From 1d1e9f04216b379000128392b11edd7f5d0ebed1 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Wed, 6 Jan 2010 15:55:12 +1100 Subject: [IA64] cpumask_of_node() should handle -1 as a node pcibus_to_node can return -1 if we cannot determine which node a pci bus is on. If passed -1, cpumask_of_node will negatively index the lookup array and pull in random data: # cat /sys/devices/pci0000:00/0000:00:01.0/local_cpus 00000000,00000003,00000000,00000000 # cat /sys/devices/pci0000:00/0000:00:01.0/local_cpulist 64-65 Change cpumask_of_node to check for -1 and return cpu_all_mask in this case: # cat /sys/devices/pci0000:00/0000:00:01.0/local_cpus ffffffff,ffffffff,ffffffff,ffffffff # cat /sys/devices/pci0000:00/0000:00:01.0/local_cpulist 0-127 Signed-off-by: Anton Blanchard Signed-off-by: Tony Luck --- arch/ia64/include/asm/topology.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/ia64/include/asm/topology.h b/arch/ia64/include/asm/topology.h index 3ddb4e709dba..d323071d0f91 100644 --- a/arch/ia64/include/asm/topology.h +++ b/arch/ia64/include/asm/topology.h @@ -33,7 +33,9 @@ /* * Returns a bitmask of CPUs on Node 'node'. */ -#define cpumask_of_node(node) (&node_to_cpu_mask[node]) +#define cpumask_of_node(node) ((node) == -1 ? \ + cpu_all_mask : \ + &node_to_cpu_mask[node]) /* * Returns the number of the node containing Node 'nid'. -- cgit v1.2.3 From 02b763b8ccc88d030117851f2b76a119932f109e Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Wed, 6 Jan 2010 16:24:30 +0100 Subject: [IA64] use helpers for rlimits Make sure compiler won't do weird things with limits. E.g. fetching them twice may return 2 different values after writable limits are implemented. I.e. either use rlimit helpers added in 3e10e716abf3c71bdb5d86b8f507f9e72236c9cd or ACCESS_ONCE if not applicable. Signed-off-by: Jiri Slaby Cc: Fenghua Yu Signed-off-by: Tony Luck --- arch/ia64/kernel/perfmon.c | 2 +- arch/ia64/mm/init.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'arch') diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c index 5246285a95fb..6bcbe215b9a4 100644 --- a/arch/ia64/kernel/perfmon.c +++ b/arch/ia64/kernel/perfmon.c @@ -2293,7 +2293,7 @@ pfm_smpl_buffer_alloc(struct task_struct *task, struct file *filp, pfm_context_t * if ((mm->total_vm << PAGE_SHIFT) + len> task->rlim[RLIMIT_AS].rlim_cur) * return -ENOMEM; */ - if (size > task->signal->rlim[RLIMIT_MEMLOCK].rlim_cur) + if (size > task_rlimit(task, RLIMIT_MEMLOCK)) return -ENOMEM; /* diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c index b9609c69343a..7c0d4814a68d 100644 --- a/arch/ia64/mm/init.c +++ b/arch/ia64/mm/init.c @@ -91,7 +91,7 @@ dma_mark_clean(void *addr, size_t size) inline void ia64_set_rbs_bot (void) { - unsigned long stack_size = current->signal->rlim[RLIMIT_STACK].rlim_max & -16; + unsigned long stack_size = rlimit_max(RLIMIT_STACK) & -16; if (stack_size > MAX_USER_STACK_SIZE) stack_size = MAX_USER_STACK_SIZE; -- cgit v1.2.3 From 410dc0aac63d1500faeabcbaecce4f4266380ed1 Mon Sep 17 00:00:00 2001 From: Tony Luck Date: Wed, 6 Jan 2010 15:52:35 -0800 Subject: [IA64] sanity in #include files. Move fnptr to types.h Signed-off-by: Tony Luck --- arch/ia64/include/asm/ftrace.h | 1 - arch/ia64/include/asm/kprobes.h | 5 ----- arch/ia64/include/asm/types.h | 5 +++++ 3 files changed, 5 insertions(+), 6 deletions(-) (limited