEFINE(ASM_PGD_PMD_OFFSET, -(PAGE_SIZE << PGD_ORDER));
 	DEFINE(ASM_PMD_ENTRY, ((PAGE_OFFSET & PMD_MASK) >> PMD_SHIFT));
 	DEFINE(ASM_PGD_ENTRY, PAGE_OFFSET >> PGDIR_SHIFT);
 	DEFINE(ASM_PGD_ENTRY_SIZE, PGD_ENTRY_SIZE);
diff --git a/arch/parisc/kernel/entry.S b/arch/parisc/kernel/entry.S
index 4d37cc9cba37..9f939afe6b88 100644
--- a/arch/parisc/kernel/entry.S
+++ b/arch/parisc/kernel/entry.S
@@ -35,10 +35,9 @@
 	.level 2.0
 #endif
 
-	.import		pa_tlb_lock,data
-	.macro  load_pa_tlb_lock reg
-	mfctl		%cr25,\reg
-	addil		L%(PAGE_SIZE << (PGD_ALLOC_ORDER - 1)),\reg
+	/* Get aligned page_table_lock address for this mm from cr28/tr4 */
+	.macro  get_ptl reg
+	mfctl	%cr28,\reg
 	.endm
 
 	/* space_to_prot macro creates a prot id from a space id */
@@ -407,7 +406,9 @@
 # endif
 #endif
 	dep             %r0,31,PAGE_SHIFT,\pmd  /* clear offset */
+#if CONFIG_PGTABLE_LEVELS < 3
 	copy		%r0,\pte
+#endif
 	ldw,s		\index(\pmd),\pmd
 	bb,>=,n		\pmd,_PxD_PRESENT_BIT,\fault
 	dep		%r0,31,PxD_FLAG_SHIFT,\pmd /* clear flags */
@@ -417,38 +418,23 @@
 	shladd		\index,BITS_PER_PTE_ENTRY,\pmd,\pmd /* pmd is now pte */
 	.endm
 
-	/* Look up PTE in a 3-Level scheme.
-	 *
-	 * Here we implement a Hybrid L2/L3 scheme: we allocate the
-	 * first pmd adjacent to the pgd.  This means that we can
-	 * subtract a constant offset to get to it.  The pmd and pgd
-	 * sizes are arranged so that a single pmd covers 4GB (giving
-	 * a full LP64 process access to 8TB) so our lookups are
-	 * effectively L2 for the first 4GB of the kernel (i.e. for
-	 * all ILP32 processes and all the kernel for machines with
-	 * under 4GB of memory) */
+	/* Look up PTE in a 3-Level scheme. */
 	.macro		L3_ptep pgd,pte,index,va,fault
-#if CONFIG_PGTABLE_LEVELS == 3 /* we might have a 2-Level scheme, e.g. with 16kb page size */
+#if CONFIG_PGTABLE_LEVELS == 3
+	copy		%r0,\pte
 	extrd,u		\va,63-ASM_PGDIR_SHIFT,ASM_BITS_PER_PGD,\index
-	extrd,u,*=	\va,63-ASM_PGDIR_SHIFT,64-ASM_PGDIR_SHIFT,%r0
 	ldw,s		\index(\pgd),\pgd
-	extrd,u,*=	\va,63-ASM_PGDIR_SHIFT,64-ASM_PGDIR_SHIFT,%r0
 	bb,>=,n		\pgd,_PxD_PRESENT_BIT,\fault
-	extrd,u,*=	\va,63-ASM_PGDIR_SHIFT,64-ASM_PGDIR_SHIFT,%r0
-	shld		\pgd,PxD_VALUE_SHIFT,\index
-	extrd,u,*=	\va,63-ASM_PGDIR_SHIFT,64-ASM_PGDIR_SHIFT,%r0
-	copy		\index,\pgd
-	extrd,u,*<>	\va,63-ASM_PGDIR_SHIFT,64-ASM_PGDIR_SHIFT,%r0
-	ldo		ASM_PGD_PMD_OFFSET(\pgd),\pgd
+	shld		\pgd,PxD_VALUE_SHIFT,\pgd
 #endif
 	L2_ptep		\pgd,\pte,\index,\va,\fault
 	.endm
 
-	/* Acquire pa_tlb_lock lock and check page is present. */
-	.macro		tlb_lock	spc,ptp,pte,tmp,tmp1,fault
-#ifdef CONFIG_SMP
+	/* Acquire page_table_lock and check page is present. */
+	.macro		ptl_lock	spc,ptp,pte,tmp,tmp1,fault
+#ifdef CONFIG_TLB_PTLOCK
 98:	cmpib,COND(=),n	0,\spc,2f
-	load_pa_tlb_lock \tmp
+	get_ptl		\tmp
 1:	LDCW		0(\tmp),\tmp1
 	cmpib,COND(=)	0,\tmp1,1b
 	nop
@@ -463,26 +449,26 @@
 3:
 	.endm
 
