linux/arch/powerpc/kernel/head_book3s_32.S

1358 lines
35 KiB
ArmAsm
Raw Normal View History

/* SPDX-License-Identifier: GPL-2.0-or-later */
/*
* PowerPC version
* Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
*
* Rewritten by Cort Dougan (cort@cs.nmt.edu) for PReP
* Copyright (C) 1996 Cort Dougan <cort@cs.nmt.edu>
* Adapted for Power Macintosh by Paul Mackerras.
* Low-level exception handlers and MMU support
* rewritten by Paul Mackerras.
* Copyright (C) 1996 Paul Mackerras.
* MPC8xx modifications Copyright (C) 1997 Dan Malek (dmalek@jlc.net).
*
* This file contains the low-level support and setup for the
* PowerPC platform, including trap and interrupt dispatch.
* (The PPC 8xx embedded CPUs use head_8xx.S instead.)
*/
#include <linux/init.h>
mm: reorder includes after introduction of linux/pgtable.h The replacement of <asm/pgrable.h> with <linux/pgtable.h> made the include of the latter in the middle of asm includes. Fix this up with the aid of the below script and manual adjustments here and there. import sys import re if len(sys.argv) is not 3: print "USAGE: %s <file> <header>" % (sys.argv[0]) sys.exit(1) hdr_to_move="#include <linux/%s>" % sys.argv[2] moved = False in_hdrs = False with open(sys.argv[1], "r") as f: lines = f.readlines() for _line in lines: line = _line.rstrip(' ') if line == hdr_to_move: continue if line.startswith("#include <linux/"): in_hdrs = True elif not moved and in_hdrs: moved = True print hdr_to_move print line Signed-off-by: Mike Rapoport <rppt@linux.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Borislav Petkov <bp@alien8.de> Cc: Brian Cain <bcain@codeaurora.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chris Zankel <chris@zankel.net> Cc: "David S. Miller" <davem@davemloft.net> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Greentime Hu <green.hu@gmail.com> Cc: Greg Ungerer <gerg@linux-m68k.org> Cc: Guan Xuetao <gxt@pku.edu.cn> Cc: Guo Ren <guoren@kernel.org> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Helge Deller <deller@gmx.de> Cc: Ingo Molnar <mingo@redhat.com> Cc: Ley Foon Tan <ley.foon.tan@intel.com> Cc: Mark Salter <msalter@redhat.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Michal Simek <monstr@monstr.eu> Cc: Nick Hu <nickhu@andestech.com> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Richard Weinberger <richard@nod.at> Cc: Rich Felker <dalias@libc.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Stafford Horne <shorne@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tony Luck <tony.luck@intel.com> Cc: Vincent Chen <deanbo422@gmail.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Link: http://lkml.kernel.org/r/20200514170327.31389-4-rppt@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-06-09 07:32:42 +03:00
#include <linux/pgtable.h>
#include <asm/reg.h>
#include <asm/page.h>
#include <asm/mmu.h>
#include <asm/cputable.h>
#include <asm/cache.h>
#include <asm/thread_info.h>
#include <asm/ppc_asm.h>
#include <asm/asm-offsets.h>
#include <asm/ptrace.h>
#include <asm/bug.h>
#include <asm/kvm_book3s_asm.h>
#include <asm/export.h>
#include <asm/feature-fixups.h>
#include "head_32.h"
#define LOAD_BAT(n, reg, RA, RB) \
/* see the comment for clear_bats() -- Cort */ \
li RA,0; \
mtspr SPRN_IBAT##n##U,RA; \
mtspr SPRN_DBAT##n##U,RA; \
lwz RA,(n*16)+0(reg); \
lwz RB,(n*16)+4(reg); \
mtspr SPRN_IBAT##n##U,RA; \
mtspr SPRN_IBAT##n##L,RB; \
lwz RA,(n*16)+8(reg); \
lwz RB,(n*16)+12(reg); \
mtspr SPRN_DBAT##n##U,RA; \
mtspr SPRN_DBAT##n##L,RB
__HEAD
.stabs "arch/powerpc/kernel/",N_SO,0,0,0f
.stabs "head_book3s_32.S",N_SO,0,0,0f
0:
_ENTRY(_stext);
/*
* _start is defined this way because the XCOFF loader in the OpenFirmware
* on the powermac expects the entry point to be a procedure descriptor.
*/
_ENTRY(_start);
/*
* These are here for legacy reasons, the kernel used to
* need to look like a coff function entry for the pmac
* but we're always started by some kind of bootloader now.
* -- Cort
*/
nop /* used by __secondary_hold on prep (mtx) and chrp smp */
nop /* used by __secondary_hold on prep (mtx) and chrp smp */
nop
/* PMAC
* Enter here with the kernel text, data and bss loaded starting at
* 0, running with virtual == physical mapping.
* r5 points to the prom entry point (the client interface handler
* address). Address translation is turned on, with the prom
* managing the hash table. Interrupts are disabled. The stack
* pointer (r1) points to just below the end of the half-meg region
* from 0x380000 - 0x400000, which is mapped in already.
*
* If we are booted from MacOS via BootX, we enter with the kernel
* image loaded somewhere, and the following values in registers:
* r3: 'BooX' (0x426f6f58)
* r4: virtual address of boot_infos_t
* r5: 0
*
* PREP
* This is jumped to on prep systems right after the kernel is relocated
* to its proper place in memory by the boot loader. The expected layout
* of the regs is:
* r3: ptr to residual data
* r4: initrd_start or if no initrd then 0
* r5: initrd_end - unused if r4 is 0
* r6: Start of command line string
* r7: End of command line string
*
* This just gets a minimal mmu environment setup so we can call
* start_here() to do the real work.
* -- Cort
*/
.globl __start
__start:
/*
* We have to do any OF calls before we map ourselves to KERNELBASE,
* because OF may have I/O devices mapped into that area
* (particularly on CHRP).
