summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--arch/alpha/kernel/process.c2
-rw-r--r--arch/arc/kernel/process.c2
-rw-r--r--arch/arm/kernel/process.c2
-rw-r--r--arch/arm64/kernel/process.c2
-rw-r--r--arch/csky/kernel/process.c2
-rw-r--r--arch/h8300/kernel/process.c2
-rw-r--r--arch/hexagon/kernel/process.c2
-rw-r--r--arch/ia64/kernel/process.c2
-rw-r--r--arch/m68k/kernel/process.c2
-rw-r--r--arch/microblaze/kernel/process.c2
-rw-r--r--arch/mips/kernel/process.c2
-rw-r--r--arch/nds32/kernel/process.c2
-rw-r--r--arch/nios2/kernel/process.c2
-rw-r--r--arch/openrisc/kernel/process.c2
-rw-r--r--arch/parisc/kernel/process.c2
-rw-r--r--arch/powerpc/kernel/process.c2
-rw-r--r--arch/riscv/kernel/process.c2
-rw-r--r--arch/s390/kernel/process.c2
-rw-r--r--arch/sh/kernel/process_32.c2
-rw-r--r--arch/sparc/kernel/process_32.c2
-rw-r--r--arch/sparc/kernel/process_64.c2
-rw-r--r--arch/um/kernel/process.c2
-rw-r--r--arch/x86/kernel/process.c2
-rw-r--r--arch/xtensa/kernel/process.c2
-rw-r--r--fs/io-wq.c621
-rw-r--r--fs/io-wq.h35
-rw-r--r--fs/io_uring.c1043
-rw-r--r--fs/proc/self.c7
-rw-r--r--fs/proc/thread_self.c7
-rw-r--r--include/linux/io_uring.h22
-rw-r--r--include/linux/net.h3
-rw-r--r--include/linux/sched.h3
-rw-r--r--include/uapi/linux/io_uring.h1
-rw-r--r--kernel/ptrace.c2
-rw-r--r--kernel/signal.c4
-rw-r--r--net/ipv4/af_inet.c1
-rw-r--r--net/ipv6/af_inet6.c1
-rw-r--r--net/socket.c10
38 files changed, 694 insertions, 1114 deletions
diff --git a/arch/alpha/kernel/process.c b/arch/alpha/kernel/process.c
index 6c71554206cc..5112ab996394 100644
--- a/arch/alpha/kernel/process.c
+++ b/arch/alpha/kernel/process.c
@@ -249,7 +249,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,
childti->pcb.ksp = (unsigned long) childstack;
childti->pcb.flags = 1; /* set FEN, clear everything else */
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
/* kernel thread */
memset(childstack, 0,
sizeof(struct switch_stack) + sizeof(struct pt_regs));
diff --git a/arch/arc/kernel/process.c b/arch/arc/kernel/process.c
index 37f724ad5e39..d838d0d57696 100644
--- a/arch/arc/kernel/process.c
+++ b/arch/arc/kernel/process.c
@@ -191,7 +191,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,
childksp[0] = 0; /* fp */
childksp[1] = (unsigned long)ret_from_fork; /* blink */
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(c_regs, 0, sizeof(struct pt_regs));
c_callee->r13 = kthread_arg;
diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c
index ee3aee69e444..5199a2bb4111 100644
--- a/arch/arm/kernel/process.c
+++ b/arch/arm/kernel/process.c
@@ -243,7 +243,7 @@ int copy_thread(unsigned long clone_flags, unsigned long stack_start,
thread->cpu_domain = get_domain();
#endif
- if (likely(!(p->flags & PF_KTHREAD))) {
+ if (likely(!(p->flags & (PF_KTHREAD | PF_IO_WORKER)))) {
*childregs = *current_pt_regs();
childregs->ARM_r0 = 0;
if (stack_start)
diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
index 4cc1ccc8d6ab..325c83b1a24d 100644
--- a/arch/arm64/kernel/process.c
+++ b/arch/arm64/kernel/process.c
@@ -398,7 +398,7 @@ int copy_thread(unsigned long clone_flags, unsigned long stack_start,
ptrauth_thread_init_kernel(p);
- if (likely(!(p->flags & PF_KTHREAD))) {
+ if (likely(!(p->flags & (PF_KTHREAD | PF_IO_WORKER)))) {
*childregs = *current_pt_regs();
childregs->regs[0] = 0;
diff --git a/arch/csky/kernel/process.c b/arch/csky/kernel/process.c
index 69af6bc87e64..3d0ca22cd0e2 100644
--- a/arch/csky/kernel/process.c
+++ b/arch/csky/kernel/process.c
@@ -49,7 +49,7 @@ int copy_thread(unsigned long clone_flags,
/* setup thread.sp for switch_to !!! */
p->thread.sp = (unsigned long)childstack;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(childregs, 0, sizeof(struct pt_regs));
childstack->r15 = (unsigned long) ret_from_kernel_thread;
childstack->r10 = kthread_arg;
