renegade-project/linux-next
2
0
Fork 0
mirror of https://github.com/edk2-porting/linux-next.git synced 2024-12-26 06:04:14 +08:00
Code
2d78d4beb6
linux-next/include/asm-sparc64/mmu_context.h

153 lines
4.4 KiB
C
Raw Normal View History

Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* $Id: mmu_context.h,v 1.54 2002/02/09 19:49:31 davem Exp $ */
#ifndef __SPARC64_MMU_CONTEXT_H
#define __SPARC64_MMU_CONTEXT_H
/* Derived heavily from Linus's Alpha/AXP ASN code... */
#ifndef __ASSEMBLY__
#include <linux/spinlock.h>
#include <asm/system.h>
#include <asm/spitfire.h>
static inline void enter_lazy_tlb(struct mm_struct *mm, struct task_struct *tsk)
{
}
extern spinlock_t ctx_alloc_lock;
extern unsigned long tlb_context_cache;
extern unsigned long mmu_context_bmap[];
extern void get_new_mmu_context(struct mm_struct *mm);
[SPARC64]: Fix TLB context allocation with SMT style shared TLBs. The context allocation scheme we use depends upon there being a 1<-->1 mapping from cpu to physical TLB for correctness. Chips like Niagara break this assumption. So what we do is notify all cpus with a cross call when the context version number changes, and if necessary this makes them allocate a valid context for the address space they are running at the time. Stress tested with make -j1024, make -j2048, and make -j4096 kernel builds on a 32-strand, 8 core, T2000 with 16GB of ram. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-24 06:19:28 +08:00
#ifdef CONFIG_SMP
extern void smp_new_mmu_context_version(void);
#else
#define smp_new_mmu_context_version() do { } while (0)
#endif
[SPARC64]: TSB refinements. Move {init_new,destroy}_context() out of line. Do not put huge pages into the TSB, only base page size translations. There are some clever things we could do here, but for now let's be correct instead of fancy. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-01 10:31:06 +08:00
extern int init_new_context(struct task_struct *tsk, struct mm_struct *mm);
extern void destroy_context(struct mm_struct *mm);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[SPARC64]: Hypervisor TSB context switching. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-10 09:21:53 +08:00
extern void __tsb_context_switch(unsigned long pgd_pa,
[SPARC64]: Add a secondary TSB for hugepage mappings. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-22 16:49:59 +08:00
struct tsb_config *tsb_base,
struct tsb_config *tsb_huge,
[SPARC64]: Hypervisor TSB context switching. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-10 09:21:53 +08:00
unsigned long tsb_descr_pa);
[SPARC64]: Add infrastructure for dynamic TSB sizing. This also cleans up tsb_context_switch(). The assembler routine is now __tsb_context_switch() and the former is an inline function that picks out the bits from the mm_struct and passes it into the assembler code as arguments. setup_tsb_parms() computes the locked TLB entry to map the TSB. Later when we support using the physical address quad load instructions of Cheetah+ and later, we'll simply use the physical address for the TSB register value and set the map virtual and PTE both to zero. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-01 10:31:20 +08:00
static inline void tsb_context_switch(struct mm_struct *mm)
{
[SPARC64]: Add a secondary TSB for hugepage mappings. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-22 16:49:59 +08:00
__tsb_context_switch(__pa(mm->pgd),
&mm->context.tsb_block[0],
#ifdef CONFIG_HUGETLB_PAGE
(mm->context.tsb_block[1].tsb ?
