qemu/job.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1280 lines
32 KiB
C
Raw Normal View History

/*
* Background jobs (long-running operations)
*
* Copyright (c) 2011 IBM Corp.
* Copyright (c) 2012, 2018 Red Hat, Inc.
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include "qemu/osdep.h"
#include "qapi/error.h"
#include "qemu/job.h"
#include "qemu/id.h"
#include "qemu/main-loop.h"
#include "block/aio-wait.h"
#include "trace/trace-root.h"
#include "qapi/qapi-events-job.h"
/*
* The job API is composed of two categories of functions.
*
* The first includes functions used by the monitor. The monitor is
* peculiar in that it accesses the job list with job_get, and
* therefore needs consistency across job_get and the actual operation
* (e.g. job_user_cancel). To achieve this consistency, the caller
* calls job_lock/job_unlock itself around the whole operation.
*
*
* The second includes functions used by the job drivers and sometimes
* by the core block layer. These delegate the locking to the callee instead.
*/
/*
* job_mutex protects the jobs list, but also makes the
* struct job fields thread-safe.
*/
QemuMutex job_mutex;
/* Protected by job_mutex */
static QLIST_HEAD(, Job) jobs = QLIST_HEAD_INITIALIZER(jobs);
/* Job State Transition Table */
bool JobSTT[JOB_STATUS__MAX][JOB_STATUS__MAX] = {
/* U, C, R, P, Y, S, W, D, X, E, N */
/* U: */ [JOB_STATUS_UNDEFINED] = {0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0},
/* C: */ [JOB_STATUS_CREATED] = {0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 1},
/* R: */ [JOB_STATUS_RUNNING] = {0, 0, 0, 1, 1, 0, 1, 0, 1, 0, 0},
/* P: */ [JOB_STATUS_PAUSED] = {0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0},
/* Y: */ [JOB_STATUS_READY] = {0, 0, 0, 0, 0, 1, 1, 0, 1, 0, 0},
/* S: */ [JOB_STATUS_STANDBY] = {0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0},
/* W: */ [JOB_STATUS_WAITING] = {0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0},
/* D: */ [JOB_STATUS_PENDING] = {0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0},
/* X: */ [JOB_STATUS_ABORTING] = {0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0},
/* E: */ [JOB_STATUS_CONCLUDED] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
/* N: */ [JOB_STATUS_NULL] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
};
bool JobVerbTable[JOB_VERB__MAX][JOB_STATUS__MAX] = {
/* U, C, R, P, Y, S, W, D, X, E, N */
[JOB_VERB_CANCEL] = {0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0},
[JOB_VERB_PAUSE] = {0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0},
[JOB_VERB_RESUME] = {0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0},
[JOB_VERB_SET_SPEED] = {0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0},
[JOB_VERB_COMPLETE] = {0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0},
[JOB_VERB_FINALIZE] = {0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0},
[JOB_VERB_DISMISS] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0},
};
/* Transactional group of jobs */
struct JobTxn {
/* Is this txn being cancelled? */
bool aborting;
/* List of jobs */
QLIST_HEAD(, Job) jobs;
/* Reference count */
int refcnt;
};
void job_lock(void)
{
qemu_mutex_lock(&job_mutex);
}
void job_unlock(void)
{
qemu_mutex_unlock(&job_mutex);
}
static void __attribute__((__constructor__)) job_init(void)
{
qemu_mutex_init(&job_mutex);
}
JobTxn *job_txn_new(void)
{
JobTxn *txn = g_new0(JobTxn, 1);
QLIST_INIT(&txn->jobs);
txn->refcnt = 1;
return txn;
}
/* Called with job_mutex held. */
static void job_txn_ref_locked(JobTxn *txn)
{
txn->refcnt++;
}
void job_txn_unref_locked(JobTxn *txn)
{
if (txn && --txn->refcnt == 0) {
g_free(txn);
}
}
void job_txn_unref(JobTxn *txn)
{
JOB_LOCK_GUARD();
job_txn_unref_locked(txn);
}
/**
* @txn: The transaction (may be NULL)
* @job: Job to add to the transaction
*
* Add @job to the transaction. The @job must not already be in a transaction.
* The caller must call either job_txn_unref() or job_completed() to release
* the reference that is automatically grabbed here.
*
* If @txn is NULL, the function does nothing.
*
* Called with job_mutex held.