diff --git a/arch/h8300/kernel/process.c b/arch/h8300/kernel/process.c
index bc1364db58fe..46b1342ce515 100644
--- a/arch/h8300/kernel/process.c
+++ b/arch/h8300/kernel/process.c
@@ -112,7 +112,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,
childregs = (struct pt_regs *) (THREAD_SIZE + task_stack_page(p)) - 1;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(childregs, 0, sizeof(struct pt_regs));
childregs->retpc = (unsigned long) ret_from_kernel_thread;
childregs->er4 = topstk; /* arg */
diff --git a/arch/hexagon/kernel/process.c b/arch/hexagon/kernel/process.c
index 6a980cba7b29..c61165c99ae0 100644
--- a/arch/hexagon/kernel/process.c
+++ b/arch/hexagon/kernel/process.c
@@ -73,7 +73,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, unsigned long arg,
sizeof(*ss));
ss->lr = (unsigned long)ret_from_fork;
p->thread.switch_sp = ss;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(childregs, 0, sizeof(struct pt_regs));
/* r24 <- fn, r25 <- arg */
ss->r24 = usp;
diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c
index 4ebbfa076a26..7e1a1525e202 100644
--- a/arch/ia64/kernel/process.c
+++ b/arch/ia64/kernel/process.c
@@ -338,7 +338,7 @@ copy_thread(unsigned long clone_flags, unsigned long user_stack_base,
ia64_drop_fpu(p); /* don't pick up stale state from a CPU's fph */
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
if (unlikely(!user_stack_base)) {
/* fork_idle() called us */
return 0;
diff --git a/arch/m68k/kernel/process.c b/arch/m68k/kernel/process.c
index 08359a6e058f..da83cc83e791 100644
--- a/arch/m68k/kernel/process.c
+++ b/arch/m68k/kernel/process.c
@@ -157,7 +157,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, unsigned long arg,
*/
p->thread.fs = get_fs().seg;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
/* kernel thread */
memset(frame, 0, sizeof(struct fork_frame));
frame->regs.sr = PS_S;
diff --git a/arch/microblaze/kernel/process.c b/arch/microblaze/kernel/process.c
index 657c2beb665e..62aa237180b6 100644
--- a/arch/microblaze/kernel/process.c
+++ b/arch/microblaze/kernel/process.c
@@ -59,7 +59,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, unsigned long arg,
struct pt_regs *childregs = task_pt_regs(p);
struct thread_info *ti = task_thread_info(p);
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
/* if we're creating a new kernel thread then just zeroing all
* the registers. That's OK for a brand new thread.*/
memset(childregs, 0, sizeof(struct pt_regs));
diff --git a/arch/mips/kernel/process.c b/arch/mips/kernel/process.c
index af4c862ec5ff..7efa0d1a4c2b 100644
--- a/arch/mips/kernel/process.c
+++ b/arch/mips/kernel/process.c
@@ -120,7 +120,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,
/* Put the stack after the struct pt_regs. */
childksp = (unsigned long) childregs;
p->thread.cp0_status = (read_c0_status() & ~(ST0_CU2|ST0_CU1)) | ST0_KERNEL_CUMASK;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
/* kernel thread */
unsigned long status = p->thread.cp0_status;
memset(childregs, 0, sizeof(struct pt_regs));
diff --git a/arch/nds32/kernel/process.c b/arch/nds32/kernel/process.c
index e01ad5d17224..c1327e552ec6 100644
--- a/arch/nds32/kernel/process.c
+++ b/arch/nds32/kernel/process.c
@@ -156,7 +156,7 @@ int copy_thread(unsigned long clone_flags, unsigned long stack_start,
memset(&p->thread.cpu_context, 0, sizeof(struct cpu_context));
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(childregs, 0, sizeof(struct pt_regs));
/* kernel thread fn */
p->thread.cpu_context.r6 = stack_start;
diff --git a/arch/nios2/kernel/process.c b/arch/nios2/kernel/process.c
index 50b4eb19a6cc..c5f916ca6845 100644
--- a/arch/nios2/kernel/process.c
+++ b/arch/nios2/kernel/process.c
@@ -109,7 +109,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, unsigned long arg,
struct switch_stack *childstack =