&mm->context.tsb_block[1] :
NULL)
#else
NULL
#endif
, __pa(&mm->context.tsb_descr[0]));
[SPARC64]: Add infrastructure for dynamic TSB sizing. This also cleans up tsb_context_switch(). The assembler routine is now __tsb_context_switch() and the former is an inline function that picks out the bits from the mm_struct and passes it into the assembler code as arguments. setup_tsb_parms() computes the locked TLB entry to map the TSB. Later when we support using the physical address quad load instructions of Cheetah+ and later, we'll simply use the physical address for the TSB register value and set the map virtual and PTE both to zero. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-01 10:31:20 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[SPARC64]: Add a secondary TSB for hugepage mappings. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-22 16:49:59 +08:00
extern void tsb_grow(struct mm_struct *mm, unsigned long tsb_index, unsigned long mm_rss);
[SPARC64]: Dynamically grow TSB in response to RSS growth. As the RSS grows, grow the TSB in order to reduce the likelyhood of hash collisions and thus poor hit rates in the TSB. This definitely needs some serious tuning. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-01 10:31:38 +08:00
#ifdef CONFIG_SMP
extern void smp_tsb_sync(struct mm_struct *mm);
#else
#define smp_tsb_sync(__mm) do { } while (0)
#endif
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* Set MMU context in the actual hardware. */
#define load_secondary_context(__mm) \
[SPARC64]: Patch up mmu context register writes for sun4v. sun4v uses ASI_MMU instead of ASI_DMMU Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-08 14:13:05 +08:00
__asm__ __volatile__( \
"\n661: stxa %0, [%1] %2\n" \
" .section .sun4v_1insn_patch, \"ax\"\n" \
" .word 661b\n" \
" stxa %0, [%1] %3\n" \
" .previous\n" \
" flush %%g6\n" \
: /* No outputs */ \
: "r" (CTX_HWBITS((__mm)->context)), \
"r" (SECONDARY_CONTEXT), "i" (ASI_DMMU), "i" (ASI_MMU))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
extern void __flush_tlb_mm(unsigned long, unsigned long);
[SPARC64]: Fix TLB context allocation with SMT style shared TLBs. The context allocation scheme we use depends upon there being a 1<-->1 mapping from cpu to physical TLB for correctness. Chips like Niagara break this assumption. So what we do is notify all cpus with a cross call when the context version number changes, and if necessary this makes them allocate a valid context for the address space they are running at the time. Stress tested with make -j1024, make -j2048, and make -j4096 kernel builds on a 32-strand, 8 core, T2000 with 16GB of ram. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-24 06:19:28 +08:00
/* Switch the current MM context. Interrupts are disabled. */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static inline void switch_mm(struct mm_struct *old_mm, struct mm_struct *mm, struct task_struct *tsk)
{
[SPARC64]: Bulletproof MMU context locking. 1) Always spin_lock_init() in init_context(). The caller essentially clears it out, or copies the mm info from the parent. In both cases we need to explicitly initialize the spinlock. 2) Always do explicit IRQ disabling while taking mm->context.lock and ctx_alloc_lock. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-07 11:59:50 +08:00
unsigned long ctx_valid, flags;
[SPARC64] mm: context switch ptlock sparc64 is unique among architectures in taking the page_table_lock in its context switch (well, cris does too, but erroneously, and it's not yet SMP anyway). This seems to be a private affair between switch_mm and activate_mm, using page_table_lock as a per-mm lock, without any relation to its uses elsewhere. That's fine, but comment it as such; and unlock sooner in switch_mm, more like in activate_mm (preemption is disabled here). There is a block of "if (0)"ed code in smp_flush_tlb_pending which would have liked to rely on the page_table_lock, in switch_mm and elsewhere; but its comment explains how dup_mmap's flush_tlb_mm defeated it. And though that could have been changed at any time over the past few years, now the chance vanishes as we push the page_table_lock downwards, and perhaps split it per page table page. Just delete that block of code. Which leaves the mysterious spin_unlock_wait(&oldmm->page_table_lock) in kernel/fork.c copy_mm. Textual analysis (supported by Nick Piggin) suggests that the comment was written by DaveM, and that it relates to the defeated approach in the sparc64 smp_flush_tlb_pending. Just delete this block too. Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-11-08 06:09:01 +08:00
int cpu;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[SPARC64]: Bulletproof MMU context locking. 1) Always spin_lock_init() in init_context(). The caller essentially clears it out, or copies the mm info from the parent. In both cases we need to explicitly initialize the spinlock. 2) Always do explicit IRQ disabling while taking mm->context.lock and ctx_alloc_lock. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-07 11:59:50 +08:00
