openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity

io_uring-worker.v3-2021-02-25 

-----BEGIN PGP SIGNATURE-----
 
 iQJEBAABCAAuFiEEwPw5LcreJtl1+l5K99NY+ylx4KYFAmA4JRkQHGF4Ym9lQGtl
 cm5lbC5kawAKCRD301j7KXHgpoWqD/9dbbqe8L701U6May1A/4hRsqL4THTA2flx
 vNCNRBl6XV3l/wBCtL6waKy6tyO4lyM8XdUdEvo3Kxl2kGPb8eVfpyYL/+77HqyH
 ctT4RMrs+84Mxn+5N6cM97hS1qVI2moTxxyvOEl/JTB7BYrutz9gvAoeY3/Dto47
 J66oSaPeuqJ32TyihxfQHVxQopJcqFzDjyoYHGDu6ATio1PXfaIdTu8ywVYSECAh
 pWI4rwnqdurGuHMNpxyL1bA6CT/jC7s+sqU7bUYUCgtYI3eG0u3V0bp5gAQQIgl9
 5sxxE3DidYGAkYZsosrelshBtzGddLdz4Qrt2ungMYv8RsGNpFQ095jDPKDwFaZj
 bSvSsfplCo7iFsJByb1TtpNEOW8eAwi81PmBDVQ9Oq5P5ygTYno9GBDc/20ql0Fk
 q6wcX28coE3IBw44ne0hIwvBOtXV4WJyluG/gqOxfbTH+kOy3pDsN8lWcY/P4X0U
 yzdU2MLHe8BNMyYlUiBF47Amzt4ltr85P4XD3WZ4bX71iwri6HvrdGWLuuKwX+Ie
 66QiIDDQIYZQ6NMMJWS9DGW3y3DBizpSXGxONbOw1J2bQdNmtToR0D2UnK/9UnKp
 msnvkUNk8fkYGS4aptpJ6HxbmjMEG5YtbiGlPj6fz5/7MTvhRjPxt7A0LWrUIdqR
 f88+sHUMqg==
 =oc8u
 -----END PGP SIGNATURE-----

Merge tag 'io_uring-worker.v3-2021-02-25' of git://git.kernel.dk/linux-block

Pull io_uring thread rewrite from Jens Axboe:
 "This converts the io-wq workers to be forked off the tasks in question
  instead of being kernel threads that assume various bits of the
  original task identity.

  This kills > 400 lines of code from io_uring/io-wq, and it's the worst
  part of the code. We've had several bugs in this area, and the worry
  is always that we could be missing some pieces for file types doing
  unusual things (recent /dev/tty example comes to mind, userfaultfd
  reads installing file descriptors is another fun one... - both of
  which need special handling, and I bet it's not the last weird oddity
  we'll find).

  With these identical workers, we can have full confidence that we're
  never missing anything. That, in itself, is a huge win. Outside of
  that, it's also more efficient since we're not wasting space and code
  on tracking state, or switching between different states.

  I'm sure we're going to find little things to patch up after this
  series, but testing has been pretty thorough, from the usual
  regression suite to production. Any issue that may crop up should be
  manageable.

  There's also a nice series of further reductions we can do on top of
  this, but I wanted to get the meat of it out sooner rather than later.
  The general worry here isn't that it's fundamentally broken. Most of
  the little issues we've found over the last week have been related to
  just changes in how thread startup/exit is done, since that's the main
  difference between using kthreads and these kinds of threads. In fact,
  if all goes according to plan, I want to get this into the 5.10 and
  5.11 stable branches as well.

  That said, the changes outside of io_uring/io-wq are:

   - arch setup, simple one-liner to each arch copy_thread()
     implementation.

   - Removal of net and proc restrictions for io_uring, they are no
     longer needed or useful"

* tag 'io_uring-worker.v3-2021-02-25' of git://git.kernel.dk/linux-block: (30 commits)
  io-wq: remove now unused IO_WQ_BIT_ERROR
  io_uring: fix SQPOLL thread handling over exec
  io-wq: improve manager/worker handling over exec
  io_uring: ensure SQPOLL startup is triggered before error shutdown
  io-wq: make buffered file write hashed work map per-ctx
  io-wq: fix race around io_worker grabbing
  io-wq: fix races around manager/worker creation and task exit
  io_uring: ensure io-wq context is always destroyed for tasks
  arch: ensure parisc/powerpc handle PF_IO_WORKER in copy_thread()
  io_uring: cleanup ->user usage
  io-wq: remove nr_process accounting
  io_uring: flag new native workers with IORING_FEAT_NATIVE_WORKERS
  net: remove cmsg restriction from io_uring based send/recvmsg calls
  Revert "proc: don't allow async path resolution of /proc/self components"
  Revert "proc: don't allow async path resolution of /proc/thread-self components"
  io_uring: move SQPOLL thread io-wq forked worker
  io-wq: make io_wq_fork_thread() available to other users
  io-wq: only remove worker from free_list, if it was there
  io_uring: remove io_identity
  io_uring: remove any grabbing of context
  ...
This commit is contained in:
Linus Torvalds 2021-02-27 08:29:02 -08:00
parent 5ceabb6078 d6ce7f6761
commit 5695e51619
38 changed files with 698 additions and 1118 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
arch/alpha/kernel/process.c
View File