*/
cmpwi 0,r5,0
beq 1f
#ifdef CONFIG_PPC_OF_BOOT_TRAMPOLINE
/* find out where we are now */
bcl 20,31,$+4
0: mflr r8 /* r8 = runtime addr here */
addis r8,r8,(_stext - 0b)@ha
addi r8,r8,(_stext - 0b)@l /* current runtime base addr */
bl prom_init
#endif /* CONFIG_PPC_OF_BOOT_TRAMPOLINE */
/* We never return. We also hit that trap if trying to boot
* from OF while CONFIG_PPC_OF_BOOT_TRAMPOLINE isn't selected */
trap
/*
* Check for BootX signature when supporting PowerMac and branch to
* appropriate trampoline if it's present
*/
#ifdef CONFIG_PPC_PMAC
1: lis r31,0x426f
ori r31,r31,0x6f58
cmpw 0,r3,r31
bne 1f
bl bootx_init
trap
#endif /* CONFIG_PPC_PMAC */
1: mr r31,r3 /* save device tree ptr */
li r24,0 /* cpu # */
/*
* early_init() does the early machine identification and does
* the necessary low-level setup and clears the BSS
* -- Cort <cort@fsmlabs.com>
*/
bl early_init
/* Switch MMU off, clear BATs and flush TLB. At this point, r3 contains
* the physical address we are running at, returned by early_init()
*/
bl mmu_off
__after_mmu_off:
bl clear_bats
bl flush_tlbs
bl initial_bats
bl load_segment_registers
bl reloc_offset
bl early_hash_table
#if defined(CONFIG_BOOTX_TEXT)
bl setup_disp_bat
#endif
#ifdef CONFIG_PPC_EARLY_DEBUG_CPM
bl setup_cpm_bat
#endif
#ifdef CONFIG_PPC_EARLY_DEBUG_USBGECKO
bl setup_usbgecko_bat
#endif
/*
* Call setup_cpu for CPU 0 and initialize 6xx Idle
*/
bl reloc_offset
li r24,0 /* cpu# */
bl call_setup_cpu /* Call setup_cpu for this CPU */
bl reloc_offset
bl init_idle_6xx
/*
* We need to run with _start at physical address 0.
* On CHRP, we are loaded at 0x10000 since OF on CHRP uses
* the exception vectors at 0 (and therefore this copy
* overwrites OF's exception vectors with our own).
* The MMU is off at this point.
*/
bl reloc_offset
mr r26,r3
addis r4,r3,KERNELBASE@h /* current address of _start */
lis r5,PHYSICAL_START@h
cmplw 0,r4,r5 /* already running at PHYSICAL_START? */
bne relocate_kernel
/*
* we now have the 1st 16M of ram mapped with the bats.
* prep needs the mmu to be turned on here, but pmac already has it on.
* this shouldn't bother the pmac since it just gets turned on again
* as we jump to our code at KERNELBASE. -- Cort
* Actually no, pmac doesn't have it on any more. BootX enters with MMU
* off, and in other cases, we now turn it off before changing BATs above.
*/
turn_on_mmu:
mfmsr r0
ori r0,r0,MSR_DR|MSR_IR|MSR_RI
mtspr SPRN_SRR1,r0
lis r0,start_here@h
ori r0,r0,start_here@l
mtspr SPRN_SRR0,r0
rfi /* enables MMU */
/*
* We need __secondary_hold as a place to hold the other cpus on
* an SMP machine, even when we are running a UP kernel.
*/
. = 0xc0 /* for prep bootloader */
li r3,1 /* MTX only has 1 cpu */
.globl __secondary_hold
__secondary_hold:
/* tell the master we're here */
stw r3,__secondary_hold_acknowledge@l(0)
#ifdef CONFIG_SMP
100: lwz r4,0(0)
/* wait until we're told to start */
cmpw 0,r4,r3
bne 100b
/* our cpu # was at addr 0 - go */
mr r24,r3 /* cpu # */
b __secondary_start
#else
b .
#endif /* CONFIG_SMP */
.globl __secondary_hold_spinloop
__secondary_hold_spinloop:
.long 0
.globl __secondary_hold_acknowledge
__secondary_hold_acknowledge:
.long -1
/* System reset */
/* core99 pmac starts the seconary here by changing the vector, and
putting it back to what it was (unknown_async_exception) when done. */
EXCEPTION(0x100, Reset, unknown_async_exception, EXC_XFER_STD)
/* Machine check */
/*
* On CHRP, this is complicated by the fact that we could get a
* machine check inside RTAS, and we have no guarantee that certain
* critical registers will have the values we expect. The set of
* registers that might have bad values includes all the GPRs
* and all the BATs. We indicate that we are in RTAS by putting
* a non-zero value, the address of the exception frame to use,
* in thread.rtas_sp. The machine check handler checks thread.rtas_sp
* and uses its value if it is non-zero.
* (Other exception handlers assume that r1 is a valid kernel stack
* pointer when we take an exception from supervisor mode.)
* -- paulus.
*/
. = 0x200
DO_KVM 0x200
MachineCheck:
EXCEPTION_PROLOG_0
#ifdef CONFIG_PPC_CHRP
#ifdef CONFIG_VMAP_STACK
mtspr SPRN_SPRG_SCRATCH2,r1
mfspr r1, SPRN_SPRG_THREAD
lwz r1, RTAS_SP(r1)
cmpwi cr1, r1, 0
bne cr1, 7f
mfspr r1, SPRN_SPRG_SCRATCH2
#else
mfspr r11, SPRN_SPRG_THREAD
lwz r11, RTAS_SP(r11)
cmpwi cr1, r11, 0
bne cr1, 7f
#endif
#endif /* CONFIG_PPC_CHRP */
EXCEPTION_PROLOG_1 for_rtas=1
7: EXCEPTION_PROLOG_2
addi r3,r1,STACK_FRAME_OVERHEAD
#ifdef CONFIG_PPC_CHRP
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
beq cr1, machine_check_tramp
twi 31, 0, 0
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
#else
b machine_check_tramp
#endif
/* Data access exception. */
. = 0x300
DO_KVM 0x300
DataAccess:
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
#ifdef CONFIG_VMAP_STACK
#ifdef CONFIG_PPC_BOOK3S_604
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
BEGIN_MMU_FTR_SECTION
mtspr SPRN_SPRG_SCRATCH2,r10