((struct switch_stack *)childregs) - 1;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(childstack, 0,
sizeof(struct switch_stack) + sizeof(struct pt_regs));
diff --git a/arch/openrisc/kernel/process.c b/arch/openrisc/kernel/process.c
index 181448f74316..eb62429681fc 100644
--- a/arch/openrisc/kernel/process.c
+++ b/arch/openrisc/kernel/process.c
@@ -174,7 +174,7 @@ copy_thread(unsigned long clone_flags, unsigned long usp, unsigned long arg,
sp -= sizeof(struct pt_regs);
kregs = (struct pt_regs *)sp;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(kregs, 0, sizeof(struct pt_regs));
kregs->gpr[20] = usp; /* fn, kernel thread */
kregs->gpr[22] = arg;
diff --git a/arch/parisc/kernel/process.c b/arch/parisc/kernel/process.c
index fda1c1a6a444..b144fbe29bc1 100644
--- a/arch/parisc/kernel/process.c
+++ b/arch/parisc/kernel/process.c
@@ -200,7 +200,7 @@ copy_thread(unsigned long clone_flags, unsigned long usp,
extern void * const ret_from_kernel_thread;
extern void * const child_return;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
/* kernel thread */
memset(cregs, 0, sizeof(struct pt_regs));
if (!usp) /* idle thread */
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 924d023dad0a..3231c2df9e26 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -1670,7 +1670,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,
/* Copy registers */
sp -= sizeof(struct pt_regs);
childregs = (struct pt_regs *) sp;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
/* kernel thread */
memset(childregs, 0, sizeof(struct pt_regs));
childregs->gpr[1] = sp + sizeof(struct pt_regs);
diff --git a/arch/riscv/kernel/process.c b/arch/riscv/kernel/process.c
index 19f4688f2f36..6f728e731bed 100644
--- a/arch/riscv/kernel/process.c
+++ b/arch/riscv/kernel/process.c
@@ -124,7 +124,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, unsigned long arg,
struct pt_regs *childregs = task_pt_regs(p);
/* p->thread holds context to be restored by __switch_to() */
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
/* Kernel thread */
memset(childregs, 0, sizeof(struct pt_regs));
childregs->gp = gp_in_global;
diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c
index 367bd000f6d1..e20bed1ed34a 100644
--- a/arch/s390/kernel/process.c
+++ b/arch/s390/kernel/process.c
@@ -130,7 +130,7 @@ int copy_thread(unsigned long clone_flags, unsigned long new_stackp,
frame->sf.gprs[9] = (unsigned long)frame;
/* Store access registers to kernel stack of new process. */
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
/* kernel thread */
memset(&frame->childregs, 0, sizeof(struct pt_regs));
frame->childregs.psw.mask = PSW_KERNEL_BITS | PSW_MASK_DAT |
diff --git a/arch/sh/kernel/process_32.c b/arch/sh/kernel/process_32.c
index 80a5d1c66a51..1aa508eb0823 100644
--- a/arch/sh/kernel/process_32.c
+++ b/arch/sh/kernel/process_32.c
@@ -114,7 +114,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, unsigned long arg,
childregs = task_pt_regs(p);
p->thread.sp = (unsigned long) childregs;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(childregs, 0, sizeof(struct pt_regs));
p->thread.pc = (unsigned long) ret_from_kernel_thread;
childregs->regs[4] = arg;
diff --git a/arch/sparc/kernel/process_32.c b/arch/sparc/kernel/process_32.c
index 7649b14d69f8..b91e88058e0c 100644
--- a/arch/sparc/kernel/process_32.c
+++ b/arch/sparc/kernel/process_32.c
@@ -309,7 +309,7 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, unsigned long arg,
ti->ksp = (unsigned long) new_stack;
p->thread.kregs = childregs;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
extern int nwindows;
unsigned long psr;
memset(new_stack, 0, STACKFRAME_SZ + TRACEREG_SZ);
diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c
index 6f8c7822fc06..7afd0a859a78 100644
--- a/arch/sparc/kernel/process_64.c
+++ b/arch/sparc/kernel/process_64.c