@ -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));

2
arch/arc/kernel/process.c
View File

@ -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;

2
arch/arm/kernel/process.c
View File

@ -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)

2
arch/arm64/kernel/process.c
View File

@ -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;

2
arch/csky/kernel/process.c
View File

@ -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;

2
arch/h8300/kernel/process.c
View File

@ -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 */

2
arch/hexagon/kernel/process.c
View File

@ -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;

2
arch/ia64/kernel/process.c
View File

@ -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;

2
arch/m68k/kernel/process.c
View File

@ -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;

2
arch/microblaze/kernel/process.c
View File

@ -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));

2
arch/mips/kernel/process.c
View File

@ -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));

2
arch/nds32/kernel/process.c
View File

@ -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;

2
arch/nios2/kernel/process.c
View File

@ -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));

2
arch/openrisc/kernel/process.c
View File

@ -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;

2
arch/parisc/kernel/process.c
View File

@ -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 */

2
arch/powerpc/kernel/process.c
View File

@ -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);

2
arch/riscv/kernel/process.c
View File

@ -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;

2
arch/s390/kernel/process.c
View File

@ -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 |

2
arch/sh/kernel/process_32.c
View File

@ -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;

2
arch/sparc/kernel/process_32.c
View File

@ -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);

2
arch/sparc/kernel/process_64.c
View File

@ -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;

2
arch/um/kernel/process.c
View File

@ -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;

2
arch/x86/kernel/process.c
View File

@ -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;

2
arch/xtensa/kernel/process.c
View File

@ -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];