mfspr r10, SPRN_SPRG_THREAD
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
stw r11, THR11(r10)
mfspr r10, SPRN_DSISR
mfcr r11
andis. r10, r10, (DSISR_BAD_FAULT_32S | DSISR_DABRMATCH)@h
mfspr r10, SPRN_SPRG_THREAD
beq hash_page_dsi
.Lhash_page_dsi_cont:
mtcr r11
lwz r11, THR11(r10)
mfspr r10, SPRN_SPRG_SCRATCH2
MMU_FTR_SECTION_ELSE
b 1f
ALT_MMU_FTR_SECTION_END_IFSET(MMU_FTR_HPTE_TABLE)
#endif
1: EXCEPTION_PROLOG_0 handle_dar_dsisr=1
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
EXCEPTION_PROLOG_1
b handle_page_fault_tramp_1
#else /* CONFIG_VMAP_STACK */
EXCEPTION_PROLOG handle_dar_dsisr=1
get_and_save_dar_dsisr_on_stack r4, r5, r11
#ifdef CONFIG_PPC_BOOK3S_604
BEGIN_MMU_FTR_SECTION
andis. r0, r5, (DSISR_BAD_FAULT_32S | DSISR_DABRMATCH)@h
bne handle_page_fault_tramp_2 /* if not, try to put a PTE */
rlwinm r3, r5, 32 - 15, 21, 21 /* DSISR_STORE -> _PAGE_RW */
bl hash_page
b handle_page_fault_tramp_1
MMU_FTR_SECTION_ELSE
#endif
b handle_page_fault_tramp_2
#ifdef CONFIG_PPC_BOOK3S_604
ALT_MMU_FTR_SECTION_END_IFSET(MMU_FTR_HPTE_TABLE)
#endif
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
#endif /* CONFIG_VMAP_STACK */
/* Instruction access exception. */
. = 0x400
DO_KVM 0x400
InstructionAccess:
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
#ifdef CONFIG_VMAP_STACK
mtspr SPRN_SPRG_SCRATCH0,r10
mtspr SPRN_SPRG_SCRATCH1,r11
mfspr r10, SPRN_SPRG_THREAD
mfspr r11, SPRN_SRR0
stw r11, SRR0(r10)
mfspr r11, SPRN_SRR1 /* check whether user or kernel */
stw r11, SRR1(r10)
mfcr r10
#ifdef CONFIG_PPC_BOOK3S_604
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
BEGIN_MMU_FTR_SECTION
andis. r11, r11, SRR1_ISI_NOPT@h /* no pte found? */
bne hash_page_isi
.Lhash_page_isi_cont:
mfspr r11, SPRN_SRR1 /* check whether user or kernel */
END_MMU_FTR_SECTION_IFSET(MMU_FTR_HPTE_TABLE)
#endif
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
andi. r11, r11, MSR_PR
EXCEPTION_PROLOG_1
EXCEPTION_PROLOG_2
#else /* CONFIG_VMAP_STACK */
EXCEPTION_PROLOG
andis. r0,r9,SRR1_ISI_NOPT@h /* no pte found? */
beq 1f /* if so, try to put a PTE */
li r3,0 /* into the hash table */
mr r4,r12 /* SRR0 is fault address */
#ifdef CONFIG_PPC_BOOK3S_604
BEGIN_MMU_FTR_SECTION
bl hash_page
END_MMU_FTR_SECTION_IFSET(MMU_FTR_HPTE_TABLE)
#endif
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
#endif /* CONFIG_VMAP_STACK */
andis. r5,r9,DSISR_SRR1_MATCH_32S@h /* Filter relevant SRR1 bits */
stw r5, _DSISR(r11)
stw r12, _DAR(r11)
EXC_XFER_LITE(0x400, handle_page_fault)
/* External interrupt */
EXCEPTION(0x500, HardwareInterrupt, do_IRQ, EXC_XFER_LITE)
/* Alignment exception */
. = 0x600
DO_KVM 0x600
Alignment:
EXCEPTION_PROLOG handle_dar_dsisr=1
save_dar_dsisr_on_stack r4, r5, r11
addi r3,r1,STACK_FRAME_OVERHEAD
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
b alignment_exception_tramp
/* Program check exception */
EXCEPTION(0x700, ProgramCheck, program_check_exception, EXC_XFER_STD)
/* Floating-point unavailable */
. = 0x800
DO_KVM 0x800
FPUnavailable:
#ifdef CONFIG_PPC_FPU
BEGIN_FTR_SECTION
/*
* Certain Freescale cores don't have a FPU and treat fp instructions
* as a FP Unavailable exception. Redirect to illegal/emulation handling.
*/
b ProgramCheck
END_FTR_SECTION_IFSET(CPU_FTR_FPU_UNAVAILABLE)
EXCEPTION_PROLOG
beq 1f
bl load_up_fpu /* if from user, just load it up */
b fast_exception_return
1: addi r3,r1,STACK_FRAME_OVERHEAD
EXC_XFER_LITE(0x800, kernel_fp_unavailable_exception)
#else
b ProgramCheck
#endif
/* Decrementer */
EXCEPTION(0x900, Decrementer, timer_interrupt, EXC_XFER_LITE)
EXCEPTION(0xa00, Trap_0a, unknown_exception, EXC_XFER_STD)
EXCEPTION(0xb00, Trap_0b, unknown_exception, EXC_XFER_STD)
/* System call */
. = 0xc00
DO_KVM 0xc00
SystemCall:
SYSCALL_ENTRY 0xc00
EXCEPTION(0xd00, SingleStep, single_step_exception, EXC_XFER_STD)
EXCEPTION(0xe00, Trap_0e, unknown_exception, EXC_XFER_STD)
/*
* The Altivec unavailable trap is at 0x0f20. Foo.
* We effectively remap it to 0x3000.
* We include an altivec unavailable exception vector even if
* not configured for Altivec, so that you can't panic a
* non-altivec kernel running on a machine with altivec just
* by executing an altivec instruction.
*/
. = 0xf00
DO_KVM 0xf00
b PerformanceMonitor
. = 0xf20
DO_KVM 0xf20
b AltiVecUnavailable
/*
* Handle TLB miss for instruction on 603/603e.
* Note: we get an alternate set of r0 - r3 to use automatically.
*/
. = 0x1000
InstructionTLBMiss:
/*
* r0: scratch
* r1: linux style pte ( later becomes ppc hardware pte )
* r2: ptr to linux-style pte
* r3: scratch
*/
/* Get PTE (linux-style) and check access */
mfspr r3,SPRN_IMISS
#ifdef CONFIG_MODULES
lis r1, TASK_SIZE@h /* check if kernel address */
cmplw 0,r1,r3
#endif
mfspr r2, SPRN_SDR1
li r1,_PAGE_PRESENT | _PAGE_ACCESSED | _PAGE_EXEC
rlwinm r2, r2, 28, 0xfffff000
#ifdef CONFIG_MODULES
bgt- 112f
lis r2, (swapper_pg_dir - PAGE_OFFSET)@ha /* if kernel address, use */
addi r2, r2, (swapper_pg_dir - PAGE_OFFSET)@l /* kernel page table */
#endif