@@ -597,7 +597,7 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, unsigned long arg,
sizeof(struct sparc_stackf));
t->fpsaved[0] = 0;
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(child_trap_frame, 0, child_stack_sz);
__thread_flag_byte_ptr(t)[TI_FLAG_BYTE_CWP] =
(current_pt_regs()->tstate + 1) & TSTATE_CWP;
diff --git a/arch/um/kernel/process.c b/arch/um/kernel/process.c
index 81d508daf67c..c5011064b5dd 100644
--- a/arch/um/kernel/process.c
+++ b/arch/um/kernel/process.c
@@ -157,7 +157,7 @@ int copy_thread(unsigned long clone_flags, unsigned long sp,
unsigned long arg, struct task_struct * p, unsigned long tls)
{
void (*handler)(void);
- int kthread = current->flags & PF_KTHREAD;
+ int kthread = current->flags & (PF_KTHREAD | PF_IO_WORKER);
int ret = 0;
p->thread = (struct thread_struct) INIT_THREAD;
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index 145a7ac0c19a..9c214d7085a4 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -161,7 +161,7 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, unsigned long arg,
#endif
/* Kernel thread ? */
- if (unlikely(p->flags & PF_KTHREAD)) {
+ if (unlikely(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
memset(childregs, 0, sizeof(struct pt_regs));
kthread_frame_init(frame, sp, arg);
return 0;
diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c
index 397a7de56377..9534ef515d74 100644
--- a/arch/xtensa/kernel/process.c
+++ b/arch/xtensa/kernel/process.c
@@ -217,7 +217,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp_thread_fn,
p->thread.sp = (unsigned long)childregs;
- if (!(p->flags & PF_KTHREAD)) {
+ if (!(p->flags & (PF_KTHREAD | PF_IO_WORKER))) {
struct pt_regs *regs = current_pt_regs();
unsigned long usp = usp_thread_fn ?
usp_thread_fn : regs->areg[1];
diff --git a/fs/io-wq.c b/fs/io-wq.c
index c36bbcd823ce..44e20248805a 100644
--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -13,13 +13,9 @@
#include <linux/sched/mm.h>
#include <linux/percpu.h>
#include <linux/slab.h>
-#include <linux/kthread.h>
#include <linux/rculist_nulls.h>
-#include <linux/fs_struct.h>
-#include <linux/task_work.h>
-#include <linux/blk-cgroup.h>
-#include <linux/audit.h>
#include <linux/cpu.h>
+#include <linux/tracehook.h>
#include "../kernel/sched/sched.h"
#include "io-wq.h"
@@ -36,7 +32,6 @@ enum {
enum {
IO_WQ_BIT_EXIT = 0, /* wq exiting */
- IO_WQ_BIT_ERROR = 1, /* error on setup */
};
enum {
@@ -57,14 +52,12 @@ struct io_worker {
struct io_wq_work *cur_work;
spinlock_t lock;
- struct rcu_head rcu;
- struct mm_struct *mm;
-#ifdef CONFIG_BLK_CGROUP
- struct cgroup_subsys_state *blkcg_css;
-#endif
const struct cred *cur_creds;
const struct cred *saved_creds;
- struct nsproxy *restore_nsproxy;
+
+ struct completion ref_done;
+
+ struct rcu_head rcu;
};
#if BITS_PER_LONG == 64
@@ -93,7 +86,6 @@ struct io_wqe {
struct {
raw_spinlock_t lock;
struct io_wq_work_list work_list;
- unsigned long hash_map;
unsigned flags;
} ____cacheline_aligned_in_smp;
@@ -103,6 +95,8 @@ struct io_wqe {
struct hlist_nulls_head free_list;
struct list_head all_list;
+ struct wait_queue_entry wait;
+
struct io_wq *wq;
struct io_wq_work *hash_tail[IO_WQ_NR_HASH_BUCKETS];
};
@@ -119,12 +113,15 @@ struct io_wq {
struct task_struct *manager;
struct user_struct *user;
+
+ struct io_wq_hash *hash;
+
refcount_t refs;
struct completion done;
struct hlist_node cpuhp_node;
- refcount_t use_refs;
+ pid_t task_pid;
};
static enum cpuhp_state io_wq_online;
@@ -137,62 +134,7 @@ static bool io_worker_get(struct io_worker *worker)
static void io_worker_release(struct io_worker *worker)
{
if (refcount_dec_and_test(&worker->ref))
- wake_up_process(worker->task);
-}
-
-/*
- * Note: drops the wqe->lock if returning true! The caller must re-acquire
- * the lock in that case. Some callers need to restart handling if this
- * happens, so we can't just re-acquire the lock on behalf of the caller.