*/
static void job_txn_add_job_locked(JobTxn *txn, Job *job)
{
if (!txn) {
return;
}
assert(!job->txn);
job->txn = txn;
QLIST_INSERT_HEAD(&txn->jobs, job, txn_list);
job_txn_ref_locked(txn);
}
/* Called with job_mutex held. */
static void job_txn_del_job_locked(Job *job)
{
if (job->txn) {
QLIST_REMOVE(job, txn_list);
job_txn_unref_locked(job->txn);
job->txn = NULL;
}
}
/* Called with job_mutex held, but releases it temporarily. */
static int job_txn_apply_locked(Job *job, int fn(Job *))
{
Job *other_job, *next;
JobTxn *txn = job->txn;
int rc = 0;
/*
* Similar to job_completed_txn_abort, we take each job's lock before
* applying fn, but since we assume that outer_ctx is held by the caller,
* we need to release it here to avoid holding the lock twice - which would
* break AIO_WAIT_WHILE from within fn.
*/
job_ref_locked(job);
QLIST_FOREACH_SAFE(other_job, &txn->jobs, txn_list, next) {
rc = fn(other_job);
if (rc) {
break;
}
}
job_unref_locked(job);
return rc;
}
bool job_is_internal(Job *job)
{
return (job->id == NULL);
}
/* Called with job_mutex held. */
static void job_state_transition_locked(Job *job, JobStatus s1)
{
JobStatus s0 = job->status;
assert(s1 >= 0 && s1 < JOB_STATUS__MAX);
trace_job_state_transition(job, job->ret,
JobSTT[s0][s1] ? "allowed" : "disallowed",
JobStatus_str(s0), JobStatus_str(s1));
assert(JobSTT[s0][s1]);
job->status = s1;
if (!job_is_internal(job) && s1 != s0) {
qapi_event_send_job_status_change(job->id, job->status);
}
}
int job_apply_verb_locked(Job *job, JobVerb verb, Error **errp)
{
JobStatus s0 = job->status;
assert(verb >= 0 && verb < JOB_VERB__MAX);
trace_job_apply_verb(job, JobStatus_str(s0), JobVerb_str(verb),
JobVerbTable[verb][s0] ? "allowed" : "prohibited");
if (JobVerbTable[verb][s0]) {
return 0;
}
error_setg(errp, "Job '%s' in state '%s' cannot accept command verb '%s'",
job->id, JobStatus_str(s0), JobVerb_str(verb));
return -EPERM;
}
JobType job_type(const Job *job)
{
return job->driver->job_type;
}
const char *job_type_str(const Job *job)
{
return JobType_str(job_type(job));
}
bool job_is_cancelled_locked(Job *job)
job: Add job_cancel_requested() Most callers of job_is_cancelled() actually want to know whether the job is on its way to immediate termination. For example, we refuse to pause jobs that are cancelled; but this only makes sense for jobs that are really actually cancelled. A mirror job that is cancelled during READY with force=false should absolutely be allowed to pause. This "cancellation" (which is actually a kind of completion) may take an indefinite amount of time, and so should behave like any job during normal operation. For example, with on-target-error=stop, the job should stop on write errors. (In contrast, force-cancelled jobs should not get write errors, as they should just terminate and not do further I/O.) Therefore, redefine job_is_cancelled() to only return true for jobs that are force-cancelled (which as of HEAD^ means any job that interprets the cancellation request as a request for immediate termination), and add job_cancel_requested() as the general variant, which returns true for any jobs which have been requested to be cancelled, whether it be immediately or after an arbitrarily long completion phase. Finally, here is a justification for how different job_is_cancelled() invocations are treated by this patch: - block/mirror.c (mirror_run()): - The first invocation is a while loop that should loop until the job has been cancelled or scheduled for completion. What kind of cancel does not matter, only the fact that the job is supposed to end. - The second invocation wants to know whether the job has been soft-cancelled. Calling job_cancel_requested() is a bit too broad, but if the job were force-cancelled, we should leave the main loop as soon as possible anyway, so this should not matter here. - The last two invocations already check force_cancel, so they should continue to use job_is_cancelled(). - block/backup.c, block/commit.c, block/stream.c, anything in tests/: These jobs know only force-cancel, so there is no difference between job_is_cancelled() and job_cancel_requested(). We