linux_dsm_epyc7002/arch/powerpc/mm/slb.c

339 lines
10 KiB
C
Raw Normal View History

/*
* PowerPC64 SLB support.
*
* Copyright (C) 2004 David Gibson <dwg@au.ibm.com>, IBM
* Based on earlier code written by:
* Dave Engebretsen and Mike Corrigan {engebret|mikejc}@us.ibm.com
* Copyright (c) 2001 Dave Engebretsen
* Copyright (C) 2002 Anton Blanchard <anton@au.ibm.com>, IBM
*
*
* 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/pgtable.h>
#include <asm/mmu.h>
#include <asm/mmu_context.h>
#include <asm/paca.h>
#include <asm/cputable.h>
#include <asm/cacheflush.h>
#include <asm/smp.h>
#include <asm/firmware.h>
#include <linux/compiler.h>
#include <asm/udbg.h>
#include <asm/code-patching.h>
extern void slb_allocate_realmode(unsigned long ea);
extern void slb_allocate_user(unsigned long ea);
static void slb_allocate(unsigned long ea)
{
/* Currently, we do real mode for all SLBs including user, but
* that will change if we bring back dynamic VSIDs
*/
slb_allocate_realmode(ea);
}
[POWERPC] Bolt in SLB entry for kernel stack on secondary cpus This fixes a regression reported by Kamalesh Bulabel where a POWER4 machine would crash because of an SLB miss at a point where the SLB miss exception was unrecoverable. This regression is tracked at: http://bugzilla.kernel.org/show_bug.cgi?id=10082 SLB misses at such points shouldn't happen because the kernel stack is the only memory accessed other than things in the first segment of the linear mapping (which is mapped at all times by entry 0 of the SLB). The context switch code ensures that SLB entry 2 covers the kernel stack, if it is not already covered by entry 0. None of entries 0 to 2 are ever replaced by the SLB miss handler. Where this went wrong is that the context switch code assumes it doesn't have to write to SLB entry 2 if the new kernel stack is in the same segment as the old kernel stack, since entry 2 should already be correct. However, when we start up a secondary cpu, it calls slb_initialize, which doesn't set up entry 2. This is correct for the boot cpu, where we will be using a stack in the kernel BSS at this point (i.e. init_thread_union), but not necessarily for secondary cpus, whose initial stack can be allocated anywhere. This doesn't cause any immediate problem since the SLB miss handler will just create an SLB entry somewhere else to cover the initial stack. In fact it's possible for the cpu to go quite a long time without SLB entry 2 being valid. Eventually, though, the entry created by the SLB miss handler will get overwritten by some other entry, and if the next access to the stack is at an unrecoverable point, we get the crash. This fixes the problem by making slb_initialize create a suitable entry for the kernel stack, if we are on a secondary cpu and the stack isn't covered by SLB entry 0. This requires initializing the get_paca()->kstack field earlier, so I do that in smp_create_idle where the current field is initialized. This also abstracts a bit of the computation that mk_esid_data in slb.c does so that it can be used in slb_initialize. Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-05-02 11:29:12 +07:00
#define slb_esid_mask(ssize) \
(((ssize) == MMU_SEGSIZE_256M)? ESID_MASK: ESID_MASK_1T)
static inline unsigned long mk_esid_data(unsigned long ea, int ssize,
unsigned long slot)
{
[POWERPC] Bolt in SLB entry for kernel stack on secondary cpus This fixes a regression reported by Kamalesh Bulabel where a POWER4 machine would crash because of an SLB miss at a point where the SLB miss exception was unrecoverable. This regression is tracked at: http://bugzilla.kernel.org/show_bug.cgi?id=10082 SLB misses at such points shouldn't happen because the kernel stack is the only memory accessed other than things in the first segment of the linear mapping (which is mapped at all times by entry 0 of the SLB). The context switch code ensures that SLB entry 2 covers the kernel stack, if it is not already covered by entry 0. None of entries 0 to 2 are ever replaced by the SLB miss handler. Where this went wrong is that the context switch code assumes it doesn't have to write to SLB entry 2 if the new kernel stack is in the same segment as the old kernel stack, since entry 2 should already be correct. However, when we start up a secondary cpu, it calls slb_initialize, which doesn't set up entry 2. This is correct for the boot cpu, where we will be using a stack in the kernel BSS at this point (i.e. init_thread_union), but not necessarily for secondary cpus, whose initial stack can be allocated anywhere. This doesn't cause any immediate problem since the SLB miss handler will just create an SLB entry somewhere else to cover the initial stack. In fact it's possible for the cpu to go quite a long time without SLB entry 2 being valid. Eventually, though, the entry created by the SLB miss handler will get overwritten by some other entry, and if the next access to the stack is at an unrecoverable point, we get the crash. This fixes the problem by making slb_initialize create a suitable entry for the kernel stack, if we are on a secondary cpu and the stack isn't covered by SLB entry 0. This requires initializing the get_paca()->kstack field earlier, so I do that in smp_create_idle where the current field is initialized. This also abstracts a bit of the computation that mk_esid_data in slb.c does so that it can be used in slb_initialize. Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-05-02 11:29:12 +07:00
return (ea & slb_esid_mask(ssize)) | SLB_ESID_V | slot;
}
#define slb_vsid_shift(ssize) \
((ssize) == MMU_SEGSIZE_256M? SLB_VSID_SHIFT: SLB_VSID_SHIFT_1T)
static inline unsigned long mk_vsid_data(unsigned long ea, int ssize,
unsigned long flags)
{
return (get_kernel_vsid(ea, ssize) << slb_vsid_shift(ssize)) | flags |
((unsigned long) ssize << SLB_VSID_SSIZE_SHIFT);
}
static inline void slb_shadow_update(unsigned long ea, int ssize,
unsigned long flags,
unsigned long entry)
{
/*
* Clear the ESID first so the entry is not valid while we are
* updating it. No write barriers are needed here, provided
* we only update the current CPU's SLB shadow buffer.