629
fs/io-wq.c
View File

@ -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();
raw_spin_lock_irq(&wqe->lock);
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);
if (worker->saved_creds) {
revert_creds(worker->saved_creds);
worker->cur_creds = worker->saved_creds = NULL;
}
raw_spin_lock_irq(&wqe->lock);
if (flags & IO_WORKER_F_FREE)
hlist_nulls_del_rcu(&worker->nulls_node);
list_del_rcu(&worker->all_list);
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;
}
if (stall_hash != -1U) {
raw_spin_unlock(&wqe->lock);
io_wait_on_hash(wqe, stall_hash);
raw_spin_lock(&wqe->lock);
}
return NULL;
}
static void io_wq_switch_mm(struct io_worker *worker, struct io_wq_work *work)
static void io_flush_signals(void)
{
if (worker->mm) {
kthread_unuse_mm(worker->mm);
mmput(worker->mm);
worker->mm = NULL;
if (unlikely(test_tsk_thread_flag(current, TIF_NOTIFY_SIGNAL))) {
if (current->task_works)
task_work_run();
clear_tsk_thread_flag(current, TIF_NOTIFY_SIGNAL);
}
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;
}
static inline void io_wq_switch_blkcg(struct io_worker *worker,
struct io_wq_work *work)
{
#ifdef CONFIG_BLK_CGROUP
if (!(work->flags & IO_WQ_WORK_BLKCG))
return;
if (work->identity->blkcg_css != worker->blkcg_css) {
kthread_associate_blkcg(work->identity->blkcg_css);
worker->blkcg_css = work->identity->blkcg_css;
}
#endif
}
static void io_wq_switch_creds(struct io_worker *worker,
struct io_wq_work *work)
{
const struct cred *old_creds = override_creds(work->identity->creds);
const struct cred *old_creds = override_creds(work->creds);
worker->cur_creds = work->identity->creds;
worker->cur_creds = work->creds;
if (worker->saved_creds)
put_cred(old_creds); /* creds set by previous switch */
else
worker->saved_creds = old_creds;
}
static void io_impersonate_work(struct io_worker *worker,
struct io_wq_work *work)
{
if ((work->flags & IO_WQ_WORK_FILES) &&
current->files != work->identity->files) {
task_lock(current);
current->files = work->identity->files;
current->nsproxy = work->identity->nsproxy;
task_unlock(current);
if (!work->identity->files) {
/* failed grabbing files, ensure work gets cancelled */
work->flags |= IO_WQ_WORK_CANCEL;
}
}
if ((work->flags & IO_WQ_WORK_FS) && current->fs != work->identity->fs)
current->fs = work->identity->fs;
if ((work->flags & IO_WQ_WORK_MM) && work->identity->mm != worker->mm)
io_wq_switch_mm(worker, work);
if ((work->flags & IO_WQ_WORK_CREDS) &&
worker->cur_creds != work->identity->creds)
io_wq_switch_creds(worker, work);
if (work->flags & IO_WQ_WORK_FSIZE)
current->signal->rlim[RLIMIT_FSIZE].rlim_cur = work->identity->fsize;
else if (current->signal->rlim[RLIMIT_FSIZE].rlim_cur != RLIM_INFINITY)
current->signal->rlim[RLIMIT_FSIZE].rlim_cur = RLIM_INFINITY;
io_wq_switch_blkcg(worker, work);
#ifdef CONFIG_AUDIT
current->loginuid = work->identity->loginuid;
current->sessionid = work->identity->sessionid;
#endif
}
static void io_assign_current_work(struct io_worker *worker,
struct io_wq_work *work)
{
if (work) {
/* flush pending signals before assigning new work */
if (signal_pending(current))
flush_signals(current);
io_flush_signals();
cond_resched();
}
#ifdef CONFIG_AUDIT
current->loginuid = KUIDT_INIT(AUDIT_UID_UNSET);
current->sessionid = AUDIT_SID_UNSET;
#endif
spin_lock_irq(&worker->lock);
worker->cur_work = work;
spin_unlock_irq(&worker->lock);
@ -550,6 +450,7 @@ static void io_worker_handle_work(struct io_worker *worker)
if (!work)
break;
io_assign_current_work(worker, work);
__set_current_state(TASK_RUNNING);
/* handle a whole dependent link */
do {
@ -557,7 +458,8 @@ static void io_worker_handle_work(struct io_worker *worker)
unsigned int hash = io_get_work_hash(work);
next_hashed = wq_next_work(work);
io_impersonate_work(worker, work);
if (work->creds && worker->cur_creds != work->creds)
io_wq_switch_creds(worker, work);
wq->do_work(work);
io_assign_current_work(worker, NULL);
@ -572,8 +474,10 @@ static void io_worker_handle_work(struct io_worker *worker)
io_wqe_enqueue(wqe, linked);
if (hash != -1U && !next_hashed) {
clear_bit(hash, &wq->hash->map);
if (wq_has_sleeper(&wq->hash->wait))
wake_up(&wq->hash->wait);
raw_spin_lock_irq(&wqe->lock);
wqe->hash_map &= ~BIT_ULL(hash);
wqe->flags &= ~IO_WQE_FLAG_STALLED;