can continue using job_is_cancelled(). - job.c: - job_pause_point(), job_yield(), job_sleep_ns(): Only force-cancelled jobs should be prevented from being paused. Continue using job_is_cancelled(). - job_update_rc(), job_finalize_single(), job_finish_sync(): These functions are all called after the job has left its main loop. The mirror job (the only job that can be soft-cancelled) will clear .cancelled before leaving the main loop if it has been soft-cancelled. Therefore, these functions will observe .cancelled to be true only if the job has been force-cancelled. We can continue to use job_is_cancelled(). (Furthermore, conceptually, a soft-cancelled mirror job should not report to have been cancelled. It should report completion (see also the block-job-cancel QAPI documentation). Therefore, it makes sense for these functions not to distinguish between a soft-cancelled mirror job and a job that has completed as normal.) - job_completed_txn_abort(): All jobs other than @job have been force-cancelled. job_is_cancelled() must be true for them. Regarding @job itself: job_completed_txn_abort() is mostly called when the job's return value is not 0. A soft-cancelled mirror has a return value of 0, and so will not end up here then. However, job_cancel() invokes job_completed_txn_abort() if the job has been deferred to the main loop, which is mostly the case for completed jobs (which skip the assertion), but not for sure. To be safe, use job_cancel_requested() in this assertion. - job_complete(): This is function eventually invoked by the user (through qmp_block_job_complete() or qmp_job_complete(), or job_complete_sync(), which comes from qemu-img). The intention here is to prevent a user from invoking job-complete after the job has been cancelled. This should also apply to soft cancelling: After a mirror job has been soft-cancelled, the user should not be able to decide otherwise and have it complete as normal (i.e. pivoting to the target). - job_cancel(): Both functions are equivalent (see comment there), but we want to use job_is_cancelled(), because this shows that we call job_completed_txn_abort() only for force-cancelled jobs. (As explained for job_update_rc(), soft-cancelled jobs should be treated as if they have completed as normal.) Buglink: https://gitlab.com/qemu-project/qemu/-/issues/462 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20211006151940.214590-9-hreitz@redhat.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2021-10-06 23:19:35 +08:00
{
/* force_cancel may be true only if cancelled is true, too */
assert(job->cancelled || !job->force_cancel);
return job->force_cancel;
job: Add job_cancel_requested() Most callers of job_is_cancelled() actually want to know whether the job is on its way to immediate termination. For example, we refuse to pause jobs that are cancelled; but this only makes sense for jobs that are really actually cancelled. A mirror job that is cancelled during READY with force=false should absolutely be allowed to pause. This "cancellation" (which is actually a kind of completion) may take an indefinite amount of time, and so should behave like any job during normal operation. For example, with on-target-error=stop, the job should stop on write errors. (In contrast, force-cancelled jobs should not get write errors, as they should just terminate and not do further I/O.) Therefore, redefine job_is_cancelled() to only return true for jobs that are force-cancelled (which as of HEAD^ means any job that interprets the cancellation request as a request for immediate termination), and add job_cancel_requested() as the general variant, which returns true for any jobs which have been requested to be cancelled, whether it be immediately or after an arbitrarily long completion phase. Finally, here is a justification for how different job_is_cancelled() invocations are treated by this patch: - block/mirror.c (mirror_run()): - The first invocation is a while loop that should loop until the job has been cancelled or scheduled for completion. What kind of cancel does not matter, only the fact that the job is supposed to