*/
get_slb_shadow()->save_area[entry].esid = 0;
get_slb_shadow()->save_area[entry].vsid = mk_vsid_data(ea, ssize, flags);
get_slb_shadow()->save_area[entry].esid = mk_esid_data(ea, ssize, entry);
}
static inline void slb_shadow_clear(unsigned long entry)
{
get_slb_shadow()->save_area[entry].esid = 0;
}
static inline void create_shadowed_slbe(unsigned long ea, int ssize,
unsigned long flags,
unsigned long entry)
{
/*
* Updating the shadow buffer before writing the SLB ensures
* we don't get a stale entry here if we get preempted by PHYP
* between these two statements.
*/
slb_shadow_update(ea, ssize, flags, entry);
asm volatile("slbmte %0,%1" :
: "r" (mk_vsid_data(ea, ssize, flags)),
"r" (mk_esid_data(ea, ssize, entry))
: "memory" );
}
powerpc: Allow perf_counters to access user memory at interrupt time This provides a mechanism to allow the perf_counters code to access user memory in a PMU interrupt routine. Such an access can cause various kinds of interrupt: SLB miss, MMU hash table miss, segment table miss, or TLB miss, depending on the processor. This commit only deals with 64-bit classic/server processors, which use an MMU hash table. 32-bit processors are already able to access user memory at interrupt time. Since we don't soft-disable on 32-bit, we avoid the possibility of reentering hash_page or the TLB miss handlers, since they run with interrupts disabled. On 64-bit processors, an SLB miss interrupt on a user address will update the slb_cache and slb_cache_ptr fields in the paca. This is OK except in the case where a PMU interrupt occurs in switch_slb, which also accesses those fields. To prevent this, we hard-disable interrupts in switch_slb. Interrupts are already soft-disabled at this point, and will get hard-enabled when they get soft-enabled later. This also reworks slb_flush_and_rebolt: to avoid hard-disabling twice, and to make sure that it clears the slb_cache_ptr when called from other callers than switch_slb, the existing routine is renamed to __slb_flush_and_rebolt, which is called by switch_slb and the new version of slb_flush_and_rebolt. Similarly, switch_stab (used on POWER3 and RS64 processors) gets a hard_irq_disable() to protect the per-cpu variables used there and in ste_allocate. If a MMU hashtable miss interrupt occurs, normally we would call hash_page to look up the Linux PTE for the address and create a HPTE. However, hash_page is fairly complex and takes some locks, so to avoid the possibility of deadlock, we check the preemption count to see if we are in a (pseudo-)NMI handler, and if so, we don't call hash_page but instead treat it like a bad access that will get reported up through the exception table mechanism. An interrupt whose handler runs even though the interrupt occurred when soft-disabled (such as the PMU interrupt) is considered a pseudo-NMI handler, which should use nmi_enter()/nmi_exit() rather than irq_enter()/irq_exit(). Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2009-08-17 12:17:54 +07:00
static void __slb_flush_and_rebolt(void)
{
/* If you change this make sure you change SLB_NUM_BOLTED
* appropriately too. */
powerpc: Use 64k pages without needing cache-inhibited large pages Some POWER5+ machines can do 64k hardware pages for normal memory but not for cache-inhibited pages. This patch lets us use 64k hardware pages for most user processes on such machines (assuming the kernel has been configured with CONFIG_PPC_64K_PAGES=y). User processes start out using 64k pages and get switched to 4k pages if they use any non-cacheable mappings. With this, we use 64k pages for the vmalloc region and 4k pages for the imalloc region. If anything creates a non-cacheable mapping in the vmalloc region, the vmalloc region will get switched to 4k pages. I don't know of any driver other than the DRM that would do this, though, and these machines don't have AGP. When a region gets switched from 64k pages to 4k pages, we do not have to clear out all the 64k HPTEs from the hash table immediately. We use the _PAGE_COMBO bit in the Linux PTE to indicate whether the page was hashed in as a 64k page or a set of 4k pages. If hash_page is trying to insert a 4k page for a Linux PTE and it sees that it has already been inserted as a 64k page, it first invalidates the 64k HPTE before inserting the 4k HPTE. The hash invalidation routines also use the _PAGE_COMBO bit, to determine whether to look for a 64k HPTE or a set of 4k HPTEs to remove. With those two changes, we can tolerate a mix of 4k and 64k HPTEs in the hash table, and they will all get removed when the address space is torn down. Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-06-15 07:45:18 +07:00
unsigned long linear_llp, vmalloc_llp, lflags, vflags;
unsigned long ksp_esid_data, ksp_vsid_data;
linear_llp = mmu_psize_defs[mmu_linear_psize].sllp;
powerpc: Use 64k pages without needing cache-inhibited large pages Some POWER5+ machines can do 64k hardware pages for normal memory but not for cache-inhibited pages. This patch lets us use 64k hardware pages for most user processes on such machines (assuming the kernel has been configured with CONFIG_PPC_64K_PAGES=y). User processes start out using 64k pages and get switched to 4k pages if they use any non-cacheable mappings. With this, we use 64k pages for the vmalloc region and 4k pages for the imalloc region. If anything creates a non-cacheable mapping in the vmalloc region, the vmalloc region will get switched to 4k pages. I don't know of any driver other than the DRM that would do this, though, and these machines don't have AGP. When a region gets switched from 64k pages to 4k pages, we do not have to clear out all the 64k HPTEs from the hash table immediately. We use the _PAGE_COMBO bit in the Linux PTE to indicate whether the page was hashed in as a 64k page or a set of 4k pages. If hash_page is trying to insert a 4k page for a Linux PTE and it sees that it has already been inserted as a 64k page, it first invalidates the 64k HPTE before inserting the 4k HPTE. The hash invalidation routines also use the _PAGE_COMBO bit, to determine whether to look for a 64k HPTE or a set of 4k HPTEs to remove. With those two changes, we can tolerate a mix of 4k and 64k HPTEs in the hash table, and they will all get removed when the address space is torn down. Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-06-15 07:45:18 +07:00