to 'arch') diff --git a/arch/ia64/include/asm/ftrace.h b/arch/ia64/include/asm/ftrace.h index d20db3c2a656..fbd1a2470cae 100644 --- a/arch/ia64/include/asm/ftrace.h +++ b/arch/ia64/include/asm/ftrace.h @@ -8,7 +8,6 @@ extern void _mcount(unsigned long pfs, unsigned long r1, unsigned long b0, unsigned long r0); #define mcount _mcount -#include /* In IA64, MCOUNT_ADDR is set in link time, so it's not a constant at compile time */ #define MCOUNT_ADDR (((struct fnptr *)mcount)->ip) #define FTRACE_ADDR (((struct fnptr *)ftrace_caller)->ip) diff --git a/arch/ia64/include/asm/kprobes.h b/arch/ia64/include/asm/kprobes.h index dbf83fb28db3..d5505d6f2382 100644 --- a/arch/ia64/include/asm/kprobes.h +++ b/arch/ia64/include/asm/kprobes.h @@ -103,11 +103,6 @@ typedef struct kprobe_opcode { bundle_t bundle; } kprobe_opcode_t; -struct fnptr { - unsigned long ip; - unsigned long gp; -}; - /* Architecture specific copy of original instruction*/ struct arch_specific_insn { /* copy of the instruction to be emulated */ diff --git a/arch/ia64/include/asm/types.h b/arch/ia64/include/asm/types.h index bcd260e597de..b8e5d97be158 100644 --- a/arch/ia64/include/asm/types.h +++ b/arch/ia64/include/asm/types.h @@ -30,6 +30,11 @@ typedef unsigned int umode_t; +struct fnptr { + unsigned long ip; + unsigned long gp; +}; + /* * These aren't exported outside the kernel to avoid name space clashes */ -- cgit v1.2.3 From 6c57a332901f851bd092aba7a2b4d8ef4e643829 Mon Sep 17 00:00:00 2001 From: Tony Luck Date: Thu, 7 Jan 2010 16:10:57 -0800 Subject: [IA64] __per_cpu_idtrs[] is a memory hog __per_cpu_idtrs is statically allocated ... on CONFIG_NR_CPUS=4096 systems it hogs 16MB of memory. This is way too much for a quite probably unused facility (only KVM uses dynamic TR registers). Change to an array of pointers, and allocate entries as needed on a per cpu basis. Change the name too as the __per_cpu_ prefix is confusing (this isn't a classic type object). Signed-off-by: Tony Luck --- arch/ia64/include/asm/tlb.h | 2 +- arch/ia64/kernel/mca.c | 5 ++++- arch/ia64/mm/tlb.c | 32 +++++++++++++++++++------------- 3 files changed, 24 insertions(+), 15 deletions(-) (limited to 'arch') diff --git a/arch/ia64/include/asm/tlb.h b/arch/ia64/include/asm/tlb.h index 85d965cb19a0..23cce999eb1c 100644 --- a/arch/ia64/include/asm/tlb.h +++ b/arch/ia64/include/asm/tlb.h @@ -74,7 +74,7 @@ struct ia64_tr_entry { extern int ia64_itr_entry(u64 target_mask, u64 va, u64 pte, u64 log_size); extern void ia64_ptr_entry(u64 target_mask, int slot); -extern struct ia64_tr_entry __per_cpu_idtrs[NR_CPUS][2][IA64_TR_ALLOC_MAX]; +extern struct ia64_tr_entry *ia64_idtrs[NR_CPUS]; /* region register macros diff --git a/arch/ia64/kernel/mca.c b/arch/ia64/kernel/mca.c index 32f2639e9b0a..378b4833024f 100644 --- a/arch/ia64/kernel/mca.c +++ b/arch/ia64/kernel/mca.c @@ -1225,9 +1225,12 @@ static void mca_insert_tr(u64 iord) unsigned long psr; int cpu = smp_processor_id(); + if (!ia64_idtrs[cpu]) + return; + psr = ia64_clear_ic(); for (i = IA64_TR_ALLOC_BASE; i < IA64_TR_ALLOC_MAX; i++) { - p = &__per_cpu_idtrs[cpu][iord-1][i]; + p = ia64_idtrs[cpu] + (iord - 1) * IA64_TR_ALLOC_MAX; if (p->pte & 0x1) { old_rr = ia64_get_rr(p->ifa); if (old_rr != p->rr) { diff --git a/arch/ia64/mm/tlb.c b/arch/ia64/mm/tlb.c index ee09d261f2e6..f3de9d7a98b4 100644 --- a/arch/ia64/mm/tlb.c +++ b/arch/ia64/mm/tlb.c @@ -48,7 +48,7 @@ DEFINE_PER_CPU(u8, ia64_need_tlb_flush); DEFINE_PER_CPU(u8, ia64_tr_num); /*Number of TR slots in current processor*/ DEFINE_PER_CPU(u8, ia64_tr_used); /*Max Slot number used by kernel*/ -struct ia64_tr_entry __per_cpu_idtrs[NR_CPUS][2][IA64_TR_ALLOC_MAX]; +struct ia64_tr_entry *ia64_idtrs[NR_CPUS]; /* * Initializes the ia64_ctx.bitmap array based on max_ctx+1. @@ -429,10 +429,16 @@ int ia64_itr_entry(u64 target_mask, u64 va, u64 pte, u64 log_size) struct ia64_tr_entry *p; int cpu = smp_processor_id(); + if (!ia64_idtrs[cpu]) { + ia64_idtrs[cpu] = kmalloc(2 * IA64_TR_ALLOC_MAX * + sizeof (struct ia64_tr_entry), GFP_KERNEL); + if (!ia64_idtrs[cpu]) + return -ENOMEM; + } r = -EINVAL; /*Check overlap with existing TR entries*/ if (target_mask & 0x1) { - p = &__per_cpu_idtrs[cpu][0][0]; + p = ia64_idtrs[cpu]; for (i = IA64_TR_ALLOC_BASE; i <= per_cpu(ia64_tr_used, cpu); i++, p++) { if (p->pte & 0x1) @@ -444,7 +450,7 @@ int ia64_itr_entry(u64 target_mask, u64 va, u64 pte, u64 log_size) } } if (target_mask & 0x2) { - p = &__per_cpu_idtrs[cpu][1][0]; + p = ia64_idtrs[cpu] + IA64_TR_ALLOC_MAX; for (i = IA64_TR_ALLOC_BASE; i <= per_cpu(ia64_tr_used, cpu); i++, p++) { if (p->pte & 0x1) @@ -459,16 +465,16 @@ int ia64_itr_entry(u64 target_mask, u64 va, u64 pte, u64 log_size) for (i = IA64_TR_ALLOC_BASE; i < per_cpu(ia64_tr_num, cpu); i++) { switch (target_mask & 0x3) { case 1: - if (!(__per_cpu_idtrs[cpu][0][i].pte & 0x1)) + if (!((ia64_idtrs[cpu] + i)->pte & 0x1)) goto found; continue; case 2: - if (!(__per_cpu_idtrs[cpu][1][i].pte & 0x1)) + if (!((ia64_idtrs[cpu] + IA64_TR_ALLOC_MAX + i)->pte & 0x1)) goto found; continue; case 3: - if (!(__per_cpu_idtrs[cpu][0][i].pte & 0x1) && - !(__per_cpu_idtrs[cpu][1][i].pte & 0x1)) + if (!((ia64_idtrs[cpu] + i)->pte & 0x1) && + !((ia64_idtrs[cpu] + IA64_TR_ALLOC_MAX + i)->pte & 0x1)) goto found; continue; default: @@ -488,7 +494,7 @@ found: if (target_mask & 0x1) { ia64_itr(0x1, i, va, pte, log_size); ia64_srlz_i(); - p = &__per_cpu_idtrs[cpu][0][i]; + p = ia64_idtrs[cpu] + i; p->ifa = va; p->pte = pte; p->itir = log_size << 2; @@ -497,7 +503,7 @@ found: if (target_mask & 0x2) { ia64_itr(0x2, i, va, pte, log_size); ia64_srlz_i(); - p = &__per_cpu_idtrs[cpu][1][i]; + p = ia64_idtrs[cpu] + IA64_TR_ALLOC_MAX + i; p->ifa = va; p->pte = pte; p->itir = log_size << 2; @@ -528,7 +534,7 @@ void ia64_ptr_entry(u64 target_mask, int slot) return; if (target_mask & 0x1) { - p = &__per_cpu_idtrs[cpu][0][slot]; + p = ia64_idtrs[cpu] + slot; if ((p->pte&0x1) && is_tr_overlap(p, p->ifa, p->itir>>2)) { p->pte = 0; ia64_ptr(0x1, p->ifa, p->itir>>2); @@ -537,7 +543,7 @@ void ia64_ptr_entry(u64 target_mask, int slot) } if (target_mask & 0x2) { - p = &__per_cpu_idtrs[cpu][1][slot]; + p = ia64_idtrs[cpu] + IA64_TR_ALLOC_MAX + slot; if ((p->pte & 0x1) && is_tr_overlap(p, p->ifa, p->itir>>2)) { p->pte = 0; ia64_ptr(0x2, p->ifa, p->itir>>2); @@ -546,8 +552,8 @@ void ia64_ptr_entry(u64 target_mask, int slot) } for (i = per_cpu(ia64_tr_used, cpu); i >= IA64_TR_ALLOC_BASE; i--) { - if ((__per_cpu_idtrs[cpu][0][i].pte & 0x1) || - (__per_cpu_idtrs[cpu][1][i].pte & 0x1)) + if (((ia64_idtrs[cpu] + i)->pte & 0x1) || + ((ia64_idtrs[cpu] + IA64_TR_ALLOC_MAX + i)->pte & 0x1)) break; } per_cpu(ia64_tr_used, cpu) = i; -- cgit v1.2.3