end. - The second invocation wants to know whether the job has been soft-cancelled. Calling job_cancel_requested() is a bit too broad, but if the job were force-cancelled, we should leave the main loop as soon as possible anyway, so this should not matter here. - The last two invocations already check force_cancel, so they should continue to use job_is_cancelled(). - block/backup.c, block/commit.c, block/stream.c, anything in tests/: These jobs know only force-cancel, so there is no difference between job_is_cancelled() and job_cancel_requested(). We can continue using job_is_cancelled(). - job.c: - job_pause_point(), job_yield(), job_sleep_ns(): Only force-cancelled jobs should be prevented from being paused. Continue using job_is_cancelled(). - job_update_rc(), job_finalize_single(), job_finish_sync(): These functions are all called after the job has left its main loop. The mirror job (the only job that can be soft-cancelled) will clear .cancelled before leaving the main loop if it has been soft-cancelled. Therefore, these functions will observe .cancelled to be true only if the job has been force-cancelled. We can continue to use job_is_cancelled(). (Furthermore, conceptually, a soft-cancelled mirror job should not report to have been cancelled. It should report completion (see also the block-job-cancel QAPI documentation). Therefore, it makes sense for these functions not to distinguish between a soft-cancelled mirror job and a job that has completed as normal.) - job_completed_txn_abort(): All jobs other than @job have been force-cancelled. job_is_cancelled() must be true for them. Regarding @job itself: job_completed_txn_abort() is mostly called when the job's return value is not 0. A soft-cancelled mirror has a return value of 0, and so will not end up here then. However, job_cancel() invokes job_completed_txn_abort() if the job has been deferred to the main loop, which is mostly the case for completed jobs (which skip the assertion), but not for sure. To be safe, use job_cancel_requested() in this assertion. - job_complete(): This is function eventually invoked by the user (through qmp_block_job_complete() or qmp_job_complete(), or job_complete_sync(), which comes from qemu-img). The intention here is to prevent a user from invoking job-complete after the job has been cancelled. This should also apply to soft cancelling: After a mirror job has been soft-cancelled, the user should not be able to decide otherwise and have it complete as normal (i.e. pivoting to the target). - job_cancel(): Both functions are equivalent (see comment there), but we want to use job_is_cancelled(), because this shows that we call job_completed_txn_abort() only for force-cancelled jobs. (As explained for job_update_rc(), soft-cancelled jobs should be treated as if they have completed as normal.) Buglink: https://gitlab.com/qemu-project/qemu/-/issues/462 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20211006151940.214590-9-hreitz@redhat.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2021-10-06 23:19:35 +08:00
}
bool job_is_cancelled(Job *job)
{
JOB_LOCK_GUARD();
return job_is_cancelled_locked(job);
}
/* Called with job_mutex held. */
static bool job_cancel_requested_locked(Job *job)
{
return job->cancelled;
}
bool job_cancel_requested(Job *job)
{
JOB_LOCK_GUARD();
return job_cancel_requested_locked(job);
}
bool job_is_ready_locked(Job *job)
{
switch (job->status) {
case JOB_STATUS_UNDEFINED:
case JOB_STATUS_CREATED:
case JOB_STATUS_RUNNING:
case JOB_STATUS_PAUSED:
case JOB_STATUS_WAITING:
case JOB_STATUS_PENDING:
case JOB_STATUS_ABORTING:
case JOB_STATUS_CONCLUDED:
case JOB_STATUS_NULL:
return false;
case JOB_STATUS_READY:
case JOB_STATUS_STANDBY:
return true;
default:
g_assert_not_reached();
}
return false;
}
bool job_is_ready(Job *job)
{
JOB_LOCK_GUARD();
return job_is_ready_locked(job);
}
bool job_is_completed_locked(Job *job)
{
switch (job->status) {
case JOB_STATUS_UNDEFINED:
case JOB_STATUS_CREATED:
case JOB_STATUS_RUNNING:
case JOB_STATUS_PAUSED:
case JOB_STATUS_READY:
case JOB_STATUS_STANDBY:
return false;
case JOB_STATUS_WAITING:
