kernel_optimize_test/arch/powerpc/mm/tlb_nohash_low.S
Kumar Gala 55fd766b5f powerpc/fsl-booke64: Use TLB CAMs to cover linear mapping on FSL 64-bit chips
On Freescale parts typically have TLB array for large mappings that we can
bolt the linear mapping into.  We utilize the code that already exists
on PPC32 on the 64-bit side to setup the linear mapping to be cover by
bolted TLB entries.  We utilize a quarter of the variable size TLB array
for this purpose.

Additionally, we limit the amount of memory to what we can cover via
bolted entries so we don't get secondary faults in the TLB miss
handlers.  We should fix this limitation in the future.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
2010-10-14 00:55:14 -05:00

397 lines
8.6 KiB
ArmAsm

/*
* This file contains low-level functions for performing various
* types of TLB invalidations on various processors with no hash
* table.
*
* This file implements the following functions for all no-hash
* processors. Some aren't implemented for some variants. Some
* are inline in tlbflush.h
*
* - tlbil_va
* - tlbil_pid
* - tlbil_all
* - tlbivax_bcast
*
* Code mostly moved over from misc_32.S
*
* Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
*
* Partially rewritten by Cort Dougan (cort@cs.nmt.edu)
* Paul Mackerras, Kumar Gala and Benjamin Herrenschmidt.
*
* 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/reg.h>
#include <asm/page.h>
#include <asm/cputable.h>
#include <asm/mmu.h>
#include <asm/ppc_asm.h>
#include <asm/asm-offsets.h>
#include <asm/processor.h>
#include <asm/bug.h>
#if defined(CONFIG_40x)
/*
* 40x implementation needs only tlbil_va
*/
_GLOBAL(__tlbil_va)
/* We run the search with interrupts disabled because we have to change
* the PID and I don't want to preempt when that happens.
*/
mfmsr r5
mfspr r6,SPRN_PID
wrteei 0
mtspr SPRN_PID,r4
tlbsx. r3, 0, r3
mtspr SPRN_PID,r6
wrtee r5
bne 1f
sync
/* There are only 64 TLB entries, so r3 < 64, which means bit 25 is
* clear. Since 25 is the V bit in the TLB_TAG, loading this value
* will invalidate the TLB entry. */
tlbwe r3, r3, TLB_TAG
isync
1: blr
#elif defined(CONFIG_8xx)
/*
* Nothing to do for 8xx, everything is inline
*/
#elif defined(CONFIG_44x) /* Includes 47x */
/*
* 440 implementation uses tlbsx/we for tlbil_va and a full sweep
* of the TLB for everything else.
*/
_GLOBAL(__tlbil_va)
mfspr r5,SPRN_MMUCR
mfmsr r10
/*
* We write 16 bits of STID since 47x supports that much, we
* will never be passed out of bounds values on 440 (hopefully)
*/
rlwimi r5,r4,0,16,31
/* We have to run the search with interrupts disabled, otherwise
* an interrupt which causes a TLB miss can clobber the MMUCR
* between the mtspr and the tlbsx.
*
* Critical and Machine Check interrupts take care of saving
* and restoring MMUCR, so only normal interrupts have to be
* taken care of.
*/
wrteei 0
mtspr SPRN_MMUCR,r5
tlbsx. r6,0,r3
bne 10f
sync
BEGIN_MMU_FTR_SECTION
b 2f
END_MMU_FTR_SECTION_IFSET(MMU_FTR_TYPE_47x)
/* On 440 There are only 64 TLB entries, so r3 < 64, which means bit
* 22, is clear. Since 22 is the V bit in the TLB_PAGEID, loading this
* value will invalidate the TLB entry.
