Files
jianzhou 4be6354a11 Merge android-4.9.194(f360bd46) into msm-4.9
* refs/heads/tmp-f360bd46:
  Linux 4.9.194
  net_sched: let qdisc_put() accept NULL pointer
  ARC: export "abort" for modules
  media: technisat-usb2: break out of loop at end of buffer
  floppy: fix usercopy direction
  iommu/amd: Fix race in increase_address_space()
  keys: Fix missing null pointer check in request_key_auth_describe()
  x86/uaccess: Don't leak the AC flags into __get_user() argument evaluation
  dmaengine: ti: omap-dma: Add cleanup in omap_dma_probe()
  dmaengine: ti: dma-crossbar: Fix a memory leak bug
  net: seeq: Fix the function used to release some memory in an error handling path
  tools/power turbostat: fix buffer overrun
  perf/x86/amd/ibs: Fix sample bias for dispatched micro-ops
  perf/x86/intel: Restrict period on Nehalem
  sky2: Disable MSI on yet another ASUS boards (P6Xxxx)
  ARM: 8901/1: add a criteria for pfn_valid of arm
  cifs: Use kzfree() to zero out the password
  cifs: set domainName when a domain-key is used in multiuser
  NFSv2: Fix write regression
  NFSv2: Fix eof handling
  netfilter: nf_conntrack_ftp: Fix debug output
  x86/apic: Fix arch_dynirq_lower_bound() bug for DT enabled machines
  r8152: Set memory to all 0xFFs on failed reg reads
  batman-adv: Only read OGM2 tvlv_len after buffer len check
  ARM: 8874/1: mm: only adjust sections of valid mm structures
  qed: Add cleanup in qed_slowpath_start()
  Kconfig: Fix the reference to the IDT77105 Phy driver in the description of ATM_NICSTAR_USE_IDT77105
  NFS: Fix initialisation of I/O result struct in nfs_pgio_rpcsetup
  NFSv4: Fix return values for nfs4_file_open()
  s390/bpf: use 32-bit index for tail calls
  ARM: OMAP2+: Fix omap4 errata warning on other SoCs
  s390/bpf: fix lcgr instruction encoding
  ARM: OMAP2+: Fix missing SYSC_HAS_RESET_STATUS for dra7 epwmss
  mwifiex: Fix three heap overflow at parsing element in cfg80211_ap_settings
  tty/serial: atmel: reschedule TX after RX was started
  serial: sprd: correct the wrong sequence of arguments
  KVM: coalesced_mmio: add bounds checking
  xen-netfront: do not assume sk_buff_head list is empty in error handling
  x86/boot: Add missing bootparam that breaks boot on some platforms
  powerpc/mm/radix: Use the right page size for vmemmap mapping
  media: tm6000: double free if usb disconnect while streaming
  USB: usbcore: Fix slab-out-of-bounds bug during device reset
  x86/build: Add -Wnoaddress-of-packed-member to REALMODE_CFLAGS, to silence GCC9 build warning
  nvmem: Use the same permissions for eeprom as for nvmem
  drm/mediatek: mtk_drm_drv.c: Add of_node_put() before goto
  crypto: talitos - HMAC SNOOP NO AFEU mode requires SW icv checking.
  crypto: talitos - Do not modify req->cryptlen on decryption.
  crypto: talitos - fix ECB algs ivsize
  crypto: talitos - check data blocksize in ablkcipher.