112: rlwimi r2,r3,12,20,29 /* insert top 10 bits of address */
lwz r2,0(r2) /* get pmd entry */
rlwinm. r2,r2,0,0,19 /* extract address of pte page */
beq- InstructionAddressInvalid /* return if no mapping */
rlwimi r2,r3,22,20,29 /* insert next 10 bits of address */
lwz r0,0(r2) /* get linux-style pte */
andc. r1,r1,r0 /* check access & ~permission */
bne- InstructionAddressInvalid /* return if access not permitted */
/* Convert linux-style PTE to low word of PPC-style PTE */
rlwimi r0,r0,32-2,31,31 /* _PAGE_USER -> PP lsb */
ori r1, r1, 0xe06 /* clear out reserved bits */
andc r1, r0, r1 /* PP = user? 1 : 0 */
BEGIN_FTR_SECTION
rlwinm r1,r1,0,~_PAGE_COHERENT /* clear M (coherence not required) */
END_FTR_SECTION_IFCLR(CPU_FTR_NEED_COHERENT)
mtspr SPRN_RPA,r1
tlbli r3
mfspr r3,SPRN_SRR1 /* Need to restore CR0 */
mtcrf 0x80,r3
rfi
InstructionAddressInvalid:
mfspr r3,SPRN_SRR1
rlwinm r1,r3,9,6,6 /* Get load/store bit */
addis r1,r1,0x2000
mtspr SPRN_DSISR,r1 /* (shouldn't be needed) */
andi. r2,r3,0xFFFF /* Clear upper bits of SRR1 */
or r2,r2,r1
mtspr SPRN_SRR1,r2
mfspr r1,SPRN_IMISS /* Get failing address */
rlwinm. r2,r2,0,31,31 /* Check for little endian access */
rlwimi r2,r2,1,30,30 /* change 1 -> 3 */
xor r1,r1,r2
mtspr SPRN_DAR,r1 /* Set fault address */
mfmsr r0 /* Restore "normal" registers */
xoris r0,r0,MSR_TGPR>>16
mtcrf 0x80,r3 /* Restore CR0 */
mtmsr r0
b InstructionAccess
/*
* Handle TLB miss for DATA Load operation on 603/603e
*/
. = 0x1100
DataLoadTLBMiss:
/*
* r0: scratch
* r1: linux style pte ( later becomes ppc hardware pte )
* r2: ptr to linux-style pte
* r3: scratch
*/
/* Get PTE (linux-style) and check access */
mfspr r3,SPRN_DMISS
lis r1, TASK_SIZE@h /* check if kernel address */
cmplw 0,r1,r3
mfspr r2, SPRN_SDR1
li r1, _PAGE_PRESENT | _PAGE_ACCESSED
rlwinm r2, r2, 28, 0xfffff000
bgt- 112f
lis r2, (swapper_pg_dir - PAGE_OFFSET)@ha /* if kernel address, use */
addi r2, r2, (swapper_pg_dir - PAGE_OFFSET)@l /* kernel page table */
112: rlwimi r2,r3,12,20,29 /* insert top 10 bits of address */
lwz r2,0(r2) /* get pmd entry */
rlwinm. r2,r2,0,0,19 /* extract address of pte page */
beq- DataAddressInvalid /* return if no mapping */
rlwimi r2,r3,22,20,29 /* insert next 10 bits of address */
lwz r0,0(r2) /* get linux-style pte */
andc. r1,r1,r0 /* check access & ~permission */
bne- DataAddressInvalid /* return if access not permitted */
/*
* NOTE! We are assuming this is not an SMP system, otherwise
* we would need to update the pte atomically with lwarx/stwcx.
*/
/* Convert linux-style PTE to low word of PPC-style PTE */
rlwinm r1,r0,32-9,30,30 /* _PAGE_RW -> PP msb */
rlwimi r0,r0,32-1,30,30 /* _PAGE_USER -> PP msb */
rlwimi r0,r0,32-1,31,31 /* _PAGE_USER -> PP lsb */
ori r1,r1,0xe04 /* clear out reserved bits */
andc r1,r0,r1 /* PP = user? rw? 1: 3: 0 */
BEGIN_FTR_SECTION
rlwinm r1,r1,0,~_PAGE_COHERENT /* clear M (coherence not required) */
END_FTR_SECTION_IFCLR(CPU_FTR_NEED_COHERENT)
mtspr SPRN_RPA,r1
mfspr r2,SPRN_SRR1 /* Need to restore CR0 */
mtcrf 0x80,r2
BEGIN_MMU_FTR_SECTION
li r0,1
mfspr r1,SPRN_SPRG_603_LRU
rlwinm r2,r3,20,27,31 /* Get Address bits 15:19 */
slw r0,r0,r2
xor r1,r0,r1
srw r0,r1,r2
mtspr SPRN_SPRG_603_LRU,r1
mfspr r2,SPRN_SRR1
rlwimi r2,r0,31-14,14,14
mtspr SPRN_SRR1,r2
END_MMU_FTR_SECTION_IFSET(MMU_FTR_NEED_DTLB_SW_LRU)
tlbld r3
rfi
DataAddressInvalid:
mfspr r3,SPRN_SRR1
rlwinm r1,r3,9,6,6 /* Get load/store bit */
addis r1,r1,0x2000
mtspr SPRN_DSISR,r1
andi. r2,r3,0xFFFF /* Clear upper bits of SRR1 */
mtspr SPRN_SRR1,r2
mfspr r1,SPRN_DMISS /* Get failing address */
rlwinm. r2,r2,0,31,31 /* Check for little endian access */
beq 20f /* Jump if big endian */
xori r1,r1,3
20: mtspr SPRN_DAR,r1 /* Set fault address */
mfmsr r0 /* Restore "normal" registers */
xoris r0,r0,MSR_TGPR>>16
mtcrf 0x80,r3 /* Restore CR0 */
mtmsr r0
b DataAccess
/*
* Handle TLB miss for DATA Store on 603/603e
*/
. = 0x1200
DataStoreTLBMiss:
/*
* r0: scratch
* r1: linux style pte ( later becomes ppc hardware pte )
* r2: ptr to linux-style pte
* r3: scratch
*/
/* Get PTE (linux-style) and check access */
mfspr r3,SPRN_DMISS
lis r1, TASK_SIZE@h /* check if kernel address */
cmplw 0,r1,r3
mfspr r2, SPRN_SDR1
li r1, _PAGE_RW | _PAGE_DIRTY | _PAGE_PRESENT | _PAGE_ACCESSED
rlwinm r2, r2, 28, 0xfffff000
bgt- 112f
lis r2, (swapper_pg_dir - PAGE_OFFSET)@ha /* if kernel address, use */
addi r2, r2, (swapper_pg_dir - PAGE_OFFSET)@l /* kernel page table */
112: rlwimi r2,r3,12,20,29 /* insert top 10 bits of address */
lwz r2,0(r2) /* get pmd entry */
rlwinm. r2,r2,0,0,19 /* extract address of pte page */
beq- DataAddressInvalid /* return if no mapping */
rlwimi r2,r3,22,20,29 /* insert next 10 bits of address */
lwz r0,0(r2) /* get linux-style pte */
andc. r1,r1,r0 /* check access & ~permission */
bne- DataAddressInvalid /* return if access not permitted */
/*
* NOTE! We are assuming this is not an SMP system, otherwise
* we would need to update the pte atomically with lwarx/stwcx.