case JOB_STATUS_PENDING:
case JOB_STATUS_ABORTING:
case JOB_STATUS_CONCLUDED:
case JOB_STATUS_NULL:
return true;
default:
g_assert_not_reached();
}
return false;
}
static bool job_is_completed(Job *job)
{
JOB_LOCK_GUARD();
return job_is_completed_locked(job);
}
static bool job_started_locked(Job *job)
{
return job->co;
}
/* Called with job_mutex held. */
static bool job_should_pause_locked(Job *job)
{
return job->pause_count > 0;
}
Job *job_next_locked(Job *job)
{
if (!job) {
return QLIST_FIRST(&jobs);
}
return QLIST_NEXT(job, job_list);
}
Job *job_next(Job *job)
{
JOB_LOCK_GUARD();
return job_next_locked(job);
}
Job *job_get_locked(const char *id)
{
Job *job;
QLIST_FOREACH(job, &jobs, job_list) {
if (job->id && !strcmp(id, job->id)) {
return job;
}
}
return NULL;
}
void job_set_aio_context(Job *job, AioContext *ctx)
{
/* protect against read in job_finish_sync_locked and job_start */
GLOBAL_STATE_CODE();
/* protect against read in job_do_yield_locked */
JOB_LOCK_GUARD();
/* ensure the job is quiescent while the AioContext is changed */
assert(job->paused || job_is_completed_locked(job));
job->aio_context = ctx;
}
/* Called with job_mutex *not* held. */
static void job_sleep_timer_cb(void *opaque)
{
Job *job = opaque;
job_enter(job);
}
void *job_create(const char *job_id, const JobDriver *driver, JobTxn *txn,
AioContext *ctx, int flags, BlockCompletionFunc *cb,
void *opaque, Error **errp)
{
Job *job;
JOB_LOCK_GUARD();
if (job_id) {
if (flags & JOB_INTERNAL) {
error_setg(errp, "Cannot specify job ID for internal job");
return NULL;
}
if (!id_wellformed(job_id)) {
error_setg(errp, "Invalid job ID '%s'", job_id);
return NULL;
}
if (job_get_locked(job_id)) {
error_setg(errp, "Job ID '%s' already in use", job_id);
return NULL;
}
} else if (!(flags & JOB_INTERNAL)) {
error_setg(errp, "An explicit job ID is required");
return NULL;
}
job = g_malloc0(driver->instance_size);
job->driver = driver;
job->id = g_strdup(job_id);
job->refcnt = 1;
job->aio_context = ctx;
job->busy = false;
job->paused = true;
job->pause_count = 1;
job->auto_finalize = !(flags & JOB_MANUAL_FINALIZE);
job->auto_dismiss = !(flags & JOB_MANUAL_DISMISS);
job->cb = cb;
job->opaque = opaque;
progress_init(&job->progress);
notifier_list_init(&job->on_finalize_cancelled);
notifier_list_init(&job->on_finalize_completed);
notifier_list_init(&job->on_pending);
notifier_list_init(&job->on_ready);
notifier_list_init(&job->on_idle);
job_state_transition_locked(job, JOB_STATUS_CREATED);
aio_timer_init(qemu_get_aio_context(), &job->sleep_timer,
QEMU_CLOCK_REALTIME, SCALE_NS,
job_sleep_timer_cb, job);
QLIST_INSERT_HEAD(&jobs, job, job_list);
/* Single jobs are modeled as single-job transactions for sake of
* consolidating the job management logic */
if (!txn) {
txn = job_txn_new();
job_txn_add_job_locked(txn, job);
job_txn_unref_locked(txn);
} else {
job_txn_add_job_locked(txn, job);
}
return job;
}
void job_ref_locked(Job *job)
{
++job->refcnt;
}
void job_unref_locked(Job *job)
{
GLOBAL_STATE_CODE();
if (--job->refcnt == 0) {
assert(job->status == JOB_STATUS_NULL);
assert(!timer_pending(&job->sleep_timer));
assert(!job->txn);
if (job->driver->free) {
AioContext *aio_context = job->aio_context;
job_unlock();
/* FIXME: aiocontext lock is required because cb calls blk_unref */
aio_context_acquire(aio_context);
job->driver->free(job);
aio_context_release(aio_context);
job_lock();
}
QLIST_REMOVE(job, job_list);
progress_destroy(&job->progress);
error_free(job->err);
g_free(job->id);
g_free(job);
}
}
void job_progress_update(Job *job, uint64_t done)
{
progress_work_done(&job->progress, done);
}
void job_progress_set_remaining(Job *job, uint64_t remaining)
{
progress_set_remaining(&job->progress, remaining);
}
void job_progress_increase_remaining(Job *job, uint64_t delta)
{
progress_increase_remaining(&job->progress, delta);
}
/**
* To be called when a cancelled job is finalised.
* Called with job_mutex held.
*/
static void job_event_cancelled_locked(Job *job)
{
notifier_list_notify(&job->on_finalize_cancelled, job);
}
/**
* To be called when a successfully completed job is finalised.
* Called with job_mutex held.