/* skip unnecessary unlock-lock wqe->lock */
if (!work)
@ -592,27 +496,23 @@ static int io_wqe_worker(void *data)
struct io_wqe *wqe = worker->wqe;
struct io_wq *wq = wqe->wq;
io_worker_start(wqe, worker);
io_worker_start(worker);
while (!test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
set_current_state(TASK_INTERRUPTIBLE);
loop:
raw_spin_lock_irq(&wqe->lock);
if (io_wqe_run_queue(wqe)) {
__set_current_state(TASK_RUNNING);
io_worker_handle_work(worker);
goto loop;
}
/* drops the lock on success, retry */
if (__io_worker_idle(wqe, worker)) {
__release(&wqe->lock);
goto loop;
}
__io_worker_idle(wqe, worker);
raw_spin_unlock_irq(&wqe->lock);
if (signal_pending(current))
flush_signals(current);
io_flush_signals();
if (schedule_timeout(WORKER_IDLE_TIMEOUT))
continue;
if (fatal_signal_pending(current))
break;
/* timed out, exit unless we're the fixed worker */
if (test_bit(IO_WQ_BIT_EXIT, &wq->state) ||
!(worker->flags & IO_WORKER_F_FIXED))
@ -636,15 +536,16 @@ static int io_wqe_worker(void *data)
*/
void io_wq_worker_running(struct task_struct *tsk)
{
struct io_worker *worker = kthread_data(tsk);
struct io_wqe *wqe = worker->wqe;
struct io_worker *worker = tsk->pf_io_worker;
if (!worker)
return;
if (!(worker->flags & IO_WORKER_F_UP))
return;
if (worker->flags & IO_WORKER_F_RUNNING)
return;
worker->flags |= IO_WORKER_F_RUNNING;
io_wqe_inc_running(wqe, worker);
io_wqe_inc_running(worker);
}
/*
@ -654,9 +555,10 @@ void io_wq_worker_running(struct task_struct *tsk)
*/
void io_wq_worker_sleeping(struct task_struct *tsk)
{
struct io_worker *worker = kthread_data(tsk);
struct io_wqe *wqe = worker->wqe;
struct io_worker *worker = tsk->pf_io_worker;
if (!worker)
return;
if (!(worker->flags & IO_WORKER_F_UP))
return;
if (!(worker->flags & IO_WORKER_F_RUNNING))
@ -664,32 +566,27 @@ void io_wq_worker_sleeping(struct task_struct *tsk)
worker->flags &= ~IO_WORKER_F_RUNNING;
raw_spin_lock_irq(&wqe->lock);
io_wqe_dec_running(wqe, worker);
raw_spin_unlock_irq(&wqe->lock);
raw_spin_lock_irq(&worker->wqe->lock);
io_wqe_dec_running(worker);
raw_spin_unlock_irq(&worker->wqe->lock);
}
static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index)
static int task_thread(void *data, int index)
{
struct io_worker *worker = data;
struct io_wqe *wqe = worker->wqe;
struct io_wqe_acct *acct = &wqe->acct[index];
struct io_worker *worker;
struct io_wq *wq = wqe->wq;
char buf[TASK_COMM_LEN];
worker = kzalloc_node(sizeof(*worker), GFP_KERNEL, wqe->node);
if (!worker)
return false;
sprintf(buf, "iou-wrk-%d", wq->task_pid);
set_task_comm(current, buf);
refcount_set(&worker->ref, 1);
worker->nulls_node.pprev = NULL;
worker->wqe = wqe;
spin_lock_init(&worker->lock);
current->pf_io_worker = worker;
worker->task = current;
worker->task = kthread_create_on_node(io_wqe_worker, worker, wqe->node,
"io_wqe_worker-%d/%d", index, wqe->node);
if (IS_ERR(worker->task)) {
kfree(worker);
return false;
}
kthread_bind_mask(worker->task, cpumask_of_node(wqe->node));
set_cpus_allowed_ptr(current, cpumask_of_node(wqe->node));
current->flags |= PF_NO_SETAFFINITY;
raw_spin_lock_irq(&wqe->lock);
hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
@ -702,11 +599,63 @@ static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index)
acct->nr_workers++;
raw_spin_unlock_irq(&wqe->lock);
if (index == IO_WQ_ACCT_UNBOUND)
atomic_inc(&wq->user->processes);
io_wqe_worker(data);
do_exit(0);
}
static int task_thread_bound(void *data)
{
return task_thread(data, IO_WQ_ACCT_BOUND);
}
static int task_thread_unbound(void *data)
{
return task_thread(data, IO_WQ_ACCT_UNBOUND);
}
pid_t io_wq_fork_thread(int (*fn)(void *), void *arg)
{
unsigned long flags = CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_THREAD|
CLONE_IO|SIGCHLD;
struct kernel_clone_args args = {
.flags = ((lower_32_bits(flags) | CLONE_VM |
CLONE_UNTRACED) & ~CSIGNAL),
.exit_signal = (lower_32_bits(flags) & CSIGNAL),
.stack = (unsigned long)fn,
.stack_size = (unsigned long)arg,
};
return kernel_clone(&args);
}
static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index)
{
struct io_worker *worker;
pid_t pid;
__set_current_state(TASK_RUNNING);
worker = kzalloc_node(sizeof(*worker), GFP_KERNEL, wqe->node);
if (!worker)
return false;
refcount_set(&worker->ref, 1);
worker->nulls_node.pprev = NULL;