*/
/* Convert linux-style PTE to low word of PPC-style PTE */
rlwimi r0,r0,32-2,31,31 /* _PAGE_USER -> PP lsb */
li r1,0xe06 /* clear out reserved bits & PP msb */
andc r1,r0,r1 /* PP = user? 1: 0 */
BEGIN_FTR_SECTION
rlwinm r1,r1,0,~_PAGE_COHERENT /* clear M (coherence not required) */
END_FTR_SECTION_IFCLR(CPU_FTR_NEED_COHERENT)
mtspr SPRN_RPA,r1
mfspr r2,SPRN_SRR1 /* Need to restore CR0 */
mtcrf 0x80,r2
BEGIN_MMU_FTR_SECTION
li r0,1
mfspr r1,SPRN_SPRG_603_LRU
rlwinm r2,r3,20,27,31 /* Get Address bits 15:19 */
slw r0,r0,r2
xor r1,r0,r1
srw r0,r1,r2
mtspr SPRN_SPRG_603_LRU,r1
mfspr r2,SPRN_SRR1
rlwimi r2,r0,31-14,14,14
mtspr SPRN_SRR1,r2
END_MMU_FTR_SECTION_IFSET(MMU_FTR_NEED_DTLB_SW_LRU)
tlbld r3
rfi
#ifndef CONFIG_ALTIVEC
#define altivec_assist_exception unknown_exception
#endif
#ifndef CONFIG_TAU_INT
#define TAUException unknown_async_exception
#endif
EXCEPTION(0x1300, Trap_13, instruction_breakpoint_exception, EXC_XFER_STD)
EXCEPTION(0x1400, SMI, SMIException, EXC_XFER_STD)
EXCEPTION(0x1500, Trap_15, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x1600, Trap_16, altivec_assist_exception, EXC_XFER_STD)
EXCEPTION(0x1700, Trap_17, TAUException, EXC_XFER_STD)
EXCEPTION(0x1800, Trap_18, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x1900, Trap_19, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x1a00, Trap_1a, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x1b00, Trap_1b, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x1c00, Trap_1c, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x1d00, Trap_1d, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x1e00, Trap_1e, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x1f00, Trap_1f, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2000, RunMode, RunModeException, EXC_XFER_STD)
EXCEPTION(0x2100, Trap_21, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2200, Trap_22, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2300, Trap_23, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2400, Trap_24, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2500, Trap_25, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2600, Trap_26, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2700, Trap_27, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2800, Trap_28, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2900, Trap_29, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2a00, Trap_2a, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2b00, Trap_2b, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2c00, Trap_2c, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2d00, Trap_2d, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2e00, Trap_2e, unknown_exception, EXC_XFER_STD)
EXCEPTION(0x2f00, Trap_2f, unknown_exception, EXC_XFER_STD)
. = 0x3000
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
machine_check_tramp:
EXC_XFER_STD(0x200, machine_check_exception)
alignment_exception_tramp:
EXC_XFER_STD(0x600, alignment_exception)
handle_page_fault_tramp_1:
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
#ifdef CONFIG_VMAP_STACK
EXCEPTION_PROLOG_2 handle_dar_dsisr=1
#endif
lwz r5, _DSISR(r11)
/* fall through */
handle_page_fault_tramp_2:
andis. r0, r5, DSISR_DABRMATCH@h
bne- 1f
EXC_XFER_LITE(0x300, handle_page_fault)
1: EXC_XFER_STD(0x300, do_break)
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
#ifdef CONFIG_VMAP_STACK
#ifdef CONFIG_PPC_BOOK3S_604
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
.macro save_regs_thread thread
stw r0, THR0(\thread)
stw r3, THR3(\thread)
stw r4, THR4(\thread)
stw r5, THR5(\thread)
stw r6, THR6(\thread)
stw r8, THR8(\thread)
stw r9, THR9(\thread)
mflr r0
stw r0, THLR(\thread)
mfctr r0
stw r0, THCTR(\thread)
.endm
.macro restore_regs_thread thread
lwz r0, THLR(\thread)
mtlr r0
lwz r0, THCTR(\thread)
mtctr r0
lwz r0, THR0(\thread)
lwz r3, THR3(\thread)
lwz r4, THR4(\thread)
lwz r5, THR5(\thread)
lwz r6, THR6(\thread)
lwz r8, THR8(\thread)
lwz r9, THR9(\thread)
.endm
hash_page_dsi:
save_regs_thread r10
mfdsisr r3
mfdar r4
mfsrr0 r5
mfsrr1 r9
rlwinm r3, r3, 32 - 15, _PAGE_RW /* DSISR_STORE -> _PAGE_RW */
bl hash_page
mfspr r10, SPRN_SPRG_THREAD
restore_regs_thread r10
b .Lhash_page_dsi_cont
hash_page_isi:
mr r11, r10
mfspr r10, SPRN_SPRG_THREAD
save_regs_thread r10
li r3, 0
lwz r4, SRR0(r10)
lwz r9, SRR1(r10)
bl hash_page
mfspr r10, SPRN_SPRG_THREAD
restore_regs_thread r10
mr r10, r11
b .Lhash_page_isi_cont
.globl fast_hash_page_return
fast_hash_page_return:
andis. r10, r9, SRR1_ISI_NOPT@h /* Set on ISI, cleared on DSI */
mfspr r10, SPRN_SPRG_THREAD
restore_regs_thread r10
bne 1f
/* DSI */
mtcr r11
lwz r11, THR11(r10)
mfspr r10, SPRN_SPRG_SCRATCH2
rfi
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
1: /* ISI */
mtcr r11
mfspr r11, SPRN_SPRG_SCRATCH1
mfspr r10, SPRN_SPRG_SCRATCH0
rfi
#endif /* CONFIG_PPC_BOOK3S_604 */
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
stack_overflow:
vmap_stack_overflow_exception
powerpc/32s: Fix DSI and ISI exceptions for CONFIG_VMAP_STACK hash_page() needs to read page tables from kernel memory. When entire kernel memory is mapped by BATs, which is normally the case when CONFIG_STRICT_KERNEL_RWX is not set, it works even if the page hosting the page table is not referenced in the MMU hash table. However, if the page where the page table resides is not covered by a BAT, a DSI fault can be encountered from hash_page(), and it loops forever. This can happen when CONFIG_STRICT_KERNEL_RWX is selected and the alignment of the different regions is too small to allow covering the entire memory with BATs. This also happens when CONFIG_DEBUG_PAGEALLOC is selected or when booting with 'nobats' flag. Also, if the page containing the kernel stack is not present in the MMU hash table, registers cannot be saved and a recursive DSI fault is encountered. To allow hash_page() to properly do its job at all time and load the MMU hash table whenever needed, it must run with data MMU disabled. This means it must be called before re-enabling data MMU. To allow this, registers clobbered by hash_page() and create_hpte() have to be saved in the thread struct together with SRR0, SSR1, DAR and DSISR. It is also necessary to ensure that DSI prolog doesn't overwrite regs saved by prolog of the current running exception. That means: - DSI can only use SPRN_SPRG_SCRATCH0 - Exceptions must free SPRN_SPRG_SCRATCH0 before writing to the stack. This also fixes the Oops reported by Erhard when create_hpte() is called by add_hash_page(). Due to prolog size increase, a few more exceptions had to get split in two parts. Fixes: cd08f109e262 ("powerpc/32s: Enable CONFIG_VMAP_STACK") Reported-by: Erhard F. <erhard_f@mailbox.org> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Tested-by: Erhard F. <erhard_f@mailbox.org> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://bugzilla.kernel.org/show_bug.cgi?id=206501 Link: https://lore.kernel.org/r/64a4aa44686e9fd4b01333401367029771d9b231.1581761633.git.christophe.leroy@c-s.fr
2020-02-15 13:14:25 +03:00
#endif
AltiVecUnavailable:
EXCEPTION_PROLOG
#ifdef CONFIG_ALTIVEC
beq 1f
bl load_up_altivec /* if from user, just load it up */
b fast_exception_return
#endif /* CONFIG_ALTIVEC */
1: addi r3,r1,STACK_FRAME_OVERHEAD
EXC_XFER_LITE(0xf20, altivec_unavailable_exception)
PerformanceMonitor:
EXCEPTION_PROLOG
addi r3,r1,STACK_FRAME_OVERHEAD
EXC_XFER_STD(0xf00, performance_monitor_exception)
/*
* This code is jumped to from the startup code to copy
* the kernel image to physical address PHYSICAL_START.