  crypto: talitos - fix CTR alg blocksize
  crypto: talitos - check AES key size
  driver core: Fix use-after-free and double free on glue directory
  mtd: rawnand: mtk: Fix wrongly assigned OOB buffer pointer issue
  clk: rockchip: Don't yell about bad mmc phases when getting
  MIPS: VDSO: Use same -m%-float cflag as the kernel proper
  MIPS: VDSO: Prevent use of smp_processor_id()
  KVM: nVMX: handle page fault in vmread
  KVM: x86: work around leak of uninitialized stack contents
  KVM: s390: Do not leak kernel stack data in the KVM_S390_INTERRUPT ioctl
  genirq: Prevent NULL pointer dereference in resend_irqs()
  Btrfs: fix assertion failure during fsync and use of stale transaction
  gpio: fix line flag validation in lineevent_create
  gpio: fix line flag validation in linehandle_create
  Revert "MIPS: SiByte: Enable swiotlb for SWARM, LittleSur and BigSur"
  tun: fix use-after-free when register netdev failed
  tipc: add NULL pointer check before calling kfree_rcu
  tcp: fix tcp_ecn_withdraw_cwr() to clear TCP_ECN_QUEUE_CWR
  sctp: use transport pf_retrans in sctp_do_8_2_transport_strike
  sctp: Fix the link time qualifier of 'sctp_ctrlsock_exit()'
  sch_hhf: ensure quantum and hhf_non_hh_weight are non-zero
  net: gso: Fix skb_segment splat when splitting gso_size mangled skb having linear-headed frag_list
  net: Fix null de-reference of device refcount
  isdn/capi: check message length in capi_write()
  ipv6: Fix the link time qualifier of 'ping_v6_proc_exit_net()'
  cdc_ether: fix rndis support for Mediatek based smartphones
  bridge/mdb: remove wrong use of NLM_F_MULTI
  ANDROID: regression introduced override_creds=off
  UPSTREAM: psi: get poll_work to run when calling poll syscall next time
  UPSTREAM: sched/psi: Do not require setsched permission from the trigger creator
  UPSTREAM: sched/psi: Reduce psimon FIFO priority
  ANDROID: Revert "ANDROID: sched: Disallow WALT with CFS bandwidth control"
  ANDROID: sched: WALT: Add support for CFS_BANDWIDTH
  ANDROID: sched: WALT: Refactor cumulative runnable average fixup
  BACKPORT: arch: add pidfd and io_uring syscalls everywhere
  UPSTREAM: pidfd: fix a poll race when setting exit_state
  BACKPORT: arch: wire-up pidfd_open()
  BACKPORT: pid: add pidfd_open()
  UPSTREAM: pidfd: add polling support
  UPSTREAM: signal: improve comments
  BACKPORT: fork: do not release lock that wasn't taken
  BACKPORT: signal: support CLONE_PIDFD with pidfd_send_signal
  BACKPORT: clone: add CLONE_PIDFD
  UPSTREAM: Make anon_inodes unconditional
  UPSTREAM: signal: use fdget() since we don't allow O_PATH
  UPSTREAM: signal: don't silently convert SI_USER signals to non-current pidfd
  BACKPORT: signal: add pidfd_send_signal() syscall
  ANDROID: sched: Disallow WALT with CFS bandwidth control
  ANDROID: fiq_debugger: remove
  ANDROID: fix kernelci build-break in lowmemorykiller
  ANDROID: Fixes to locking around handle_lmk_event
  UPSTREAM: net/ipv6: allow sysctl to change link-local address generation mode
  ANDROID: fix binder change in merge of 4.9.188
  fs/crypto: Fix 4.9.186 missmerge
  UPSTREAM: net-ipv6-ndisc: add support for RFC7710 RA Captive Portal Identifier
  ANDROID: fix up 9p filesystem due to CFI non-upstream patches
  ANDROID: enable CONFIG_RTC_DRV_TEST on cuttlefish
  ANDROID: xfrm: remove in_compat_syscall() checks
  UPSTREAM: binder: Set end of SG buffer area properly.
  ANDROID: Avoid taking multiple locks in handle_lmk_event
  efi/libstub: remove duplicate nokaslr
  UPSTREAM: binder: check for overflow when alloc for security context
  BACKPORT: binder: fix race between munmap() and direct reclaim
  Revert "fib_rules: return 0 directly if an exactly same rule exists when NLM_F_EXCL not supplied"
  Revert "fib_rules: fix error in backport of e9919a24d302 ("fib_rules: return 0...")"
  Revert "x86/build: Move _etext to actual end of .text"
  UPSTREAM: fib_rules: fix error in backport of e9919a24d302 ("fib_rules: return 0...")
  ANDROID: cuttlefish 4.9: enable CONFIG_CRYPTO_AES_NI_INTEL=y
  UPSTREAM: bpf: relax inode permission check for retrieving bpf program
  ANDROID: cuttlefish_defconfig: Disable DEVTMPFS
  ANDROID: block/cfq-iosched: make group_idle per io cgroup tunable
  ANDROID: Move from clang r349610 to r353983c.
