openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity
linux/arch/powerpc/mm/tlb_low_64e.S

1219 lines
35 KiB
ArmAsm
Raw Normal View History

powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
/*
powerpc: Fix typo s/leve/level/ in TLB code Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2010-01-17 19:23:14 +08:00
* Low level TLB miss handlers for Book3E
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
*
* Copyright (C) 2008-2009
* Ben. Herrenschmidt (benh@kernel.crashing.org), IBM Corp.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#include <asm/processor.h>
#include <asm/reg.h>
#include <asm/page.h>
#include <asm/mmu.h>
#include <asm/ppc_asm.h>
#include <asm/asm-offsets.h>
#include <asm/cputable.h>
#include <asm/pgtable.h>
#include <asm/exception-64e.h>
#include <asm/ppc-opcode.h>
powerpc/booke64: Add DO_KVM kernel hooks Hook DO_KVM macro into 64-bit booke for KVM integration. Extend interrupt handlers' parameter list with interrupt vector numbers to accomodate the macro. Only the bolted version of tlb miss handers is addressed now. Signed-off-by: Mihai Caraman <mihai.caraman@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-08-06 11:27:05 +08:00
#include <asm/kvm_asm.h>
#include <asm/kvm_booke_hv_asm.h>
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
#ifdef CONFIG_PPC_64K_PAGES
#define VPTE_PMD_SHIFT (PTE_INDEX_SIZE+1)
#else
#define VPTE_PMD_SHIFT (PTE_INDEX_SIZE)
#endif
#define VPTE_PUD_SHIFT (VPTE_PMD_SHIFT + PMD_INDEX_SIZE)
#define VPTE_PGD_SHIFT (VPTE_PUD_SHIFT + PUD_INDEX_SIZE)
#define VPTE_INDEX_SIZE (VPTE_PGD_SHIFT + PGD_INDEX_SIZE)
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
/**********************************************************************
* *
* TLB miss handling for Book3E with a bolted linear mapping *
* No virtual page table, no nested TLB misses *
* *
**********************************************************************/
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
/*
* Note that, unlike non-bolted handlers, TLB_EXFRAME is not
* modified by the TLB miss handlers themselves, since the TLB miss
* handler code will not itself cause a recursive TLB miss.
*
* TLB_EXFRAME will be modified when crit/mc/debug exceptions are
* entered/exited.
*/
powerpc/booke64: Add DO_KVM kernel hooks Hook DO_KVM macro into 64-bit booke for KVM integration. Extend interrupt handlers' parameter list with interrupt vector numbers to accomodate the macro. Only the bolted version of tlb miss handers is addressed now. Signed-off-by: Mihai Caraman <mihai.caraman@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-08-06 11:27:05 +08:00
.macro tlb_prolog_bolted intnum addr
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
mtspr SPRN_SPRG_GEN_SCRATCH,r12
mfspr r12,SPRN_SPRG_TLB_EXFRAME
std r13,EX_TLB_R13(r12)
std r10,EX_TLB_R10(r12)
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
mfspr r13,SPRN_SPRG_PACA
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
mfcr r10
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
std r11,EX_TLB_R11(r12)
powerpc/booke64: Add DO_KVM kernel hooks Hook DO_KVM macro into 64-bit booke for KVM integration. Extend interrupt handlers' parameter list with interrupt vector numbers to accomodate the macro. Only the bolted version of tlb miss handers is addressed now. Signed-off-by: Mihai Caraman <mihai.caraman@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-08-06 11:27:05 +08:00
#ifdef CONFIG_KVM_BOOKE_HV
BEGIN_FTR_SECTION
mfspr r11, SPRN_SRR1
END_FTR_SECTION_IFSET(CPU_FTR_EMB_HV)
#endif
DO_KVM \intnum, SPRN_SRR1
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
std r16,EX_TLB_R16(r12)
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
mfspr r16,\addr /* get faulting address */
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
std r14,EX_TLB_R14(r12)
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
ld r14,PACAPGD(r13)
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
std r15,EX_TLB_R15(r12)
std r10,EX_TLB_CR(r12)
TLB_MISS_PROLOG_STATS
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
.endm
.macro tlb_epilog_bolted
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
ld r14,EX_TLB_CR(r12)
ld r10,EX_TLB_R10(r12)
ld r11,EX_TLB_R11(r12)
ld r13,EX_TLB_R13(r12)
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
mtcr r14
powerpc/booke64: Use SPRG_TLB_EXFRAME on bolted handlers While bolted handlers (including e6500) do not need to deal with a TLB miss recursively causing another TLB miss, nested TLB misses can still happen with crit/mc/debug exceptions -- so we still need to honor SPRG_TLB_EXFRAME. We don't need to spend time modifying it in the TLB miss fastpath, though -- the special level exception will handle that. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com> Cc: kvm-ppc@vger.kernel.org
2014-03-11 06:29:38 +08:00
ld r14,EX_TLB_R14(r12)
ld r15,EX_TLB_R15(r12)
TLB_MISS_RESTORE_STATS
ld r16,EX_TLB_R16(r12)
mfspr r12,SPRN_SPRG_GEN_SCRATCH
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
.endm
/* Data TLB miss */
START_EXCEPTION(data_tlb_miss_bolted)
powerpc/booke64: Add DO_KVM kernel hooks Hook DO_KVM macro into 64-bit booke for KVM integration. Extend interrupt handlers' parameter list with interrupt vector numbers to accomodate the macro. Only the bolted version of tlb miss handers is addressed now. Signed-off-by: Mihai Caraman <mihai.caraman@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-08-06 11:27:05 +08:00
tlb_prolog_bolted BOOKE_INTERRUPT_DTLB_MISS SPRN_DEAR
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
/* We need _PAGE_PRESENT and _PAGE_ACCESSED set */
/* We do the user/kernel test for the PID here along with the RW test
*/
/* We pre-test some combination of permissions to avoid double
* faults:
*
* We move the ESR:ST bit into the position of _PAGE_BAP_SW in the PTE
* ESR_ST is 0x00800000
* _PAGE_BAP_SW is 0x00000010
* So the shift is >> 19. This tests for supervisor writeability.
* If the page happens to be supervisor writeable and not user
* writeable, we will take a new fault later, but that should be
* a rare enough case.
*
* We also move ESR_ST in _PAGE_DIRTY position
* _PAGE_DIRTY is 0x00001000 so the shift is >> 11
*
* MAS1 is preset for all we need except for TID that needs to
* be cleared for kernel translations
*/
mfspr r11,SPRN_ESR
srdi r15,r16,60 /* get region */
rldicl. r10,r16,64-PGTABLE_EADDR_SIZE,PGTABLE_EADDR_SIZE+4
powerpc: Whitespace/comment changes to tlb_low_64e.S I happened to comment this code while I was digging through it; we might as well commit that. I also made some whitespace changes - the existing code had a lot of unnecessary newlines that I found annoying when I was working on my tiny laptop. No functional changes. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:41 +08:00
bne- dtlb_miss_fault_bolted /* Bail if fault addr is invalid */
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
rlwinm r10,r11,32-19,27,27
rlwimi r10,r11,32-16,19,19
powerpc: Whitespace/comment changes to tlb_low_64e.S I happened to comment this code while I was digging through it; we might as well commit that. I also made some whitespace changes - the existing code had a lot of unnecessary newlines that I found annoying when I was working on my tiny laptop. No functional changes. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:41 +08:00
cmpwi r15,0 /* user vs kernel check */
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
ori r10,r10,_PAGE_PRESENT
oris r11,r10,_PAGE_ACCESSED@h
TLB_MISS_STATS_SAVE_INFO_BOLTED
bne tlb_miss_kernel_bolted
tlb_miss_common_bolted:
/*
* This is the guts of the TLB miss handler for bolted-linear.
* We are entered with:
*
* r16 = faulting address
* r15 = crap (free to use)
* r14 = page table base
* r13 = PACA
* r11 = PTE permission mask
* r10 = crap (free to use)
*/
rldicl r15,r16,64-PGDIR_SHIFT+3,64-PGD_INDEX_SIZE-3
cmpldi cr0,r14,0
clrrdi r15,r15,3
powerpc: Whitespace/comment changes to tlb_low_64e.S I happened to comment this code while I was digging through it; we might as well commit that. I also made some whitespace changes - the existing code had a lot of unnecessary newlines that I found annoying when I was working on my tiny laptop. No functional changes. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:41 +08:00
beq tlb_miss_fault_bolted /* No PGDIR, bail */
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
BEGIN_MMU_FTR_SECTION
/* Set the TLB reservation and search for existing entry. Then load
* the entry.
