From 9674f35b1ec17577163897f052f405c1e9e5893d Mon Sep 17 00:00:00 2001 From: Cliff Wickman Date: Fri, 3 Apr 2009 08:34:05 -0500 Subject: x86: UV BAU and nodes with no memory This patch fixes BAU initialization for systems containing nodes with no memory and for systems with non-consecutive node numbers. Fixes and clarifies situations where pnode should be used instead of node id. Tested on the UV hardware simulator. Signed-off-by: Cliff Wickman LKML-Reference: Signed-off-by: Ingo Molnar --- arch/x86/kernel/tlb_uv.c | 108 ++++++++++++++++++++++++++--------------------- 1 file changed, 61 insertions(+), 47 deletions(-) (limited to 'arch/x86/kernel/tlb_uv.c') diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index 79c073247284..b833bc634d17 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -31,6 +31,34 @@ static unsigned long uv_mmask __read_mostly; static DEFINE_PER_CPU(struct ptc_stats, ptcstats); static DEFINE_PER_CPU(struct bau_control, bau_control); +/* + * Determine the first node on a blade. + */ +static int __init blade_to_first_node(int blade) +{ + int node, b; + + for_each_online_node(node) { + b = uv_node_to_blade_id(node); + if (blade == b) + return node; + } + BUG(); +} + +/* + * Determine the apicid of the first cpu on a blade. + */ +static int __init blade_to_first_apicid(int blade) +{ + int cpu; + + for_each_present_cpu(cpu) + if (blade == uv_cpu_to_blade_id(cpu)) + return per_cpu(x86_cpu_to_apicid, cpu); + return -1; +} + /* * Free a software acknowledge hardware resource by clearing its Pending * bit. This will return a reply to the sender. @@ -67,7 +95,7 @@ static void uv_bau_process_message(struct bau_payload_queue_entry *msg, msp = __get_cpu_var(bau_control).msg_statuses + msg_slot; cpu = uv_blade_processor_id(); msg->number_of_cpus = - uv_blade_nr_online_cpus(uv_node_to_blade_id(numa_node_id())); + uv_blade_nr_online_cpus(uv_node_to_blade_id(numa_node_id())); this_cpu_mask = 1UL << cpu; if (msp->seen_by.bits & this_cpu_mask) return; @@ -215,14 +243,14 @@ static int uv_wait_completion(struct bau_desc *bau_desc, * Returns @flush_mask if some remote flushing remains to be done. The * mask will have some bits still set. */ -const struct cpumask *uv_flush_send_and_wait(int cpu, int this_blade, +const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode, struct bau_desc *bau_desc, struct cpumask *flush_mask) { int completion_status = 0; int right_shift; int tries = 0; - int blade; + int pnode; int bit; unsigned long mmr_offset; unsigned long index; @@ -265,8 +293,8 @@ const struct cpumask *uv_flush_send_and_wait(int cpu, int this_blade, * use the IPI method of shootdown on them. */ for_each_cpu(bit, flush_mask) { - blade = uv_cpu_to_blade_id(bit); - if (blade == this_blade) + pnode = uv_cpu_to_pnode(bit); + if (pnode == this_pnode) continue; cpumask_clear_cpu(bit, flush_mask); } @@ -308,16 +336,16 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask, struct cpumask *flush_mask = &__get_cpu_var(flush_tlb_mask); int i; int bit; - int blade; + int pnode; int uv_cpu; - int this_blade; + int this_pnode; int locals = 0; struct bau_desc *bau_desc; cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu)); uv_cpu = uv_blade_processor_id(); - this_blade = uv_numa_blade_id(); + this_pnode = uv_hub_info->pnode; bau_desc = __get_cpu_var(bau_control).descriptor_base; bau_desc += UV_ITEMS_PER_DESCRIPTOR * uv_cpu; @@ -325,13 +353,13 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask, i = 