From f39b2dd9d065151a04f5996656d1f27a7eb32d45 Mon Sep 17 00:00:00 2001 From: Philip Rakity Date: Thu, 7 Jul 2011 09:04:55 -0700 Subject: mmc: core: Bus width testing needs to handle suspend/resume On reading the ext_csd for the first time (in 1 bit mode), save the ext_csd information needed for bus width compare. On every pass we make re-reading the ext_csd, compare the data against the saved ext_csd data. This fixes a regression introduced in 3.0-rc1 by 08ee80cc397ac1a3 ("mmc: core: eMMC bus width may not work on all platforms"), which incorrectly assumed we would be re-reading the ext_csd at resume- time. Signed-off-by: Philip Rakity Tested-by: Jaehoon Chung Signed-off-by: Chris Ball --- include/linux/mmc/card.h | 13 +++++++++++++ 1 file changed, 13 insertions(+) (limited to 'include/linux') diff --git a/include/linux/mmc/card.h b/include/linux/mmc/card.h index c6927a4d157f..6ad43554ac05 100644 --- a/include/linux/mmc/card.h +++ b/include/linux/mmc/card.h @@ -64,6 +64,19 @@ struct mmc_ext_csd { unsigned long long enhanced_area_offset; /* Units: Byte */ unsigned int enhanced_area_size; /* Units: KB */ unsigned int boot_size; /* in bytes */ + u8 raw_partition_support; /* 160 */ + u8 raw_erased_mem_count; /* 181 */ + u8 raw_ext_csd_structure; /* 194 */ + u8 raw_card_type; /* 196 */ + u8 raw_s_a_timeout; /* 217 */ + u8 raw_hc_erase_gap_size; /* 221 */ + u8 raw_erase_timeout_mult; /* 223 */ + u8 raw_hc_erase_grp_size; /* 224 */ + u8 raw_sec_trim_mult; /* 229 */ + u8 raw_sec_erase_mult; /* 230 */ + u8 raw_sec_feature_support;/* 231 */ + u8 raw_trim_mult; /* 232 */ + u8 raw_sectors[4]; /* 212 - 4 bytes */ }; struct sd_scr { -- cgit v1.2.3-71-gd317 From 62f2a3a48bdc99822a24356e667e52c30df287c9 Mon Sep 17 00:00:00 2001 From: Michał Mirosław Date: Wed, 13 Jul 2011 14:10:29 +0000 Subject: net: remove NETIF_F_ALL_TX_OFFLOADS MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit There is no software fallback implemented for SCTP or FCoE checksumming, and so it should not be passed on by software devices like bridge or bonding. For VLAN devices, this is different. First, the driver for underlying device should be prepared to get offloaded packets even when the feature is disabled (especially if it advertises it in vlan_features). Second, devices under VLANs do not get replaced without tearing down the VLAN first. This fixes a mess I accidentally introduced while converting bonding to ndo_fix_features. NETIF_F_SOFT_FEATURES are removed from BOND_VLAN_FEATURES because they are unused as of commit 712ae51afd. Signed-off-by: Michał Mirosław Signed-off-by: David S. Miller --- drivers/net/bonding/bond_main.c | 6 +++--- include/linux/netdevice.h | 6 ------ net/8021q/vlan_dev.c | 6 +++++- 3 files changed, 8 insertions(+), 10 deletions(-) (limited to 'include/linux') diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c index eafe44a528ac..63c22b0bb5ad 100644 --- a/drivers/net/bonding/bond_main.c +++ b/drivers/net/bonding/bond_main.c @@ -1428,9 +1428,9 @@ out: return features; } -#define BOND_VLAN_FEATURES (NETIF_F_ALL_TX_OFFLOADS | \ - NETIF_F_SOFT_FEATURES | \ - NETIF_F_LRO) +#define BOND_VLAN_FEATURES (NETIF_F_ALL_CSUM | NETIF_F_SG | \ + NETIF_F_FRAGLIST | NETIF_F_ALL_TSO | \ + NETIF_F_HIGHDMA | NETIF_F_LRO) static void bond_compute_features(struct bonding *bond) { diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index 54b8b4d7b68f..9e19477991ad 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -1097,12 +1097,6 @@ struct net_device { #define NETIF_F_ALL_FCOE (NETIF_F_FCOE_CRC | NETIF_F_FCOE_MTU | \ NETIF_F_FSO) -#define NETIF_F_ALL_TX_OFFLOADS (NETIF_F_ALL_CSUM | NETIF_F_SG | \ - NETIF_F_FRAGLIST | NETIF_F_ALL_TSO | \ - NETIF_F_HIGHDMA | \ - NETIF_F_SCTP_CSUM | \ - NETIF_F_ALL_FCOE) - /* * If one device supports one of these features, then enable them * for all in netdev_increment_features. diff --git a/net/8021q/vlan_dev.c b/net/8021q/vlan_dev.c index 86bff9b1ac47..6e82148edfc8 100644 --- a/net/8021q/vlan_dev.c +++ b/net/8021q/vlan_dev.c @@ -528,7 +528,11 @@ static int vlan_dev_init(struct net_device *dev) (1<<__LINK_STATE_DORMANT))) | (1<<__LINK_STATE_PRESENT); - dev->hw_features = NETIF_F_ALL_TX_OFFLOADS; + dev->hw_features = NETIF_F_ALL_CSUM | NETIF_F_SG | + NETIF_F_FRAGLIST | NETIF_F_ALL_TSO | + NETIF_F_HIGHDMA | NETIF_F_SCTP_CSUM | + NETIF_F_ALL_FCOE; + dev->features |= real_dev->vlan_features | NETIF_F_LLTX; dev->gso_max_size = real_dev->gso_max_size; -- cgit v1.2.3-71-gd317 From a07c7964a29b6dc515b120f1e1c223ac2f8666f5 Mon Sep 17 00:00:00 2001 From: WANG Cong Date: Sat, 16 Jul 2011 22:22:20 +0000 Subject: include/linux/sdla.h: remove the prototype of sdla() `make headers_check` complains that linux-2.6/usr/include/linux/sdla.h:116: userspace cannot reference function or variable defined in the kernel this is due to that there is no such a kernel