worker->wqe = wqe;
spin_lock_init(&worker->lock);
init_completion(&worker->ref_done);
refcount_inc(&wq->refs);
wake_up_process(worker->task);
if (index == IO_WQ_ACCT_BOUND)
pid = io_wq_fork_thread(task_thread_bound, worker);
else
pid = io_wq_fork_thread(task_thread_unbound, worker);
if (pid < 0) {
io_wq_put(wq);
kfree(worker);
return false;
}
return true;
}
@ -752,93 +701,57 @@ static bool io_wq_worker_wake(struct io_worker *worker, void *data)
return false;
}
static void io_wq_check_workers(struct io_wq *wq)
{
int node;
for_each_node(node) {
struct io_wqe *wqe = wq->wqes[node];
bool fork_worker[2] = { false, false };
if (!node_online(node))
continue;
raw_spin_lock_irq(&wqe->lock);
if (io_wqe_need_worker(wqe, IO_WQ_ACCT_BOUND))
fork_worker[IO_WQ_ACCT_BOUND] = true;
if (io_wqe_need_worker(wqe, IO_WQ_ACCT_UNBOUND))
fork_worker[IO_WQ_ACCT_UNBOUND] = true;
raw_spin_unlock_irq(&wqe->lock);
if (fork_worker[IO_WQ_ACCT_BOUND])
create_io_worker(wq, wqe, IO_WQ_ACCT_BOUND);
if (fork_worker[IO_WQ_ACCT_UNBOUND])
create_io_worker(wq, wqe, IO_WQ_ACCT_UNBOUND);
}
}
/*
* Manager thread. Tasked with creating new workers, if we need them.
*/
static int io_wq_manager(void *data)
{
struct io_wq *wq = data;
int node;
char buf[TASK_COMM_LEN];
/* create fixed workers */
refcount_set(&wq->refs, 1);
for_each_node(node) {
if (!node_online(node))
continue;
if (create_io_worker(wq, wq->wqes[node], IO_WQ_ACCT_BOUND))
continue;
set_bit(IO_WQ_BIT_ERROR, &wq->state);
set_bit(IO_WQ_BIT_EXIT, &wq->state);
goto out;
}
sprintf(buf, "iou-mgr-%d", wq->task_pid);
set_task_comm(current, buf);
current->flags |= PF_IO_WORKER;
wq->manager = current;
complete(&wq->done);
while (!kthread_should_stop()) {
if (current->task_works)
task_work_run();
for_each_node(node) {
struct io_wqe *wqe = wq->wqes[node];
bool fork_worker[2] = { false, false };
if (!node_online(node))
continue;
raw_spin_lock_irq(&wqe->lock);
if (io_wqe_need_worker(wqe, IO_WQ_ACCT_BOUND))
fork_worker[IO_WQ_ACCT_BOUND] = true;
if (io_wqe_need_worker(wqe, IO_WQ_ACCT_UNBOUND))
fork_worker[IO_WQ_ACCT_UNBOUND] = true;
raw_spin_unlock_irq(&wqe->lock);
if (fork_worker[IO_WQ_ACCT_BOUND])
create_io_worker(wq, wqe, IO_WQ_ACCT_BOUND);
if (fork_worker[IO_WQ_ACCT_UNBOUND])
create_io_worker(wq, wqe, IO_WQ_ACCT_UNBOUND);
}
do {
set_current_state(TASK_INTERRUPTIBLE);
io_wq_check_workers(wq);
schedule_timeout(HZ);
}
if (fatal_signal_pending(current))
set_bit(IO_WQ_BIT_EXIT, &wq->state);
} while (!test_bit(IO_WQ_BIT_EXIT, &wq->state));
if (current->task_works)
task_work_run();
out:
if (refcount_dec_and_test(&wq->refs)) {
complete(&wq->done);
return 0;
}
/* if ERROR is set and we get here, we have workers to wake */
if (test_bit(IO_WQ_BIT_ERROR, &wq->state)) {
rcu_read_lock();
for_each_node(node)
io_wq_for_each_worker(wq->wqes[node], io_wq_worker_wake, NULL);
rcu_read_unlock();
}
return 0;
}
static bool io_wq_can_queue(struct io_wqe *wqe, struct io_wqe_acct *acct,
struct io_wq_work *work)
{
bool free_worker;
if (!(work->flags & IO_WQ_WORK_UNBOUND))
return true;
if (atomic_read(&acct->nr_running))
return true;
rcu_read_lock();
free_worker = !hlist_nulls_empty(&wqe->free_list);
rcu_read_unlock();
if (free_worker)
return true;
if (atomic_read(&wqe->wq->user->processes) >= acct->max_workers &&
!(capable(CAP_SYS_RESOURCE) || capable(CAP_SYS_ADMIN)))
return false;
return true;
io_wq_check_workers(wq);
wq->manager = NULL;
io_wq_put(wq);
do_exit(0);
}
static void io_run_cancel(struct io_wq_work *work, struct io_wqe *wqe)
@ -872,20 +785,37 @@ static void io_wqe_insert_work(struct io_wqe *wqe, struct io_wq_work *work)
wq_list_add_after(&work->list, &tail->list, &wqe->work_list);
}
static int io_wq_fork_manager(struct io_wq *wq)
{
int ret;
if (wq->manager)
return 0;
clear_bit(IO_WQ_BIT_EXIT, &wq->state);
refcount_inc(&wq->refs);
current->flags |= PF_IO_WORKER;
ret = io_wq_fork_thread(io_wq_manager, wq);
current->flags &= ~PF_IO_WORKER;
if (ret >= 0) {
wait_for_completion(&wq->done);
return 0;
}
io_wq_put(wq);
return ret;
}
static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work)
{
struct io_wqe_acct *acct = io_work_get_acct(wqe, work);
int work_flags;
unsigned long flags;
/*
* Do early check to see if we need a new unbound worker, and if we do,
* if we're allowed to do so. This isn't 100% accurate as there's a