vmalloc_llp = mmu_psize_defs[mmu_vmalloc_psize].sllp;
lflags = SLB_VSID_KERNEL | linear_llp;
powerpc: Use 64k pages without needing cache-inhibited large pages Some POWER5+ machines can do 64k hardware pages for normal memory but not for cache-inhibited pages. This patch lets us use 64k hardware pages for most user processes on such machines (assuming the kernel has been configured with CONFIG_PPC_64K_PAGES=y). User processes start out using 64k pages and get switched to 4k pages if they use any non-cacheable mappings. With this, we use 64k pages for the vmalloc region and 4k pages for the imalloc region. If anything creates a non-cacheable mapping in the vmalloc region, the vmalloc region will get switched to 4k pages. I don't know of any driver other than the DRM that would do this, though, and these machines don't have AGP. When a region gets switched from 64k pages to 4k pages, we do not have to clear out all the 64k HPTEs from the hash table immediately. We use the _PAGE_COMBO bit in the Linux PTE to indicate whether the page was hashed in as a 64k page or a set of 4k pages. If hash_page is trying to insert a 4k page for a Linux PTE and it sees that it has already been inserted as a 64k page, it first invalidates the 64k HPTE before inserting the 4k HPTE. The hash invalidation routines also use the _PAGE_COMBO bit, to determine whether to look for a 64k HPTE or a set of 4k HPTEs to remove. With those two changes, we can tolerate a mix of 4k and 64k HPTEs in the hash table, and they will all get removed when the address space is torn down. Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-06-15 07:45:18 +07:00
vflags = SLB_VSID_KERNEL | vmalloc_llp;
ksp_esid_data = mk_esid_data(get_paca()->kstack, mmu_kernel_ssize, 2);
if ((ksp_esid_data & ~0xfffffffUL) <= PAGE_OFFSET) {
ksp_esid_data &= ~SLB_ESID_V;
ksp_vsid_data = 0;
slb_shadow_clear(2);
} else {
/* Update stack entry; others don't change */
slb_shadow_update(get_paca()->kstack, mmu_kernel_ssize, lflags, 2);
ksp_vsid_data = get_slb_shadow()->save_area[2].vsid;
}
/* We need to do this all in asm, so we're sure we don't touch
* the stack between the slbia and rebolting it. */
asm volatile("isync\n"
"slbia\n"
/* Slot 1 - first VMALLOC segment */
"slbmte %0,%1\n"
/* Slot 2 - kernel stack */
"slbmte %2,%3\n"
"isync"
:: "r"(mk_vsid_data(VMALLOC_START, mmu_kernel_ssize, vflags)),
"r"(mk_esid_data(VMALLOC_START, mmu_kernel_ssize, 1)),
"r"(ksp_vsid_data),
"r"(ksp_esid_data)
: "memory");
}
powerpc: Allow perf_counters to access user memory at interrupt time This provides a mechanism to allow the perf_counters code to access user memory in a PMU interrupt routine. Such an access can cause various kinds of interrupt: SLB miss, MMU hash table miss, segment table miss, or TLB miss, depending on the processor. This commit only deals with 64-bit classic/server processors, which use an MMU hash table. 32-bit processors are already able to access user memory at interrupt time. Since we don't soft-disable on 32-bit, we avoid the possibility of reentering hash_page or the TLB miss handlers, since they run with interrupts disabled. On 64-bit processors, an SLB miss interrupt on a user address will update the slb_cache and slb_cache_ptr fields in the paca. This is OK except in the case where a PMU interrupt occurs in switch_slb, which also accesses those fields. To prevent this, we hard-disable interrupts in switch_slb. Interrupts are already soft-disabled at this point, and will get hard-enabled when they get soft-enabled later. This also reworks slb_flush_and_rebolt: to avoid hard-disabling twice, and to make sure that it clears the slb_cache_ptr when called from other callers than switch_slb, the existing routine is renamed to __slb_flush_and_rebolt, which is called by switch_slb and the new version of slb_flush_and_rebolt. Similarly, switch_stab (used on POWER3 and RS64 processors) gets a hard_irq_disable() to protect the per-cpu variables used there and in ste_allocate. If a MMU hashtable miss interrupt occurs, normally we would call hash_page to look up the Linux PTE for the address and create a HPTE. However, hash_page is fairly complex and takes some locks, so to avoid the possibility of deadlock, we check the preemption count to see if we are in a (pseudo-)NMI handler, and if so, we don't call hash_page but instead treat it like a bad access that will get reported up through the exception table mechanism. An interrupt whose handler runs even though the interrupt occurred when soft-disabled (such as the PMU interrupt) is considered a pseudo-NMI handler, which should use nmi_enter()/nmi_exit() rather than irq_enter()/irq_exit(). Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2009-08-17 12:17:54 +07:00
void slb_flush_and_rebolt(void)
{
WARN_ON(!irqs_disabled());
/*
* We can't take a PMU exception in the following code, so hard
* disable interrupts.