function, void sdla(void *cfg_info, char *dev, struct frad_conf *conf, int quiet); I don't know why we have it in a kernel header, so remove it. Signed-off-by: WANG Cong Signed-off-by: David S. Miller --- include/linux/sdla.h | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'include/linux') diff --git a/include/linux/sdla.h b/include/linux/sdla.h index 564acd3a71c1..9995c7fc3f60 100644 --- a/include/linux/sdla.h +++ b/include/linux/sdla.h @@ -112,11 +112,7 @@ struct sdla_dlci_conf { short Tb_max; }; -#ifndef __KERNEL__ - -void sdla(void *cfg_info, char *dev, struct frad_conf *conf, int quiet); - -#else +#ifdef __KERNEL__ /* important Z80 window addresses */ #define SDLA_CONTROL_WND 0xE000 -- cgit v1.2.3-71-gd317 From 7765be2fec0f476fcd61812d5f9406b04c765020 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Thu, 14 Jul 2011 12:24:11 -0700 Subject: rcu: Fix RCU_BOOST race handling current->rcu_read_unlock_special The RCU_BOOST commits for TREE_PREEMPT_RCU introduced an other-task write to a new RCU_READ_UNLOCK_BOOSTED bit in the task_struct structure's ->rcu_read_unlock_special field, but, as noted by Steven Rostedt, without correctly synchronizing all accesses to ->rcu_read_unlock_special. This could result in bits in ->rcu_read_unlock_special being spuriously set and cleared due to conflicting accesses, which in turn could result in deadlocks between the rcu_node structure's ->lock and the scheduler's rq and pi locks. These deadlocks would result from RCU incorrectly believing that the just-ended RCU read-side critical section had been preempted and/or boosted. If that RCU read-side critical section was executed with either rq or pi locks held, RCU's ensuing (incorrect) calls to the scheduler would cause the scheduler to attempt to once again acquire the rq and pi locks, resulting in deadlock. More complex deadlock cycles are also possible, involving multiple rq and pi locks as well as locks from multiple rcu_node structures. This commit fixes synchronization by creating ->rcu_boosted field in task_struct that is accessed and modified only when holding the ->lock in the rcu_node structure on which the task is queued (on that rcu_node structure's ->blkd_tasks list). This results in tasks accessing only their own current->rcu_read_unlock_special fields, making unsynchronized access once again legal, and keeping the rcu_read_unlock() fastpath free of atomic instructions and memory barriers. The reason that the rcu_read_unlock() fastpath does not need to access the new current->rcu_boosted field is that this new field cannot be non-zero unless the RCU_READ_UNLOCK_BLOCKED bit is set in the current->rcu_read_unlock_special field. Therefore, rcu_read_unlock() need only test current->rcu_read_unlock_special: if that is zero, then current->rcu_boosted must also be zero. This bug does not affect TINY_PREEMPT_RCU because this implementation of RCU accesses current->rcu_read_unlock_special with irqs disabled, thus preventing races on the !SMP systems that TINY_PREEMPT_RCU runs on. Maybe-reported-by: Dave Jones Maybe-reported-by: Sergey Senozhatsky Reported-by: Steven Rostedt Signed-off-by: Paul E. McKenney Signed-off-by: Paul E. McKenney Reviewed-by: Steven Rostedt --- include/linux/sched.h | 3 +++ kernel/rcutree_plugin.h | 8 ++++++-- 2 files changed, 9 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/sched.h b/include/linux/sched.h index 496770a96487..76676a407e4a 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1254,6 +1254,9 @@ struct task_struct { #ifdef CONFIG_PREEMPT_RCU int rcu_read_lock_nesting; char rcu_read_unlock_special; +#if defined(CONFIG_RCU_BOOST) && defined(CONFIG_TREE_PREEMPT_RCU) + int rcu_boosted; +#endif /* #if defined(CONFIG_RCU_BOOST) && defined(CONFIG_TREE_PREEMPT_RCU) */ struct list_head rcu_node_entry; #endif /* #ifdef CONFIG_PREEMPT_RCU */ #ifdef CONFIG_TREE_PREEMPT_RCU diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 6abef3cfcbc1..3a0ae0355222 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -342,6 +342,11 @@ static void rcu_read_unlock_special(struct task_struct *t) #ifdef CONFIG_RCU_BOOST if (&t->rcu_node_entry == rnp->boost_tasks) rnp->boost_tasks = np; + /* Snapshot and clear ->rcu_boosted with rcu_node lock held. */ + if (t->rcu_boosted) { + special |= RCU_READ_UNLOCK_BOOSTED; + t->rcu_boosted = 0; + } #endif /* #ifdef CONFIG_RCU_BOOST */ t->rcu_blocked_node = NULL; @@ -358,7 +363,6 @@ static void rcu_read_unlock_special(struct task_struct *t) #ifdef CONFIG_RCU_BOOST /* Unboost if we were boosted. */ if (special & RCU_READ_UNLOCK_BOOSTED) { - t->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_BOOSTED; rt_mutex_unlock(t->rcu_boost_mutex); t->rcu_boost_mutex = NULL; } @@ -1176,7 +1180,7 @@ static int rcu_boost(struct rcu_node *rnp) t = container_of(tb, struct