* gap between this check and incrementing the value, but that's OK.
* It's close enough to not be an issue, fork() has the same delay.
*/
if (unlikely(!io_wq_can_queue(wqe, acct, work))) {
io_run_cancel(work, wqe);
/* Can only happen if manager creation fails after exec */
if (unlikely(io_wq_fork_manager(wqe->wq))) {
work->flags |= IO_WQ_WORK_CANCEL;
wqe->wq->do_work(work);
return;
}
@ -939,7 +869,7 @@ static bool io_wq_worker_cancel(struct io_worker *worker, void *data)
spin_lock_irqsave(&worker->lock, flags);
if (worker->cur_work &&
match->fn(worker->cur_work, match->data)) {
send_sig(SIGINT, worker->task, 1);
set_notify_signal(worker->task);
match->nr_running++;
}
spin_unlock_irqrestore(&worker->lock, flags);
@ -1043,6 +973,24 @@ enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
return IO_WQ_CANCEL_NOTFOUND;
}
static int io_wqe_hash_wake(struct wait_queue_entry *wait, unsigned mode,
int sync, void *key)
{
struct io_wqe *wqe = container_of(wait, struct io_wqe, wait);
int ret;
list_del_init(&wait->entry);
rcu_read_lock();
ret = io_wqe_activate_free_worker(wqe);
rcu_read_unlock();
if (!ret)
wake_up_process(wqe->wq->manager);
return 1;
}
struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
{
int ret = -ENOMEM, node;
@ -1063,12 +1011,11 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
if (ret)
goto err_wqes;
refcount_inc(&data->hash->refs);
wq->hash = data->hash;
wq->free_work = data->free_work;
wq->do_work = data->do_work;
/* caller must already hold a reference to this */
wq->user = data->user;
ret = -ENOMEM;
for_each_node(node) {
struct io_wqe *wqe;
@ -1083,11 +1030,11 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
wqe->node = alloc_node;
wqe->acct[IO_WQ_ACCT_BOUND].max_workers = bounded;
atomic_set(&wqe->acct[IO_WQ_ACCT_BOUND].nr_running, 0);
if (wq->user) {
wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers =
wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers =
task_rlimit(current, RLIMIT_NPROC);
}
atomic_set(&wqe->acct[IO_WQ_ACCT_UNBOUND].nr_running, 0);
wqe->wait.func = io_wqe_hash_wake;
INIT_LIST_HEAD(&wqe->wait.entry);
wqe->wq = wq;
raw_spin_lock_init(&wqe->lock);
INIT_WQ_LIST(&wqe->work_list);
@ -1095,23 +1042,16 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
INIT_LIST_HEAD(&wqe->all_list);
}
wq->task_pid = current->pid;
init_completion(&wq->done);
refcount_set(&wq->refs, 1);
wq->manager = kthread_create(io_wq_manager, wq, "io_wq_manager");
if (!IS_ERR(wq->manager)) {
wake_up_process(wq->manager);
wait_for_completion(&wq->done);
if (test_bit(IO_WQ_BIT_ERROR, &wq->state)) {
ret = -ENOMEM;
goto err;
}
refcount_set(&wq->use_refs, 1);
reinit_completion(&wq->done);
ret = io_wq_fork_manager(wq);
if (!ret)
return wq;
}
ret = PTR_ERR(wq->manager);
complete(&wq->done);
io_wq_put(wq);
io_wq_put_hash(data->hash);
err:
cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node);
for_each_node(node)
@ -1123,15 +1063,7 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
return ERR_PTR(ret);
}
bool io_wq_get(struct io_wq *wq, struct io_wq_data *data)
{
if (data->free_work != wq->free_work || data->do_work != wq->do_work)
return false;
return refcount_inc_not_zero(&wq->use_refs);
}
static void __io_wq_destroy(struct io_wq *wq)
static void io_wq_destroy(struct io_wq *wq)
{
int node;
@ -1139,30 +1071,31 @@ static void __io_wq_destroy(struct io_wq *wq)
set_bit(IO_WQ_BIT_EXIT, &wq->state);
if (wq->manager)
kthread_stop(wq->manager);
wake_up_process(wq->manager);
rcu_read_lock();
for_each_node(node)
io_wq_for_each_worker(wq->wqes[node], io_wq_worker_wake, NULL);
rcu_read_unlock();
wait_for_completion(&wq->done);
spin_lock_irq(&wq->hash->wait.lock);
for_each_node(node) {
struct io_wqe *wqe = wq->wqes[node];
for_each_node(node)
kfree(wq->wqes[node]);
list_del_init(&wqe->wait.entry);
kfree(wqe);
}
spin_unlock_irq(&wq->hash->wait.lock);
io_wq_put_hash(wq->hash);
kfree(wq->wqes);
kfree(wq);
}
void io_wq_destroy(struct io_wq *wq)
void io_wq_put(struct io_wq *wq)
{
if (refcount_dec_and_test(&wq->use_refs))
__io_wq_destroy(wq);
}
struct task_struct *io_wq_get_task(struct io_wq *wq)
{
return wq->manager;
if (refcount_dec_and_test(&wq->refs))
io_wq_destroy(wq);
}
static bool io_wq_worker_affinity(struct io_worker *worker, void *data)