*/
relocate_kernel:
addis r9,r26,klimit@ha /* fetch klimit */
lwz r25,klimit@l(r9)
addis r25,r25,-KERNELBASE@h
lis r3,PHYSICAL_START@h /* Destination base address */
li r6,0 /* Destination offset */
li r5,0x4000 /* # bytes of memory to copy */
bl copy_and_flush /* copy the first 0x4000 bytes */
addi r0,r3,4f@l /* jump to the address of 4f */
mtctr r0 /* in copy and do the rest. */
bctr /* jump to the copy */
4: mr r5,r25
bl copy_and_flush /* copy the rest */
b turn_on_mmu
/*
* Copy routine used to copy the kernel to start at physical address 0
* and flush and invalidate the caches as needed.
* r3 = dest addr, r4 = source addr, r5 = copy limit, r6 = start offset
* on exit, r3, r4, r5 are unchanged, r6 is updated to be >= r5.
*/
_ENTRY(copy_and_flush)
addi r5,r5,-4
addi r6,r6,-4
4: li r0,L1_CACHE_BYTES/4
mtctr r0
3: addi r6,r6,4 /* copy a cache line */
lwzx r0,r6,r4
stwx r0,r6,r3
bdnz 3b
dcbst r6,r3 /* write it to memory */
sync
icbi r6,r3 /* flush the icache line */
cmplw 0,r6,r5
blt 4b
sync /* additional sync needed on g4 */
isync
addi r5,r5,4
addi r6,r6,4
blr
#ifdef CONFIG_SMP
.globl __secondary_start_mpc86xx
__secondary_start_mpc86xx:
mfspr r3, SPRN_PIR
stw r3, __secondary_hold_acknowledge@l(0)
mr r24, r3 /* cpu # */
b __secondary_start
.globl __secondary_start_pmac_0
__secondary_start_pmac_0:
/* NB the entries for cpus 0, 1, 2 must each occupy 8 bytes. */
li r24,0
b 1f
li r24,1
b 1f
li r24,2
b 1f
li r24,3
1:
/* on powersurge, we come in here with IR=0 and DR=1, and DBAT 0
set to map the 0xf0000000 - 0xffffffff region */
mfmsr r0
rlwinm r0,r0,0,28,26 /* clear DR (0x10) */
mtmsr r0
isync
.globl __secondary_start
__secondary_start:
/* Copy some CPU settings from CPU 0 */
bl __restore_cpu_setup
lis r3,-KERNELBASE@h
mr r4,r24
bl call_setup_cpu /* Call setup_cpu for this CPU */
lis r3,-KERNELBASE@h
bl init_idle_6xx
/* get current's stack and current */
lis r2,secondary_current@ha
tophys(r2,r2)
lwz r2,secondary_current@l(r2)
tophys(r1,r2)
lwz r1,TASK_STACK(r1)
/* stack */
addi r1,r1,THREAD_SIZE-STACK_FRAME_OVERHEAD
li r0,0
tophys(r3,r1)
stw r0,0(r3)
/* load up the MMU */
bl load_segment_registers
bl load_up_mmu
/* ptr to phys current thread */
tophys(r4,r2)
addi r4,r4,THREAD /* phys address of our thread_struct */
mtspr SPRN_SPRG_THREAD,r4
BEGIN_MMU_FTR_SECTION
lis r4, (swapper_pg_dir - PAGE_OFFSET)@h
ori r4, r4, (swapper_pg_dir - PAGE_OFFSET)@l
rlwinm r4, r4, 4, 0xffff01ff
mtspr SPRN_SDR1, r4
END_MMU_FTR_SECTION_IFCLR(MMU_FTR_HPTE_TABLE)
/* enable MMU and jump to start_secondary */
li r4,MSR_KERNEL
lis r3,start_secondary@h
ori r3,r3,start_secondary@l
mtspr SPRN_SRR0,r3
mtspr SPRN_SRR1,r4
rfi
#endif /* CONFIG_SMP */
#ifdef CONFIG_KVM_BOOK3S_HANDLER
#include "../kvm/book3s_rmhandlers.S"
#endif
/*
* Load stuff into the MMU. Intended to be called with
* IR=0 and DR=0.
*/
early_hash_table:
sync /* Force all PTE updates to finish */
isync
tlbia /* Clear all TLB entries */
sync /* wait for tlbia/tlbie to finish */
TLBSYNC /* ... on all CPUs */
/* Load the SDR1 register (hash table base & size) */
lis r6, early_hash - PAGE_OFFSET@h
ori r6, r6, 3 /* 256kB table */
mtspr SPRN_SDR1, r6
blr
load_up_mmu:
sync /* Force all PTE updates to finish */
isync
tlbia /* Clear all TLB entries */
sync /* wait for tlbia/tlbie to finish */
TLBSYNC /* ... on all CPUs */
BEGIN_MMU_FTR_SECTION
/* Load the SDR1 register (hash table base & size) */
lis r6,_SDR1@ha
tophys(r6,r6)
lwz r6,_SDR1@l(r6)
mtspr SPRN_SDR1,r6
END_MMU_FTR_SECTION_IFSET(MMU_FTR_HPTE_TABLE)
/* Load the BAT registers with the values set up by MMU_init. */
lis r3,BATS@ha
addi r3,r3,BATS@l
tophys(r3,r3)
LOAD_BAT(0,r3,r4,r5)
LOAD_BAT(1,r3,r4,r5)
LOAD_BAT(2,r3,r4,r5)
LOAD_BAT(3,r3,r4,r5)
BEGIN_MMU_FTR_SECTION
LOAD_BAT(4,r3,r4,r5)
LOAD_BAT(5,r3,r4,r5)
LOAD_BAT(6,r3,r4,r5)
LOAD_BAT(7,r3,r4,r5)
END_MMU_FTR_SECTION_IFSET(MMU_FTR_USE_HIGH_BATS)
blr
_GLOBAL(load_segment_registers)
li r0, NUM_USER_SEGMENTS /* load up user segment register values */
mtctr r0 /* for context 0 */
li r3, 0 /* Kp = 0, Ks = 0, VSID = 0 */
#ifdef CONFIG_PPC_KUEP
oris r3, r3, SR_NX@h /* Set Nx */
#endif
#ifdef CONFIG_PPC_KUAP
oris r3, r3, SR_KS@h /* Set Ks */
#endif
li r4, 0
3: mtsrin r3, r4
addi r3, r3, 0x111 /* increment VSID */
addis r4, r4, 0x1000 /* address of next segment */
bdnz 3b
li r0, 16 - NUM_USER_SEGMENTS /* load up kernel segment registers */
mtctr r0 /* for context 0 */
rlwinm r3, r3, 0, ~SR_NX /* Nx = 0 */
rlwinm r3, r3, 0, ~SR_KS /* Ks = 0 */
oris r3, r3, SR_KP@h /* Kp = 1 */
3: mtsrin r3, r4
addi r3, r3, 0x111 /* increment VSID */
addis r4, r4, 0x1000 /* address of next segment */
bdnz 3b
blr
/*
* This is where the main kernel code starts.