*/
hard_irq_disable();
__slb_flush_and_rebolt();
get_paca()->slb_cache_ptr = 0;
}
void slb_vmalloc_update(void)
{
unsigned long vflags;
vflags = SLB_VSID_KERNEL | mmu_psize_defs[mmu_vmalloc_psize].sllp;
slb_shadow_update(VMALLOC_START, mmu_kernel_ssize, vflags, 1);
slb_flush_and_rebolt();
}
/* Helper function to compare esids. There are four cases to handle.
* 1. The system is not 1T segment size capable. Use the GET_ESID compare.
* 2. The system is 1T capable, both addresses are < 1T, use the GET_ESID compare.
* 3. The system is 1T capable, only one of the two addresses is > 1T. This is not a match.
* 4. The system is 1T capable, both addresses are > 1T, use the GET_ESID_1T macro to compare.
*/
static inline int esids_match(unsigned long addr1, unsigned long addr2)
{
int esid_1t_count;
/* System is not 1T segment size capable. */
if (!mmu_has_feature(MMU_FTR_1T_SEGMENT))
return (GET_ESID(addr1) == GET_ESID(addr2));
esid_1t_count = (((addr1 >> SID_SHIFT_1T) != 0) +
((addr2 >> SID_SHIFT_1T) != 0));
/* both addresses are < 1T */
if (esid_1t_count == 0)
return (GET_ESID(addr1) == GET_ESID(addr2));
/* One address < 1T, the other > 1T. Not a match */
if (esid_1t_count == 1)
return 0;
/* Both addresses are > 1T. */
return (GET_ESID_1T(addr1) == GET_ESID_1T(addr2));
}
/* Flush all user entries from the segment table of the current processor. */
void switch_slb(struct task_struct *tsk, struct mm_struct *mm)
{
powerpc: Allow perf_counters to access user memory at interrupt time This provides a mechanism to allow the perf_counters code to access user memory in a PMU interrupt routine. Such an access can cause various kinds of interrupt: SLB miss, MMU hash table miss, segment table miss, or TLB miss, depending on the processor. This commit only deals with 64-bit classic/server processors, which use an MMU hash table. 32-bit processors are already able to access user memory at interrupt time. Since we don't soft-disable on 32-bit, we avoid the possibility of reentering hash_page or the TLB miss handlers, since they run with interrupts disabled. On 64-bit processors, an SLB miss interrupt on a user address will update the slb_cache and slb_cache_ptr fields in the paca. This is OK except in the case where a PMU interrupt occurs in switch_slb, which also accesses those fields. To prevent this, we hard-disable interrupts in switch_slb. Interrupts are already soft-disabled at this point, and will get hard-enabled when they get soft-enabled later. This also reworks slb_flush_and_rebolt: to avoid hard-disabling twice, and to make sure that it clears the slb_cache_ptr when called from other callers than switch_slb, the existing routine is renamed to __slb_flush_and_rebolt, which is called by switch_slb and the new version of slb_flush_and_rebolt. Similarly, switch_stab (used on POWER3 and RS64 processors) gets a hard_irq_disable() to protect the per-cpu variables used there and in ste_allocate. If a MMU hashtable miss interrupt occurs, normally we would call hash_page to look up the Linux PTE for the address and create a HPTE. However, hash_page is fairly complex and takes some locks, so to avoid the possibility of deadlock, we check the preemption count to see if we are in a (pseudo-)NMI handler, and if so, we don't call hash_page but instead treat it like a bad access that will get reported up through the exception table mechanism. An interrupt whose handler runs even though the interrupt occurred when soft-disabled (such as the PMU interrupt) is considered a pseudo-NMI handler, which should use nmi_enter()/nmi_exit() rather than irq_enter()/irq_exit(). Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2009-08-17 12:17:54 +07:00
unsigned long offset;
unsigned long slbie_data = 0;
unsigned long pc = KSTK_EIP(tsk);
unsigned long stack = KSTK_ESP(tsk);
unsigned long exec_base;
powerpc: Allow perf_counters to access user memory at interrupt time This provides a mechanism to allow the perf_counters code to access user memory in a PMU interrupt routine. Such an access can cause various kinds of interrupt: SLB miss, MMU hash table miss, segment table miss, or TLB miss, depending on the processor. This commit only deals with 64-bit classic/server processors, which use an MMU hash table. 