- */
-static bool __io_worker_unuse(struct io_wqe *wqe, struct io_worker *worker)
-{
- bool dropped_lock = false;
-
- if (worker->saved_creds) {
- revert_creds(worker->saved_creds);
- worker->cur_creds = worker->saved_creds = NULL;
- }
-
- if (current->files) {
- __acquire(&wqe->lock);
- raw_spin_unlock_irq(&wqe->lock);
- dropped_lock = true;
-
- task_lock(current);
- current->files = NULL;
- current->nsproxy = worker->restore_nsproxy;
- task_unlock(current);
- }
-
- if (current->fs)
- current->fs = NULL;
-
- /*
- * If we have an active mm, we need to drop the wq lock before unusing
- * it. If we do, return true and let the caller retry the idle loop.
- */
- if (worker->mm) {
- if (!dropped_lock) {
- __acquire(&wqe->lock);
- raw_spin_unlock_irq(&wqe->lock);
- dropped_lock = true;
- }
- __set_current_state(TASK_RUNNING);
- kthread_unuse_mm(worker->mm);
- mmput(worker->mm);
- worker->mm = NULL;
- }
-
-#ifdef CONFIG_BLK_CGROUP
- if (worker->blkcg_css) {
- kthread_associate_blkcg(NULL);
- worker->blkcg_css = NULL;
- }
-#endif
- if (current->signal->rlim[RLIMIT_FSIZE].rlim_cur != RLIM_INFINITY)
- current->signal->rlim[RLIMIT_FSIZE].rlim_cur = RLIM_INFINITY;
- return dropped_lock;
+ complete(&worker->ref_done);
}
static inline struct io_wqe_acct *io_work_get_acct(struct io_wqe *wqe,
@@ -204,9 +146,10 @@ static inline struct io_wqe_acct *io_work_get_acct(struct io_wqe *wqe,
return &wqe->acct[IO_WQ_ACCT_BOUND];
}
-static inline struct io_wqe_acct *io_wqe_get_acct(struct io_wqe *wqe,
- struct io_worker *worker)
+static inline struct io_wqe_acct *io_wqe_get_acct(struct io_worker *worker)
{
+ struct io_wqe *wqe = worker->wqe;
+
if (worker->flags & IO_WORKER_F_BOUND)
return &wqe->acct[IO_WQ_ACCT_BOUND];
@@ -216,39 +159,36 @@ static inline struct io_wqe_acct *io_wqe_get_acct(struct io_wqe *wqe,
static void io_worker_exit(struct io_worker *worker)
{
struct io_wqe *wqe = worker->wqe;
- struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
+ struct io_wqe_acct *acct = io_wqe_get_acct(worker);
+ unsigned flags;
- /*
- * If we're not at zero, someone else is holding a brief reference
- * to the worker. Wait for that to go away.