*/
start_here:
/* ptr to current */
lis r2,init_task@h
ori r2,r2,init_task@l
/* Set up for using our exception vectors */
/* ptr to phys current thread */
tophys(r4,r2)
addi r4,r4,THREAD /* init task's THREAD */
mtspr SPRN_SPRG_THREAD,r4
BEGIN_MMU_FTR_SECTION
lis r4, (swapper_pg_dir - PAGE_OFFSET)@h
ori r4, r4, (swapper_pg_dir - PAGE_OFFSET)@l
rlwinm r4, r4, 4, 0xffff01ff
mtspr SPRN_SDR1, r4
END_MMU_FTR_SECTION_IFCLR(MMU_FTR_HPTE_TABLE)
/* stack */
lis r1,init_thread_union@ha
addi r1,r1,init_thread_union@l
li r0,0
stwu r0,THREAD_SIZE-STACK_FRAME_OVERHEAD(r1)
/*
* Do early platform-specific initialization,
* and set up the MMU.
*/
#ifdef CONFIG_KASAN
bl kasan_early_init
#endif
li r3,0
mr r4,r31
bl machine_init
bl __save_cpu_setup
bl MMU_init
bl MMU_init_hw_patch
/*
* Go back to running unmapped so we can load up new values
* for SDR1 (hash table pointer) and the segment registers
* and change to using our exception vectors.
*/
lis r4,2f@h
ori r4,r4,2f@l
tophys(r4,r4)
li r3,MSR_KERNEL & ~(MSR_IR|MSR_DR)
.align 4
mtspr SPRN_SRR0,r4
mtspr SPRN_SRR1,r3
rfi
/* Load up the kernel context */
2: bl load_up_mmu
#ifdef CONFIG_BDI_SWITCH
/* Add helper information for the Abatron bdiGDB debugger.
* We do this here because we know the mmu is disabled, and
* will be enabled for real in just a few instructions.
*/
lis r5, abatron_pteptrs@h
ori r5, r5, abatron_pteptrs@l
stw r5, 0xf0(0) /* This much match your Abatron config */
lis r6, swapper_pg_dir@h
ori r6, r6, swapper_pg_dir@l
tophys(r5, r5)
stw r6, 0(r5)
#endif /* CONFIG_BDI_SWITCH */
/* Now turn on the MMU for real! */
li r4,MSR_KERNEL
lis r3,start_kernel@h
ori r3,r3,start_kernel@l
mtspr SPRN_SRR0,r3
mtspr SPRN_SRR1,r4
rfi
/*
* void switch_mmu_context(struct mm_struct *prev, struct mm_struct *next);
*
* Set up the segment registers for a new context.
*/
_ENTRY(switch_mmu_context)
lwz r3,MMCONTEXTID(r4)
cmpwi cr0,r3,0
blt- 4f
mulli r3,r3,897 /* multiply context by skew factor */
rlwinm r3,r3,4,8,27 /* VSID = (context & 0xfffff) << 4 */
#ifdef CONFIG_PPC_KUEP
oris r3, r3, SR_NX@h /* Set Nx */
#endif
#ifdef CONFIG_PPC_KUAP
oris r3, r3, SR_KS@h /* Set Ks */
#endif
li r0,NUM_USER_SEGMENTS
mtctr r0
#ifdef CONFIG_BDI_SWITCH
/* Context switch the PTE pointer for the Abatron BDI2000.
* The PGDIR is passed as second argument.
*/
lwz r4, MM_PGD(r4)
lis r5, abatron_pteptrs@ha
stw r4, abatron_pteptrs@l + 0x4(r5)
#endif
BEGIN_MMU_FTR_SECTION
#ifndef CONFIG_BDI_SWITCH
lwz r4, MM_PGD(r4)
#endif
tophys(r4, r4)
rlwinm r4, r4, 4, 0xffff01ff
mtspr SPRN_SDR1, r4
END_MMU_FTR_SECTION_IFCLR(MMU_FTR_HPTE_TABLE)
li r4,0
isync
3:
mtsrin r3,r4
addi r3,r3,0x111 /* next VSID */
rlwinm r3,r3,0,8,3 /* clear out any overflow from VSID field */
addis r4,r4,0x1000 /* address of next segment */
bdnz 3b
sync
isync
blr
4: trap
EMIT_BUG_ENTRY 4b,__FILE__,__LINE__,0
blr
EXPORT_SYMBOL(switch_mmu_context)
/*
* An undocumented "feature" of 604e requires that the v bit
* be cleared before changing BAT values.
*
* Also, newer IBM firmware does not clear bat3 and 4 so
* this makes sure it's done.
* -- Cort
*/
clear_bats:
li r10,0
mtspr SPRN_DBAT0U,r10
mtspr SPRN_DBAT0L,r10
mtspr SPRN_DBAT1U,r10
mtspr SPRN_DBAT1L,r10
mtspr SPRN_DBAT2U,r10
mtspr SPRN_DBAT2L,r10
mtspr SPRN_DBAT3U,r10
mtspr SPRN_DBAT3L,r10
mtspr SPRN_IBAT0U,r10
mtspr SPRN_IBAT0L,r10
mtspr SPRN_IBAT1U,r10
mtspr SPRN_IBAT1L,r10
mtspr SPRN_IBAT2U,r10
mtspr SPRN_IBAT2L,r10
mtspr SPRN_IBAT3U,r10
mtspr SPRN_IBAT3L,r10
BEGIN_MMU_FTR_SECTION
/* Here's a tweak: at this point, CPU setup have
* not been called yet, so HIGH_BAT_EN may not be
* set in HID0 for the 745x processors. However, it
* seems that doesn't affect our ability to actually
* write to these SPRs.