spin_lock_irqsave(&mm->context.lock, flags);
[SPARC64] mm: context switch ptlock sparc64 is unique among architectures in taking the page_table_lock in its context switch (well, cris does too, but erroneously, and it's not yet SMP anyway). This seems to be a private affair between switch_mm and activate_mm, using page_table_lock as a per-mm lock, without any relation to its uses elsewhere. That's fine, but comment it as such; and unlock sooner in switch_mm, more like in activate_mm (preemption is disabled here). There is a block of "if (0)"ed code in smp_flush_tlb_pending which would have liked to rely on the page_table_lock, in switch_mm and elsewhere; but its comment explains how dup_mmap's flush_tlb_mm defeated it. And though that could have been changed at any time over the past few years, now the chance vanishes as we push the page_table_lock downwards, and perhaps split it per page table page. Just delete that block of code. Which leaves the mysterious spin_unlock_wait(&oldmm->page_table_lock) in kernel/fork.c copy_mm. Textual analysis (supported by Nick Piggin) suggests that the comment was written by DaveM, and that it relates to the defeated approach in the sparc64 smp_flush_tlb_pending. Just delete this block too. Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-11-08 06:09:01 +08:00
ctx_valid = CTX_VALID(mm->context);
if (!ctx_valid)
get_new_mmu_context(mm);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[SPARC64]: Fix and re-enable dynamic TSB sizing. This is good for up to %50 performance improvement of some test cases. The problem has been the race conditions, and hopefully I've plugged them all up here. 1) There was a serious race in switch_mm() wrt. lazy TLB switching to and from kernel threads. We could erroneously skip a tsb_context_switch() and thus use a stale TSB across a TSB grow event. There is a big comment now in that function describing exactly how it can happen. 2) All code paths that do something with the TSB need to be guarded with the mm->context.lock spinlock. This makes page table flushing paths properly synchronize with both TSB growing and TLB context changes. 3) TSB growing events are moved to the end of successful fault processing. Previously it was in update_mmu_cache() but that is deadlock prone. At the end of do_sparc64_fault() we hold no spinlocks that could deadlock the TSB grow sequence. We also have dropped the address space semaphore. While we're here, add prefetching to the copy_tsb() routine and put it in assembler into the tsb.S file. This piece of code is quite time critical. There are some small negative side effects to this code which can be improved upon. In particular we grab the mm->context.lock even for the tsb insert done by update_mmu_cache() now and that's a bit excessive. We can get rid of that locking, and the same lock taking in flush_tsb_user(), by disabling PSTATE_IE around the whole operation including the capturing of the tsb pointer and tsb_nentries value. That would work because anyone growing the TSB won't free up the old TSB until all cpus respond to the TSB change cross call. I'm not quite so confident in that optimization to put it in right now, but eventually we might be able to and the description is here for reference. This code seems very solid now. It passes several parallel GCC bootstrap builds, and our favorite "nut cruncher" stress test which is a full "make -j8192" build of a "make allmodconfig" kernel. That puts about 256 processes on each cpu's run queue, makes lots of process cpu migrations occur, causes lots of page table and TLB flushing activity, incurs many context version number changes, and it swaps the machine real far out to disk even though there is 16GB of ram on this test system. :-) Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-16 18:02:32 +08:00
/* We have to be extremely careful here or else we will miss
* a TSB grow if we switch back and forth between a kernel
* thread and an address space which has it's TSB size increased
* on another processor.
*
* It is possible to play some games in order to optimize the
* switch, but the safest thing to do is to unconditionally
* perform the secondary context load and the TSB context switch.
*
* For reference the bad case is, for address space "A":
*
* CPU 0 CPU 1
* run address space A
* set cpu0's bits in cpu_vm_mask
* switch to kernel thread, borrow
* address space A via entry_lazy_tlb
* run address space A
* set cpu1's bit in cpu_vm_mask
* flush_tlb_pending()
* reset cpu_vm_mask to just cpu1
* TSB grow
* run address space A
* context was valid, so skip
* TSB context switch
*
* At that point cpu0 continues to use a stale TSB, the one from
* before the TSB grow performed on cpu1. cpu1 did not cross-call
* cpu0 to update it's TSB because at that point the cpu_vm_mask
* only had cpu1 set in it.