task_struct, rcu_node_entry); rt_mutex_init_proxy_locked(&mtx, t); t->rcu_boost_mutex = &mtx; - t->rcu_read_unlock_special |= RCU_READ_UNLOCK_BOOSTED; + t->rcu_boosted = 1; raw_spin_unlock_irqrestore(&rnp->lock, flags); rt_mutex_lock(&mtx); /* Side effect: boosts task t's priority. */ rt_mutex_unlock(&mtx); /* Keep lockdep happy. */ -- cgit v1.2.3-71-gd317 From 9c3f75cbd144014bea6af866a154cc2e73ab2287 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 14 Jul 2011 13:00:06 +0200 Subject: sched: Break out cpu_power from the sched_group structure In order to prepare for non-unique sched_groups per domain, we need to carry the cpu_power elsewhere, so put a level of indirection in. Reported-and-tested-by: Anton Blanchard Signed-off-by: Peter Zijlstra Cc: Linus Torvalds Cc: Andrew Morton Link: http://lkml.kernel.org/n/tip-qkho2byuhe4482fuknss40ad@git.kernel.org Signed-off-by: Ingo Molnar --- include/linux/sched.h | 14 +++++++++----- kernel/sched.c | 32 ++++++++++++++++++++++++++------ kernel/sched_fair.c | 46 +++++++++++++++++++++++----------------------- 3 files changed, 58 insertions(+), 34 deletions(-) (limited to 'include/linux') diff --git a/include/linux/sched.h b/include/linux/sched.h index 496770a96487..2e5b3c8e2d3e 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -893,16 +893,20 @@ static inline int sd_power_saving_flags(void) return 0; } -struct sched_group { - struct sched_group *next; /* Must be a circular list */ - atomic_t ref; - +struct sched_group_power { /* * CPU power of this group, SCHED_LOAD_SCALE being max power for a * single CPU. */ - unsigned int cpu_power, cpu_power_orig; + unsigned int power, power_orig; +}; + +struct sched_group { + struct sched_group *next; /* Must be a circular list */ + atomic_t ref; + unsigned int group_weight; + struct sched_group_power *sgp; /* * The CPUs this group covers. diff --git a/kernel/sched.c b/kernel/sched.c index 3dc716f6d8ad..36c10d25d4cd 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -6557,7 +6557,7 @@ static int sched_domain_debug_one(struct sched_domain *sd, int cpu, int level, break; } - if (!group->cpu_power) { + if (!group->sgp->power) { printk(KERN_CONT "\n"); printk(KERN_ERR "ERROR: domain->cpu_power not " "set\n"); @@ -6581,9 +6581,9 @@ static int sched_domain_debug_one(struct sched_domain *sd, int cpu, int level, cpulist_scnprintf(str, sizeof(str), sched_group_cpus(group)); printk(KERN_CONT " %s", str); - if (group->cpu_power != SCHED_POWER_SCALE) { + if (group->sgp->power != SCHED_POWER_SCALE) { printk(KERN_CONT " (cpu_power = %d)", - group->cpu_power); + group->sgp->power); } group = group->next; @@ -6777,8 +6777,10 @@ static struct root_domain *alloc_rootdomain(void) static void free_sched_domain(struct rcu_head *rcu) { struct sched_domain *sd = container_of(rcu, struct sched_domain, rcu); - if (atomic_dec_and_test(&sd->groups->ref)) + if (atomic_dec_and_test(&sd->groups->ref)) { + kfree(sd->groups->sgp); kfree(sd->groups); + } kfree(sd); } @@ -6945,6 +6947,7 @@ int sched_smt_power_savings = 0, sched_mc_power_savings = 0; struct sd_data { struct sched_domain **__percpu sd; struct sched_group **__percpu sg; + struct sched_group_power **__percpu sgp; }; struct s_data { @@ -6981,8 +6984,10 @@ static int get_group(int cpu, struct sd_data *sdd, struct sched_group **sg) if (child) cpu = cpumask_first(sched_domain_span(child)); - if (sg) + if (sg) { *sg = *per_cpu_ptr(sdd->sg, cpu); + (*sg)->sgp = *per_cpu_ptr(sdd->sgp, cpu); + } return cpu; } @@ -7020,7 +7025,7 @@ build_sched_groups(struct sched_domain *sd) continue; cpumask_clear(sched_group_cpus(sg)); - sg->cpu_power = 0; + sg->sgp->power = 0; for_each_cpu(j, span) { if (get_group(j, sdd, NULL) != group) @@ -7185,6 +7190,7 @@ static void claim_allocations(int cpu, struct sched_domain *sd) if (cpu == cpumask_first(sched_group_cpus(sg))) { WARN_ON_ONCE(*per_cpu_ptr(sdd->sg, cpu) != sg); *per_cpu_ptr(sdd->sg, cpu) = NULL; + *per_cpu_ptr(sdd->sgp, cpu) = NULL; } } @@ -7234,9 +7240,14 @@ static int __sdt_alloc(const struct cpumask *cpu_map) if (!sdd->sg) return -ENOMEM; + sdd->sgp = alloc_percpu(struct sched_group_power *); + if (!sdd->sgp) + return -ENOMEM; + for_each_cpu(j, cpu_map) { struct sched_domain *sd; struct sched_group *sg; + struct sched_group_power *sgp; sd = kzalloc_node(sizeof(struct sched_domain) + cpumask_size(), GFP_KERNEL, cpu_to_node(j)); @@ -7251,6 +7262,13 @@ static int __sdt_alloc(const struct cpumask *cpu_map) return -ENOMEM; *per_cpu_ptr(sdd->sg, j) = sg; + + sgp = kzalloc_node(sizeof(struct sched_group_power), + GFP_KERNEL, cpu_to_node(j)); + if (!sgp) + return -ENOMEM; + + *per_cpu_ptr(sdd->sgp, j) = sgp; } } @@ -7268,9 +7286,11 @@ static void __sdt_free(const struct cpumask *cpu_map) for_each_cpu(j, cpu_map) { kfree(*per_cpu_ptr(sdd->sd, j)); kfree(*per_cpu_ptr(sdd->sg, j)); + kfree(*per_cpu_ptr(sdd->sgp, j)); } free_percpu(sdd->sd); free_percpu(sdd->sg); + free_percpu(sdd->sgp); } } diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 