35
fs/io-wq.h
View File

@ -1,6 +1,7 @@
#ifndef INTERNAL_IO_WQ_H
#define INTERNAL_IO_WQ_H
#include <linux/refcount.h>
#include <linux/io_uring.h>
struct io_wq;
@ -11,13 +12,6 @@ enum {
IO_WQ_WORK_UNBOUND = 4,
IO_WQ_WORK_CONCURRENT = 16,
IO_WQ_WORK_FILES = 32,
IO_WQ_WORK_FS = 64,
IO_WQ_WORK_MM = 128,
IO_WQ_WORK_CREDS = 256,
IO_WQ_WORK_BLKCG = 512,
IO_WQ_WORK_FSIZE = 1024,
IO_WQ_HASH_SHIFT = 24, /* upper 8 bits are used for hash key */
};
@ -85,7 +79,7 @@ static inline void wq_list_del(struct io_wq_work_list *list,
struct io_wq_work {
struct io_wq_work_node list;
struct io_identity *identity;
const struct cred *creds;
unsigned flags;
};
@ -100,20 +94,32 @@ static inline struct io_wq_work *wq_next_work(struct io_wq_work *work)
typedef struct io_wq_work *(free_work_fn)(struct io_wq_work *);
typedef void (io_wq_work_fn)(struct io_wq_work *);
struct io_wq_data {
struct user_struct *user;
struct io_wq_hash {
refcount_t refs;
unsigned long map;
struct wait_queue_head wait;
};
static inline void io_wq_put_hash(struct io_wq_hash *hash)
{
if (refcount_dec_and_test(&hash->refs))
kfree(hash);
}
struct io_wq_data {
struct io_wq_hash *hash;
io_wq_work_fn *do_work;
free_work_fn *free_work;
};
struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data);
bool io_wq_get(struct io_wq *wq, struct io_wq_data *data);
void io_wq_destroy(struct io_wq *wq);
void io_wq_put(struct io_wq *wq);
void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work);
void io_wq_hash_work(struct io_wq_work *work, void *val);
pid_t io_wq_fork_thread(int (*fn)(void *), void *arg);
static inline bool io_wq_is_hashed(struct io_wq_work *work)
{
return work->flags & IO_WQ_WORK_HASHED;
@ -124,8 +130,6 @@ typedef bool (work_cancel_fn)(struct io_wq_work *, void *);
enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
void *data, bool cancel_all);
struct task_struct *io_wq_get_task(struct io_wq *wq);
#if defined(CONFIG_IO_WQ)
extern void io_wq_worker_sleeping(struct task_struct *);
extern void io_wq_worker_running(struct task_struct *);
@ -140,6 +144,7 @@ static inline void io_wq_worker_running(struct task_struct *tsk)
static inline bool io_wq_current_is_worker(void)
{
return in_task() && (current->flags & PF_IO_WORKER);
return in_task() && (current->flags & PF_IO_WORKER) &&
current->pf_io_worker;
}
#endif

1043
fs/io_uring.c
View File

File diff suppressed because it is too large Load Diff

7
fs/proc/self.c
View File

@ -16,13 +16,6 @@ static const char *proc_self_get_link(struct dentry *dentry,
pid_t tgid = task_tgid_nr_ns(current, ns);
char *name;
/*
* Not currently supported. Once we can inherit all of struct pid,
* we can allow this.
*/
if (current->flags & PF_IO_WORKER)
return ERR_PTR(-EOPNOTSUPP);
if (!tgid)
return ERR_PTR(-ENOENT);
/* max length of unsigned int in decimal + NULL term */

7
fs/proc/thread_self.c
View File

@ -17,13 +17,6 @@ static const char *proc_thread_self_get_link(struct dentry *dentry,
pid_t pid = task_pid_nr_ns(current, ns);
char *name;
/*
* Not currently supported. Once we can inherit all of struct pid,
* we can allow this.
*/
if (current->flags & PF_IO_WORKER)
return ERR_PTR(-EOPNOTSUPP);
if (!pid)
return ERR_PTR(-ENOENT);
name = kmalloc(10 + 6 + 10 + 1, dentry ? GFP_KERNEL : GFP_ATOMIC);

22
include/linux/io_uring.h
View File

@ -5,23 +5,6 @@
#include <linux/sched.h>
#include <linux/xarray.h>
struct io_identity {
struct files_struct *files;
struct mm_struct *mm;
#ifdef CONFIG_BLK_CGROUP
struct cgroup_subsys_state *blkcg_css;
#endif
const struct cred *creds;
struct nsproxy *nsproxy;
struct fs_struct *fs;
unsigned long fsize;
#ifdef CONFIG_AUDIT
kuid_t loginuid;
unsigned int sessionid;
#endif
refcount_t count;
};
struct io_wq_work_node {
struct io_wq_work_node *next;
};
@ -36,9 +19,8 @@ struct io_uring_task {
struct xarray xa;
struct wait_queue_head wait;
struct file *last;
void *io_wq;
struct percpu_counter inflight;
struct io_identity __identity;
struct io_identity *identity;
atomic_t in_idle;
bool sqpoll;
@ -61,7 +43,7 @@ static inline void io_uring_task_cancel(void)
}
static inline void io_uring_files_cancel(struct files_struct *files)
{
if (current->io_uring && !xa_empty(&current->io_uring->xa))
if (current->io_uring)
__io_uring_files_cancel(files);
}
static inline void io_uring_free(struct task_struct *tsk)