*/
load_secondary_context(mm);
tsb_context_switch(mm);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[SPARC64]: Fix and re-enable dynamic TSB sizing. This is good for up to %50 performance improvement of some test cases. The problem has been the race conditions, and hopefully I've plugged them all up here. 1) There was a serious race in switch_mm() wrt. lazy TLB switching to and from kernel threads. We could erroneously skip a tsb_context_switch() and thus use a stale TSB across a TSB grow event. There is a big comment now in that function describing exactly how it can happen. 2) All code paths that do something with the TSB need to be guarded with the mm->context.lock spinlock. This makes page table flushing paths properly synchronize with both TSB growing and TLB context changes. 3) TSB growing events are moved to the end of successful fault processing. Previously it was in update_mmu_cache() but that is deadlock prone. At the end of do_sparc64_fault() we hold no spinlocks that could deadlock the TSB grow sequence. We also have dropped the address space semaphore. While we're here, add prefetching to the copy_tsb() routine and put it in assembler into the tsb.S file. This piece of code is quite time critical. There are some small negative side effects to this code which can be improved upon. In particular we grab the mm->context.lock even for the tsb insert done by update_mmu_cache() now and that's a bit excessive. We can get rid of that locking, and the same lock taking in flush_tsb_user(), by disabling PSTATE_IE around the whole operation including the capturing of the tsb pointer and tsb_nentries value. That would work because anyone growing the TSB won't free up the old TSB until all cpus respond to the TSB change cross call. I'm not quite so confident in that optimization to put it in right now, but eventually we might be able to and the description is here for reference. This code seems very solid now. It passes several parallel GCC bootstrap builds, and our favorite "nut cruncher" stress test which is a full "make -j8192" build of a "make allmodconfig" kernel. That puts about 256 processes on each cpu's run queue, makes lots of process cpu migrations occur, causes lots of page table and TLB flushing activity, incurs many context version number changes, and it swaps the machine real far out to disk even though there is 16GB of ram on this test system. :-) Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-16 18:02:32 +08:00
/* Any time a processor runs a context on an address space
* for the first time, we must flush that context out of the
* local TLB.
[SPARC64] mm: context switch ptlock sparc64 is unique among architectures in taking the page_table_lock in its context switch (well, cris does too, but erroneously, and it's not yet SMP anyway). This seems to be a private affair between switch_mm and activate_mm, using page_table_lock as a per-mm lock, without any relation to its uses elsewhere. That's fine, but comment it as such; and unlock sooner in switch_mm, more like in activate_mm (preemption is disabled here). There is a block of "if (0)"ed code in smp_flush_tlb_pending which would have liked to rely on the page_table_lock, in switch_mm and elsewhere; but its comment explains how dup_mmap's flush_tlb_mm defeated it. And though that could have been changed at any time over the past few years, now the chance vanishes as we push the page_table_lock downwards, and perhaps split it per page table page. Just delete that block of code. Which leaves the mysterious spin_unlock_wait(&oldmm->page_table_lock) in kernel/fork.c copy_mm. Textual analysis (supported by Nick Piggin) suggests that the comment was written by DaveM, and that it relates to the defeated approach in the sparc64 smp_flush_tlb_pending. Just delete this block too. Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-11-08 06:09:01 +08:00
*/
cpu = smp_processor_id();
if (!ctx_valid || !cpu_isset(cpu, mm->cpu_vm_mask)) {
cpu_set(cpu, mm->cpu_vm_mask);
__flush_tlb_mm(CTX_HWBITS(mm->context),
SECONDARY_CONTEXT);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[SPARC64]: Fix and re-enable dynamic TSB sizing. This is good for up to %50 performance improvement of some test cases. The problem has been the race conditions, and hopefully I've plugged them all up here. 1) There was a serious race in switch_mm() wrt. lazy TLB switching to and from kernel threads. We could erroneously skip a tsb_context_switch() and thus use a stale TSB across a TSB grow event. There is a big comment now in that function describing exactly how it can happen. 2) All code paths that do something with the TSB need to be guarded with the mm->context.lock spinlock. This makes page table flushing paths properly synchronize with both TSB growing and TLB context changes. 3) TSB growing events are moved to the end of successful fault processing. Previously it was in update_mmu_cache() but that is deadlock prone. At the end of do_sparc64_fault() we hold no spinlocks that could deadlock the TSB grow sequence. We also have dropped the address space semaphore. While we're here, add prefetching to the copy_tsb() routine and put it in assembler into the tsb.S file. This piece of code is quite time critical. There are some small negative side effects to this code which can be improved upon. In particular we grab the mm->context.lock even for the tsb insert done by update_mmu_cache() now and that's a bit excessive. We can get rid of that locking, and the same lock taking in flush_tsb_user(), by disabling PSTATE_IE around the whole operation including the capturing of the tsb pointer and tsb_nentries value. That would work because anyone growing the TSB won't free up the old TSB until all cpus respond to the TSB change cross call. I'm not quite so confident in that optimization to put it in right now, but eventually we might be able to and the description is here for reference. This code seems very solid now. It passes several parallel GCC bootstrap builds, and our favorite "nut cruncher" stress test which is a full "make -j8192" build of a "make allmodconfig" kernel. That puts about 256 processes on each cpu's run queue, makes lots of process cpu migrations occur, causes lots of page table and TLB flushing activity, incurs many context version number changes, and it swaps the machine real far out to disk even though there is 16GB of ram on this test system. :-) Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-16 18:02:32 +08:00