*/
powerpc: Enforce usage of RA 0-R31 where possible Some macros use RA where when RA=R0 the values is 0, so make this the enforced mnemonic in the macro. Idea suggested by Andreas Schwab. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-06-25 21:33:25 +08:00
PPC_TLBSRX_DOT(0,R16)
powerpc: Whitespace/comment changes to tlb_low_64e.S I happened to comment this code while I was digging through it; we might as well commit that. I also made some whitespace changes - the existing code had a lot of unnecessary newlines that I found annoying when I was working on my tiny laptop. No functional changes. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:41 +08:00
ldx r14,r14,r15 /* grab pgd entry */
powerpc/booke-64: fix tlbsrx. path in bolted tlb handler It was branching to the cleanup part of the non-bolted handler, which would have been bad if there were any chips with tlbsrx. that use the bolted handler. Signed-off-by: Scott Wood <scottwood@freescale.com>
2012-06-13 06:02:32 +08:00
beq tlb_miss_done_bolted /* tlb exists already, bail */
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
MMU_FTR_SECTION_ELSE
powerpc: Whitespace/comment changes to tlb_low_64e.S I happened to comment this code while I was digging through it; we might as well commit that. I also made some whitespace changes - the existing code had a lot of unnecessary newlines that I found annoying when I was working on my tiny laptop. No functional changes. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:41 +08:00
ldx r14,r14,r15 /* grab pgd entry */
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
ALT_MMU_FTR_SECTION_END_IFSET(MMU_FTR_USE_TLBRSRV)
#ifndef CONFIG_PPC_64K_PAGES
rldicl r15,r16,64-PUD_SHIFT+3,64-PUD_INDEX_SIZE-3
clrrdi r15,r15,3
powerpc: Add hugepage support to 64-bit tablewalk code for FSL_BOOK3E Before hugetlb, at each level of the table, we test for !0 to determine if we have a valid table entry. With hugetlb, this compare becomes: < 0 is a normal entry 0 is an invalid entry > 0 is huge This works because the hugepage code pulls the top bit off the entry (which for non-huge entries always has the top bit set) as an indicator that we have a hugepage. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:42 +08:00
cmpdi cr0,r14,0
bge tlb_miss_fault_bolted /* Bad pgd entry or hugepage; bail */
powerpc: Whitespace/comment changes to tlb_low_64e.S I happened to comment this code while I was digging through it; we might as well commit that. I also made some whitespace changes - the existing code had a lot of unnecessary newlines that I found annoying when I was working on my tiny laptop. No functional changes. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:41 +08:00
ldx r14,r14,r15 /* grab pud entry */
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
#endif /* CONFIG_PPC_64K_PAGES */
rldicl r15,r16,64-PMD_SHIFT+3,64-PMD_INDEX_SIZE-3
clrrdi r15,r15,3
powerpc: Add hugepage support to 64-bit tablewalk code for FSL_BOOK3E Before hugetlb, at each level of the table, we test for !0 to determine if we have a valid table entry. With hugetlb, this compare becomes: < 0 is a normal entry 0 is an invalid entry > 0 is huge This works because the hugepage code pulls the top bit off the entry (which for non-huge entries always has the top bit set) as an indicator that we have a hugepage. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:42 +08:00
cmpdi cr0,r14,0
bge tlb_miss_fault_bolted
powerpc: Whitespace/comment changes to tlb_low_64e.S I happened to comment this code while I was digging through it; we might as well commit that. I also made some whitespace changes - the existing code had a lot of unnecessary newlines that I found annoying when I was working on my tiny laptop. No functional changes. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:41 +08:00
ldx r14,r14,r15 /* Grab pmd entry */
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
rldicl r15,r16,64-PAGE_SHIFT+3,64-PTE_INDEX_SIZE-3
clrrdi r15,r15,3
powerpc: Add hugepage support to 64-bit tablewalk code for FSL_BOOK3E Before hugetlb, at each level of the table, we test for !0 to determine if we have a valid table entry. With hugetlb, this compare becomes: < 0 is a normal entry 0 is an invalid entry > 0 is huge This works because the hugepage code pulls the top bit off the entry (which for non-huge entries always has the top bit set) as an indicator that we have a hugepage. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-10-10 18:50:42 +08:00
cmpdi cr0,r14,0
bge tlb_miss_fault_bolted
ldx r14,r14,r15 /* Grab PTE, normal (!huge) page */
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
/* Check if required permissions are met */
andc. r15,r11,r14
rldicr r15,r14,64-(PTE_RPN_SHIFT-PAGE_SHIFT),63-PAGE_SHIFT
bne- tlb_miss_fault_bolted
/* Now we build the MAS:
*
* MAS 0 : Fully setup with defaults in MAS4 and TLBnCFG
* MAS 1 : Almost fully setup
* - PID already updated by caller if necessary
* - TSIZE need change if !base page size, not
* yet implemented for now
* MAS 2 : Defaults not useful, need to be redone
* MAS 3+7 : Needs to be done
*/
clrrdi r11,r16,12 /* Clear low crap in EA */
clrldi r15,r15,12 /* Clear crap at the top */
rlwimi r11,r14,32-19,27,31 /* Insert WIMGE */
rlwimi r15,r14,32-8,22,25 /* Move in U bits */
mtspr SPRN_MAS2,r11
andi. r11,r14,_PAGE_DIRTY
rlwimi r15,r14,32-2,26,31 /* Move in BAP bits */
/* Mask out SW and UW if !DIRTY (XXX optimize this !) */
bne 1f
li r11,MAS3_SW|MAS3_UW
andc r15,r15,r11
1:
mtspr SPRN_MAS7_MAS3,r15
tlbwe
powerpc/booke-64: fix tlbsrx. path in bolted tlb handler It was branching to the cleanup part of the non-bolted handler, which would have been bad if there were any chips with tlbsrx. that use the bolted handler. Signed-off-by: Scott Wood <scottwood@freescale.com>
2012-06-13 06:02:32 +08:00
tlb_miss_done_bolted:
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
TLB_MISS_STATS_X(MMSTAT_TLB_MISS_NORM_OK)
tlb_epilog_bolted
rfi
itlb_miss_kernel_bolted:
li r11,_PAGE_PRESENT|_PAGE_BAP_SX /* Base perm */
oris r11,r11,_PAGE_ACCESSED@h
tlb_miss_kernel_bolted:
mfspr r10,SPRN_MAS1
ld r14,PACA_KERNELPGD(r13)
cmpldi cr0,r15,8 /* Check for vmalloc region */
rlwinm r10,r10,0,16,1 /* Clear TID */
mtspr SPRN_MAS1,r10
beq+ tlb_miss_common_bolted
tlb_miss_fault_bolted:
/* We need to check if it was an instruction miss */
andi. r10,r11,_PAGE_EXEC|_PAGE_BAP_SX
bne itlb_miss_fault_bolted
dtlb_miss_fault_bolted:
TLB_MISS_STATS_D(MMSTAT_TLB_MISS_NORM_FAULT)
tlb_epilog_bolted
b exc_data_storage_book3e
itlb_miss_fault_bolted:
TLB_MISS_STATS_I(MMSTAT_TLB_MISS_NORM_FAULT)
tlb_epilog_bolted
b exc_instruction_storage_book3e
/* Instruction TLB miss */
START_EXCEPTION(instruction_tlb_miss_bolted)
powerpc/booke64: Add DO_KVM kernel hooks Hook DO_KVM macro into 64-bit booke for KVM integration. Extend interrupt handlers' parameter list with interrupt vector numbers to accomodate the macro. Only the bolted version of tlb miss handers is addressed now. Signed-off-by: Mihai Caraman <mihai.caraman@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-08-06 11:27:05 +08:00
tlb_prolog_bolted BOOKE_INTERRUPT_ITLB_MISS SPRN_SRR0
powerpc/book3e-64: use a separate TLB handler when linear map is bolted On MMUs such as FSL where we can guarantee the entire linear mapping is bolted, we don't need to worry about linear TLB misses. If on top of that we do a full table walk, we get rid of all recursive TLB faults, and can dispense with some state saving. This gains a few percent on TLB-miss-heavy workloads, and around 50% on a benchmark that had a high rate of virtual page table faults under the normal handler. While touching the EX_TLB layout, remove EX_TLB_MMUCR0, EX_TLB_SRR0, and EX_TLB_SRR1 as they're not used. [BenH: Fixed build with 64K pages (wsp config)] Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-22 19:25:42 +08:00
rldicl. r10,r16,64-PGTABLE_EADDR_SIZE,PGTABLE_EADDR_SIZE+4
srdi r15,r16,60 /* get region */
TLB_MISS_STATS_SAVE_INFO_BOLTED
bne- itlb_miss_fault_bolted
li r11,_PAGE_PRESENT|_PAGE_EXEC /* Base perm */
/* We do the user/kernel test for the PID here along with the RW test
*/
cmpldi cr0,r15,0 /* Check for user region */
oris r11,r11,_PAGE_ACCESSED@h
beq tlb_miss_common_bolted
b itlb_miss_kernel_bolted
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
powerpc/booke64: Guard e6500 tlb handler with CONFIG_PPC_FSL_BOOK3E ...and make CONFIG_PPC_FSL_BOOK3E conflict with CONFIG_PPC_64K_PAGES. This fixes a build break with CONFIG_PPC_64K_PAGES on 64-bit book3e, that was introduced by commit 28efc35fe68dacbddc4b12c2fa8f2df1593a4ad3 ("powerpc/e6500: TLB miss handler with hardware tablewalk support"). Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-01-18 08:36:38 +08:00
#ifdef CONFIG_PPC_FSL_BOOK3E
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
/*
* TLB miss handling for e6500 and derivatives, using hardware tablewalk.
*
* Linear mapping is bolted: no virtual page table or nested TLB misses
* Indirect entries in TLB1, hardware loads resulting direct entries
* into TLB0
* No HES or NV hint on TLB1, so we need to do software round-robin
* No tlbsrx. so we need a spinlock, and we have to deal
* with MAS-damage caused by tlbsx
* 4K pages only
*/
START_EXCEPTION(instruction_tlb_miss_e6500)
tlb_prolog_bolted BOOKE_INTERRUPT_ITLB_MISS SPRN_SRR0
ld r11,PACA_TCD_PTR(r13)
srdi. r15,r16,60 /* get region */
ori r16,r16,1
TLB_MISS_STATS_SAVE_INFO_BOLTED
bne tlb_miss_kernel_e6500 /* user/kernel test */
b tlb_miss_common_e6500
START_EXCEPTION(data_tlb_miss_e6500)
tlb_prolog_bolted BOOKE_INTERRUPT_DTLB_MISS SPRN_DEAR
ld r11,PACA_TCD_PTR(r13)
srdi. r15,r16,60 /* get region */
rldicr r16,r16,0,62
TLB_MISS_STATS_SAVE_INFO_BOLTED
bne tlb_miss_kernel_e6500 /* user vs kernel check */
/*
* This is the guts of the TLB miss handler for e6500 and derivatives.
* We are entered with:
*
* r16 = page of faulting address (low bit 0 if data, 1 if instruction)
* r15 = crap (free to use)
* r14 = page table base
* r13 = PACA
* r11 = tlb_per_core ptr
powerpc/e6500: hw tablewalk: fix recursive tlb lock on cpu 0 Commit 82d86de25b9c99db546e17c6f7ebf9a691da557e "TLB lock recursive" introduced a bug whereby cpu 0 uses the same value for "lock held" as is used to indicate that the lock is free. This means that cpu 1 can acquire the lock whenever it wants, regardless of whether cpu 0 has it locked, which in turn means we can get duplicate TLB entries. Add one to the CPU value to ensure we do not use zero as a "lock held" value. Signed-off-by: Scott Wood <scottwood@freescale.com> Reported-by: Ed Swarthout <ed.swarthout@freescale.com>
2014-05-23 05:10:35 +08:00
* r10 = crap (free to use)
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
*/
tlb_miss_common_e6500:
powerpc/e6500: Work around erratum A-008139 Erratum A-008139 can cause duplicate TLB entries if an indirect entry is overwritten using tlbwe while the other thread is using it to do a lookup. Work around this by using tlbilx to invalidate prior to overwriting. To avoid the need to save another register to hold MAS1 during the workaround code, TID clearing has been moved from tlb_miss_kernel_e6500 until after the SMT section. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-06-12 05:09:32 +08:00
crmove cr2*4+2,cr0*4+2 /* cr2.eq != 0 if kernel address */
BEGIN_FTR_SECTION /* CPU_FTR_SMT */
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
/*
* Search if we already have an indirect entry for that virtual
* address, and if we do, bail out.