32-bit processors are already able to access user memory at interrupt time. Since we don't soft-disable on 32-bit, we avoid the possibility of reentering hash_page or the TLB miss handlers, since they run with interrupts disabled. On 64-bit processors, an SLB miss interrupt on a user address will update the slb_cache and slb_cache_ptr fields in the paca. This is OK except in the case where a PMU interrupt occurs in switch_slb, which also accesses those fields. To prevent this, we hard-disable interrupts in switch_slb. Interrupts are already soft-disabled at this point, and will get hard-enabled when they get soft-enabled later. This also reworks slb_flush_and_rebolt: to avoid hard-disabling twice, and to make sure that it clears the slb_cache_ptr when called from other callers than switch_slb, the existing routine is renamed to __slb_flush_and_rebolt, which is called by switch_slb and the new version of slb_flush_and_rebolt. Similarly, switch_stab (used on POWER3 and RS64 processors) gets a hard_irq_disable() to protect the per-cpu variables used there and in ste_allocate. If a MMU hashtable miss interrupt occurs, normally we would call hash_page to look up the Linux PTE for the address and create a HPTE. However, hash_page is fairly complex and takes some locks, so to avoid the possibility of deadlock, we check the preemption count to see if we are in a (pseudo-)NMI handler, and if so, we don't call hash_page but instead treat it like a bad access that will get reported up through the exception table mechanism. An interrupt whose handler runs even though the interrupt occurred when soft-disabled (such as the PMU interrupt) is considered a pseudo-NMI handler, which should use nmi_enter()/nmi_exit() rather than irq_enter()/irq_exit(). Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2009-08-17 12:17:54 +07:00
/*
* We need interrupts hard-disabled here, not just soft-disabled,
* so that a PMU interrupt can't occur, which might try to access
* user memory (to get a stack trace) and possible cause an SLB miss
* which would update the slb_cache/slb_cache_ptr fields in the PACA.
*/
hard_irq_disable();
offset = get_paca()->slb_cache_ptr;
if (!mmu_has_feature(MMU_FTR_NO_SLBIE_B) &&
offset <= SLB_CACHE_ENTRIES) {
int i;
asm volatile("isync" : : : "memory");
for (i = 0; i < offset; i++) {
slbie_data = (unsigned long)get_paca()->slb_cache[i]
<< SID_SHIFT; /* EA */
slbie_data |= user_segment_size(slbie_data)
<< SLBIE_SSIZE_SHIFT;
slbie_data |= SLBIE_C; /* C set for user addresses */
asm volatile("slbie %0" : : "r" (slbie_data));
}
asm volatile("isync" : : : "memory");
} else {
powerpc: Allow perf_counters to access user memory at interrupt time This provides a mechanism to allow the perf_counters code to access user memory in a PMU interrupt routine. Such an access can cause various kinds of interrupt: SLB miss, MMU hash table miss, segment table miss, or TLB miss, depending on the processor. This commit only deals with 64-bit classic/server processors, which use an MMU hash table. 32-bit processors are already able to access user memory at interrupt time. Since we don't soft-disable on 32-bit, we avoid the possibility of reentering hash_page or the TLB miss handlers, since they run with interrupts disabled. On 64-bit processors, an SLB miss interrupt on a user address will update the slb_cache and slb_cache_ptr fields in the paca. This is OK except in the case where a PMU interrupt occurs in switch_slb, which also accesses those fields. To prevent this, we hard-disable interrupts in switch_slb. Interrupts are already soft-disabled at this point, and will get hard-enabled when they get soft-enabled later. This also reworks slb_flush_and_rebolt: to avoid hard-disabling twice, and to make sure that it clears the slb_cache_ptr when called from other callers than switch_slb, the existing routine is renamed to __slb_flush_and_rebolt, which is called by switch_slb and the new version of slb_flush_and_rebolt. Similarly, switch_stab (used on POWER3 and RS64 processors) gets a hard_irq_disable() to protect the per-cpu variables used there and in ste_allocate. If a MMU hashtable miss interrupt occurs, normally we would call hash_page to look up the Linux PTE for the address and create a HPTE. However, hash_page is fairly complex and takes some locks, so to avoid the possibility of deadlock, we check the preemption count to see if we are in a (pseudo-)NMI handler, and if so, we don't call hash_page but instead treat it like a bad access that will get reported up through the exception table mechanism. An interrupt whose handler runs even though the interrupt occurred when soft-disabled (such as the PMU interrupt) is considered a pseudo-NMI handler, which should use nmi_enter()/nmi_exit() rather than irq_enter()/irq_exit(). Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2009-08-17 12:17:54 +07:00
__slb_flush_and_rebolt();
}
/* Workaround POWER5 < DD2.1 issue */
if (offset == 1 || offset > SLB_CACHE_ENTRIES)
asm volatile("slbie %0" : : "r" (slbie_data));
get_paca()->slb_cache_ptr = 0;
get_paca()->context = mm->context;
/*
* preload some userspace segments into the SLB.