0; for_each_cpu(bit, flush_mask) { - blade = uv_cpu_to_blade_id(bit); - BUG_ON(blade > (UV_DISTRIBUTION_SIZE - 1)); - if (blade == this_blade) { + pnode = uv_cpu_to_pnode(bit); + BUG_ON(pnode > (UV_DISTRIBUTION_SIZE - 1)); + if (pnode == this_pnode) { locals++; continue; } - bau_node_set(blade, &bau_desc->distribution); + bau_node_set(pnode, &bau_desc->distribution); i++; } if (i == 0) { @@ -349,7 +377,7 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask, bau_desc->payload.address = va; bau_desc->payload.sending_cpu = cpu; - return uv_flush_send_and_wait(uv_cpu, this_blade, bau_desc, flush_mask); + return uv_flush_send_and_wait(uv_cpu, this_pnode, bau_desc, flush_mask); } /* @@ -481,8 +509,7 @@ static int uv_ptc_seq_show(struct seq_file *file, void *data) stat->requestee, stat->onetlb, stat->alltlb, stat->s_retry, stat->d_retry, stat->ptc_i); seq_printf(file, "%lx %ld %ld %ld %ld %ld %ld\n", - uv_read_global_mmr64(uv_blade_to_pnode - (uv_cpu_to_blade_id(cpu)), + uv_read_global_mmr64(uv_cpu_to_pnode(cpu), UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE), stat->sflush, stat->dflush, stat->retriesok, stat->nomsg, @@ -616,16 +643,18 @@ static struct bau_control * __init uv_table_bases_init(int blade, int node) * finish the initialization of the per-blade control structures */ static void __init -uv_table_bases_finish(int blade, int node, int cur_cpu, +uv_table_bases_finish(int blade, struct bau_control *bau_tablesp, struct bau_desc *adp) { struct bau_control *bcp; - int i; + int cpu; - for (i = cur_cpu; i < cur_cpu + uv_blade_nr_possible_cpus(blade); i++) { - bcp = (struct bau_control *)&per_cpu(bau_control, i); + for_each_present_cpu(cpu) { + if (blade != uv_cpu_to_blade_id(cpu)) + continue; + bcp = (struct bau_control *)&per_cpu(bau_control, cpu); bcp->bau_msg_head = bau_tablesp->va_queue_first; bcp->va_queue_first = bau_tablesp->va_queue_first; bcp->va_queue_last = bau_tablesp->va_queue_last; @@ -648,8 +677,7 @@ uv_activation_descriptor_init(int node, int pnode) struct bau_desc *adp; struct bau_desc *ad2; - adp = (struct bau_desc *) - kmalloc_node(16384, GFP_KERNEL, node); + adp = (struct bau_desc *)kmalloc_node(16384, GFP_KERNEL, node); BUG_ON(!adp); pa = __pa((unsigned long)adp); @@ -666,8 +694,7 @@ uv_activation_descriptor_init(int node, int pnode) for (i = 0, ad2 = adp; i < UV_ACTIVATION_DESCRIPTOR_SIZE; i++, ad2++) { memset(ad2, 0, sizeof(struct bau_desc)); ad2->header.sw_ack_flag = 1; - ad2->header.base_dest_nodeid = - uv_blade_to_pnode(uv_cpu_to_blade_id(0)); + ad2->header.base_dest_nodeid = uv_cpu_to_pnode(0); ad2->header.command = UV_NET_ENDPOINT_INTD; ad2->header.int_both = 1; /* @@ -714,8 +741,9 @@ uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp) /* * Initialization of each UV blade's structures */ -static int __init uv_init_blade(int blade, int node, int cur_cpu) +static int __init uv_init_blade(int blade) { + int node; int pnode; unsigned long pa; unsigned long apicid; @@ -723,16 +751,17 @@ static int __init uv_init_blade(int blade, int node, int cur_cpu) struct bau_payload_queue_entry *pqp; struct bau_control *bau_tablesp; + node = blade_to_first_node(blade); bau_tablesp = uv_table_bases_init(blade, node); pnode = uv_blade_to_pnode(blade); adp = uv_activation_descriptor_init(node, pnode); pqp = uv_payload_queue_init(node, pnode, bau_tablesp); - uv_table_bases_finish(blade, node, cur_cpu, bau_tablesp, adp); + uv_table_bases_finish(blade, bau_tablesp, adp); /* * the below initialization can't be in firmware