*
* MAS6:IND should be already set based on MAS4
*/
powerpc/e6500: Make TLB lock recursive Once special level interrupts are supported, we may take nested TLB misses -- so allow the same thread to acquire the lock recursively. The lock will not be effective against the nested TLB miss handler trying to write the same entry as the interrupted TLB miss handler, but that's also a problem on non-threaded CPUs that lack TLB write conditional. This will be addressed in the patch that enables crit/mc support by invalidating the TLB on return from level exceptions. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-03-08 04:48:35 +08:00
1: lbarx r15,0,r11
lhz r10,PACAPACAINDEX(r13)
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
cmpdi r15,0
powerpc/e6500: Make TLB lock recursive Once special level interrupts are supported, we may take nested TLB misses -- so allow the same thread to acquire the lock recursively. The lock will not be effective against the nested TLB miss handler trying to write the same entry as the interrupted TLB miss handler, but that's also a problem on non-threaded CPUs that lack TLB write conditional. This will be addressed in the patch that enables crit/mc support by invalidating the TLB on return from level exceptions. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-03-08 04:48:35 +08:00
cmpdi cr1,r15,1 /* set cr1.eq = 0 for non-recursive */
powerpc/e6500: hw tablewalk: fix recursive tlb lock on cpu 0 Commit 82d86de25b9c99db546e17c6f7ebf9a691da557e "TLB lock recursive" introduced a bug whereby cpu 0 uses the same value for "lock held" as is used to indicate that the lock is free. This means that cpu 1 can acquire the lock whenever it wants, regardless of whether cpu 0 has it locked, which in turn means we can get duplicate TLB entries. Add one to the CPU value to ensure we do not use zero as a "lock held" value. Signed-off-by: Scott Wood <scottwood@freescale.com> Reported-by: Ed Swarthout <ed.swarthout@freescale.com>
2014-05-23 05:10:35 +08:00
addi r10,r10,1
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
bne 2f
powerpc/e6500: Make TLB lock recursive Once special level interrupts are supported, we may take nested TLB misses -- so allow the same thread to acquire the lock recursively. The lock will not be effective against the nested TLB miss handler trying to write the same entry as the interrupted TLB miss handler, but that's also a problem on non-threaded CPUs that lack TLB write conditional. This will be addressed in the patch that enables crit/mc support by invalidating the TLB on return from level exceptions. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-03-08 04:48:35 +08:00
stbcx. r10,0,r11
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
bne 1b
powerpc/e6500: Make TLB lock recursive Once special level interrupts are supported, we may take nested TLB misses -- so allow the same thread to acquire the lock recursively. The lock will not be effective against the nested TLB miss handler trying to write the same entry as the interrupted TLB miss handler, but that's also a problem on non-threaded CPUs that lack TLB write conditional. This will be addressed in the patch that enables crit/mc support by invalidating the TLB on return from level exceptions. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-03-08 04:48:35 +08:00
3:
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
.subsection 1
powerpc/e6500: Make TLB lock recursive Once special level interrupts are supported, we may take nested TLB misses -- so allow the same thread to acquire the lock recursively. The lock will not be effective against the nested TLB miss handler trying to write the same entry as the interrupted TLB miss handler, but that's also a problem on non-threaded CPUs that lack TLB write conditional. This will be addressed in the patch that enables crit/mc support by invalidating the TLB on return from level exceptions. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-03-08 04:48:35 +08:00
2: cmpd cr1,r15,r10 /* recursive lock due to mcheck/crit/etc? */
beq cr1,3b /* unlock will happen if cr1.eq = 0 */
lbz r15,0(r11)
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
cmpdi r15,0
bne 2b
b 1b
.previous
powerpc/e6500: Work around erratum A-008139 Erratum A-008139 can cause duplicate TLB entries if an indirect entry is overwritten using tlbwe while the other thread is using it to do a lookup. Work around this by using tlbilx to invalidate prior to overwriting. To avoid the need to save another register to hold MAS1 during the workaround code, TID clearing has been moved from tlb_miss_kernel_e6500 until after the SMT section. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-06-12 05:09:32 +08:00
/*
* Erratum A-008139 says that we can't use tlbwe to change
* an indirect entry in any way (including replacing or
* invalidating) if the other thread could be in the process
* of a lookup. The workaround is to invalidate the entry
* with tlbilx before overwriting.
*/
lbz r15,TCD_ESEL_NEXT(r11)
rlwinm r10,r15,16,0xff0000
oris r10,r10,MAS0_TLBSEL(1)@h
mtspr SPRN_MAS0,r10
isync
tlbre
powerpc/e6500: hw tablewalk: clear TID in kernel indirect entries Previously TID was being cleared before the tlbsx, but not after. This can lead to a multiway hit between a TLB entry with TID=0 (previously inserted when PID=0) and a TLB entry with TID!=0 that matches PID. This can theoretically result in undefined behavior, though we probably get lucky due to the details of the overlap. It also results in the inability to use multihit detection to detect other conflicting TLB entries, as well as poorer TLB utilization due to duplicating kernel TLB entries. Rather than try to patch up MAS1 after tlbsx, the entire value is saved/restored as with MAS2. I observed a slight improvement in TLB miss performance with this patch applied. Signed-off-by: Scott Wood <scottwood@freescale.com> Reported-by: Ed Swarthout <ed.swarthout@freescale.com>
2014-05-20 12:04:55 +08:00
mfspr r15,SPRN_MAS1
powerpc/e6500: Work around erratum A-008139 Erratum A-008139 can cause duplicate TLB entries if an indirect entry is overwritten using tlbwe while the other thread is using it to do a lookup. Work around this by using tlbilx to invalidate prior to overwriting. To avoid the need to save another register to hold MAS1 during the workaround code, TID clearing has been moved from tlb_miss_kernel_e6500 until after the SMT section. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-06-12 05:09:32 +08:00
andis. r15,r15,MAS1_VALID@h
beq 5f
BEGIN_FTR_SECTION_NESTED(532)
mfspr r10,SPRN_MAS8
rlwinm r10,r10,0,0x80000fff /* tgs,tlpid -> sgs,slpid */
mtspr SPRN_MAS5,r10
END_FTR_SECTION_NESTED(CPU_FTR_EMB_HV,CPU_FTR_EMB_HV,532)
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
mfspr r10,SPRN_MAS1
powerpc/e6500: Work around erratum A-008139 Erratum A-008139 can cause duplicate TLB entries if an indirect entry is overwritten using tlbwe while the other thread is using it to do a lookup. Work around this by using tlbilx to invalidate prior to overwriting. To avoid the need to save another register to hold MAS1 during the workaround code, TID clearing has been moved from tlb_miss_kernel_e6500 until after the SMT section. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-06-12 05:09:32 +08:00
rlwinm r15,r10,0,0x3fff0000 /* tid -> spid */
rlwimi r15,r10,20,0x00000003 /* ind,ts -> sind,sas */
mfspr r10,SPRN_MAS6
mtspr SPRN_MAS6,r15
mfspr r15,SPRN_MAS2
isync
tlbilxva 0,r15
isync
powerpc/e6500: hw tablewalk: clear TID in kernel indirect entries Previously TID was being cleared before the tlbsx, but not after. This can lead to a multiway hit between a TLB entry with TID=0 (previously inserted when PID=0) and a TLB entry with TID!=0 that matches PID. This can theoretically result in undefined behavior, though we probably get lucky due to the details of the overlap. It also results in the inability to use multihit detection to detect other conflicting TLB entries, as well as poorer TLB utilization due to duplicating kernel TLB entries. Rather than try to patch up MAS1 after tlbsx, the entire value is saved/restored as with MAS2. I observed a slight improvement in TLB miss performance with this patch applied. Signed-off-by: Scott Wood <scottwood@freescale.com> Reported-by: Ed Swarthout <ed.swarthout@freescale.com>
2014-05-20 12:04:55 +08:00
powerpc/e6500: Work around erratum A-008139 Erratum A-008139 can cause duplicate TLB entries if an indirect entry is overwritten using tlbwe while the other thread is using it to do a lookup. Work around this by using tlbilx to invalidate prior to overwriting. To avoid the need to save another register to hold MAS1 during the workaround code, TID clearing has been moved from tlb_miss_kernel_e6500 until after the SMT section. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-06-12 05:09:32 +08:00
mtspr SPRN_MAS6,r10
5:
BEGIN_FTR_SECTION_NESTED(532)
li r10,0
mtspr SPRN_MAS8,r10
mtspr SPRN_MAS5,r10
END_FTR_SECTION_NESTED(CPU_FTR_EMB_HV,CPU_FTR_EMB_HV,532)
tlbsx 0,r16
mfspr r10,SPRN_MAS1
andis. r15,r10,MAS1_VALID@h
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
bne tlb_miss_done_e6500
powerpc/e6500: Work around erratum A-008139 Erratum A-008139 can cause duplicate TLB entries if an indirect entry is overwritten using tlbwe while the other thread is using it to do a lookup. Work around this by using tlbilx to invalidate prior to overwriting. To avoid the need to save another register to hold MAS1 during the workaround code, TID clearing has been moved from tlb_miss_kernel_e6500 until after the SMT section. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-06-12 05:09:32 +08:00
FTR_SECTION_ELSE
mfspr r10,SPRN_MAS1
ALT_FTR_SECTION_END_IFSET(CPU_FTR_SMT)
oris r10,r10,MAS1_VALID@h
beq cr2,4f
rlwinm r10,r10,0,16,1 /* Clear TID */
4: mtspr SPRN_MAS1,r10
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
/* Now, we need to walk the page tables. First check if we are in
* range.