- */
- set_current_state(TASK_INTERRUPTIBLE);
- if (!refcount_dec_and_test(&worker->ref))
- schedule();
- __set_current_state(TASK_RUNNING);
+ if (refcount_dec_and_test(&worker->ref))
+ complete(&worker->ref_done);
+ wait_for_completion(&worker->ref_done);
preempt_disable();
current->flags &= ~PF_IO_WORKER;
- if (worker->flags & IO_WORKER_F_RUNNING)
+ flags = worker->flags;
+ worker->flags = 0;
+ if (flags & IO_WORKER_F_RUNNING)
atomic_dec(&acct->nr_running);
- if (!(worker->flags & IO_WORKER_F_BOUND))
- atomic_dec(&wqe->wq->user->processes);
worker->flags = 0;
preempt_enable();
+ if (worker->saved_creds) {
+ revert_creds(worker->saved_creds);
+ worker->cur_creds = worker->saved_creds = NULL;
+ }
+
raw_spin_lock_irq(&wqe->lock);
- hlist_nulls_del_rcu(&worker->nulls_node);
+ if (flags & IO_WORKER_F_FREE)
+ hlist_nulls_del_rcu(&worker->nulls_node);
list_del_rcu(&worker->all_list);
- if (__io_worker_unuse(wqe, worker)) {
- __release(&wqe->lock);
- raw_spin_lock_irq(&wqe->lock);
- }
acct->nr_workers--;
raw_spin_unlock_irq(&wqe->lock);
kfree_rcu(worker, rcu);
- if (refcount_dec_and_test(&wqe->wq->refs))
- complete(&wqe->wq->done);
+ io_wq_put(wqe->wq);
}
static inline bool io_wqe_run_queue(struct io_wqe *wqe)
@@ -306,33 +246,27 @@ static void io_wqe_wake_worker(struct io_wqe *wqe, struct io_wqe_acct *acct)
wake_up_process(wqe->wq->manager);
}
-static void io_wqe_inc_running(struct io_wqe *wqe, struct io_worker *worker)
+static void io_wqe_inc_running(struct io_worker *worker)
{
- struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
+ struct io_wqe_acct *acct = io_wqe_get_acct(worker);
atomic_inc(&acct->nr_running);
}
-static void io_wqe_dec_running(struct io_wqe *wqe, struct io_worker *worker)
+static void io_wqe_dec_running(struct io_worker *worker)
__must_hold(wqe->lock)
{
- struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
+ struct io_wqe_acct *acct = io_wqe_get_acct(worker);
+ struct io_wqe *wqe = worker->wqe;
if (atomic_dec_and_test(&acct->nr_running) && io_wqe_run_queue(wqe))
io_wqe_wake_worker(wqe, acct);
}
-static void io_worker_start(struct io_wqe *wqe, struct io_worker *worker)
+static void io_worker_start(struct io_worker *worker)
{
- allow_kernel_signal(SIGINT);
-
- current->flags |= PF_IO_WORKER;
- current->fs = NULL;
- current->files = NULL;
-
worker->flags |= (IO_WORKER_F_UP | IO_WORKER_F_RUNNING);
- worker->restore_nsproxy = current->nsproxy;
- io_wqe_inc_running(wqe, worker);
+ io_wqe_inc_running(worker);
}
/*
@@ -357,19 +291,17 @@ static void __io_worker_busy(struct io_wqe *wqe, struct io_worker *worker,
worker_bound = (worker->flags & IO_WORKER_F_BOUND) != 0;
work_bound = (work->flags & IO_WQ_WORK_UNBOUND) == 0;
if (worker_bound != work_bound) {
- io_wqe_dec_running(wqe, worker);
+ io_wqe_dec_running(worker);
if (work_bound) {
worker->flags |= IO_WORKER_F_BOUND;
wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers--;
wqe->acct[IO_WQ_ACCT_BOUND].nr_workers++;
- atomic_dec(&wqe->wq->user->processes);
} else {
worker->flags &= ~IO_WORKER_F_BOUND;
wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers++;
wqe->acct[IO_WQ_ACCT_BOUND].nr_workers--;
- atomic_inc(&wqe->wq->user->processes);
}
- io_wqe_inc_running(wqe, worker);
+ io_wqe_inc_running(worker);
}
}
@@ -380,15 +312,17 @@ static void __io_worker_busy(struct io_wqe *wqe, struct io_worker *worker,
* retry the loop in that case (we changed task state), we don't regrab
* the lock if we return success.