  ANDROID: cuttlefish_defconfig: Enable CONFIG_CPUSETS and CONFIG_CGROUP_SCHEDTUNE
  ANDROID: Communicates LMK events to userland where they can be logged
  ANDROID: cuttlefish_defconfig: Enable CONFIG_XFRM_STATISTICS
  ANDROID: cuttlefish_defconfig: Enable L2TP/PPTP
  Make arm64 serial port config compatible with crosvm
  ANDROID: Makefile: Properly resolve 4.9.169 merge
  ANDROID: cuttlefish_defconfig: Enable CONFIG_OVERLAY_FS
  ANDROID: cuttlefish: enable CONFIG_NET_SCH_INGRESS=y
  UPSTREAM: virt_wifi: Remove REGULATORY_WIPHY_SELF_MANAGED

Change-Id: I6167a0e2c957a504c02f3e6bd4c25cb047f0c416
Signed-off-by: jianzhou <jianzhou@codeaurora.org>
2019-12-04 15:00:19 +08:00

397 lines
10 KiB
C
Executable File

/*
* Copyright (c) 2016-2018, The Linux Foundation. All rights reserved.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 and
* only version 2 as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*/
#ifndef __WALT_H
#define __WALT_H
#ifdef CONFIG_SCHED_WALT
#include <linux/sched/sysctl.h>
#define WINDOW_STATS_RECENT 0
#define WINDOW_STATS_MAX 1
#define WINDOW_STATS_MAX_RECENT_AVG 2
#define WINDOW_STATS_AVG 3
#define WINDOW_STATS_INVALID_POLICY 4
#define EXITING_TASK_MARKER 0xdeaddead
#define FREQ_REPORT_MAX_CPU_LOAD_TOP_TASK 0
#define FREQ_REPORT_CPU_LOAD 1
#define FREQ_REPORT_TOP_TASK 2
#define for_each_related_thread_group(grp) \
list_for_each_entry(grp, &active_related_thread_groups, list)
#define SCHED_NEW_TASK_WINDOWS 5
extern unsigned int sched_ravg_window;
extern unsigned int max_possible_efficiency;
extern unsigned int min_possible_efficiency;
extern unsigned int max_possible_freq;
extern unsigned int sched_major_task_runtime;
extern unsigned int __read_mostly sched_load_granule;
extern struct mutex cluster_lock;
extern rwlock_t related_thread_group_lock;
extern __read_mostly unsigned int sched_ravg_hist_size;
extern __read_mostly unsigned int sched_freq_aggregate;
extern __read_mostly int sched_freq_aggregate_threshold;
extern __read_mostly unsigned int sched_window_stats_policy;
extern __read_mostly unsigned int sched_group_upmigrate;
extern __read_mostly unsigned int sched_group_downmigrate;
extern struct sched_cluster init_cluster;
extern void update_task_ravg(struct task_struct *p, struct rq *rq, int event,
u64 wallclock, u64 irqtime);
extern unsigned int nr_eligible_big_tasks(int cpu);
static inline void
inc_nr_big_task(struct walt_sched_stats *stats, struct task_struct *p)
{
if (sched_disable_window_stats)
return;
if (p->misfit)
stats->nr_big_tasks++;
}
static inline void
dec_nr_big_task(struct walt_sched_stats *stats, struct task_struct *p)
{
if (sched_disable_window_stats)
return;
if (p->misfit)
stats->nr_big_tasks--;
BUG_ON(stats->nr_big_tasks < 0);
}
static inline void
walt_adjust_nr_big_tasks(struct rq *rq, int delta, bool inc)
{
if (sched_disable_window_stats)
return;
sched_update_nr_prod(cpu_of(rq), 0, true);
rq->walt_stats.nr_big_tasks += inc ? delta : -delta;
BUG_ON(rq->walt_stats.nr_big_tasks < 0);
}
static inline void
fixup_cumulative_runnable_avg(struct walt_sched_stats *stats,
s64 task_load_delta, s64 pred_demand_delta)
{
if (sched_disable_window_stats)
return;
stats->cumulative_runnable_avg += task_load_delta;
BUG_ON((s64)stats->cumulative_runnable_avg < 0);
stats->pred_demands_sum += pred_demand_delta;
BUG_ON((s64)stats->pred_demands_sum < 0);
}
static inline void
walt_inc_cumulative_runnable_avg(struct rq *rq, struct task_struct *p)
{
if (sched_disable_window_stats)
return;
fixup_cumulative_runnable_avg(&rq->walt_stats, p->ravg.demand,
p->ravg.pred_demand);
/*
* Add a task's contribution to the cumulative window demand when
*
* (1) task is enqueued with on_rq = 1 i.e migration,
* prio/cgroup/class change.