*/
rldicl. r10,r16,64-PGTABLE_EADDR_SIZE,PGTABLE_EADDR_SIZE+4
bne- tlb_miss_fault_e6500
rldicl r15,r16,64-PGDIR_SHIFT+3,64-PGD_INDEX_SIZE-3
cmpldi cr0,r14,0
clrrdi r15,r15,3
beq- tlb_miss_fault_e6500 /* No PGDIR, bail */
ldx r14,r14,r15 /* grab pgd entry */
rldicl r15,r16,64-PUD_SHIFT+3,64-PUD_INDEX_SIZE-3
clrrdi r15,r15,3
cmpdi cr0,r14,0
bge tlb_miss_fault_e6500 /* Bad pgd entry or hugepage; bail */
ldx r14,r14,r15 /* grab pud entry */
rldicl r15,r16,64-PMD_SHIFT+3,64-PMD_INDEX_SIZE-3
clrrdi r15,r15,3
cmpdi cr0,r14,0
bge tlb_miss_fault_e6500
ldx r14,r14,r15 /* Grab pmd entry */
mfspr r10,SPRN_MAS0
cmpdi cr0,r14,0
bge tlb_miss_fault_e6500
/* Now we build the MAS for a 2M indirect page:
*
* MAS 0 : ESEL needs to be filled by software round-robin
* MAS 1 : Fully set up
* - PID already updated by caller if necessary
* - TSIZE for now is base ind page size always
* - TID already cleared if necessary
* MAS 2 : Default not 2M-aligned, need to be redone
* MAS 3+7 : Needs to be done
*/
ori r14,r14,(BOOK3E_PAGESZ_4K << MAS3_SPSIZE_SHIFT)
mtspr SPRN_MAS7_MAS3,r14
clrrdi r15,r16,21 /* make EA 2M-aligned */
mtspr SPRN_MAS2,r15
lbz r15,TCD_ESEL_NEXT(r11)
lbz r16,TCD_ESEL_MAX(r11)
lbz r14,TCD_ESEL_FIRST(r11)
rlwimi r10,r15,16,0x00ff0000 /* insert esel_next into MAS0 */
addi r15,r15,1 /* increment esel_next */
mtspr SPRN_MAS0,r10
cmpw r15,r16
iseleq r15,r14,r15 /* if next == last use first */
stb r15,TCD_ESEL_NEXT(r11)
tlbwe
tlb_miss_done_e6500:
.macro tlb_unlock_e6500
powerpc/booke64: wrap tlb lock and search in htw miss with FTR_SMT Virtualized environments may expose a e6500 dual-threaded core as two single-threaded e6500 cores. Take advantage of this and get rid of the tlb lock and the trap-causing tlbsx in the htw miss handler by guarding with CPU_FTR_SMT, as it's already being done in the bolted tlb1 miss handler. As seen in the results below, measurements done with lmbench random memory access latency test running under Freescale's Embedded Hypervisor, there is a ~34% improvement. Memory latencies in nanoseconds - smaller is better (WARNING - may not be correct, check graphs) ---------------------------------------------------- Host Mhz L1 $ L2 $ Main mem Rand mem --------- --- ---- ---- -------- -------- smt 1665 1.8020 13.2 83.0 1149.7 nosmt 1665 1.8020 13.2 83.0 758.1 Signed-off-by: Laurentiu Tudor <Laurentiu.Tudor@freescale.com> Cc: Scott Wood <scottwood@freescale.com> [scottwood@freescale.com: commit message tweak] Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-05-30 22:59:15 +08:00
BEGIN_FTR_SECTION
powerpc/e6500: Make TLB lock recursive Once special level interrupts are supported, we may take nested TLB misses -- so allow the same thread to acquire the lock recursively. The lock will not be effective against the nested TLB miss handler trying to write the same entry as the interrupted TLB miss handler, but that's also a problem on non-threaded CPUs that lack TLB write conditional. This will be addressed in the patch that enables crit/mc support by invalidating the TLB on return from level exceptions. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-03-08 04:48:35 +08:00
beq cr1,1f /* no unlock if lock was recursively grabbed */
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
li r15,0
isync
powerpc/e6500: Make TLB lock recursive Once special level interrupts are supported, we may take nested TLB misses -- so allow the same thread to acquire the lock recursively. The lock will not be effective against the nested TLB miss handler trying to write the same entry as the interrupted TLB miss handler, but that's also a problem on non-threaded CPUs that lack TLB write conditional. This will be addressed in the patch that enables crit/mc support by invalidating the TLB on return from level exceptions. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-03-08 04:48:35 +08:00
stb r15,0(r11)
1:
powerpc/booke64: wrap tlb lock and search in htw miss with FTR_SMT Virtualized environments may expose a e6500 dual-threaded core as two single-threaded e6500 cores. Take advantage of this and get rid of the tlb lock and the trap-causing tlbsx in the htw miss handler by guarding with CPU_FTR_SMT, as it's already being done in the bolted tlb1 miss handler. As seen in the results below, measurements done with lmbench random memory access latency test running under Freescale's Embedded Hypervisor, there is a ~34% improvement. Memory latencies in nanoseconds - smaller is better (WARNING - may not be correct, check graphs) ---------------------------------------------------- Host Mhz L1 $ L2 $ Main mem Rand mem --------- --- ---- ---- -------- -------- smt 1665 1.8020 13.2 83.0 1149.7 nosmt 1665 1.8020 13.2 83.0 758.1 Signed-off-by: Laurentiu Tudor <Laurentiu.Tudor@freescale.com> Cc: Scott Wood <scottwood@freescale.com> [scottwood@freescale.com: commit message tweak] Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-05-30 22:59:15 +08:00
END_FTR_SECTION_IFSET(CPU_FTR_SMT)
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
.endm
tlb_unlock_e6500
TLB_MISS_STATS_X(MMSTAT_TLB_MISS_NORM_OK)
tlb_epilog_bolted
rfi
tlb_miss_kernel_e6500:
ld r14,PACA_KERNELPGD(r13)
powerpc/e6500: Work around erratum A-008139 Erratum A-008139 can cause duplicate TLB entries if an indirect entry is overwritten using tlbwe while the other thread is using it to do a lookup. Work around this by using tlbilx to invalidate prior to overwriting. To avoid the need to save another register to hold MAS1 during the workaround code, TID clearing has been moved from tlb_miss_kernel_e6500 until after the SMT section. Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-06-12 05:09:32 +08:00
cmpldi cr1,r15,8 /* Check for vmalloc region */
beq+ cr1,tlb_miss_common_e6500
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
tlb_miss_fault_e6500:
tlb_unlock_e6500
/* We need to check if it was an instruction miss */
andi. r16,r16,1
bne itlb_miss_fault_e6500
dtlb_miss_fault_e6500:
TLB_MISS_STATS_D(MMSTAT_TLB_MISS_NORM_FAULT)
tlb_epilog_bolted
b exc_data_storage_book3e
itlb_miss_fault_e6500:
TLB_MISS_STATS_I(MMSTAT_TLB_MISS_NORM_FAULT)
tlb_epilog_bolted
b exc_instruction_storage_book3e
powerpc/booke64: Guard e6500 tlb handler with CONFIG_PPC_FSL_BOOK3E ...and make CONFIG_PPC_FSL_BOOK3E conflict with CONFIG_PPC_64K_PAGES. This fixes a build break with CONFIG_PPC_64K_PAGES on 64-bit book3e, that was introduced by commit 28efc35fe68dacbddc4b12c2fa8f2df1593a4ad3 ("powerpc/e6500: TLB miss handler with hardware tablewalk support"). Signed-off-by: Scott Wood <scottwood@freescale.com>
2014-01-18 08:36:38 +08:00
#endif /* CONFIG_PPC_FSL_BOOK3E */
powerpc/e6500: TLB miss handler with hardware tablewalk support There are a few things that make the existing hw tablewalk handlers unsuitable for e6500: - Indirect entries go in TLB1 (though the resulting direct entries go in TLB0). - It has threads, but no "tlbsrx." -- so we need a spinlock and a normal "tlbsx". Because we need this lock, hardware tablewalk is mandatory on e6500 unless we want to add spinlock+tlbsx to the normal bolted TLB miss handler. - TLB1 has no HES (nor next-victim hint) so we need software round robin (TODO: integrate this round robin data with hugetlb/KVM) - The existing tablewalk handlers map half of a page table at a time, because IBM hardware has a fixed 1MiB indirect page size. e6500 has variable size indirect entries, with a minimum of 2MiB. So we can't do the half-page indirect mapping, and even if we could it would be less efficient than mapping the full page. - Like on e5500, the linear mapping is bolted, so we don't need the overhead of supporting nested tlb misses. Note that hardware tablewalk does not work in rev1 of e6500. We do not expect to support e6500 rev1 in mainline Linux. Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Mihai Caraman <mihai.caraman@freescale.com>
2013-10-12 08:22:38 +08:00
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
/**********************************************************************
* *
* TLB miss handling for Book3E with TLB reservation and HES support *
* *
**********************************************************************/
/* Data TLB miss */
START_EXCEPTION(data_tlb_miss)
TLB_MISS_PROLOG
/* Now we handle the fault proper. We only save DEAR in normal
* fault case since that's the only interesting values here.
* We could probably also optimize by not saving SRR0/1 in the
* linear mapping case but I'll leave that for later
*/
mfspr r14,SPRN_ESR
mfspr r16,SPRN_DEAR /* get faulting address */
srdi r15,r16,60 /* get region */
cmpldi cr0,r15,0xc /* linear mapping ? */
TLB_MISS_STATS_SAVE_INFO
beq tlb_load_linear /* yes -> go to linear map load */
/* The page tables are mapped virtually linear. At this point, though,
* we don't know whether we are trying to fault in a first level
* virtual address or a virtual page table address. We can get that
* from bit 0x1 of the region ID which we have set for a page table
*/
andi. r10,r15,0x1
bne- virt_page_table_tlb_miss
std r14,EX_TLB_ESR(r12); /* save ESR */
std r16,EX_TLB_DEAR(r12); /* save DEAR */
/* We need _PAGE_PRESENT and _PAGE_ACCESSED set */
li r11,_PAGE_PRESENT
oris r11,r11,_PAGE_ACCESSED@h
/* We do the user/kernel test for the PID here along with the RW test
*/
cmpldi cr0,r15,0 /* Check for user region */
/* We pre-test some combination of permissions to avoid double
* faults:
*
* We move the ESR:ST bit into the position of _PAGE_BAP_SW in the PTE
* ESR_ST is 0x00800000
* _PAGE_BAP_SW is 0x00000010
* So the shift is >> 19. This tests for supervisor writeability.