*/
tlbwe r6,r6,PPC44x_TLB_PAGEID
isync
10: wrtee r10
blr
2:
#ifdef CONFIG_PPC_47x
oris r7,r6,0x8000 /* specify way explicitely */
clrrwi r4,r3,12 /* get an EPN for the hashing with V = 0 */
ori r4,r4,PPC47x_TLBE_SIZE
tlbwe r4,r7,0 /* write it */
isync
wrtee r10
blr
#else /* CONFIG_PPC_47x */
1: trap
EMIT_BUG_ENTRY 1b,__FILE__,__LINE__,0;
#endif /* !CONFIG_PPC_47x */
_GLOBAL(_tlbil_all)
_GLOBAL(_tlbil_pid)
BEGIN_MMU_FTR_SECTION
b 2f
END_MMU_FTR_SECTION_IFSET(MMU_FTR_TYPE_47x)
li r3,0
sync
/* Load high watermark */
lis r4,tlb_44x_hwater@ha
lwz r5,tlb_44x_hwater@l(r4)
1: tlbwe r3,r3,PPC44x_TLB_PAGEID
addi r3,r3,1
cmpw 0,r3,r5
ble 1b
isync
blr
2:
#ifdef CONFIG_PPC_47x
/* 476 variant. There's not simple way to do this, hopefully we'll
* try to limit the amount of such full invalidates
*/
mfmsr r11 /* Interrupts off */
wrteei 0
li r3,-1 /* Current set */
lis r10,tlb_47x_boltmap@h
ori r10,r10,tlb_47x_boltmap@l
lis r7,0x8000 /* Specify way explicitely */
b 9f /* For each set */
1: li r9,4 /* Number of ways */
li r4,0 /* Current way */
li r6,0 /* Default entry value 0 */
andi. r0,r8,1 /* Check if way 0 is bolted */
mtctr r9 /* Load way counter */
bne- 3f /* Bolted, skip loading it */
2: /* For each way */
or r5,r3,r4 /* Make way|index for tlbre */
rlwimi r5,r5,16,8,15 /* Copy index into position */
tlbre r6,r5,0 /* Read entry */
3: addis r4,r4,0x2000 /* Next way */
andi. r0,r6,PPC47x_TLB0_VALID /* Valid entry ? */
beq 4f /* Nope, skip it */
rlwimi r7,r5,0,1,2 /* Insert way number */
rlwinm r6,r6,0,21,19 /* Clear V */
tlbwe r6,r7,0 /* Write it */
4: bdnz 2b /* Loop for each way */
srwi r8,r8,1 /* Next boltmap bit */
9: cmpwi cr1,r3,255 /* Last set done ? */
addi r3,r3,1 /* Next set */
beq cr1,1f /* End of loop */
andi. r0,r3,0x1f /* Need to load a new boltmap word ? */
bne 1b /* No, loop */
lwz r8,0(r10) /* Load boltmap entry */
addi r10,r10,4 /* Next word */
b 1b /* Then loop */
1: isync /* Sync shadows */
wrtee r11
#else /* CONFIG_PPC_47x */
1: trap
EMIT_BUG_ENTRY 1b,__FILE__,__LINE__,0;
#endif /* !CONFIG_PPC_47x */
blr
#ifdef CONFIG_PPC_47x
/*
* _tlbivax_bcast is only on 47x. We don't bother doing a runtime
* check though, it will blow up soon enough if we mistakenly try
* to use it on a 440.
*/
_GLOBAL(_tlbivax_bcast)
mfspr r5,SPRN_MMUCR
mfmsr r10
rlwimi r5,r4,0,16,31
wrteei 0
mtspr SPRN_MMUCR,r5
isync
/* tlbivax 0,r3 - use .long to avoid binutils deps */
.long 0x7c000624 | (r3 << 11)
isync
eieio
tlbsync
sync
wrtee r10
blr
#endif /* CONFIG_PPC_47x */
#elif defined(CONFIG_FSL_BOOKE)
/*
* FSL BookE implementations.
*
* Since feature sections are using _SECTION_ELSE we need
* to have the larger code path before the _SECTION_ELSE
*/
/*
* Flush MMU TLB on the local processor
*/
_GLOBAL(_tlbil_all)
BEGIN_MMU_FTR_SECTION
li r3,(MMUCSR0_TLBFI)@l
mtspr SPRN_MMUCSR0, r3
1:
mfspr r3,SPRN_MMUCSR0
andi. r3,r3,MMUCSR0_TLBFI@l
bne 1b
MMU_FTR_SECTION_ELSE
PPC_TLBILX_ALL(0,0)