433491c2dc8f..c768588e180b 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1585,7 +1585,7 @@ find_idlest_group(struct sched_domain *sd, struct task_struct *p, } /* Adjust by relative CPU power of the group */ - avg_load = (avg_load * SCHED_POWER_SCALE) / group->cpu_power; + avg_load = (avg_load * SCHED_POWER_SCALE) / group->sgp->power; if (local_group) { this_load = avg_load; @@ -2631,7 +2631,7 @@ static void update_cpu_power(struct sched_domain *sd, int cpu) power >>= SCHED_POWER_SHIFT; } - sdg->cpu_power_orig = power; + sdg->sgp->power_orig = power; if (sched_feat(ARCH_POWER)) power *= arch_scale_freq_power(sd, cpu); @@ -2647,7 +2647,7 @@ static void update_cpu_power(struct sched_domain *sd, int cpu) power = 1; cpu_rq(cpu)->cpu_power = power; - sdg->cpu_power = power; + sdg->sgp->power = power; } static void update_group_power(struct sched_domain *sd, int cpu) @@ -2665,11 +2665,11 @@ static void update_group_power(struct sched_domain *sd, int cpu) group = child->groups; do { - power += group->cpu_power; + power += group->sgp->power; group = group->next; } while (group != child->groups); - sdg->cpu_power = power; + sdg->sgp->power = power; } /* @@ -2691,7 +2691,7 @@ fix_small_capacity(struct sched_domain *sd, struct sched_group *group) /* * If ~90% of the cpu_power is still there, we're good. */ - if (group->cpu_power * 32 > group->cpu_power_orig * 29) + if (group->sgp->power * 32 > group->sgp->power_orig * 29) return 1; return 0; @@ -2771,7 +2771,7 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, } /* Adjust by relative CPU power of the group */ - sgs->avg_load = (sgs->group_load*SCHED_POWER_SCALE) / group->cpu_power; + sgs->avg_load = (sgs->group_load*SCHED_POWER_SCALE) / group->sgp->power; /* * Consider the group unbalanced when the imbalance is larger @@ -2788,7 +2788,7 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, if ((max_cpu_load - min_cpu_load) >= avg_load_per_task && max_nr_running > 1) sgs->group_imb = 1; - sgs->group_capacity = DIV_ROUND_CLOSEST(group->cpu_power, + sgs->group_capacity = DIV_ROUND_CLOSEST(group->sgp->power, SCHED_POWER_SCALE); if (!sgs->group_capacity) sgs->group_capacity = fix_small_capacity(sd, group); @@ -2877,7 +2877,7 @@ static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu, return; sds->total_load += sgs.group_load; - sds->total_pwr += sg->cpu_power; + sds->total_pwr += sg->sgp->power; /* * In case the child domain prefers tasks go to siblings @@ -2962,7 +2962,7 @@ static int check_asym_packing(struct sched_domain *sd, if (this_cpu > busiest_cpu) return 0; - *imbalance = DIV_ROUND_CLOSEST(sds->max_load * sds->busiest->cpu_power, + *imbalance = DIV_ROUND_CLOSEST(sds->max_load * sds->busiest->sgp->power, SCHED_POWER_SCALE); return 1; } @@ -2993,7 +2993,7 @@ static inline void fix_small_imbalance(struct sd_lb_stats *sds, scaled_busy_load_per_task = sds->busiest_load_per_task * SCHED_POWER_SCALE; - scaled_busy_load_per_task /= sds->busiest->cpu_power; + scaled_busy_load_per_task /= sds->busiest->sgp->power; if (sds->max_load - sds->this_load + scaled_busy_load_per_task >= (scaled_busy_load_per_task * imbn)) { @@ -3007,28 +3007,28 @@ static inline void fix_small_imbalance(struct sd_lb_stats *sds, * moving them. */ - pwr_now += sds->busiest->cpu_power * + pwr_now += sds->busiest->sgp->power * min(sds->busiest_load_per_task, sds->max_load); - pwr_now += sds->this->cpu_power * + pwr_now += sds->this->sgp->power * min(sds->this_load_per_task, sds->this_load); pwr_now /= SCHED_POWER_SCALE; /* Amount of load we'd subtract */ tmp = (sds->busiest_load_per_task * SCHED_POWER_SCALE) / - sds->busiest->cpu_power; + sds->busiest->sgp->power; if (sds->max_load > tmp) - pwr_move += sds->busiest->cpu_power * + pwr_move += sds->busiest->sgp->power * min(sds->busiest_load_per_task, sds->max_load - tmp); /* Amount of load we'd add */ - if (sds->max_load * sds->busiest->cpu_power < + if (sds->max_load * sds->busiest->sgp->power < sds->busiest_load_per_task * SCHED_POWER_SCALE) - tmp = (sds->max_load * sds->busiest->cpu_power) / - sds->this->cpu_power; + tmp = (sds->max_load * sds->busiest->sgp->power) / + sds->this->sgp->power; else tmp = (sds->busiest_load_per_task * SCHED_POWER_SCALE) / - sds->this->cpu_power; - pwr_move += sds->this->cpu_power * + sds->this->sgp->power; + pwr_move += sds->this->sgp->power * min(sds->this_load_per_task, sds->this_load + tmp); pwr_move /= SCHED_POWER_SCALE; @@ -3074,7 +3074,7 @@ static inline void calculate_imbalance(struct sd_lb_stats *sds, int this_cpu, load_above_capacity *= (SCHED_LOAD_SCALE * SCHED_POWER_SCALE); - load_above_capacity /= sds->busiest->cpu_power; + load_above_capacity /= sds->busiest->sgp->power; } /* @@ -3090,8 +3090,8 @@ static inline void calculate_imbalance(struct