*/
mtspr SPRN_DBAT4U,r10
mtspr SPRN_DBAT4L,r10
mtspr SPRN_DBAT5U,r10
mtspr SPRN_DBAT5L,r10
mtspr SPRN_DBAT6U,r10
mtspr SPRN_DBAT6L,r10
mtspr SPRN_DBAT7U,r10
mtspr SPRN_DBAT7L,r10
mtspr SPRN_IBAT4U,r10
mtspr SPRN_IBAT4L,r10
mtspr SPRN_IBAT5U,r10
mtspr SPRN_IBAT5L,r10
mtspr SPRN_IBAT6U,r10
mtspr SPRN_IBAT6L,r10
mtspr SPRN_IBAT7U,r10
mtspr SPRN_IBAT7L,r10
END_MMU_FTR_SECTION_IFSET(MMU_FTR_USE_HIGH_BATS)
blr
_ENTRY(update_bats)
lis r4, 1f@h
ori r4, r4, 1f@l
tophys(r4, r4)
mfmsr r6
mflr r7
li r3, MSR_KERNEL & ~(MSR_IR | MSR_DR)
rlwinm r0, r6, 0, ~MSR_RI
rlwinm r0, r0, 0, ~MSR_EE
mtmsr r0
.align 4
mtspr SPRN_SRR0, r4
mtspr SPRN_SRR1, r3
rfi
1: bl clear_bats
lis r3, BATS@ha
addi r3, r3, BATS@l
tophys(r3, r3)
LOAD_BAT(0, r3, r4, r5)
LOAD_BAT(1, r3, r4, r5)
LOAD_BAT(2, r3, r4, r5)
LOAD_BAT(3, r3, r4, r5)
BEGIN_MMU_FTR_SECTION
LOAD_BAT(4, r3, r4, r5)
LOAD_BAT(5, r3, r4, r5)
LOAD_BAT(6, r3, r4, r5)
LOAD_BAT(7, r3, r4, r5)
END_MMU_FTR_SECTION_IFSET(MMU_FTR_USE_HIGH_BATS)
li r3, MSR_KERNEL & ~(MSR_IR | MSR_DR | MSR_RI)
mtmsr r3
mtspr SPRN_SRR0, r7
mtspr SPRN_SRR1, r6
rfi
flush_tlbs:
lis r10, 0x40
1: addic. r10, r10, -0x1000
tlbie r10
bgt 1b
sync
blr
mmu_off:
addi r4, r3, __after_mmu_off - _start
mfmsr r3
andi. r0,r3,MSR_DR|MSR_IR /* MMU enabled? */
beqlr
andc r3,r3,r0
.align 4
mtspr SPRN_SRR0,r4
mtspr SPRN_SRR1,r3
sync
rfi
/* We use one BAT to map up to 256M of RAM at _PAGE_OFFSET */
initial_bats:
lis r11,PAGE_OFFSET@h
tophys(r8,r11)
#ifdef CONFIG_SMP
ori r8,r8,0x12 /* R/W access, M=1 */
#else
ori r8,r8,2 /* R/W access */
#endif /* CONFIG_SMP */
ori r11,r11,BL_256M<<2|0x2 /* set up BAT registers for 604 */
mtspr SPRN_DBAT0L,r8 /* N.B. 6xx have valid */
mtspr SPRN_DBAT0U,r11 /* bit in upper BAT register */
mtspr SPRN_IBAT0L,r8
mtspr SPRN_IBAT0U,r11
isync
blr
#ifdef CONFIG_BOOTX_TEXT
setup_disp_bat:
/*
* setup the display bat prepared for us in prom.c
*/
mflr r8
bl reloc_offset
mtlr r8
addis r8,r3,disp_BAT@ha
addi r8,r8,disp_BAT@l
cmpwi cr0,r8,0
beqlr
lwz r11,0(r8)
lwz r8,4(r8)
mtspr SPRN_DBAT3L,r8
mtspr SPRN_DBAT3U,r11
blr
#endif /* CONFIG_BOOTX_TEXT */
#ifdef CONFIG_PPC_EARLY_DEBUG_CPM
setup_cpm_bat:
lis r8, 0xf000
ori r8, r8, 0x002a
mtspr SPRN_DBAT1L, r8
lis r11, 0xf000
ori r11, r11, (BL_1M << 2) | 2
mtspr SPRN_DBAT1U, r11
blr
#endif
#ifdef CONFIG_PPC_EARLY_DEBUG_USBGECKO
setup_usbgecko_bat:
/* prepare a BAT for early io */
#if defined(CONFIG_GAMECUBE)
lis r8, 0x0c00
#elif defined(CONFIG_WII)
lis r8, 0x0d00
#else
#error Invalid platform for USB Gecko based early debugging.
#endif
/*
* The virtual address used must match the virtual address
* associated to the fixmap entry FIX_EARLY_DEBUG_BASE.
*/
lis r11, 0xfffe /* top 128K */
ori r8, r8, 0x002a /* uncached, guarded ,rw */
ori r11, r11, 0x2 /* 128K, Vs=1, Vp=0 */
mtspr SPRN_DBAT1L, r8
mtspr SPRN_DBAT1U, r11
blr
#endif
#ifdef CONFIG_8260
/* Jump into the system reset for the rom.
* We first disable the MMU, and then jump to the ROM reset address.
*
* r3 is the board info structure, r4 is the location for starting.
* I use this for building a small kernel that can load other kernels,
* rather than trying to write or rely on a rom monitor that can tftp load.
*/
.globl m8260_gorom
m8260_gorom:
mfmsr r0
rlwinm r0,r0,0,17,15 /* clear MSR_EE in r0 */
sync
mtmsr r0
sync
mfspr r11, SPRN_HID0
lis r10, 0
ori r10,r10,HID0_ICE|HID0_DCE
andc r11, r11, r10
mtspr SPRN_HID0, r11
isync
li r5, MSR_ME|MSR_RI
lis r6,2f@h
addis r6,r6,-KERNELBASE@h
ori r6,r6,2f@l
mtspr SPRN_SRR0,r6
mtspr SPRN_SRR1,r5
isync
sync
rfi
2:
mtlr r4
blr
#endif
/*
* We put a few things here that have to be page-aligned.
* This stuff goes at the beginning of the data segment,
* which is page-aligned.
*/
.data
.globl sdata
sdata:
.globl empty_zero_page
empty_zero_page:
.space 4096
EXPORT_SYMBOL(empty_zero_page)
.globl swapper_pg_dir
swapper_pg_dir:
.space PGD_TABLE_SIZE
/* Room for two PTE pointers, usually the kernel and current user pointers
* to their respective root page table.
*/
abatron_pteptrs:
.space 8