* If the page happens to be supervisor writeable and not user
* writeable, we will take a new fault later, but that should be
* a rare enough case.
*
* We also move ESR_ST in _PAGE_DIRTY position
* _PAGE_DIRTY is 0x00001000 so the shift is >> 11
*
* MAS1 is preset for all we need except for TID that needs to
* be cleared for kernel translations
*/
rlwimi r11,r14,32-19,27,27
rlwimi r11,r14,32-16,19,19
beq normal_tlb_miss
/* XXX replace the RMW cycles with immediate loads + writes */
1: mfspr r10,SPRN_MAS1
cmpldi cr0,r15,8 /* Check for vmalloc region */
rlwinm r10,r10,0,16,1 /* Clear TID */
mtspr SPRN_MAS1,r10
beq+ normal_tlb_miss
/* We got a crappy address, just fault with whatever DEAR and ESR
* are here
*/
TLB_MISS_STATS_D(MMSTAT_TLB_MISS_NORM_FAULT)
TLB_MISS_EPILOG_ERROR
b exc_data_storage_book3e
/* Instruction TLB miss */
START_EXCEPTION(instruction_tlb_miss)
TLB_MISS_PROLOG
/* If we take a recursive fault, the second level handler may need
* to know whether we are handling a data or instruction fault in
* order to get to the right store fault handler. We provide that
* info by writing a crazy value in ESR in our exception frame
*/
li r14,-1 /* store to exception frame is done later */
/* Now we handle the fault proper. We only save DEAR in the non
* linear mapping case since we know the linear mapping case will
* not re-enter. We could indeed optimize and also not save SRR0/1
* in the linear mapping case but I'll leave that for later
*
* Faulting address is SRR0 which is already in r16
*/
srdi r15,r16,60 /* get region */
cmpldi cr0,r15,0xc /* linear mapping ? */
TLB_MISS_STATS_SAVE_INFO
beq tlb_load_linear /* yes -> go to linear map load */
/* We do the user/kernel test for the PID here along with the RW test
*/
powerpc/mm: Cleanup handling of execute permission This is an attempt at cleaning up a bit the way we handle execute permission on powerpc. _PAGE_HWEXEC is gone, _PAGE_EXEC is now only defined by CPUs that can do something with it, and the myriad of #ifdef's in the I$/D$ coherency code is reduced to 2 cases that hopefully should cover everything. The logic on BookE is a little bit different than what it was though not by much. Since now, _PAGE_EXEC will be set by the generic code for executable pages, we need to filter out if they are unclean and recover it. However, I don't expect the code to be more bloated than it already was in that area due to that change. I could boast that this brings proper enforcing of per-page execute permissions to all BookE and 40x but in fact, we've had that now for some time as a side effect of my previous rework in that area (and I didn't even know it :-) We would only enable execute permission if the page was cache clean and we would only cache clean it if we took and exec fault. Since we now enforce that the later only work if VM_EXEC is part of the VMA flags, we de-fact already enforce per-page execute permissions... Unless I missed something Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-19 03:00:34 +08:00
li r11,_PAGE_PRESENT|_PAGE_EXEC /* Base perm */
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
oris r11,r11,_PAGE_ACCESSED@h
cmpldi cr0,r15,0 /* Check for user region */
std r14,EX_TLB_ESR(r12) /* write crazy -1 to frame */
beq normal_tlb_miss
powerpc/mm: Fix module instruction tlb fault handling on Book-E 64 We were seeing oops like the following when we did an rmmod on a module: Unable to handle kernel paging request for instruction fetch Faulting instruction address: 0x8000000000008010 Oops: Kernel access of bad area, sig: 11 [#1] SMP NR_CPUS=2 P5020 DS last sysfs file: /sys/devices/qman-portals.2/qman-pool.9/uevent Modules linked in: qman_tester(-) NIP: 8000000000008010 LR: c000000000074858 CTR: 8000000000008010 REGS: c00000002e29bab0 TRAP: 0400 Not tainted (2.6.34.6-00744-g2d21f14) MSR: 0000000080029000 <EE,ME,CE> CR: 24000448 XER: 00000000 TASK = c00000007a8be600[4987] 'rmmod' THREAD: c00000002e298000 CPU: 1 GPR00: 8000000000008010 c00000002e29bd30 8000000000012798 c00000000035fb28 GPR04: 0000000000000002 0000000000000002 0000000024022428 c000000000009108 GPR08: fffffffffffffffe 800000000000a618 c0000000003c13c8 0000000000000000 GPR12: 0000000022000444 c00000000fffed00 0000000000000000 0000000000000000 GPR16: 00000000100c0000 0000000000000000 00000000100dabc8 0000000010099688 GPR20: 0000000000000000 00000000100cfc28 0000000000000000 0000000010011a44 GPR24: 00000000100017b2 0000000000000000 0000000000000000 0000000000000880 GPR28: c00000000035fb28 800000000000a7b8 c000000000376d80 c0000000003cce50 NIP [8000000000008010] .test_exit+0x0/0x10 [qman_tester] LR [c000000000074858] .SyS_delete_module+0x1f8/0x2f0 Call Trace: [c00000002e29bd30] [c0000000000748b4] .SyS_delete_module+0x254/0x2f0 (unreliable) [c00000002e29be30] [c000000000000580] syscall_exit+0x0/0x2c Instruction dump: XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX 38600000 4e800020 60000000 60000000 <4e800020> 60000000 60000000 60000000 ---[ end trace 4f57124939a84dc8 ]--- This appears to be due to checking the wrong permission bits in the instruction_tlb_miss handling if the address that faulted was in vmalloc space. We need to look at the supervisor execute (_PAGE_BAP_SX) bit and not the user bit (_PAGE_BAP_UX/_PAGE_EXEC). Also removed a branch level since it did not appear to be used. Reported-by: Jeffrey Ladouceur <Jeffrey.Ladouceur@freescale.com> Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2010-11-17 15:20:32 +08:00
li r11,_PAGE_PRESENT|_PAGE_BAP_SX /* Base perm */
oris r11,r11,_PAGE_ACCESSED@h
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
/* XXX replace the RMW cycles with immediate loads + writes */
powerpc/mm: Fix module instruction tlb fault handling on Book-E 64 We were seeing oops like the following when we did an rmmod on a module: Unable to handle kernel paging request for instruction fetch Faulting instruction address: 0x8000000000008010 Oops: Kernel access of bad area, sig: 11 [#1] SMP NR_CPUS=2 P5020 DS last sysfs file: /sys/devices/qman-portals.2/qman-pool.9/uevent Modules linked in: qman_tester(-) NIP: 8000000000008010 LR: c000000000074858 CTR: 8000000000008010 REGS: c00000002e29bab0 TRAP: 0400 Not tainted (2.6.34.6-00744-g2d21f14) MSR: 0000000080029000 <EE,ME,CE> CR: 24000448 XER: 00000000 TASK = c00000007a8be600[4987] 'rmmod' THREAD: c00000002e298000 CPU: 1 GPR00: 8000000000008010 c00000002e29bd30 8000000000012798 c00000000035fb28 GPR04: 0000000000000002 0000000000000002 0000000024022428 c000000000009108 GPR08: fffffffffffffffe 800000000000a618 c0000000003c13c8 0000000000000000 GPR12: 0000000022000444 c00000000fffed00 0000000000000000 0000000000000000 GPR16: 00000000100c0000 0000000000000000 00000000100dabc8 0000000010099688 GPR20: 0000000000000000 00000000100cfc28 0000000000000000 0000000010011a44 GPR24: 00000000100017b2 0000000000000000 0000000000000000 0000000000000880 GPR28: c00000000035fb28 800000000000a7b8 c000000000376d80 c0000000003cce50 NIP [8000000000008010] .test_exit+0x0/0x10 [qman_tester] LR [c000000000074858] .SyS_delete_module+0x1f8/0x2f0 Call Trace: [c00000002e29bd30] [c0000000000748b4] .SyS_delete_module+0x254/0x2f0 (unreliable) [c00000002e29be30] [c000000000000580] syscall_exit+0x0/0x2c Instruction dump: XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX 38600000 4e800020 60000000 60000000 <4e800020> 60000000 60000000 60000000 ---[ end trace 4f57124939a84dc8 ]--- This appears to be due to checking the wrong permission bits in the instruction_tlb_miss handling if the address that faulted was in vmalloc space. We need to look at the supervisor execute (_PAGE_BAP_SX) bit and not the user bit (_PAGE_BAP_UX/_PAGE_EXEC). Also removed a branch level since it did not appear to be used. Reported-by: Jeffrey Ladouceur <Jeffrey.Ladouceur@freescale.com> Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2010-11-17 15:20:32 +08:00
mfspr r10,SPRN_MAS1
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
cmpldi cr0,r15,8 /* Check for vmalloc region */
rlwinm r10,r10,0,16,1 /* Clear TID */
mtspr SPRN_MAS1,r10
beq+ normal_tlb_miss
/* We got a crappy address, just fault */
TLB_MISS_STATS_I(MMSTAT_TLB_MISS_NORM_FAULT)
TLB_MISS_EPILOG_ERROR
b exc_instruction_storage_book3e
/*
* This is the guts of the first-level TLB miss handler for direct
* misses. We are entered with:
*
* r16 = faulting address
* r15 = region ID
* r14 = crap (free to use)
* r13 = PACA
* r12 = TLB exception frame in PACA
* r11 = PTE permission mask
* r10 = crap (free to use)
*/
normal_tlb_miss:
/* So we first construct the page table address. We do that by
* shifting the bottom of the address (not the region ID) by
* PAGE_SHIFT-3, clearing the bottom 3 bits (get a PTE ptr) and
* or'ing the fourth high bit.