because the * messaging IRQ will be determined by the OS */ - apicid = per_cpu(x86_cpu_to_apicid, cur_cpu); + apicid = blade_to_first_apicid(blade); pa = uv_read_global_mmr64(pnode, UVH_BAU_DATA_CONFIG); if ((pa & 0xff) != UV_BAU_MESSAGE) { uv_write_global_mmr64(pnode, UVH_BAU_DATA_CONFIG, @@ -747,9 +776,7 @@ static int __init uv_init_blade(int blade, int node, int cur_cpu) static int __init uv_bau_init(void) { int blade; - int node; int nblades; - int last_blade; int cur_cpu; if (!is_uv_system()) @@ -758,29 +785,16 @@ static int __init uv_bau_init(void) uv_bau_retry_limit = 1; uv_nshift = uv_hub_info->n_val; uv_mmask = (1UL << uv_hub_info->n_val) - 1; - nblades = 0; - last_blade = -1; - cur_cpu = 0; - for_each_online_node(node) { - blade = uv_node_to_blade_id(node); - if (blade == last_blade) - continue; - last_blade = blade; - nblades++; - } + nblades = uv_num_possible_blades(); + uv_bau_table_bases = (struct bau_control **) kmalloc(nblades * sizeof(struct bau_control *), GFP_KERNEL); BUG_ON(!uv_bau_table_bases); - last_blade = -1; - for_each_online_node(node) { - blade = uv_node_to_blade_id(node); - if (blade == last_blade) - continue; - last_blade = blade; - uv_init_blade(blade, node, cur_cpu); - cur_cpu += uv_blade_nr_possible_cpus(blade); - } + for (blade = 0; blade < nblades; blade++) + if (uv_blade_nr_possible_cpus(blade)) + uv_init_blade(blade); + alloc_intr_gate(UV_BAU_MESSAGE, uv_bau_message_intr1); uv_enable_timeouts(); -- cgit v1.2.3-59-g8ed1b From c4c4688f72e638708e5f6b5c259699de82a36fec Mon Sep 17 00:00:00 2001 From: Cliff Wickman Date: Fri, 3 Apr 2009 08:34:32 -0500 Subject: x86: UV BAU messaging timeouts This patch replaces a 'nop' uv_enable_timeouts() in the UV TLB shootdown code. (somehow, long ago that function got eviscerated) If any cpu in the destination node does not get interrupted by the message and post completion in a reasonable time the hardware should respond to the sender with an error. This function enables such timeouts. Tested on the UV hardware simulator. Signed-off-by: Cliff Wickman LKML-Reference: Signed-off-by: Ingo Molnar --- arch/x86/include/asm/uv/uv_mmrs.h | 5 ++++ arch/x86/kernel/tlb_uv.c | 56 +++++++++++++++++++++++++++++++-------- 2 files changed, 50 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel/tlb_uv.c') diff --git a/arch/x86/include/asm/uv/uv_mmrs.h b/arch/x86/include/asm/uv/uv_mmrs.h index db68ac8a5ac2..2cae46c7c8a2 100644 --- a/arch/x86/include/asm/uv/uv_mmrs.h +++ b/arch/x86/include/asm/uv/uv_mmrs.h @@ -17,6 +17,11 @@ /* ========================================================================= */ /* UVH_BAU_DATA_CONFIG */ /* ========================================================================= */ +#define UVH_LB_BAU_MISC_CONTROL 0x320170UL +#define UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT 15 +#define UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT 16 +#define UV_INTD_SOFT_ACK_TIMEOUT_PERIOD 0x000000000bUL +/* 1011 timebase 7 (168millisec) * 3 ticks -> 500ms */ #define UVH_BAU_DATA_CONFIG 0x61680UL #define UVH_BAU_DATA_CONFIG_32 0x0438 diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index b833bc634d17..fced96e94e22 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -445,24 +445,58 @@ void uv_bau_message_interrupt(struct pt_regs *regs) set_irq_regs(old_regs); } +/* + * uv_enable_timeouts + * + * Each target blade (i.e. blades that have cpu's) needs to have + * shootdown message timeouts enabled. The timeout does not cause + * an interrupt, but causes an error message to be returned to + * the sender. + */ static void uv_enable_timeouts(void) { - int i; int blade; - int last_blade; + int nblades; int pnode; - int cur_cpu = 0; - unsigned long apicid; + unsigned long mmr_image; + + nblades = uv_num_possible_blades(); - last_blade = -1; - for_each_online_node(i) { - blade = uv_node_to_blade_id(i); - if (blade == last_blade) + for (blade = 0; blade < nblades; blade++) { + if (!uv_blade_nr_possible_cpus(blade)) continue; - last_blade = blade; - apicid = per_cpu(x86_cpu_to_apicid, cur_cpu); + pnode = uv_blade_to_pnode(blade); - cur_cpu += uv_blade_nr_possible_cpus(i); + mmr_image = + uv_read_global_mmr64(pnode, UVH_LB_BAU_MISC_CONTROL); + /* + * Set the timeout period and then lock it in, in three + * steps; captures and locks in the period. + * + * To program the period, the SOFT_ACK_MODE must be off. + */ + mmr_image &= ~((unsigned long)1 << + UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT); + uv_write_global_mmr64 + (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image); + /* + * Set the 4-bit period. + */ + mmr_image &= ~((unsigned long)0xf << + UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT); + mmr_image |= (UV_INTD_SOFT_ACK_TIMEOUT_PERIOD << + UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT); + uv_write_global_mmr64 + (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image); + /* + * Subsequent reversals of the timebase bit (3) cause an + * immediate timeout of one or all INTD resources as + * indicated in bits 2:0 (7 causes all of them to timeout). + */ + mmr_image |= ((unsigned long)1 << + UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT); + uv_write_global_mmr64 + (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image); } } -- cgit v1.2.3-59-g8ed1b From 94ca8e4852807fc42d2f64fcaf248aafc4f2e6a7 Mon Sep 17 00:00:00 2001 From: Cliff Wickman Date: Tue, 14 Apr 2009 10:56:48 -0500 Subject: x86: UV: BAU partition-relative distribution map This patch enables each partition's BAU distribution bit map to be partition-relative. The distribution bitmap had been constructed assuming 0 as the base node number. That construct would not have allowed a total system of greater than 256 nodes. It also corrects an error that occurred when the first blade's nasid was not zero. That nasid was stored as the base node. The base node number gets added by hardware to the node numbers implied in the distribution bitmap, resulting in invalid target nasids. Tested on the UV hardware simulator. Signed-off-by: Cliff Wickman LKML-Reference: Signed-off-by: Ingo Molnar --- arch/x86/kernel/tlb_uv.c | 19 ++++++++++++++++--- 1 file changed, 16 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel/tlb_uv.c') diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index fced96e94e22..98307f953492 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -25,6 +25,8 @@ static int uv_bau_retry_limit __read_mostly; /* position of pnode (which is nasid>>1): */ static int uv_nshift __read_mostly; +/* base pnode in this partition */ +static int uv_partition_base_pnode __read_mostly; static unsigned long uv_mmask __read_mostly; @@ -43,7 +45,7 @@ static int __init blade_to_first_node(int blade) if (blade == b) return node; } - BUG(); + return -1; /* shouldn't happen */ } /* @@ -359,7 +361,8 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask, locals++; continue; } - bau_node_set(pnode, &bau_desc->distribution); + bau_node_set(pnode - uv_partition_base_pnode, + &bau_desc->distribution); i++; } if (i == 0) { @@ -728,7 +731,12 @@ uv_activation_descriptor_init(int node, int pnode) for (i = 0, ad2 = adp; i < UV_ACTIVATION_DESCRIPTOR_SIZE; i++, ad2++) { memset(ad2, 0, sizeof(struct bau_desc)); ad2->header.sw_ack_flag = 1; - ad2->header.base_dest_nodeid = uv_cpu_to_pnode(0); + /* + * base_dest_nodeid is the first node in the partition, so + * the bit map will indicate partition-relative node numbers. + * note that base_dest_nodeid is actually a nasid. + */ + ad2->header.base_dest_nodeid = uv_partition_base_pnode << 1; ad2->header.command = UV_NET_ENDPOINT_INTD; ad2->header.int_both = 1; /* @@ -825,6 +833,11 @@ static int __init uv_bau_init(void) kmalloc(nblades * sizeof(struct bau_control *), GFP_KERNEL); BUG_ON(!uv_bau_table_bases); + uv_partition_base_pnode = 0x7fffffff; + for (blade = 0; blade < nblades; blade++) + if (uv_blade_nr_possible_cpus(blade) && + (uv_blade_to_pnode(blade) < uv_partition_base_pnode)) + uv_partition_base_pnode = uv_blade_to_pnode(blade); for (blade = 0; blade < nblades; blade++) if (uv_blade_nr_possible_cpus(blade)) uv_init_blade(blade); -- cgit v1.2.3-59-g8ed1b From 4ea3c51d5bd3bb4eea7d7d3a1f80d1a48c2a6f92 Mon Sep 17 00:00:00 2001 From: Cliff Wickman Date: Thu, 16 Apr 2009 07:53:09 -0500 Subject: x86: UV BAU distribution and payload MMRs This patch correctly sets BAU memory mapped registers to point to the sending activation descriptor table and target payload table. The "Broadcast Assist Unit" is used for TLB shootdown in UV. The memory mapped registers that point to sending and receiving memory structures contain node numbers. In one case the __pa() function did not provide the node id of memory on blade zero in configurations where that id is nonzero. In another case, it was assumed that memory was allocated on the local node. That assumption is not true in a configuration in which the node has no memory. Tested on the UV hardware simulator. [ Impact: fix possible runtime crash due to incorrect TLB logic ] Signed-off-by: Cliff Wickman LKML-Reference: Signed-off-by: Ingo Molnar --- arch/x86/kernel/tlb_uv.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel/tlb_uv.c') diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index 98307f953492..78422336ddea 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -717,7 +717,7 @@ uv_activation_descriptor_init(int node, int pnode) adp = (struct bau_desc *)kmalloc_node(16384, GFP_KERNEL, node); BUG_ON(!adp); - pa = __pa((unsigned long)adp); + pa = uv_gpa(adp); /* need the real nasid*/ n = pa >> uv_nshift; m = pa & uv_mmask; @@ -754,6 +754,8 @@ static struct bau_payload_queue_entry * __init uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp) { struct bau_payload_queue_entry *pqp; + unsigned long pa; + int pn; char *cp; pqp = (struct bau_payload_queue_entry *) kmalloc_node( @@ -764,10 +766,14 @@ uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp) cp = (char *)pqp + 31; pqp = (struct bau_payload_queue_entry *)(((unsigned long)cp >> 5) << 5); bau_tablesp->va_queue_first = pqp; + /* + * need the pnode of where the memory was really allocated + */ + pa = uv_gpa(pqp); + pn = pa >> uv_nshift; uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_FIRST, - ((unsigned long)pnode << - UV_PAYLOADQ_PNODE_SHIFT) | + ((unsigned long)pn << UV_PAYLOADQ_PNODE_SHIFT) | uv_physnodeaddr(pqp)); uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_TAIL, uv_physnodeaddr(pqp)); -- cgit v1.2.3-59-g8ed1b