* (2) task is waking for the first time in this window.
*/
if (p->on_rq || (p->last_sleep_ts < rq->window_start))
walt_fixup_cum_window_demand(rq, p->ravg.demand);
}
static inline void
walt_dec_cumulative_runnable_avg(struct rq *rq, struct task_struct *p)
{
if (sched_disable_window_stats)
return;
fixup_cumulative_runnable_avg(&rq->walt_stats, -(s64)p->ravg.demand,
-(s64)p->ravg.pred_demand);
/*
* on_rq will be 1 for sleeping tasks. So check if the task
* is migrating or dequeuing in RUNNING state to change the
* prio/cgroup/class.
*/
if (task_on_rq_migrating(p) || p->state == TASK_RUNNING)
walt_fixup_cum_window_demand(rq, -(s64)p->ravg.demand);
}
extern void fixup_walt_sched_stats_common(struct rq *rq, struct task_struct *p,
u32 new_task_load,
u32 new_pred_demand);
extern void inc_rq_walt_stats(struct rq *rq, struct task_struct *p);
extern void dec_rq_walt_stats(struct rq *rq, struct task_struct *p);
extern void fixup_busy_time(struct task_struct *p, int new_cpu);
extern void init_new_task_load(struct task_struct *p);
extern void mark_task_starting(struct task_struct *p);
extern void set_window_start(struct rq *rq);
void account_irqtime(int cpu, struct task_struct *curr, u64 delta,
u64 wallclock);
void walt_fixup_cumulative_runnable_avg(struct rq *rq, struct task_struct *p,
u64 new_task_load);
extern bool do_pl_notif(struct rq *rq);
#define SCHED_HIGH_IRQ_TIMEOUT 3
static inline u64 sched_irqload(int cpu)
{
struct rq *rq = cpu_rq(cpu);
s64 delta;
delta = get_jiffies_64() - rq->irqload_ts;
/*
* Current context can be preempted by irq and rq->irqload_ts can be
* updated by irq context so that delta can be negative.
* But this is okay and we can safely return as this means there
* was recent irq occurrence.
*/
if (delta < SCHED_HIGH_IRQ_TIMEOUT)
return rq->avg_irqload;
else
return 0;
}
static inline int sched_cpu_high_irqload(int cpu)
{
return sched_irqload(cpu) >= sysctl_sched_cpu_high_irqload;
}
static inline int exiting_task(struct task_struct *p)
{
return (p->ravg.sum_history[0] == EXITING_TASK_MARKER);
}
static inline struct sched_cluster *cpu_cluster(int cpu)
{
return cpu_rq(cpu)->cluster;
}
static inline u64
scale_load_to_freq(u64 load, unsigned int src_freq, unsigned int dst_freq)
{
return div64_u64(load * (u64)src_freq, (u64)dst_freq);
}
static inline bool is_new_task(struct task_struct *p)
{
return p->ravg.active_windows < SCHED_NEW_TASK_WINDOWS;
}
static inline void clear_top_tasks_table(u8 *table)
{
memset(table, 0, NUM_LOAD_INDICES * sizeof(u8));
}
extern void update_cluster_load_subtractions(struct task_struct *p,
int cpu, u64 ws, bool new_task);
extern void sched_account_irqstart(int cpu, struct task_struct *curr,
u64 wallclock);
static inline unsigned int max_task_load(void)
{
return sched_ravg_window;
}
static inline u32 cpu_cycles_to_freq(u64 cycles, u64 period)
{
return div64_u64(cycles, period);
}
static inline unsigned int cpu_cur_freq(int cpu)
{
return cpu_rq(cpu)->cluster->cur_freq;
}
static inline void
move_list(struct list_head *dst, struct list_head *src, bool sync_rcu)
{
struct list_head *first, *last;
first = src->next;
last = src->prev;
if (sync_rcu) {
INIT_LIST_HEAD_RCU(src);
synchronize_rcu();
}
first->prev = dst;
dst->prev = last;
last->next = dst;
/* Ensure list sanity before making the head visible to all CPUs. */
smp_mb();
dst->next = first;
}
extern void reset_task_stats(struct task_struct *p);
extern void update_cluster_topology(void);