*/
static void job_event_completed_locked(Job *job)
{
notifier_list_notify(&job->on_finalize_completed, job);
}
/* Called with job_mutex held. */
static void job_event_pending_locked(Job *job)
{
notifier_list_notify(&job->on_pending, job);
}
/* Called with job_mutex held. */
static void job_event_ready_locked(Job *job)
{
notifier_list_notify(&job->on_ready, job);
}
/* Called with job_mutex held. */
static void job_event_idle_locked(Job *job)
{
notifier_list_notify(&job->on_idle, job);
}
void job_enter_cond_locked(Job *job, bool(*fn)(Job *job))
{
if (!job_started_locked(job)) {
return;
}
if (job->deferred_to_main_loop) {
return;
}
if (job->busy) {
return;
}
if (fn && !fn(job)) {
return;
}
assert(!job->deferred_to_main_loop);
timer_del(&job->sleep_timer);
job->busy = true;
job_unlock();
job: detect change of aiocontext within job coroutine We want to make sure access of job->aio_context is always done under either BQL or job_mutex. The problem is that using aio_co_enter(job->aiocontext, job->co) in job_start and job_enter_cond makes the coroutine immediately resume, so we can't hold the job lock. And caching it is not safe either, as it might change. job_start is under BQL, so it can freely read job->aiocontext, but job_enter_cond is not. We want to avoid reading job->aio_context in job_enter_cond, therefore: 1) use aio_co_wake(), since it doesn't want an aiocontext as argument but uses job->co->ctx 2) detect possible discrepancy between job->co->ctx and job->aio_context by checking right after the coroutine resumes back from yielding if job->aio_context has changed. If so, reschedule the coroutine to the new context. Calling bdrv_try_set_aio_context() will issue the following calls (simplified): * in terms of bdrv callbacks: .drained_begin -> .set_aio_context -> .drained_end * in terms of child_job functions: child_job_drained_begin -> child_job_set_aio_context -> child_job_drained_end * in terms of job functions: job_pause_locked -> job_set_aio_context -> job_resume_locked We can see that after setting the new aio_context, job_resume_locked calls again job_enter_cond, which then invokes aio_co_wake(). But while job->aiocontext has been set in job_set_aio_context, job->co->ctx has not changed, so the coroutine would be entering in the wrong aiocontext. Using aio_co_schedule in job_resume_locked() might seem as a valid alternative, but the problem is that the bh resuming the coroutine is not scheduled immediately, and if in the meanwhile another bdrv_try_set_aio_context() is run (see test_propagate_mirror() in test-block-iothread.c), we would have the first schedule in the wrong aiocontext, and the second set of drains won't even manage to schedule the coroutine, as job->busy would still be true from the previous job_resume_locked(). The solution is to stick with aio_co_wake() and detect every time the coroutine resumes back from yielding if job->aio_context has changed. If so, we can reschedule it to the new context. Check for the aiocontext change in job_do_yield_locked because: 1) aio_co_reschedule_self requires to be in the running coroutine 2) since child_job_set_aio_context allows changing the aiocontext only while the job is paused, this is the exact place where the coroutine resumes, before running JobDriver's code. Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Message-Id: <20220926093214.506243-13-eesposit@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2022-09-26 17:32:05 +08:00
aio_co_wake(job->co);
job_lock();
}
void job_enter(Job *job)
{
JOB_LOCK_GUARD();
job_enter_cond_locked(job, NULL);
}
/* Yield, and schedule a timer to reenter the coroutine after @ns nanoseconds.
* Reentering the job coroutine with job_enter() before the timer has expired
* is allowed and cancels the timer.
*
* If @ns is (uint64_t) -1, no timer is scheduled and job_enter() must be
* called explicitly.
*
* Called with job_mutex held, but releases it temporarily.
*/
static void coroutine_fn job_do_yield_locked(Job *job, uint64_t ns)
{
job: detect change of aiocontext within job coroutine We want to make sure access of job->aio_context is always done under either BQL or job_mutex. The problem is that using aio_co_enter(job->aiocontext, job->co) in job_start and job_enter_cond makes the coroutine immediately resume, so we can't hold the job lock. And caching it is not safe either, as it might change. job_start is under BQL, so it can freely read job->aiocontext, but job_enter_cond is not. We want to avoid reading job->aio_context in job_enter_cond, therefore: 1) use aio_co_wake(), since it doesn't want an aiocontext as argument but uses job->co->ctx 2) detect possible discrepancy between job->co->ctx and job->aio_context by checking right after the coroutine resumes back from yielding if job->aio_context has changed. If so, reschedule the coroutine to the new context. Calling bdrv_try_set_aio_context() will issue the following calls (simplified): * in terms of bdrv callbacks: .drained_begin -> .set_aio_context -> .drained_end * in terms of child_job functions: child_job_drained_begin -> child_job_set_aio_context -> child_job_drained_end * in terms of job functions: job_pause_locked -> job_set_aio_context -> job_resume_locked We can see that after setting the new aio_context, job_resume_locked calls again job_enter_cond, which then invokes aio_co_wake(). But while job->aiocontext has been set in job_set_aio_context, job->co->ctx has not changed, so the coroutine would be entering in the wrong aiocontext. Using aio_co_schedule in job_resume_locked() might seem as a valid alternative, but the problem is that the bh resuming the coroutine is not scheduled immediately, and if in the meanwhile another bdrv_try_set_aio_context() is run (see test_propagate_mirror() in test-block-iothread.c), we would have the first schedule in the wrong aiocontext, and the second set of drains won't even manage to schedule the coroutine, as job->busy would still be true from the previous job_resume_locked(). The solution is to stick with aio_co_wake() and detect every time the coroutine resumes back from yielding if job->aio_context has changed. If so, we can reschedule it to the new context. Check for the aiocontext change in job_do_yield_locked because: 1) aio_co_reschedule_self requires to be in the running coroutine 2) since child_job_set_aio_context allows changing the aiocontext only while the job is paused, this is the exact place where the coroutine resumes, before running JobDriver's code. Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Message-Id: <20220926093214.506243-13-eesposit@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2022-09-26 17:32:05 +08:00