* Almost all 32 and 64bit PowerPC executables are linked at
* 0x10000000 so it makes sense to preload this segment.
*/
exec_base = 0x10000000;
if (is_kernel_addr(pc) || is_kernel_addr(stack) ||
is_kernel_addr(exec_base))
return;
slb_allocate(pc);
if (!esids_match(pc, stack))
slb_allocate(stack);
if (!esids_match(pc, exec_base) &&
!esids_match(stack, exec_base))
slb_allocate(exec_base);
}
static inline void patch_slb_encoding(unsigned int *insn_addr,
unsigned int immed)
{
int insn = (*insn_addr & 0xffff0000) | immed;
patch_instruction(insn_addr, insn);
}
void slb_set_size(u16 size)
{
extern unsigned int *slb_compare_rr_to_size;
if (mmu_slb_size == size)
return;
mmu_slb_size = size;
patch_slb_encoding(slb_compare_rr_to_size, mmu_slb_size);
}
void slb_initialize(void)
{
powerpc: Use 64k pages without needing cache-inhibited large pages Some POWER5+ machines can do 64k hardware pages for normal memory but not for cache-inhibited pages. This patch lets us use 64k hardware pages for most user processes on such machines (assuming the kernel has been configured with CONFIG_PPC_64K_PAGES=y). User processes start out using 64k pages and get switched to 4k pages if they use any non-cacheable mappings. With this, we use 64k pages for the vmalloc region and 4k pages for the imalloc region. If anything creates a non-cacheable mapping in the vmalloc region, the vmalloc region will get switched to 4k pages. I don't know of any driver other than the DRM that would do this, though, and these machines don't have AGP. When a region gets switched from 64k pages to 4k pages, we do not have to clear out all the 64k HPTEs from the hash table immediately. We use the _PAGE_COMBO bit in the Linux PTE to indicate whether the page was hashed in as a 64k page or a set of 4k pages. If hash_page is trying to insert a 4k page for a Linux PTE and it sees that it has already been inserted as a 64k page, it first invalidates the 64k HPTE before inserting the 4k HPTE. The hash invalidation routines also use the _PAGE_COMBO bit, to determine whether to look for a 64k HPTE or a set of 4k HPTEs to remove. With those two changes, we can tolerate a mix of 4k and 64k HPTEs in the hash table, and they will all get removed when the address space is torn down. Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-06-15 07:45:18 +07:00
unsigned long linear_llp, vmalloc_llp, io_llp;
unsigned long lflags, vflags;
static int slb_encoding_inited;
extern unsigned int *slb_miss_kernel_load_linear;
powerpc: Use 64k pages without needing cache-inhibited large pages Some POWER5+ machines can do 64k hardware pages for normal memory but not for cache-inhibited pages. This patch lets us use 64k hardware pages for most user processes on such machines (assuming the kernel has been configured with CONFIG_PPC_64K_PAGES=y). User processes start out using 64k pages and get switched to 4k pages if they use any non-cacheable mappings. With this, we use 64k pages for the vmalloc region and 4k pages for the imalloc region. If anything creates a non-cacheable mapping in the vmalloc region, the vmalloc region will get switched to 4k pages. I don't know of any driver other than the DRM that would do this, though, and these machines don't have AGP. When a region gets switched from 64k pages to 4k pages, we do not have to clear out all the 64k HPTEs from the hash table immediately. We use the _PAGE_COMBO bit in the Linux PTE to indicate whether the page was hashed in as a 64k page or a set of 4k pages. If hash_page is trying to insert a 4k page for a Linux PTE and it sees that it has already been inserted as a 64k page, it first invalidates the 64k HPTE before inserting the 4k HPTE. The hash invalidation routines also use the _PAGE_COMBO bit, to determine whether to look for a 64k HPTE or a set of 4k HPTEs to remove. With those two changes, we can tolerate a mix of 4k and 64k HPTEs in the hash table, and they will all get removed when the address space is torn down. Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-06-15 07:45:18 +07:00
extern unsigned int *slb_miss_kernel_load_io;
extern unsigned int *slb_compare_rr_to_size;
#ifdef CONFIG_SPARSEMEM_VMEMMAP
extern unsigned int *slb_miss_kernel_load_vmemmap;
unsigned long vmemmap_llp;
#endif
/* Prepare our SLB miss handler based on our page size */
linear_llp = mmu_psize_defs[mmu_linear_psize].sllp;