extern struct list_head cluster_head;
#define for_each_sched_cluster(cluster) \
list_for_each_entry_rcu(cluster, &cluster_head, list)
extern void init_clusters(void);
extern void clear_top_tasks_bitmap(unsigned long *bitmap);
extern void sched_account_irqtime(int cpu, struct task_struct *curr,
u64 delta, u64 wallclock);
static inline void assign_cluster_ids(struct list_head *head)
{
struct sched_cluster *cluster;
int pos = 0;
list_for_each_entry(cluster, head, list) {
cluster->id = pos;
sched_cluster[pos++] = cluster;
}
}
static inline int same_cluster(int src_cpu, int dst_cpu)
{
return cpu_rq(src_cpu)->cluster == cpu_rq(dst_cpu)->cluster;
}
void walt_irq_work(struct irq_work *irq_work);
void walt_sched_init(struct rq *rq);
extern int __read_mostly min_power_cpu;
static inline int walt_start_cpu(int prev_cpu)
{
return sysctl_sched_is_big_little ? prev_cpu : min_power_cpu;
}
static inline void walt_update_last_enqueue(struct task_struct *p)
{
p->last_enqueued_ts = sched_ktime_clock();
}
extern void walt_rotate_work_init(void);
extern void walt_rotation_checkpoint(int nr_big);
extern unsigned int walt_rotation_enabled;
#else /* CONFIG_SCHED_WALT */
static inline void walt_sched_init(struct rq *rq) { }
static inline void walt_rotate_work_init(void) { }
static inline void walt_rotation_checkpoint(int nr_big) { }
static inline void walt_update_last_enqueue(struct task_struct *p) { }
static inline void walt_fixup_cumulative_runnable_avg(struct rq *rq,
struct task_struct *p,
u64 new_task_load) { }
static inline void update_task_ravg(struct task_struct *p, struct rq *rq,
int event, u64 wallclock, u64 irqtime) { }
static inline void walt_inc_cumulative_runnable_avg(struct rq *rq,
struct task_struct *p)
{
}
static inline unsigned int nr_eligible_big_tasks(int cpu)
{
return 0;
}
static inline void walt_adjust_nr_big_tasks(struct rq *rq,
int delta, bool inc)
{
}
static inline void inc_nr_big_task(struct walt_sched_stats *stats,
struct task_struct *p)
{
}
static inline void dec_nr_big_task(struct walt_sched_stats *stats,
struct task_struct *p)
{
}
static inline void walt_dec_cumulative_runnable_avg(struct rq *rq,
struct task_struct *p)
{
}
static inline void fixup_busy_time(struct task_struct *p, int new_cpu) { }
static inline void init_new_task_load(struct task_struct *p)
{
}
static inline void mark_task_starting(struct task_struct *p) { }
static inline void set_window_start(struct rq *rq) { }
static inline int sched_cpu_high_irqload(int cpu) { return 0; }
static inline void sched_account_irqstart(int cpu, struct task_struct *curr,
u64 wallclock)
{
}
static inline void update_cluster_topology(void) { }
static inline void init_clusters(void) {}
static inline void sched_account_irqtime(int cpu, struct task_struct *curr,
u64 delta, u64 wallclock)
{
}
static inline int same_cluster(int src_cpu, int dst_cpu) { return 1; }
static inline bool do_pl_notif(struct rq *rq) { return false; }
static inline void
inc_rq_walt_stats(struct rq *rq, struct task_struct *p) { }
static inline void
dec_rq_walt_stats(struct rq *rq, struct task_struct *p) { }
static inline void
fixup_walt_sched_stats_common(struct rq *rq, struct task_struct *p,
u32 new_task_load, u32 new_pred_demand)
{
}
static inline int walt_start_cpu(int prev_cpu)
{
return prev_cpu;
}
static inline u64 sched_irqload(int cpu)
{
return 0;
}
#endif /* CONFIG_SCHED_WALT */
#define walt_cpu_high_irqload(cpu) sched_cpu_high_irqload(cpu)
#endif