-	/* Release pa_tlb_lock lock without reloading lock address.
+	/* Release page_table_lock without reloading lock address.
 	   Note that the values in the register spc are limited to
 	   NR_SPACE_IDS (262144). Thus, the stw instruction always
 	   stores a nonzero value even when register spc is 64 bits.
 	   We use an ordered store to ensure all prior accesses are
 	   performed prior to releasing the lock. */
-	.macro		tlb_unlock0	spc,tmp
-#ifdef CONFIG_SMP
+	.macro		ptl_unlock0	spc,tmp
+#ifdef CONFIG_TLB_PTLOCK
 98:	or,COND(=)	%r0,\spc,%r0
 	stw,ma		\spc,0(\tmp)
 99:	ALTERNATIVE(98b, 99b, ALT_COND_NO_SMP, INSN_NOP)
 #endif
 	.endm
 
-	/* Release pa_tlb_lock lock. */
-	.macro		tlb_unlock1	spc,tmp
-#ifdef CONFIG_SMP
-98:	load_pa_tlb_lock \tmp
+	/* Release page_table_lock. */
+	.macro		ptl_unlock1	spc,tmp
+#ifdef CONFIG_TLB_PTLOCK
+98:	get_ptl		\tmp
+	ptl_unlock0	\spc,\tmp
 99:	ALTERNATIVE(98b, 99b, ALT_COND_NO_SMP, INSN_NOP)
-	tlb_unlock0	\spc,\tmp
 #endif
 	.endm
 
@@ -1165,14 +1151,14 @@ dtlb_miss_20w:
 
 	L3_ptep		ptp,pte,t0,va,dtlb_check_alias_20w
 
-	tlb_lock	spc,ptp,pte,t0,t1,dtlb_check_alias_20w
+	ptl_lock	spc,ptp,pte,t0,t1,dtlb_check_alias_20w
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot,t1
 	
 	idtlbt          pte,prot
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1191,14 +1177,14 @@ nadtlb_miss_20w:
 
 	L3_ptep		ptp,pte,t0,va,nadtlb_check_alias_20w
 
-	tlb_lock	spc,ptp,pte,t0,t1,nadtlb_check_alias_20w
+	ptl_lock	spc,ptp,pte,t0,t1,nadtlb_check_alias_20w
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot,t1
 
 	idtlbt          pte,prot
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1219,7 +1205,7 @@ dtlb_miss_11:
 
 	L2_ptep		ptp,pte,t0,va,dtlb_check_alias_11
 
-	tlb_lock	spc,ptp,pte,t0,t1,dtlb_check_alias_11
+	ptl_lock	spc,ptp,pte,t0,t1,dtlb_check_alias_11
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb_11	spc,pte,prot
@@ -1232,7 +1218,7 @@ dtlb_miss_11:
 
 	mtsp		t1, %sr1	/* Restore sr1 */
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1252,7 +1238,7 @@ nadtlb_miss_11:
 
 	L2_ptep		ptp,pte,t0,va,nadtlb_check_alias_11
 
-	tlb_lock	spc,ptp,pte,t0,t1,nadtlb_check_alias_11
+	ptl_lock	spc,ptp,pte,t0,t1,nadtlb_check_alias_11
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb_11	spc,pte,prot
@@ -1265,7 +1251,7 @@ nadtlb_miss_11:
 
 	mtsp		t1, %sr1	/* Restore sr1 */
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1285,7 +1271,7 @@ dtlb_miss_20:
 
 	L2_ptep		ptp,pte,t0,va,dtlb_check_alias_20
 
-	tlb_lock	spc,ptp,pte,t0,t1,dtlb_check_alias_20
+	ptl_lock	spc,ptp,pte,t0,t1,dtlb_check_alias_20
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot,t1
@@ -1294,7 +1280,7 @@ dtlb_miss_20:
 
 	idtlbt          pte,prot
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1313,7 +1299,7 @@ nadtlb_miss_20:
 
 	L2_ptep		ptp,pte,t0,va,nadtlb_check_alias_20
 
-	tlb_lock	spc,ptp,pte,t0,t1,nadtlb_check_alias_20
+	ptl_lock	spc,ptp,pte,t0,t1,nadtlb_check_alias_20
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot,t1
@@ -1322,7 +1308,7 @@ nadtlb_miss_20:
 	
 	idtlbt		pte,prot
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1422,14 +1408,14 @@ itlb_miss_20w:
 