AioContext *next_aio_context;
if (ns != -1) {
timer_mod(&job->sleep_timer, ns);
}
job->busy = false;
job_event_idle_locked(job);
job_unlock();
qemu_coroutine_yield();
job_lock();
job: detect change of aiocontext within job coroutine We want to make sure access of job->aio_context is always done under either BQL or job_mutex. The problem is that using aio_co_enter(job->aiocontext, job->co) in job_start and job_enter_cond makes the coroutine immediately resume, so we can't hold the job lock. And caching it is not safe either, as it might change. job_start is under BQL, so it can freely read job->aiocontext, but job_enter_cond is not. We want to avoid reading job->aio_context in job_enter_cond, therefore: 1) use aio_co_wake(), since it doesn't want an aiocontext as argument but uses job->co->ctx 2) detect possible discrepancy between job->co->ctx and job->aio_context by checking right after the coroutine resumes back from yielding if job->aio_context has changed. If so, reschedule the coroutine to the new context. Calling bdrv_try_set_aio_context() will issue the following calls (simplified): * in terms of bdrv callbacks: .drained_begin -> .set_aio_context -> .drained_end * in terms of child_job functions: child_job_drained_begin -> child_job_set_aio_context -> child_job_drained_end * in terms of job functions: job_pause_locked -> job_set_aio_context -> job_resume_locked We can see that after setting the new aio_context, job_resume_locked calls again job_enter_cond, which then invokes aio_co_wake(). But while job->aiocontext has been set in job_set_aio_context, job->co->ctx has not changed, so the coroutine would be entering in the wrong aiocontext. Using aio_co_schedule in job_resume_locked() might seem as a valid alternative, but the problem is that the bh resuming the coroutine is not scheduled immediately, and if in the meanwhile another bdrv_try_set_aio_context() is run (see test_propagate_mirror() in test-block-iothread.c), we would have the first schedule in the wrong aiocontext, and the second set of drains won't even manage to schedule the coroutine, as job->busy would still be true from the previous job_resume_locked(). The solution is to stick with aio_co_wake() and detect every time the coroutine resumes back from yielding if job->aio_context has changed. If so, we can reschedule it to the new context. Check for the aiocontext change in job_do_yield_locked because: 1) aio_co_reschedule_self requires to be in the running coroutine 2) since child_job_set_aio_context allows changing the aiocontext only while the job is paused, this is the exact place where the coroutine resumes, before running JobDriver's code. Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Message-Id: <20220926093214.506243-13-eesposit@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2022-09-26 17:32:05 +08:00
next_aio_context = job->aio_context;
/*
* Coroutine has resumed, but in the meanwhile the job AioContext
* might have changed via bdrv_try_change_aio_context(), so we need to move
job: detect change of aiocontext within job coroutine We want to make sure access of job->aio_context is always done under either BQL or job_mutex. The problem is that using aio_co_enter(job->aiocontext, job->co) in job_start and job_enter_cond makes the coroutine immediately resume, so we can't hold the job lock. And caching it is not safe either, as it might change. job_start is under BQL, so it can freely read job->aiocontext, but job_enter_cond is not. We want to avoid reading job->aio_context in job_enter_cond, therefore: 1) use aio_co_wake(), since it doesn't want an aiocontext as argument but uses job->co->ctx 2) detect possible discrepancy between job->co->ctx and job->aio_context by checking right after the coroutine resumes back from yielding if job->aio_context has changed. If so, reschedule the coroutine to the new context. Calling bdrv_try_set_aio_context() will issue the following calls (simplified): * in terms of bdrv callbacks: .drained_begin -> .set_aio_context -> .drained_end * in terms of child_job functions: child_job_drained_begin -> child_job_set_aio_context -> child_job_drained_end * in terms of job functions: job_pause_locked -> job_set_aio_context -> job_resume_locked We can see that after setting the new aio_context, job_resume_locked calls again job_enter_cond, which then invokes aio_co_wake(). But while job->aiocontext has been set in job_set_aio_context, job->co->ctx has not changed, so the coroutine would be entering in the wrong aiocontext. Using aio_co_schedule in job_resume_locked() might seem as a valid alternative, but the problem is that the bh resuming the coroutine is not scheduled immediately, and if in the meanwhile another bdrv_try_set_aio_context() is run (see test_propagate_mirror() in test-block-iothread.c), we would have the first schedule in the wrong aiocontext, and the second set of drains won't even manage to schedule the coroutine, as job->busy would still be true from the previous job_resume_locked(). The solution is to stick with aio_co_wake() and detect every time the coroutine resumes back from yielding if job->aio_context has changed. If so, we can reschedule it to the new context. Check for the aiocontext change in job_do_yield_locked because: 1) aio_co_reschedule_self requires to be in the running coroutine 2) since child_job_set_aio_context allows changing the aiocontext only while the job is paused, this is the exact place where the coroutine resumes, before running JobDriver's code. Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Message-Id: <20220926093214.506243-13-eesposit@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2022-09-26 17:32:05 +08:00
* the coroutine too in the new aiocontext.