*/
-static bool __io_worker_idle(struct io_wqe *wqe, struct io_worker *worker)
+static void __io_worker_idle(struct io_wqe *wqe, struct io_worker *worker)
__must_hold(wqe->lock)
{
if (!(worker->flags & IO_WORKER_F_FREE)) {
worker->flags |= IO_WORKER_F_FREE;
hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
}
-
- return __io_worker_unuse(wqe, worker);
+ if (worker->saved_creds) {
+ revert_creds(worker->saved_creds);
+ worker->cur_creds = worker->saved_creds = NULL;
+ }
}
static inline unsigned int io_get_work_hash(struct io_wq_work *work)
@@ -396,14 +330,31 @@ static inline unsigned int io_get_work_hash(struct io_wq_work *work)
return work->flags >> IO_WQ_HASH_SHIFT;
}
+static void io_wait_on_hash(struct io_wqe *wqe, unsigned int hash)
+{
+ struct io_wq *wq = wqe->wq;
+
+ spin_lock(&wq->hash->wait.lock);
+ if (list_empty(&wqe->wait.entry)) {
+ __add_wait_queue(&wq->hash->wait, &wqe->wait);
+ if (!test_bit(hash, &wq->hash->map)) {
+ __set_current_state(TASK_RUNNING);
+ list_del_init(&wqe->wait.entry);
+ }
+ }
+ spin_unlock(&wq->hash->wait.lock);
+}
+
static struct io_wq_work *io_get_next_work(struct io_wqe *wqe)
__must_hold(wqe->lock)
{
struct io_wq_work_node *node, *prev;
struct io_wq_work *work, *tail;
- unsigned int hash;
+ unsigned int stall_hash = -1U;
wq_list_for_each(node, prev, &wqe->work_list) {
+ unsigned int hash;
+
work = container_of(node, struct io_wq_work, list);
/* not hashed, can run anytime */
@@ -412,111 +363,60 @@ static struct io_wq_work *io_get_next_work(struct io_wqe *wqe)
return work;
}
- /* hashed, can run if not already running */
hash = io_get_work_hash(work);
- if (!(wqe->hash_map & BIT(hash))) {
- wqe->hash_map |= BIT(hash);
- /* all items with this hash lie in [work, tail] */
- tail = wqe->hash_tail[hash];
+ /* all items with this hash lie in [work, tail] */
+ tail = wqe->hash_tail[hash];
+
+ /* hashed, can run if not already running */
+ if (!test_and_set_bit(hash, &wqe->wq->hash->map)) {
wqe->hash_tail[hash] = NULL;
wq_list_cut(&wqe->work_list, &tail->list, prev);
return work;
}
+ if (stall_hash == -1U)
+ stall_hash = hash;
+ /* fast forward to a next hash, for-each will fix up @prev */
+ node = &tail->list;
}
- return NULL;
-}
-
-static void io_wq_switch_mm(struct io_worker *worker, struct io_wq_work *work)
-{
- if (worker->mm) {
- kthread_unuse_mm(worker->mm);
- mmput(worker->mm);
- worker->mm = NULL;
+ if (stall_hash != -1U) {
+ raw_spin_unlock(&wqe->lock);
+ io_wait_on_hash(wqe, stall_hash);
+ raw_spin_lock(&wqe->lock);
}
- if (mmget_not_zero(work->identity->mm)) {
- kthread_use_mm(work->identity->mm);
- worker->mm = work->identity->mm;
- return;
- }
-
- /* failed grabbing mm, ensure work gets cancelled */
- work->flags |= IO_WQ_WORK_CANCEL;
+ return NULL;
}
-static inline void io_wq_switch_blkcg(struct io_worker *worker,
- struct io_wq_work *work)
+static void io_flush_signals(void)
{
-#ifdef CONFIG_BLK_CGROUP
- if (!(work->flags & IO_WQ_WORK_BLKCG))
- return;
- if (work->identity->blkcg_css != worker->blkcg_css) {
- kthread_associate_blkcg(work-