 	L3_ptep		ptp,pte,t0,va,itlb_fault
 
-	tlb_lock	spc,ptp,pte,t0,t1,itlb_fault
+	ptl_lock	spc,ptp,pte,t0,t1,itlb_fault
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot,t1
 	
 	iitlbt          pte,prot
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1446,14 +1432,14 @@ naitlb_miss_20w:
 
 	L3_ptep		ptp,pte,t0,va,naitlb_check_alias_20w
 
-	tlb_lock	spc,ptp,pte,t0,t1,naitlb_check_alias_20w
+	ptl_lock	spc,ptp,pte,t0,t1,naitlb_check_alias_20w
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot,t1
 
 	iitlbt          pte,prot
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1474,7 +1460,7 @@ itlb_miss_11:
 
 	L2_ptep		ptp,pte,t0,va,itlb_fault
 
-	tlb_lock	spc,ptp,pte,t0,t1,itlb_fault
+	ptl_lock	spc,ptp,pte,t0,t1,itlb_fault
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb_11	spc,pte,prot
@@ -1487,7 +1473,7 @@ itlb_miss_11:
 
 	mtsp		t1, %sr1	/* Restore sr1 */
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1498,7 +1484,7 @@ naitlb_miss_11:
 
 	L2_ptep		ptp,pte,t0,va,naitlb_check_alias_11
 
-	tlb_lock	spc,ptp,pte,t0,t1,naitlb_check_alias_11
+	ptl_lock	spc,ptp,pte,t0,t1,naitlb_check_alias_11
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb_11	spc,pte,prot
@@ -1511,7 +1497,7 @@ naitlb_miss_11:
 
 	mtsp		t1, %sr1	/* Restore sr1 */
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1532,7 +1518,7 @@ itlb_miss_20:
 
 	L2_ptep		ptp,pte,t0,va,itlb_fault
 
-	tlb_lock	spc,ptp,pte,t0,t1,itlb_fault
+	ptl_lock	spc,ptp,pte,t0,t1,itlb_fault
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot,t1
@@ -1541,7 +1527,7 @@ itlb_miss_20:
 
 	iitlbt          pte,prot
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1552,7 +1538,7 @@ naitlb_miss_20:
 
 	L2_ptep		ptp,pte,t0,va,naitlb_check_alias_20
 
-	tlb_lock	spc,ptp,pte,t0,t1,naitlb_check_alias_20
+	ptl_lock	spc,ptp,pte,t0,t1,naitlb_check_alias_20
 	update_accessed	ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot,t1
@@ -1561,7 +1547,7 @@ naitlb_miss_20:
 
 	iitlbt          pte,prot
 
-	tlb_unlock1	spc,t0
+	ptl_unlock1	spc,t0
 	rfir
 	nop
 
@@ -1584,14 +1570,14 @@ dbit_trap_20w:
 
 	L3_ptep		ptp,pte,t0,va,dbit_fault
 
-	tlb_lock	spc,ptp,pte,t0,t1,dbit_fault
+	ptl_lock	spc,ptp,pte,t0,t1,dbit_fault
 	update_dirty	ptp,pte,t1
 
 	make_insert_tlb	spc,pte,prot,t1
 		
 	idtlbt          pte,prot
 
-	tlb_unlock0	spc,t0
+	ptl_unlock0	spc,t0
 	rfir
 	nop
 #else
@@ -1604,7 +1590,7 @@ dbit_trap_11:
 
 	L2_ptep		ptp,pte,t0,va,dbit_fault
 
-	tlb_lock	spc,ptp,pte,t0,t1,dbit_fault
+	ptl_lock	spc,ptp,pte,t0,t1,dbit_fault
 	update_dirty	ptp,pte,t1
 
 	make_insert_tlb_11	spc,pte,prot
@@ -1617,7 +1603,7 @@ dbit_trap_11:
 
 	mtsp            t1, %sr1     /* Restore sr1 */
 
-	tlb_unlock0	spc,t0
+	ptl_unlock0	spc,t0
 	rfir
 	nop
 
@@ -1628,7 +1614,7 @@ dbit_trap_20:
 
 	L2_ptep		ptp,pte,t0,va,dbit_fault
 
-	tlb_lock	spc,ptp,pte,t0,t1,dbit_fault
+	ptl_lock	spc,ptp,pte,t0,t1,dbit_fault
 	update_dirty	ptp,pte,t1
 
 	make_insert_tlb	spc,pte,prot,t1
@@ -1637,7 +1623,7 @@ dbit_trap_20:
 	
 	idtlbt		pte,prot
 
-	tlb_unlock0	spc,t0
+	ptl_unlock0	spc,t0
 	rfir
 	nop
 #endif