*
* NOTE: For 64K pages, we do things slightly differently in
* order to handle the weird page table format used by linux
*/
ori r10,r15,0x1
#ifdef CONFIG_PPC_64K_PAGES
/* For the top bits, 16 bytes per PTE */
rldicl r14,r16,64-(PAGE_SHIFT-4),PAGE_SHIFT-4+4
/* Now create the bottom bits as 0 in position 0x8000 and
* the rest calculated for 8 bytes per PTE
*/
rldicl r15,r16,64-(PAGE_SHIFT-3),64-15
/* Insert the bottom bits in */
rlwimi r14,r15,0,16,31
#else
rldicl r14,r16,64-(PAGE_SHIFT-3),PAGE_SHIFT-3+4
#endif
sldi r15,r10,60
clrrdi r14,r14,3
or r10,r15,r14
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
BEGIN_MMU_FTR_SECTION
Fix common misspellings Fixes generated by 'codespell' and manually reviewed. Signed-off-by: Lucas De Marchi <lucas.demarchi@profusion.mobi>
2011-03-31 09:57:33 +08:00
/* Set the TLB reservation and search for existing entry. Then load
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
* the entry.
*/
powerpc: Enforce usage of RA 0-R31 where possible Some macros use RA where when RA=R0 the values is 0, so make this the enforced mnemonic in the macro. Idea suggested by Andreas Schwab. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-06-25 21:33:25 +08:00
PPC_TLBSRX_DOT(0,R16)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
ld r14,0(r10)
beq normal_tlb_miss_done
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
MMU_FTR_SECTION_ELSE
ld r14,0(r10)
ALT_MMU_FTR_SECTION_END_IFSET(MMU_FTR_USE_TLBRSRV)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
finish_normal_tlb_miss:
/* Check if required permissions are met */
andc. r15,r11,r14
bne- normal_tlb_miss_access_fault
/* Now we build the MAS:
*
* MAS 0 : Fully setup with defaults in MAS4 and TLBnCFG
* MAS 1 : Almost fully setup
* - PID already updated by caller if necessary
* - TSIZE need change if !base page size, not
* yet implemented for now
* MAS 2 : Defaults not useful, need to be redone
* MAS 3+7 : Needs to be done
*
* TODO: mix up code below for better scheduling
*/
clrrdi r11,r16,12 /* Clear low crap in EA */
rlwimi r11,r14,32-19,27,31 /* Insert WIMGE */
mtspr SPRN_MAS2,r11
/* Check page size, if not standard, update MAS1 */
rldicl r11,r14,64-8,64-8
#ifdef CONFIG_PPC_64K_PAGES
cmpldi cr0,r11,BOOK3E_PAGESZ_64K
#else
cmpldi cr0,r11,BOOK3E_PAGESZ_4K
#endif
beq- 1f
mfspr r11,SPRN_MAS1
rlwimi r11,r14,31,21,24
rlwinm r11,r11,0,21,19
mtspr SPRN_MAS1,r11
1:
/* Move RPN in position */
rldicr r11,r14,64-(PTE_RPN_SHIFT-PAGE_SHIFT),63-PAGE_SHIFT
clrldi r15,r11,12 /* Clear crap at the top */
rlwimi r15,r14,32-8,22,25 /* Move in U bits */
rlwimi r15,r14,32-2,26,31 /* Move in BAP bits */
/* Mask out SW and UW if !DIRTY (XXX optimize this !) */
andi. r11,r14,_PAGE_DIRTY
bne 1f
li r11,MAS3_SW|MAS3_UW
andc r15,r15,r11
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
1:
BEGIN_MMU_FTR_SECTION
srdi r16,r15,32
mtspr SPRN_MAS3,r15
mtspr SPRN_MAS7,r16
MMU_FTR_SECTION_ELSE
mtspr SPRN_MAS7_MAS3,r15
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_USE_PAIRED_MAS)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
tlbwe
normal_tlb_miss_done:
/* We don't bother with restoring DEAR or ESR since we know we are
* level 0 and just going back to userland. They are only needed
* if you are going to take an access fault
*/
TLB_MISS_STATS_X(MMSTAT_TLB_MISS_NORM_OK)
TLB_MISS_EPILOG_SUCCESS
rfi
normal_tlb_miss_access_fault:
/* We need to check if it was an instruction miss */
powerpc/mm: Cleanup handling of execute permission This is an attempt at cleaning up a bit the way we handle execute permission on powerpc. _PAGE_HWEXEC is gone, _PAGE_EXEC is now only defined by CPUs that can do something with it, and the myriad of #ifdef's in the I$/D$ coherency code is reduced to 2 cases that hopefully should cover everything. The logic on BookE is a little bit different than what it was though not by much. Since now, _PAGE_EXEC will be set by the generic code for executable pages, we need to filter out if they are unclean and recover it. However, I don't expect the code to be more bloated than it already was in that area due to that change. I could boast that this brings proper enforcing of per-page execute permissions to all BookE and 40x but in fact, we've had that now for some time as a side effect of my previous rework in that area (and I didn't even know it :-) We would only enable execute permission if the page was cache clean and we would only cache clean it if we took and exec fault. Since we now enforce that the later only work if VM_EXEC is part of the VMA flags, we de-fact already enforce per-page execute permissions... Unless I missed something Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-19 03:00:34 +08:00
andi. r10,r11,_PAGE_EXEC
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
bne 1f
ld r14,EX_TLB_DEAR(r12)
ld r15,EX_TLB_ESR(r12)
mtspr SPRN_DEAR,r14
mtspr SPRN_ESR,r15
TLB_MISS_STATS_D(MMSTAT_TLB_MISS_NORM_FAULT)
TLB_MISS_EPILOG_ERROR
b exc_data_storage_book3e
1: TLB_MISS_STATS_I(MMSTAT_TLB_MISS_NORM_FAULT)
TLB_MISS_EPILOG_ERROR
b exc_instruction_storage_book3e
/*
* This is the guts of the second-level TLB miss handler for direct
* misses. We are entered with:
*
* r16 = virtual page table faulting address
* r15 = region (top 4 bits of address)
* r14 = crap (free to use)
* r13 = PACA
* r12 = TLB exception frame in PACA
* r11 = crap (free to use)
* r10 = crap (free to use)
*
* Note that this should only ever be called as a second level handler
* with the current scheme when using SW load.
* That means we can always get the original fault DEAR at
* EX_TLB_DEAR-EX_TLB_SIZE(r12)
*
* It can be re-entered by the linear mapping miss handler. However, to
* avoid too much complication, it will restart the whole fault at level
* 0 so we don't care too much about clobbers
*
* XXX That code was written back when we couldn't clobber r14. We can now,
* so we could probably optimize things a bit
*/
virt_page_table_tlb_miss:
/* Are we hitting a kernel page table ? */
andi. r10,r15,0x8
/* The cool thing now is that r10 contains 0 for user and 8 for kernel,
* and we happen to have the swapper_pg_dir at offset 8 from the user
* pgdir in the PACA :-).
*/
add r11,r10,r13
/* If kernel, we need to clear MAS1 TID */
beq 1f
/* XXX replace the RMW cycles with immediate loads + writes */
mfspr r10,SPRN_MAS1
rlwinm r10,r10,0,16,1 /* Clear TID */
mtspr SPRN_MAS1,r10
1:
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
BEGIN_MMU_FTR_SECTION
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
/* Search if we already have a TLB entry for that virtual address, and
* if we do, bail out.
*/
powerpc: Enforce usage of RA 0-R31 where possible Some macros use RA where when RA=R0 the values is 0, so make this the enforced mnemonic in the macro. Idea suggested by Andreas Schwab. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-06-25 21:33:25 +08:00
PPC_TLBSRX_DOT(0,R16)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
beq virt_page_table_tlb_miss_done
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
END_MMU_FTR_SECTION_IFSET(MMU_FTR_USE_TLBRSRV)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
/* Now, we need to walk the page tables. First check if we are in
* range.
*/
rldicl. r10,r16,64-(VPTE_INDEX_SIZE+3),VPTE_INDEX_SIZE+3+4
bne- virt_page_table_tlb_miss_fault
/* Get the PGD pointer */
ld r15,PACAPGD(r11)
cmpldi cr0,r15,0
beq- virt_page_table_tlb_miss_fault
/* Get to PGD entry */
rldicl r11,r16,64-VPTE_PGD_SHIFT,64-PGD_INDEX_SIZE-3
clrrdi r10,r11,3
ldx r15,r10,r15
powerpc: Hugetlb for BookE Enable hugepages on Freescale BookE processors. This allows the kernel to use huge TLB entries to map pages, which can greatly reduce the number of TLB misses and the amount of TLB thrashing experienced by applications with large memory footprints. Care should be taken when using this on FSL processors, as the number of large TLB entries supported by the core is low (16-64) on current processors. The supported set of hugepage sizes include 4m, 16m, 64m, 256m, and 1g. Page sizes larger than the max zone size are called "gigantic" pages and must be allocated on the command line (and cannot be deallocated). This is currently only fully implemented for Freescale 32-bit BookE processors, but there is some infrastructure in the code for 64-bit BooKE. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-28 17:54:48 +08:00
cmpdi cr0,r15,0
bge virt_page_table_tlb_miss_fault
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
#ifndef CONFIG_PPC_64K_PAGES
/* Get to PUD entry */
rldicl r11,r16,64-VPTE_PUD_SHIFT,64-PUD_INDEX_SIZE-3
clrrdi r10,r11,3
ldx r15,r10,r15
powerpc: Hugetlb for BookE Enable hugepages on Freescale BookE processors. This allows the kernel to use huge TLB entries to map pages, which can greatly reduce the number of TLB misses and the amount of TLB thrashing experienced by applications with large memory footprints. Care should be taken when using this on FSL processors, as the number of large TLB entries supported by the core is low (16-64) on current processors. The supported set of hugepage sizes include 4m, 16m, 64m, 256m, and 1g. Page sizes larger than the max zone size are called "gigantic" pages and must be allocated on the command line (and cannot be deallocated). This is currently only fully implemented for Freescale 32-bit BookE processors, but there is some infrastructure in the code for 64-bit BooKE. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-28 17:54:48 +08:00
cmpdi cr0,r15,0
bge virt_page_table_tlb_miss_fault
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
#endif /* CONFIG_PPC_64K_PAGES */
/* Get to PMD entry */
rldicl r11,r16,64-VPTE_PMD_SHIFT,64-PMD_INDEX_SIZE-3
clrrdi r10,r11,3
ldx r15,r10,r15
powerpc: Hugetlb for BookE Enable hugepages on Freescale BookE processors. This allows the kernel to use huge TLB entries to map pages, which can greatly reduce the number of TLB misses and the amount of TLB thrashing experienced by applications with large memory footprints. Care should be taken when using this on FSL processors, as the number of large TLB entries supported by the core is low (16-64) on current processors. The supported set of hugepage sizes include 4m, 16m, 64m, 256m, and 1g. Page sizes larger than the max zone size are called "gigantic" pages and must be allocated on the command line (and cannot be deallocated). This is currently only fully implemented for Freescale 32-bit BookE processors, but there is some infrastructure in the code for 64-bit BooKE. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-28 17:54:48 +08:00
cmpdi cr0,r15,0
bge virt_page_table_tlb_miss_fault
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
/* Ok, we're all right, we can now create a kernel translation for
* a 4K or 64K page from r16 -> r15.