sd_lb_stats *sds, int this_cpu, max_pull = min(sds->max_load - sds->avg_load, load_above_capacity); /* How much load to actually move to equalise the imbalance */ - *imbalance = min(max_pull * sds->busiest->cpu_power, - (sds->avg_load - sds->this_load) * sds->this->cpu_power) + *imbalance = min(max_pull * sds->busiest->sgp->power, + (sds->avg_load - sds->this_load) * sds->this->sgp->power) / SCHED_POWER_SCALE; /* -- cgit v1.2.3-71-gd317 From e3589f6c81e4764d32a25d2a2a0afe54fa344f5c Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 15 Jul 2011 10:35:52 +0200 Subject: sched: Allow for overlapping sched_domain spans Allow for sched_domain spans that overlap by giving such domains their own sched_group list instead of sharing the sched_groups amongst each-other. This is needed for machines with more than 16 nodes, because sched_domain_node_span() will generate a node mask from the 16 nearest nodes without regard if these masks have any overlap. Currently sched_domains have a sched_group that maps to their child sched_domain span, and since there is no overlap we share the sched_group between the sched_domains of the various CPUs. If however there is overlap, we would need to link the sched_group list in different ways for each cpu, and hence sharing isn't possible. In order to solve this, allocate private sched_groups for each CPU's sched_domain but have the sched_groups share a sched_group_power structure such that we can uniquely track the power. Reported-and-tested-by: Anton Blanchard Signed-off-by: Peter Zijlstra Cc: Linus Torvalds Cc: Andrew Morton Link: http://lkml.kernel.org/n/tip-08bxqw9wis3qti9u5inifh3y@git.kernel.org Signed-off-by: Ingo Molnar --- include/linux/sched.h | 2 + kernel/sched.c | 157 +++++++++++++++++++++++++++++++++++++++--------- kernel/sched_features.h | 2 + 3 files changed, 132 insertions(+), 29 deletions(-) (limited to 'include/linux') diff --git a/include/linux/sched.h b/include/linux/sched.h index 2e5b3c8e2d3e..bde99d5358dc 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -844,6 +844,7 @@ enum cpu_idle_type { #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */ #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */ #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */ +#define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */ enum powersavings_balance_level { POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */ @@ -894,6 +895,7 @@ static inline int sd_power_saving_flags(void) } struct sched_group_power { + atomic_t ref; /* * CPU power of this group, SCHED_LOAD_SCALE being max power for a * single CPU. diff --git a/kernel/sched.c b/kernel/sched.c index 36c10d25d4cd..921adf6f6fad 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -6774,10 +6774,36 @@ static struct root_domain *alloc_rootdomain(void) return rd; } +static void free_sched_groups(struct sched_group *sg, int free_sgp) +{ + struct sched_group *tmp, *first; + + if (!sg) + return; + + first = sg; + do { + tmp = sg->next; + + if (free_sgp && atomic_dec_and_test(&sg->sgp->ref)) + kfree(sg->sgp); + + kfree(sg); + sg = tmp; + } while (sg != first); +} + static void free_sched_domain(struct rcu_head *rcu) { struct sched_domain *sd = container_of(rcu, struct sched_domain, rcu); - if (atomic_dec_and_test(&sd->groups->ref)) { + + /* + * If its an overlapping domain it has private groups, iterate and + * nuke them all. + */ + if (sd->flags & SD_OVERLAP) { + free_sched_groups(sd->groups, 1); + } else if (atomic_dec_and_test(&sd->groups->ref)) { kfree(sd->groups->sgp); kfree(sd->groups); } @@ -6967,15 +6993,73 @@ struct sched_domain_topology_level; typedef struct sched_domain *(*sched_domain_init_f)(struct sched_domain_topology_level *tl, int cpu); typedef const struct cpumask *(*sched_domain_mask_f)(int cpu); +#define SDTL_OVERLAP 0x01 + struct sched_domain_topology_level { sched_domain_init_f init; sched_domain_mask_f mask; + int flags; struct sd_data data; }; -/* - * Assumes the sched_domain tree is fully constructed - */ +static int +build_overlap_sched_groups(struct sched_domain *sd, int cpu) +{ + struct sched_group *first = NULL, *last = NULL, *groups = NULL, *sg; + const struct cpumask *span = sched_domain_span(sd); + struct cpumask *covered = sched_domains_tmpmask; + struct sd_data *sdd = sd->private; + struct sched_domain *child; + int i; + + cpumask_clear(covered); + + for_each_cpu(i, span) { + struct cpumask *sg_span; + + if (cpumask_test_cpu(i, covered)) + continue; + + sg = kzalloc_node(sizeof(struct sched_group) + cpumask_size(), + GFP_KERNEL, cpu_to_node(i)); + + if (!sg) + goto fail; + + sg_span = sched_group_cpus(sg); + + child = *per_cpu_ptr(sdd->sd, i); + if (child->child) { + child = child->child; + cpumask_copy(sg_span, sched_domain_span(child)); + } else + cpumask_set_cpu(i, sg_span); + + cpumask_or(covered, covered, sg_span); + + sg->sgp = *per_cpu_ptr(sdd->sgp, cpumask_first(sg_span)); + atomic_inc(&sg->sgp->ref); + + if (cpumask_test_cpu(cpu, sg_span)) + groups = sg; + + if (!first) + first = sg; + if (last) + last->next = sg; + last = sg; + last->next = first; + } + sd->groups = groups; + + return 0; + +fail: + free_sched_groups(first, 0); + + return -ENOMEM; +} + static int get_group(int cpu, struct sd_data *sdd, struct sched_group **sg) { struct sched_domain *sd = *per_cpu_ptr(sdd->sd, cpu); @@ -6987,23 +7071,21 @@ static int get_group(int cpu, struct sd_data *sdd, struct sched_group **sg) if (sg) { *sg = *per_cpu_ptr(sdd->sg, cpu); (*sg)->sgp = *per_cpu_ptr(sdd->sgp, cpu); + atomic_set(&(*sg)->sgp->ref, 1); /* for claim_allocations */ } return cpu; } /* - * build_sched_groups takes the cpumask we wish to span, and a pointer - * to a function which identifies what group(along with sched group) a CPU - * belongs to. The return value of group_fn must be a >= 0 and < nr_cpu_ids - * (due to the fact that we keep track of groups covered with a struct cpumask). - * * build_sched_groups will build a circular linked list of the groups * covered by the given span, and will set each group's ->cpumask correctly, * and ->cpu_power to 0. + * + * Assumes the sched_domain tree is fully constructed */ -static void -build_sched_groups(struct sched_domain *sd) +static int +build_sched_groups(struct sched_domain *sd, int cpu) { struct sched_group *first = NULL, *last = NULL; struct sd_data *sdd = sd->private; @@ -7011,6 +7093,12 @@ build_sched_groups(struct sched_domain *sd) struct cpumask *covered; int i; + get_group(cpu, sdd, &sd->groups); + atomic_inc(&sd->groups->ref); + + if (cpu != cpumask_first(sched_domain_span(sd))) + return 0; + lockdep_assert_held(&sched_domains_mutex); covered = sched_domains_tmpmask; @@ -7042,6 +7130,8 @@ build_sched_groups(struct sched_domain *sd) last = sg; } last->next = first; + + return 0; } /* @@ -7056,12 +7146,17 @@ build_sched_groups(struct sched_domain *sd) */ static void init_sched_groups_power(int cpu, struct sched_domain *sd) { - WARN_ON(!sd || !sd->groups); + struct sched_group *sg = sd->groups; - if (cpu != group_first_cpu(sd->groups)) - return; + WARN_ON(!sd || !sg); + + do { + sg->group_weight = cpumask_weight(sched_group_cpus(sg)); + sg = sg->next; + } while (sg != sd->groups); - sd->groups->group_weight = cpumask_weight(sched_group_cpus(sd->groups)); + if (cpu != group_first_cpu(sg)) + return; update_group_power(sd, cpu); } @@ -7182,16 +7277,15 @@ static enum s_alloc __visit_domain_allocation_hell(struct s_data *d, static void claim_allocations(int cpu, struct sched_domain *sd) { struct sd_data *sdd = sd->private; - struct sched_group *sg = sd->groups; WARN_ON_ONCE(*per_cpu_ptr(sdd->sd, cpu) != sd); *per_cpu_ptr(sdd->sd, cpu) = NULL; - if (cpu == cpumask_first(sched_group_cpus(sg))) { - WARN_ON_ONCE(*per_cpu_ptr(sdd->sg, cpu) != sg); + if (atomic_read(&(*per_cpu_ptr(sdd->sg, cpu))->ref)) *per_cpu_ptr(sdd->sg, cpu) = NULL; + + if (atomic_read(&(*per_cpu_ptr(sdd->sgp, cpu))->ref)) *per_cpu_ptr(sdd->sgp, cpu) = NULL; - } } #ifdef CONFIG_SCHED_SMT @@ -7216,7 +7310,7 @@ static struct sched_domain_topology_level default_topology[] = { #endif { sd_init_CPU, cpu_cpu_mask, }, #ifdef CONFIG_NUMA - { sd_init_NODE, cpu_node_mask, }, + { sd_init_NODE, cpu_node_mask, SDTL_OVERLAP, }, { sd_init_ALLNODES, cpu_allnodes_mask, }, #endif { NULL, }, @@ -7284,7 +7378,9 @@ static void __sdt_free(const struct cpumask *cpu_map) struct sd_data *sdd = &tl->data; for_each_cpu(j, cpu_map) { - kfree(*per_cpu_ptr(sdd->sd, j)); + struct sched_domain *sd = *per_cpu_ptr(sdd->sd, j); + if (sd && (sd->flags & SD_OVERLAP)) + free_sched_groups(sd->groups, 0); kfree(*per_cpu_ptr(sdd->sg, j)); kfree(*per_cpu_ptr(sdd->sgp, j)); } @@ -7336,8 +7432,11 @@ static int build_sched_domains(const struct cpumask *cpu_map, struct sched_domain_topology_level *tl; sd = NULL; - for (tl = sched_domain_topology; tl->init; tl++) + for (tl = sched_domain_topology; tl->init; tl++) { sd = build_sched_domain(tl, &d, cpu_map, attr, sd, i); + if (tl->flags & SDTL_OVERLAP || sched_feat(FORCE_SD_OVERLAP)) + sd->flags |= SD_OVERLAP; + } while (sd->child) sd = sd->child; @@ -7349,13 +7448,13 @@ static int build_sched_domains(const struct cpumask *cpu_map, for_each_cpu(i, cpu_map) { for (sd = *per_cpu_ptr(d.sd, i); sd; sd = sd->parent) { sd->span_weight = cpumask_weight(sched_domain_span(sd)); - get_group(i, sd->private, &sd->groups); - atomic_inc(&sd->groups->ref); - - if (i != cpumask_first(sched_domain_span(sd))) - continue; - - build_sched_groups(sd); + if (sd->flags & SD_OVERLAP) { + if (build_overlap_sched_groups(sd, i)) + goto error; + } else { + if (build_sched_groups(sd, i)) + goto error; + } } } diff --git a/kernel/sched_features.h b/kernel/sched_features.h index be40f7371ee1..1e7066d76c26 100644 --- a/kernel/sched_features.h +++ b/kernel/sched_features.h @@ -70,3 +70,5 @@ SCHED_FEAT(NONIRQ_POWER, 1) * using the scheduler IPI. Reduces rq->lock contention/bounces. */ SCHED_FEAT(TTWU_QUEUE, 1) + +SCHED_FEAT(FORCE_SD_OVERLAP, 0) -- cgit v1.2.3-71-gd317 From 31f5396ad3bde23c8416e8d23ba425e27f413314 Mon Sep 17 00:00:00 2001 From: Jean Delvare Date: Thu, 28 Jul 2011 13:48:42 -0700 Subject: thermal: make THERMAL_HWMON implementation fully internal THERMAL_HWMON is implemented inside the thermal_sys driver and has no effect on drivers implementing thermal zones, so they shouldn't see anything related to it in . Making the THERMAL_HWMON implementation fully internal has two advantages beyond the cleaner design: * This avoids rebuilding all thermal drivers if the THERMAL_HWMON implementation changes, or if CONFIG_THERMAL_HWMON gets enabled or disabled. * This avoids breaking the thermal kABI in these cases too, which should make distributions happy. The only drawback I can see is slightly higher memory fragmentation, as the number of kzalloc() calls will increase by one per thermal zone. But I doubt it will be a problem in practice, as I've never seen a system with more than two thermal zones. Signed-off-by: Jean Delvare Cc: Rene Herman Acked-by: Guenter Roeck Signed-off-by: Andrew Morton Signed-off-by: Len Brown --- drivers/thermal/thermal_sys.c | 117 +++++++++++++++++++++++++++++++++--------- include/linux/thermal.h | 22 -------- 2 files changed, 92 insertions(+), 47 deletions(-) (limited to 'include/linux') diff --git a/drivers/thermal/thermal_sys.c b/drivers/thermal/thermal_sys.c index b6353d61fac5..708f8e92771a 100644 --- a/drivers/thermal/thermal_sys.c +++ b/drivers/thermal/thermal_sys.c @@ -420,6 +420,29 @@ thermal_cooling_device_trip_point_show(struct device *dev, /* hwmon sys I/F */ #include + +/* thermal zone devices with the same type share one hwmon device */ +struct thermal_hwmon_device { + char type[THERMAL_NAME_LENGTH]; + struct device *device; + int count; + struct list_head tz_list; + struct list_head node; +}; + +struct thermal_hwmon_attr { + struct device_attribute attr; + char name[16]; +}; + +/* one temperature input for each thermal zone */ +struct thermal_hwmon_temp { + struct list_head hwmon_node; + struct thermal_zone_device *tz; + struct thermal_hwmon_attr temp_input; /* hwmon sys attr */ + struct thermal_hwmon_attr temp_crit; /* hwmon sys attr */ +}; + static LIST_HEAD(thermal_hwmon_list); static ssize_t @@ -437,9 +460,10 @@ temp_input_show(struct device *dev, struct device_attribute *attr, char *buf) int ret; struct thermal_hwmon_attr *hwmon_attr = container_of(attr, struct thermal_hwmon_attr, attr); - struct thermal_zone_device *tz - = container_of(hwmon_attr, struct thermal_zone_device, + struct thermal_hwmon_temp *temp + = container_of(hwmon_attr, struct thermal_hwmon_temp, temp_input); + struct thermal_zone_device *tz = temp->tz; ret = tz->ops->get_temp(tz, &temperature); @@ -455,9 +479,10 @@ temp_crit_show(struct device *dev, struct device_attribute *attr, { struct thermal_hwmon_attr *hwmon_attr = container_of(attr, struct thermal_hwmon_attr, attr); - struct thermal_zone_device *tz - = container_of(hwmon_attr, struct thermal_zone_device, + struct thermal_hwmon_temp *temp + = container_of(hwmon_attr, struct thermal_hwmon_temp, temp_crit); + struct thermal_zone_device *tz = temp->tz; long temperature; int ret; @@ -485,10 +510,29 @@ thermal_hwmon_lookup_by_type(const struct thermal_zone_device *tz) return NULL; } +/* Find the temperature input matching a given thermal zone */ +static struct thermal_hwmon_temp * +thermal_hwmon_lookup_temp(const struct thermal_hwmon_device *hwmon, + const struct thermal_zone_device *tz) +{ + struct thermal_hwmon_temp *temp; + + mutex_lock(&thermal_list_lock); + list_for_each_entry(temp, &hwmon->tz_list, hwmon_node) + if (temp->tz == tz) { + mutex_unlock(&thermal_list_lock); + return temp; + } + mutex_unlock(&thermal_list_lock); + + return NULL; +} + static int thermal_add_hwmon_sysfs(struct thermal_zone_device *tz) { struct thermal_hwmon_device *hwmon; + struct thermal_hwmon_temp *temp; int new_hwmon_device = 1; int result; @@ -515,30 +559,36 @@ thermal_add_hwmon_sysfs(struct thermal_zone_device *tz) goto free_mem; register_sys_interface: - tz->hwmon = hwmon; + temp = kzalloc(sizeof(struct thermal_hwmon_temp), GFP_KERNEL); + if (!temp) { + result = -ENOMEM; + goto unregister_name; + } + + temp->tz = tz; hwmon->count++; - snprintf(tz->temp_input.name, THERMAL_NAME_LENGTH, + snprintf(temp->temp_input.name, THERMAL_NAME_LENGTH, "temp%d_input", hwmon->count); - tz->temp_input.attr.attr.name = tz->temp_input.name; - tz->temp_input.attr.attr.mode = 0444; - tz->temp_input.attr.show = temp_input_show; - sysfs_attr_init(&tz->temp_input.attr.attr); - result = device_create_file(hwmon->device, &tz->temp_input.attr); + temp->temp_input.attr.attr.name = temp->temp_input.name; + temp->temp_input.attr.attr.mode = 0444; + temp->temp_input.attr.show = temp_input_show; + sysfs_attr_init(&temp->temp_input.attr.attr); + result = device_create_file(hwmon->device, &temp->temp_input.attr); if (result) - goto unregister_name; + goto free_temp_mem; if (tz->ops->get_crit_temp) { unsigned long temperature; if (!tz->ops->get_crit_temp(tz, &temperature)) { - snprintf(tz->temp_crit.name, THERMAL_NAME_LENGTH, + snprintf(temp->temp_crit.name, THERMAL_NAME_LENGTH, "temp%d_crit", hwmon->count); - tz->temp_crit.attr.attr.name = tz->temp_crit.name; - tz->temp_crit.attr.attr.mode = 0444; - tz->temp_crit.attr.show = temp_crit_show; - sysfs_attr_init(&tz->temp_crit.attr.attr); + temp->temp_crit.attr.attr.name = temp->temp_crit.name; + temp->temp_crit.attr.attr.mode = 0444; + temp->temp_crit.attr.show = temp_crit_show; + sysfs_attr_init(&temp->temp_crit.attr.attr); result = device_create_file(hwmon->device, - &tz->temp_crit.attr); + &temp->temp_crit.attr); if (result) goto unregister_input; } @@ -547,13 +597,15 @@ thermal_add_hwmon_sysfs(struct thermal_zone_device *tz) mutex_lock(&thermal_list_lock); if (new_hwmon_device) list_add_tail(&hwmon->node, &thermal_hwmon_list); - list_add_tail(&tz->hwmon_node, &hwmon->tz_list); + list_add_tail(&temp->hwmon_node, &hwmon->tz_list); mutex_unlock(&thermal_list_lock); return 0; unregister_input: - device_remove_file(hwmon->device, &tz->temp_input.attr); + device_remove_file(hwmon->device, &temp->temp_input.attr); + free_temp_mem: + kfree(temp); unregister_name: if (new_hwmon_device) { device_remove_file(hwmon->device, &dev_attr_name); @@ -569,15 +621,30 @@ thermal_add_hwmon_sysfs(struct thermal_zone_device *tz) static void thermal_remove_hwmon_sysfs(struct thermal_zone_device *tz) { - struct thermal_hwmon_device *hwmon = tz->hwmon; + struct thermal_hwmon_device *hwmon; + struct thermal_hwmon_temp *temp; + + hwmon = thermal_hwmon_lookup_by_type(tz); + if (unlikely(!hwmon)) { + /* Should never happen... */ + dev_dbg(&tz->device, "hwmon device lookup failed!\n"); + return; + } + + temp = thermal_hwmon_lookup_temp(hwmon, tz); + if (unlikely(!temp)) { + /* Should never happen... */ + dev_dbg(&tz->device, "temperature input lookup failed!\n"); + return; + } - tz->hwmon = NULL; - device_remove_file(hwmon->device, &tz->temp_input.attr); + device_remove_file(hwmon->device, &temp->temp_input.attr); if (tz->ops->get_crit_temp) - device_remove_file(hwmon->device, &tz->temp_crit.attr); + device_remove_file(hwmon->device, &temp->temp_crit.attr); mutex_lock(&thermal_list_lock); - list_del(&tz->hwmon_node); + list_del(&temp->hwmon_node); + kfree(temp); if (!list_empty(&hwmon->tz_list)) { mutex_unlock(&thermal_list_lock); return; diff --git a/include/linux/thermal.h b/include/linux/thermal.h index d3ec89fb4122..47b4a27e6e97 100644 --- a/include/linux/thermal.h +++ b/include/linux/thermal.h @@ -85,22 +85,6 @@ struct thermal_cooling_device { ((long)t-2732+5)/10 : ((long)t-2732-5)/10) #define CELSIUS_TO_KELVIN(t) ((t)*10+2732) -#if defined(CONFIG_THERMAL_HWMON) -/* thermal zone devices with the same type share one hwmon device */ -struct thermal_hwmon_device { - char type[THERMAL_NAME_LENGTH]; - struct device *device; - int count; - struct list_head tz_list; - struct list_head node; -}; - -struct thermal_hwmon_attr { - struct device_attribute attr; - char name[16]; -}; -#endif - struct thermal_zone_device { int id; char type[THERMAL_NAME_LENGTH]; @@ -120,12 +104,6 @@ struct thermal_zone_device { struct mutex lock; /* protect cooling devices list */ struct list_head node; struct delayed_work poll_queue; -#if defined(CONFIG_THERMAL_HWMON) - struct list_head hwmon_node; - struct thermal_hwmon_device *hwmon; - struct thermal_hwmon_attr temp_input; /* hwmon sys attr */ - struct thermal_hwmon_attr temp_crit; /* hwmon sys attr */ -#endif }; /* Adding event notification support elements */ #define THERMAL_GENL_FAMILY_NAME "thermal_event" -- cgit v1.2.3-71-gd317 From f52e00c668669c9c290e84adf859c76db6d92a5a Mon Sep 17 00:00:00 2001 From: David Rientjes Date: Thu, 28 Jul 2011 13:48:43 -0700 Subject: ACPI: remove NID_INVAL b552a8c56db8 ("ACPI: remove NID_INVAL") removed the left over uses of NID_INVAL, but didn't actually remove the definition. Remove it. Signed-off-by: David Rientjes Signed-off-by: Andrew Morton Signed-off-by: Len Brown --- include/linux/acpi.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/acpi.h b/include/linux/acpi.h index 1deb2a73c2da..2312e850aab8 100644 --- a/include/linux/acpi.h +++ b/include/linux/acpi.h @@ -238,7 +238,6 @@ extern int acpi_paddr_to_node(u64 start_addr, u64 size); extern int pnpacpi_disabled; #define PXM_INVAL (-1) -#define NID_INVAL (-1) int acpi_check_resource_conflict(const struct resource *res); -- cgit v1.2.3-71-gd317