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_USE_TLBILX)
msync
isync
blr
_GLOBAL(_tlbil_pid)
BEGIN_MMU_FTR_SECTION
slwi r3,r3,16
mfmsr r10
wrteei 0
mfspr r4,SPRN_MAS6 /* save MAS6 */
mtspr SPRN_MAS6,r3
PPC_TLBILX_PID(0,0)
mtspr SPRN_MAS6,r4 /* restore MAS6 */
wrtee r10
MMU_FTR_SECTION_ELSE
li r3,(MMUCSR0_TLBFI)@l
mtspr SPRN_MMUCSR0, r3
1:
mfspr r3,SPRN_MMUCSR0
andi. r3,r3,MMUCSR0_TLBFI@l
bne 1b
ALT_MMU_FTR_SECTION_END_IFSET(MMU_FTR_USE_TLBILX)
msync
isync
blr
/*
* Flush MMU TLB for a particular address, but only on the local processor
* (no broadcast)
*/
_GLOBAL(__tlbil_va)
mfmsr r10
wrteei 0
slwi r4,r4,16
ori r4,r4,(MAS6_ISIZE(BOOK3E_PAGESZ_4K))@l
mtspr SPRN_MAS6,r4 /* assume AS=0 for now */
BEGIN_MMU_FTR_SECTION
tlbsx 0,r3
mfspr r4,SPRN_MAS1 /* check valid */
andis. r3,r4,MAS1_VALID@h
beq 1f
rlwinm r4,r4,0,1,31
mtspr SPRN_MAS1,r4
tlbwe
MMU_FTR_SECTION_ELSE
PPC_TLBILX_VA(0,r3)
ALT_MMU_FTR_SECTION_END_IFCLR(MMU_FTR_USE_TLBILX)
msync
isync
1: wrtee r10
blr
#elif defined(CONFIG_PPC_BOOK3E)
/*
* New Book3E (>= 2.06) implementation
*
* Note: We may be able to get away without the interrupt masking stuff
* if we save/restore MAS6 on exceptions that might modify it
*/
_GLOBAL(_tlbil_pid)
slwi r4,r3,MAS6_SPID_SHIFT
mfmsr r10
wrteei 0
mtspr SPRN_MAS6,r4
PPC_TLBILX_PID(0,0)
wrtee r10
msync
isync
blr
_GLOBAL(_tlbil_pid_noind)
slwi r4,r3,MAS6_SPID_SHIFT
mfmsr r10
ori r4,r4,MAS6_SIND
wrteei 0
mtspr SPRN_MAS6,r4
PPC_TLBILX_PID(0,0)
wrtee r10
msync
isync
blr
_GLOBAL(_tlbil_all)
PPC_TLBILX_ALL(0,0)
msync
isync
blr
_GLOBAL(_tlbil_va)
mfmsr r10
wrteei 0
cmpwi cr0,r6,0
slwi r4,r4,MAS6_SPID_SHIFT
rlwimi r4,r5,MAS6_ISIZE_SHIFT,MAS6_ISIZE_MASK
beq 1f
rlwimi r4,r6,MAS6_SIND_SHIFT,MAS6_SIND
1: mtspr SPRN_MAS6,r4 /* assume AS=0 for now */
PPC_TLBILX_VA(0,r3)
msync
isync
wrtee r10
blr
_GLOBAL(_tlbivax_bcast)
mfmsr r10
wrteei 0
cmpwi cr0,r6,0
slwi r4,r4,MAS6_SPID_SHIFT
rlwimi r4,r5,MAS6_ISIZE_SHIFT,MAS6_ISIZE_MASK
beq 1f
rlwimi r4,r6,MAS6_SIND_SHIFT,MAS6_SIND
1: mtspr SPRN_MAS6,r4 /* assume AS=0 for now */
PPC_TLBIVAX(0,r3)
eieio
tlbsync
sync
wrtee r10
blr
_GLOBAL(set_context)
#ifdef CONFIG_BDI_SWITCH
/* Context switch the PTE pointer for the Abatron BDI2000.
* The PGDIR is the second parameter.
*/
lis r5, abatron_pteptrs@h
ori r5, r5, abatron_pteptrs@l
stw r4, 0x4(r5)
#endif
mtspr SPRN_PID,r3
isync /* Force context change */
blr
#else
#error Unsupported processor type !
#endif
#if defined(CONFIG_PPC_FSL_BOOK3E)
/*
* extern void loadcam_entry(unsigned int index)
*
* Load TLBCAM[index] entry in to the L2 CAM MMU
*/
_GLOBAL(loadcam_entry)
LOAD_REG_ADDR(r4, TLBCAM)
mulli r5,r3,TLBCAM_SIZE
add r3,r5,r4
lwz r4,TLBCAM_MAS0(r3)
mtspr SPRN_MAS0,r4
lwz r4,TLBCAM_MAS1(r3)
mtspr SPRN_MAS1,r4
PPC_LL r4,TLBCAM_MAS2(r3)
mtspr SPRN_MAS2,r4
lwz r4,TLBCAM_MAS3(r3)
mtspr SPRN_MAS3,r4
BEGIN_MMU_FTR_SECTION
lwz r4,TLBCAM_MAS7(r3)
mtspr SPRN_MAS7,r4
END_MMU_FTR_SECTION_IFSET(MMU_FTR_BIG_PHYS)
isync
tlbwe
isync
blr
#endif