powerpc: Use 64k pages without needing cache-inhibited large pages Some POWER5+ machines can do 64k hardware pages for normal memory but not for cache-inhibited pages. This patch lets us use 64k hardware pages for most user processes on such machines (assuming the kernel has been configured with CONFIG_PPC_64K_PAGES=y). User processes start out using 64k pages and get switched to 4k pages if they use any non-cacheable mappings. With this, we use 64k pages for the vmalloc region and 4k pages for the imalloc region. If anything creates a non-cacheable mapping in the vmalloc region, the vmalloc region will get switched to 4k pages. I don't know of any driver other than the DRM that would do this, though, and these machines don't have AGP. When a region gets switched from 64k pages to 4k pages, we do not have to clear out all the 64k HPTEs from the hash table immediately. We use the _PAGE_COMBO bit in the Linux PTE to indicate whether the page was hashed in as a 64k page or a set of 4k pages. If hash_page is trying to insert a 4k page for a Linux PTE and it sees that it has already been inserted as a 64k page, it first invalidates the 64k HPTE before inserting the 4k HPTE. The hash invalidation routines also use the _PAGE_COMBO bit, to determine whether to look for a 64k HPTE or a set of 4k HPTEs to remove. With those two changes, we can tolerate a mix of 4k and 64k HPTEs in the hash table, and they will all get removed when the address space is torn down. Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-06-15 07:45:18 +07:00
io_llp = mmu_psize_defs[mmu_io_psize].sllp;
vmalloc_llp = mmu_psize_defs[mmu_vmalloc_psize].sllp;
get_paca()->vmalloc_sllp = SLB_VSID_KERNEL | vmalloc_llp;
#ifdef CONFIG_SPARSEMEM_VMEMMAP
vmemmap_llp = mmu_psize_defs[mmu_vmemmap_psize].sllp;
#endif
if (!slb_encoding_inited) {
slb_encoding_inited = 1;
patch_slb_encoding(slb_miss_kernel_load_linear,
SLB_VSID_KERNEL | linear_llp);
powerpc: Use 64k pages without needing cache-inhibited large pages Some POWER5+ machines can do 64k hardware pages for normal memory but not for cache-inhibited pages. This patch lets us use 64k hardware pages for most user processes on such machines (assuming the kernel has been configured with CONFIG_PPC_64K_PAGES=y). User processes start out using 64k pages and get switched to 4k pages if they use any non-cacheable mappings. With this, we use 64k pages for the vmalloc region and 4k pages for the imalloc region. If anything creates a non-cacheable mapping in the vmalloc region, the vmalloc region will get switched to 4k pages. I don't know of any driver other than the DRM that would do this, though, and these machines don't have AGP. When a region gets switched from 64k pages to 4k pages, we do not have to clear out all the 64k HPTEs from the hash table immediately. We use the _PAGE_COMBO bit in the Linux PTE to indicate whether the page was hashed in as a 64k page or a set of 4k pages. If hash_page is trying to insert a 4k page for a Linux PTE and it sees that it has already been inserted as a 64k page, it first invalidates the 64k HPTE before inserting the 4k HPTE. The hash invalidation routines also use the _PAGE_COMBO bit, to determine whether to look for a 64k HPTE or a set of 4k HPTEs to remove. With those two changes, we can tolerate a mix of 4k and 64k HPTEs in the hash table, and they will all get removed when the address space is torn down. Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-06-15 07:45:18 +07:00
patch_slb_encoding(slb_miss_kernel_load_io,
SLB_VSID_KERNEL | io_llp);
patch_slb_encoding(slb_compare_rr_to_size,
mmu_slb_size);
pr_devel("SLB: linear LLP = %04lx\n", linear_llp);
pr_devel("SLB: io LLP = %04lx\n", io_llp);
#ifdef CONFIG_SPARSEMEM_VMEMMAP
patch_slb_encoding(slb_miss_kernel_load_vmemmap,
SLB_VSID_KERNEL | vmemmap_llp);
pr_devel("SLB: vmemmap LLP = %04lx\n", vmemmap_llp);
#endif
}
get_paca()->stab_rr = SLB_NUM_BOLTED;
/* On iSeries the bolted entries have already been set up by
* the hypervisor from the lparMap data in head.S */
if (firmware_has_feature(FW_FEATURE_ISERIES))
return;
lflags = SLB_VSID_KERNEL | linear_llp;
powerpc: Use 64k pages without needing cache-inhibited large pages Some POWER5+ machines can do 64k hardware pages for normal memory but not for cache-inhibited pages. This patch lets us use 64k hardware pages for most user processes on such machines (assuming the kernel has been configured with CONFIG_PPC_64K_PAGES=y). User processes start out using 64k pages and get switched to 4k pages if they use any non-cacheable mappings. With this, we use 64k pages for the vmalloc region and 4k pages for the imalloc region. If anything creates a non-cacheable mapping in the vmalloc region, the vmalloc region will get switched to 4k pages. I don't know of any driver other than the DRM that would do this, though, and these machines don't have AGP. When a region gets switched from 64k pages to 4k pages, we do not have to clear out all the 64k HPTEs from the hash table immediately. We use the _PAGE_COMBO bit in the Linux PTE to indicate whether the page was hashed in as a 64k page or a set of 4k pages. If hash_page is trying to insert a 4k page for a Linux PTE and it sees that it has already been inserted as a 64k page, it first invalidates the 64k HPTE before inserting the 4k HPTE. The hash invalidation routines also use the _PAGE_COMBO bit, to determine whether to look for a 64k HPTE or a set of 4k HPTEs to remove. With those two changes, we can tolerate a mix of 4k and 64k HPTEs in the hash table, and they will all get removed when the address space is torn down. Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-06-15 07:45:18 +07:00
vflags = SLB_VSID_KERNEL | vmalloc_llp;
/* Invalidate the entire SLB (even slot 0) & all the ERATS */