3
include/linux/net.h
View File

@ -42,8 +42,6 @@ struct net;
#define SOCK_PASSCRED 3
#define SOCK_PASSSEC 4
#define PROTO_CMSG_DATA_ONLY 0x0001
#ifndef ARCH_HAS_SOCKET_TYPES
/**
* enum sock_type - Socket types
@ -138,7 +136,6 @@ typedef int (*sk_read_actor_t)(read_descriptor_t *, struct sk_buff *,
struct proto_ops {
int family;
unsigned int flags;
struct module *owner;
int (*release) (struct socket *sock);
int (*bind) (struct socket *sock,

3
include/linux/sched.h
View File

@ -895,6 +895,9 @@ struct task_struct {
/* CLONE_CHILD_CLEARTID: */
int __user *clear_child_tid;
/* PF_IO_WORKER */
void *pf_io_worker;
u64 utime;
u64 stime;
#ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME

1
include/uapi/linux/io_uring.h
View File

@ -262,6 +262,7 @@ struct io_uring_params {
#define IORING_FEAT_POLL_32BITS (1U << 6)
#define IORING_FEAT_SQPOLL_NONFIXED (1U << 7)
#define IORING_FEAT_EXT_ARG (1U << 8)
#define IORING_FEAT_NATIVE_WORKERS (1U << 9)
/*
* io_uring_register(2) opcodes and arguments

2
kernel/ptrace.c
View File

@ -375,7 +375,7 @@ static int ptrace_attach(struct task_struct *task, long request,
audit_ptrace(task);
retval = -EPERM;
if (unlikely(task->flags & PF_KTHREAD))
if (unlikely(task->flags & (PF_KTHREAD | PF_IO_WORKER)))
goto out;
if (same_thread_group(task, current))
goto out;

4
kernel/signal.c
View File

@ -91,7 +91,7 @@ static bool sig_task_ignored(struct task_struct *t, int sig, bool force)
return true;
/* Only allow kernel generated signals to this kthread */
if (unlikely((t->flags & PF_KTHREAD) &&
if (unlikely((t->flags & (PF_KTHREAD | PF_IO_WORKER)) &&
(handler == SIG_KTHREAD_KERNEL) && !force))
return true;
@ -1096,7 +1096,7 @@ static int __send_signal(int sig, struct kernel_siginfo *info, struct task_struc
/*
* Skip useless siginfo allocation for SIGKILL and kernel threads.
*/
if ((sig == SIGKILL) || (t->flags & PF_KTHREAD))
if ((sig == SIGKILL) || (t->flags & (PF_KTHREAD | PF_IO_WORKER)))
goto out_set;
/*

1
net/ipv4/af_inet.c
View File

@ -1021,7 +1021,6 @@ static int inet_compat_ioctl(struct socket *sock, unsigned int cmd, unsigned lon
const struct proto_ops inet_stream_ops = {
.family = PF_INET,
.flags = PROTO_CMSG_DATA_ONLY,
.owner = THIS_MODULE,
.release = inet_release,
.bind = inet_bind,

1
net/ipv6/af_inet6.c
View File

@ -665,7 +665,6 @@ int inet6_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
const struct proto_ops inet6_stream_ops = {
.family = PF_INET6,
.flags = PROTO_CMSG_DATA_ONLY,
.owner = THIS_MODULE,
.release = inet6_release,
.bind = inet6_bind,

10
net/socket.c
View File

@ -2413,10 +2413,6 @@ static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg,
long __sys_sendmsg_sock(struct socket *sock, struct msghdr *msg,
unsigned int flags)
{
/* disallow ancillary data requests from this path */
if (msg->msg_control || msg->msg_controllen)
return -EINVAL;
return ____sys_sendmsg(sock, msg, flags, NULL, 0);
}
@ -2625,12 +2621,6 @@ long __sys_recvmsg_sock(struct socket *sock, struct msghdr *msg,
struct user_msghdr __user *umsg,
struct sockaddr __user *uaddr, unsigned int flags)
{
if (msg->msg_control || msg->msg_controllen) {
/* disallow ancillary data reqs unless cmsg is plain data */
if (!(sock->ops->flags & PROTO_CMSG_DATA_ONLY))
return -EINVAL;
}
return ____sys_recvmsg(sock, msg, umsg, uaddr, flags, 0);
}