spin_unlock_irqrestore(&mm->context.lock, flags);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
#define deactivate_mm(tsk,mm) do { } while (0)
/* Activate a new MM instance for the current task. */
static inline void activate_mm(struct mm_struct *active_mm, struct mm_struct *mm)
{
[SPARC64]: Fix TLB context allocation with SMT style shared TLBs. The context allocation scheme we use depends upon there being a 1<-->1 mapping from cpu to physical TLB for correctness. Chips like Niagara break this assumption. So what we do is notify all cpus with a cross call when the context version number changes, and if necessary this makes them allocate a valid context for the address space they are running at the time. Stress tested with make -j1024, make -j2048, and make -j4096 kernel builds on a 32-strand, 8 core, T2000 with 16GB of ram. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-24 06:19:28 +08:00
unsigned long flags;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int cpu;
[SPARC64]: Fix TLB context allocation with SMT style shared TLBs. The context allocation scheme we use depends upon there being a 1<-->1 mapping from cpu to physical TLB for correctness. Chips like Niagara break this assumption. So what we do is notify all cpus with a cross call when the context version number changes, and if necessary this makes them allocate a valid context for the address space they are running at the time. Stress tested with make -j1024, make -j2048, and make -j4096 kernel builds on a 32-strand, 8 core, T2000 with 16GB of ram. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-24 06:19:28 +08:00
spin_lock_irqsave(&mm->context.lock, flags);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (!CTX_VALID(mm->context))
get_new_mmu_context(mm);
cpu = smp_processor_id();
if (!cpu_isset(cpu, mm->cpu_vm_mask))
cpu_set(cpu, mm->cpu_vm_mask);
load_secondary_context(mm);
__flush_tlb_mm(CTX_HWBITS(mm->context), SECONDARY_CONTEXT);
[SPARC64]: Add infrastructure for dynamic TSB sizing. This also cleans up tsb_context_switch(). The assembler routine is now __tsb_context_switch() and the former is an inline function that picks out the bits from the mm_struct and passes it into the assembler code as arguments. setup_tsb_parms() computes the locked TLB entry to map the TSB. Later when we support using the physical address quad load instructions of Cheetah+ and later, we'll simply use the physical address for the TSB register value and set the map virtual and PTE both to zero. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-02-01 10:31:20 +08:00
tsb_context_switch(mm);
[SPARC64]: Fix and re-enable dynamic TSB sizing. This is good for up to %50 performance improvement of some test cases. The problem has been the race conditions, and hopefully I've plugged them all up here. 1) There was a serious race in switch_mm() wrt. lazy TLB switching to and from kernel threads. We could erroneously skip a tsb_context_switch() and thus use a stale TSB across a TSB grow event. There is a big comment now in that function describing exactly how it can happen. 2) All code paths that do something with the TSB need to be guarded with the mm->context.lock spinlock. This makes page table flushing paths properly synchronize with both TSB growing and TLB context changes. 3) TSB growing events are moved to the end of successful fault processing. Previously it was in update_mmu_cache() but that is deadlock prone. At the end of do_sparc64_fault() we hold no spinlocks that could deadlock the TSB grow sequence. We also have dropped the address space semaphore. While we're here, add prefetching to the copy_tsb() routine and put it in assembler into the tsb.S file. This piece of code is quite time critical. There are some small negative side effects to this code which can be improved upon. In particular we grab the mm->context.lock even for the tsb insert done by update_mmu_cache() now and that's a bit excessive. We can get rid of that locking, and the same lock taking in flush_tsb_user(), by disabling PSTATE_IE around the whole operation including the capturing of the tsb pointer and tsb_nentries value. That would work because anyone growing the TSB won't free up the old TSB until all cpus respond to the TSB change cross call. I'm not quite so confident in that optimization to put it in right now, but eventually we might be able to and the description is here for reference. This code seems very solid now. It passes several parallel GCC bootstrap builds, and our favorite "nut cruncher" stress test which is a full "make -j8192" build of a "make allmodconfig" kernel. That puts about 256 processes on each cpu's run queue, makes lots of process cpu migrations occur, causes lots of page table and TLB flushing activity, incurs many context version number changes, and it swaps the machine real far out to disk even though there is 16GB of ram on this test system. :-) Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-16 18:02:32 +08:00
spin_unlock_irqrestore(&mm->context.lock, flags);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
#endif /* !(__ASSEMBLY__) */
#endif /* !(__SPARC64_MMU_CONTEXT_H) */
Copy Permalink