*/
/* Now we build the MAS:
*
* MAS 0 : Fully setup with defaults in MAS4 and TLBnCFG
* MAS 1 : Almost fully setup
* - PID already updated by caller if necessary
* - TSIZE for now is base page size always
* MAS 2 : Use defaults
* MAS 3+7 : Needs to be done
*
* So we only do MAS 2 and 3 for now...
*/
clrldi r11,r15,4 /* remove region ID from RPN */
ori r10,r11,1 /* Or-in SR */
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
BEGIN_MMU_FTR_SECTION
srdi r16,r10,32
mtspr SPRN_MAS3,r10
mtspr SPRN_MAS7,r16
MMU_FTR_SECTION_ELSE
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
mtspr SPRN_MAS7_MAS3,r10
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_USE_PAIRED_MAS)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
tlbwe
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
BEGIN_MMU_FTR_SECTION
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
virt_page_table_tlb_miss_done:
/* We have overriden MAS2:EPN but currently our primary TLB miss
* handler will always restore it so that should not be an issue,
* if we ever optimize the primary handler to not write MAS2 on
* some cases, we'll have to restore MAS2:EPN here based on the
* original fault's DEAR. If we do that we have to modify the
* ITLB miss handler to also store SRR0 in the exception frame
* as DEAR.
*
* However, one nasty thing we did is we cleared the reservation
* (well, potentially we did). We do a trick here thus if we
* are not a level 0 exception (we interrupted the TLB miss) we
* offset the return address by -4 in order to replay the tlbsrx
* instruction there
*/
subf r10,r13,r12
cmpldi cr0,r10,PACA_EXTLB+EX_TLB_SIZE
bne- 1f
ld r11,PACA_EXTLB+EX_TLB_SIZE+EX_TLB_SRR0(r13)
addi r10,r11,-4
std r10,PACA_EXTLB+EX_TLB_SIZE+EX_TLB_SRR0(r13)
1:
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
END_MMU_FTR_SECTION_IFSET(MMU_FTR_USE_TLBRSRV)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
/* Return to caller, normal case */
TLB_MISS_STATS_X(MMSTAT_TLB_MISS_PT_OK);
TLB_MISS_EPILOG_SUCCESS
rfi
virt_page_table_tlb_miss_fault:
/* If we fault here, things are a little bit tricky. We need to call
Fix common misspellings Fixes generated by 'codespell' and manually reviewed. Signed-off-by: Lucas De Marchi <lucas.demarchi@profusion.mobi>
2011-03-31 09:57:33 +08:00
* either data or instruction store fault, and we need to retrieve
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
* the original fault address and ESR (for data).
*
* The thing is, we know that in normal circumstances, this is
* always called as a second level tlb miss for SW load or as a first
* level TLB miss for HW load, so we should be able to peek at the
Documentation: fix minor typos/spelling Fix some minor typos: * informations => information * there own => their own * these => this Signed-off-by: Sylvestre Ledru <sylvestre.ledru@scilab.org> Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-04-05 06:04:46 +08:00
* relevant information in the first exception frame in the PACA.
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
*
* However, we do need to double check that, because we may just hit
* a stray kernel pointer or a userland attack trying to hit those
* areas. If that is the case, we do a data fault. (We can't get here
* from an instruction tlb miss anyway).
*
* Note also that when going to a fault, we must unwind the previous
* level as well. Since we are doing that, we don't need to clear or
* restore the TLB reservation neither.
*/
subf r10,r13,r12
cmpldi cr0,r10,PACA_EXTLB+EX_TLB_SIZE
bne- virt_page_table_tlb_miss_whacko_fault
/* We dig the original DEAR and ESR from slot 0 */
ld r15,EX_TLB_DEAR+PACA_EXTLB(r13)
ld r16,EX_TLB_ESR+PACA_EXTLB(r13)
/* We check for the "special" ESR value for instruction faults */
cmpdi cr0,r16,-1
beq 1f
mtspr SPRN_DEAR,r15
mtspr SPRN_ESR,r16
TLB_MISS_STATS_D(MMSTAT_TLB_MISS_PT_FAULT);
TLB_MISS_EPILOG_ERROR
b exc_data_storage_book3e
1: TLB_MISS_STATS_I(MMSTAT_TLB_MISS_PT_FAULT);
TLB_MISS_EPILOG_ERROR
b exc_instruction_storage_book3e
virt_page_table_tlb_miss_whacko_fault:
/* The linear fault will restart everything so ESR and DEAR will
* not have been clobbered, let's just fault with what we have
*/
TLB_MISS_STATS_X(MMSTAT_TLB_MISS_PT_FAULT);
TLB_MISS_EPILOG_ERROR
b exc_data_storage_book3e
/**************************************************************
* *
* TLB miss handling for Book3E with hw page table support *
* *
**************************************************************/
/* Data TLB miss */
START_EXCEPTION(data_tlb_miss_htw)
TLB_MISS_PROLOG
/* Now we handle the fault proper. We only save DEAR in normal
* fault case since that's the only interesting values here.
* We could probably also optimize by not saving SRR0/1 in the
* linear mapping case but I'll leave that for later
*/
mfspr r14,SPRN_ESR
mfspr r16,SPRN_DEAR /* get faulting address */
srdi r11,r16,60 /* get region */
cmpldi cr0,r11,0xc /* linear mapping ? */
TLB_MISS_STATS_SAVE_INFO
beq tlb_load_linear /* yes -> go to linear map load */
/* We do the user/kernel test for the PID here along with the RW test
*/
cmpldi cr0,r11,0 /* Check for user region */
ld r15,PACAPGD(r13) /* Load user pgdir */
beq htw_tlb_miss
/* XXX replace the RMW cycles with immediate loads + writes */
1: mfspr r10,SPRN_MAS1
cmpldi cr0,r11,8 /* Check for vmalloc region */
rlwinm r10,r10,0,16,1 /* Clear TID */
mtspr SPRN_MAS1,r10
ld r15,PACA_KERNELPGD(r13) /* Load kernel pgdir */
beq+ htw_tlb_miss
/* We got a crappy address, just fault with whatever DEAR and ESR
* are here
*/
TLB_MISS_STATS_D(MMSTAT_TLB_MISS_NORM_FAULT)
TLB_MISS_EPILOG_ERROR
b exc_data_storage_book3e
/* Instruction TLB miss */
START_EXCEPTION(instruction_tlb_miss_htw)
TLB_MISS_PROLOG
/* If we take a recursive fault, the second level handler may need
* to know whether we are handling a data or instruction fault in
* order to get to the right store fault handler. We provide that
* info by keeping a crazy value for ESR in r14
*/
li r14,-1 /* store to exception frame is done later */
/* Now we handle the fault proper. We only save DEAR in the non
* linear mapping case since we know the linear mapping case will
* not re-enter. We could indeed optimize and also not save SRR0/1
* in the linear mapping case but I'll leave that for later
*
* Faulting address is SRR0 which is already in r16
*/
srdi r11,r16,60 /* get region */
cmpldi cr0,r11,0xc /* linear mapping ? */
TLB_MISS_STATS_SAVE_INFO
beq tlb_load_linear /* yes -> go to linear map load */
/* We do the user/kernel test for the PID here along with the RW test
*/
cmpldi cr0,r11,0 /* Check for user region */
ld r15,PACAPGD(r13) /* Load user pgdir */
beq htw_tlb_miss
/* XXX replace the RMW cycles with immediate loads + writes */
1: mfspr r10,SPRN_MAS1
cmpldi cr0,r11,8 /* Check for vmalloc region */
rlwinm r10,r10,0,16,1 /* Clear TID */
mtspr SPRN_MAS1,r10
ld r15,PACA_KERNELPGD(r13) /* Load kernel pgdir */
beq+ htw_tlb_miss
/* We got a crappy address, just fault */
TLB_MISS_STATS_I(MMSTAT_TLB_MISS_NORM_FAULT)
TLB_MISS_EPILOG_ERROR
b exc_instruction_storage_book3e
/*
* This is the guts of the second-level TLB miss handler for direct
* misses. We are entered with:
*
* r16 = virtual page table faulting address
* r15 = PGD pointer
* r14 = ESR
* r13 = PACA
* r12 = TLB exception frame in PACA
* r11 = crap (free to use)
* r10 = crap (free to use)
*
* It can be re-entered by the linear mapping miss handler. However, to
* avoid too much complication, it will save/restore things for us
*/
htw_tlb_miss:
/* Search if we already have a TLB entry for that virtual address, and
* if we do, bail out.
*
* MAS1:IND should be already set based on MAS4
*/
powerpc: Enforce usage of RA 0-R31 where possible Some macros use RA where when RA=R0 the values is 0, so make this the enforced mnemonic in the macro. Idea suggested by Andreas Schwab. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-06-25 21:33:25 +08:00
PPC_TLBSRX_DOT(0,R16)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
beq htw_tlb_miss_done
/* Now, we need to walk the page tables. First check if we are in
* range.