asm volatile("isync":::"memory");
asm volatile("slbmte %0,%0"::"r" (0) : "memory");
asm volatile("isync; slbia; isync":::"memory");
create_shadowed_slbe(PAGE_OFFSET, mmu_kernel_ssize, lflags, 0);
create_shadowed_slbe(VMALLOC_START, mmu_kernel_ssize, vflags, 1);
[POWERPC] Bolt in SLB entry for kernel stack on secondary cpus This fixes a regression reported by Kamalesh Bulabel where a POWER4 machine would crash because of an SLB miss at a point where the SLB miss exception was unrecoverable. This regression is tracked at: http://bugzilla.kernel.org/show_bug.cgi?id=10082 SLB misses at such points shouldn't happen because the kernel stack is the only memory accessed other than things in the first segment of the linear mapping (which is mapped at all times by entry 0 of the SLB). The context switch code ensures that SLB entry 2 covers the kernel stack, if it is not already covered by entry 0. None of entries 0 to 2 are ever replaced by the SLB miss handler. Where this went wrong is that the context switch code assumes it doesn't have to write to SLB entry 2 if the new kernel stack is in the same segment as the old kernel stack, since entry 2 should already be correct. However, when we start up a secondary cpu, it calls slb_initialize, which doesn't set up entry 2. This is correct for the boot cpu, where we will be using a stack in the kernel BSS at this point (i.e. init_thread_union), but not necessarily for secondary cpus, whose initial stack can be allocated anywhere. This doesn't cause any immediate problem since the SLB miss handler will just create an SLB entry somewhere else to cover the initial stack. In fact it's possible for the cpu to go quite a long time without SLB entry 2 being valid. Eventually, though, the entry created by the SLB miss handler will get overwritten by some other entry, and if the next access to the stack is at an unrecoverable point, we get the crash. This fixes the problem by making slb_initialize create a suitable entry for the kernel stack, if we are on a secondary cpu and the stack isn't covered by SLB entry 0. This requires initializing the get_paca()->kstack field earlier, so I do that in smp_create_idle where the current field is initialized. This also abstracts a bit of the computation that mk_esid_data in slb.c does so that it can be used in slb_initialize. Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-05-02 11:29:12 +07:00
/* For the boot cpu, we're running on the stack in init_thread_union,
* which is in the first segment of the linear mapping, and also
* get_paca()->kstack hasn't been initialized yet.
* For secondary cpus, we need to bolt the kernel stack entry now.
*/
slb_shadow_clear(2);
[POWERPC] Bolt in SLB entry for kernel stack on secondary cpus This fixes a regression reported by Kamalesh Bulabel where a POWER4 machine would crash because of an SLB miss at a point where the SLB miss exception was unrecoverable. This regression is tracked at: http://bugzilla.kernel.org/show_bug.cgi?id=10082 SLB misses at such points shouldn't happen because the kernel stack is the only memory accessed other than things in the first segment of the linear mapping (which is mapped at all times by entry 0 of the SLB). The context switch code ensures that SLB entry 2 covers the kernel stack, if it is not already covered by entry 0. None of entries 0 to 2 are ever replaced by the SLB miss handler. Where this went wrong is that the context switch code assumes it doesn't have to write to SLB entry 2 if the new kernel stack is in the same segment as the old kernel stack, since entry 2 should already be correct. However, when we start up a secondary cpu, it calls slb_initialize, which doesn't set up entry 2. This is correct for the boot cpu, where we will be using a stack in the kernel BSS at this point (i.e. init_thread_union), but not necessarily for secondary cpus, whose initial stack can be allocated anywhere. This doesn't cause any immediate problem since the SLB miss handler will just create an SLB entry somewhere else to cover the initial stack. In fact it's possible for the cpu to go quite a long time without SLB entry 2 being valid. Eventually, though, the entry created by the SLB miss handler will get overwritten by some other entry, and if the next access to the stack is at an unrecoverable point, we get the crash. This fixes the problem by making slb_initialize create a suitable entry for the kernel stack, if we are on a secondary cpu and the stack isn't covered by SLB entry 0. This requires initializing the get_paca()->kstack field earlier, so I do that in smp_create_idle where the current field is initialized. This also abstracts a bit of the computation that mk_esid_data in slb.c does so that it can be used in slb_initialize. Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-05-02 11:29:12 +07:00
if (raw_smp_processor_id() != boot_cpuid &&
(get_paca()->kstack & slb_esid_mask(mmu_kernel_ssize)) > PAGE_OFFSET)
create_shadowed_slbe(get_paca()->kstack,
mmu_kernel_ssize, lflags, 2);
asm volatile("isync":::"memory");
}