*/
rldicl. r10,r16,64-PGTABLE_EADDR_SIZE,PGTABLE_EADDR_SIZE+4
bne- htw_tlb_miss_fault
/* Get the PGD pointer */
cmpldi cr0,r15,0
beq- htw_tlb_miss_fault
/* Get to PGD entry */
rldicl r11,r16,64-(PGDIR_SHIFT-3),64-PGD_INDEX_SIZE-3
clrrdi r10,r11,3
ldx r15,r10,r15
powerpc: Hugetlb for BookE Enable hugepages on Freescale BookE processors. This allows the kernel to use huge TLB entries to map pages, which can greatly reduce the number of TLB misses and the amount of TLB thrashing experienced by applications with large memory footprints. Care should be taken when using this on FSL processors, as the number of large TLB entries supported by the core is low (16-64) on current processors. The supported set of hugepage sizes include 4m, 16m, 64m, 256m, and 1g. Page sizes larger than the max zone size are called "gigantic" pages and must be allocated on the command line (and cannot be deallocated). This is currently only fully implemented for Freescale 32-bit BookE processors, but there is some infrastructure in the code for 64-bit BooKE. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-28 17:54:48 +08:00
cmpdi cr0,r15,0
bge htw_tlb_miss_fault
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
#ifndef CONFIG_PPC_64K_PAGES
/* Get to PUD entry */
rldicl r11,r16,64-(PUD_SHIFT-3),64-PUD_INDEX_SIZE-3
clrrdi r10,r11,3
ldx r15,r10,r15
powerpc: Hugetlb for BookE Enable hugepages on Freescale BookE processors. This allows the kernel to use huge TLB entries to map pages, which can greatly reduce the number of TLB misses and the amount of TLB thrashing experienced by applications with large memory footprints. Care should be taken when using this on FSL processors, as the number of large TLB entries supported by the core is low (16-64) on current processors. The supported set of hugepage sizes include 4m, 16m, 64m, 256m, and 1g. Page sizes larger than the max zone size are called "gigantic" pages and must be allocated on the command line (and cannot be deallocated). This is currently only fully implemented for Freescale 32-bit BookE processors, but there is some infrastructure in the code for 64-bit BooKE. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-28 17:54:48 +08:00
cmpdi cr0,r15,0
bge htw_tlb_miss_fault
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
#endif /* CONFIG_PPC_64K_PAGES */
/* Get to PMD entry */
rldicl r11,r16,64-(PMD_SHIFT-3),64-PMD_INDEX_SIZE-3
clrrdi r10,r11,3
ldx r15,r10,r15
powerpc: Hugetlb for BookE Enable hugepages on Freescale BookE processors. This allows the kernel to use huge TLB entries to map pages, which can greatly reduce the number of TLB misses and the amount of TLB thrashing experienced by applications with large memory footprints. Care should be taken when using this on FSL processors, as the number of large TLB entries supported by the core is low (16-64) on current processors. The supported set of hugepage sizes include 4m, 16m, 64m, 256m, and 1g. Page sizes larger than the max zone size are called "gigantic" pages and must be allocated on the command line (and cannot be deallocated). This is currently only fully implemented for Freescale 32-bit BookE processors, but there is some infrastructure in the code for 64-bit BooKE. Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org> Signed-off-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-06-28 17:54:48 +08:00
cmpdi cr0,r15,0
bge htw_tlb_miss_fault
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
/* Ok, we're all right, we can now create an indirect entry for
* a 1M or 256M page.
*
* The last trick is now that because we use "half" pages for
* the HTW (1M IND is 2K and 256M IND is 32K) we need to account
* for an added LSB bit to the RPN. For 64K pages, there is no
* problem as we already use 32K arrays (half PTE pages), but for
* 4K page we need to extract a bit from the virtual address and
* insert it into the "PA52" bit of the RPN.
*/
#ifndef CONFIG_PPC_64K_PAGES
rlwimi r15,r16,32-9,20,20
#endif
/* Now we build the MAS:
*
* MAS 0 : Fully setup with defaults in MAS4 and TLBnCFG
* MAS 1 : Almost fully setup
* - PID already updated by caller if necessary
* - TSIZE for now is base ind page size always
* MAS 2 : Use defaults
* MAS 3+7 : Needs to be done
*/
#ifdef CONFIG_PPC_64K_PAGES
ori r10,r15,(BOOK3E_PAGESZ_64K << MAS3_SPSIZE_SHIFT)
#else
ori r10,r15,(BOOK3E_PAGESZ_4K << MAS3_SPSIZE_SHIFT)
#endif
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
BEGIN_MMU_FTR_SECTION
srdi r16,r10,32
mtspr SPRN_MAS3,r10
mtspr SPRN_MAS7,r16
MMU_FTR_SECTION_ELSE
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
mtspr SPRN_MAS7_MAS3,r10
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_USE_PAIRED_MAS)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
tlbwe
htw_tlb_miss_done:
/* We don't bother with restoring DEAR or ESR since we know we are
* level 0 and just going back to userland. They are only needed
* if you are going to take an access fault
*/
TLB_MISS_STATS_X(MMSTAT_TLB_MISS_PT_OK)
TLB_MISS_EPILOG_SUCCESS
rfi
htw_tlb_miss_fault:
/* We need to check if it was an instruction miss. We know this
* though because r14 would contain -1
*/
cmpdi cr0,r14,-1
beq 1f
mtspr SPRN_DEAR,r16
mtspr SPRN_ESR,r14
TLB_MISS_STATS_D(MMSTAT_TLB_MISS_PT_FAULT)
TLB_MISS_EPILOG_ERROR
b exc_data_storage_book3e
1: TLB_MISS_STATS_I(MMSTAT_TLB_MISS_PT_FAULT)
TLB_MISS_EPILOG_ERROR
b exc_instruction_storage_book3e
/*
* This is the guts of "any" level TLB miss handler for kernel linear
* mapping misses. We are entered with:
*
*
* r16 = faulting address
* r15 = crap (free to use)
* r14 = ESR (data) or -1 (instruction)
* r13 = PACA
* r12 = TLB exception frame in PACA
* r11 = crap (free to use)
* r10 = crap (free to use)
*
* In addition we know that we will not re-enter, so in theory, we could
* use a simpler epilog not restoring SRR0/1 etc.. but we'll do that later.
*
* We also need to be careful about MAS registers here & TLB reservation,
* as we know we'll have clobbered them if we interrupt the main TLB miss
* handlers in which case we probably want to do a full restart at level
* 0 rather than saving / restoring the MAS.
*
* Note: If we care about performance of that core, we can easily shuffle
* a few things around
*/
tlb_load_linear:
/* For now, we assume the linear mapping is contiguous and stops at
* linear_map_top. We also assume the size is a multiple of 1G, thus
* we only use 1G pages for now. That might have to be changed in a
* final implementation, especially when dealing with hypervisors
*/
ld r11,PACATOC(r13)
ld r11,linear_map_top@got(r11)
ld r10,0(r11)
powerpc/book3e: Fix check for linear mapping in TLB miss handler The previous code added wrong TLBs and causes machine check errors if a driver accessed passed the end of the linear mapping instead of a clean page fault. Signed-off-by: Ralph E. Bellofatto <ralphbel@us.ibm.com> Signed-off-by: Benjamin Krill <ben@codiert.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2014-02-27 22:49:21 +08:00
tovirt(10,10)
cmpld cr0,r16,r10
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
bge tlb_load_linear_fault
/* MAS1 need whole new setup. */
li r15,(BOOK3E_PAGESZ_1GB<<MAS1_TSIZE_SHIFT)
oris r15,r15,MAS1_VALID@h /* MAS1 needs V and TSIZE */
mtspr SPRN_MAS1,r15
/* Already somebody there ? */
powerpc: Enforce usage of RA 0-R31 where possible Some macros use RA where when RA=R0 the values is 0, so make this the enforced mnemonic in the macro. Idea suggested by Andreas Schwab. Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-06-25 21:33:25 +08:00
PPC_TLBSRX_DOT(0,R16)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
beq tlb_load_linear_done
/* Now we build the remaining MAS. MAS0 and 2 should be fine
* with their defaults, which leaves us with MAS 3 and 7. The
* mapping is linear, so we just take the address, clear the
* region bits, and or in the permission bits which are currently
* hard wired
*/
clrrdi r10,r16,30 /* 1G page index */
clrldi r10,r10,4 /* clear region bits */
ori r10,r10,MAS3_SR|MAS3_SW|MAS3_SX
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
BEGIN_MMU_FTR_SECTION
srdi r16,r10,32
mtspr SPRN_MAS3,r10
mtspr SPRN_MAS7,r16
MMU_FTR_SECTION_ELSE
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
mtspr SPRN_MAS7_MAS3,r10
powerpc/mm: Add MMU features for TLB reservation & Paired MAS registers Support for TLB reservation (or TLB Write Conditional) and Paired MAS registers are optional for a processor implementation so we handle them via MMU feature sections. We currently only used paired MAS registers to access the full RPN + perm bits that are kept in MAS7||MAS3. We assume that if an implementation has hardware page table at this time it also implements in TLB reservations. Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-08-24 23:52:48 +08:00
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_USE_PAIRED_MAS)
powerpc: Add TLB management code for 64-bit Book3E This adds the TLB miss handler assembly, the low level TLB flush routines along with the necessary hook for dealing with our virtual page tables or indirect TLB entries that need to be flushes when PTE pages are freed. There is currently no support for hugetlbfs Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-07-24 07:15:47 +08:00
tlbwe
tlb_load_linear_done:
/* We use the "error" epilog for success as we do want to
* restore to the initial faulting context, whatever it was.
* We do that because we can't resume a fault within a TLB
* miss handler, due to MAS and TLB reservation being clobbered.
*/
TLB_MISS_STATS_X(MMSTAT_TLB_MISS_LINEAR)
TLB_MISS_EPILOG_ERROR
rfi
tlb_load_linear_fault:
/* We keep the DEAR and ESR around, this shouldn't have happened */
cmpdi cr0,r14,-1
beq 1f
TLB_MISS_EPILOG_ERROR_SPECIAL
b exc_data_storage_book3e
1: TLB_MISS_EPILOG_ERROR_SPECIAL
b exc_instruction_storage_book3e
#ifdef CONFIG_BOOK3E_MMU_TLB_STATS
.tlb_stat_inc:
1: ldarx r8,0,r9
addi r8,r8,1
stdcx. r8,0,r9
bne- 1b
blr
#endif