*/
while (qemu_get_current_aio_context() != next_aio_context) {
job_unlock();
aio_co_reschedule_self(next_aio_context);
job_lock();
next_aio_context = job->aio_context;
}
/* Set by job_enter_cond_locked() before re-entering the coroutine. */
assert(job->busy);
}
/* Called with job_mutex held, but releases it temporarily. */
static void coroutine_fn job_pause_point_locked(Job *job)
{
assert(job && job_started_locked(job));
if (!job_should_pause_locked(job)) {
return;
}
if (job_is_cancelled_locked(job)) {
return;
}
if (job->driver->pause) {
job_unlock();
job->driver->pause(job);
job_lock();
}
if (job_should_pause_locked(job) && !job_is_cancelled_locked(job)) {
JobStatus status = job->status;
job_state_transition_locked(job, status == JOB_STATUS_READY
? JOB_STATUS_STANDBY
: JOB_STATUS_PAUSED);
job->paused = true;
job_do_yield_locked(job, -1);
job->paused = false;
job_state_transition_locked(job, status);
}
if (job->driver->resume) {
job_unlock();
job->driver->resume(job);
job_lock();
}
}
void coroutine_fn job_pause_point(Job *job)
{
JOB_LOCK_GUARD();
job_pause_point_locked(job);
}
void coroutine_fn job_yield(Job *job)
{
JOB_LOCK_GUARD();
assert(job->busy);
/* Check cancellation *before* setting busy = false, too! */
if (job_is_cancelled_locked(job)) {
return;
}
if (!job_should_pause_locked(job)) {
job_do_yield_locked(job, -1);
}
job_pause_point_locked(job);
}
void coroutine_fn job_sleep_ns(Job *job, int64_t ns)
{
JOB_LOCK_GUARD();
assert(job->busy);
/* Check cancellation *before* setting busy = false, too! */
if (job_is_cancelled_locked(job)) {
return;
}
if (!job_should_pause_locked(job)) {
job_do_yield_locked(job, qemu_clock_get_ns(QEMU_CLOCK_REALTIME) + ns);
}
job_pause_point_locked(job);
}
/* Assumes the job_mutex is held */
static bool job_timer_not_pending_locked(Job *job)
{
return !timer_pending(&job->sleep_timer);
}
void job_pause_locked(Job *job)
{
job->pause_count++;
if (!job->paused) {
job_enter_cond_locked(job, NULL);
}
}
void job_pause(Job *job)
{
JOB_LOCK_GUARD();
job_pause_locked(job);
}
void job_resume_locked(Job *job)
{
assert(job->pause_count > 0);
job->pause_count--;
if (job->pause_count) {
return;
}
/* kick only if no timer is pending */
job_enter_cond_locked(job, job_timer_not_pending_locked);
}
void job_resume(Job *job)
{
JOB_LOCK_GUARD();
job_resume_locked(job);
}
void job_user_pause_locked(Job *job, Error **errp)
{
if (job_apply_verb_locked(job, JOB_VERB_PAUSE, errp)) {
return;
}
if (job->user_paused) {
error_setg(errp, "Job is already paused");
return;
}
job->user_paused = true;
job_pause_locked(job);
}
bool job_user_paused_locked(Job *job)
{
return job->user_paused;
}
void job_user_resume_locked(Job *job, Error **errp)
{
assert(job);
GLOBAL_STATE_CODE();
if (!job->user_paused || job->pause_count <= 0) {
error_setg(errp, "Can't resume a job that was not paused");
return;
}
if (job_apply_verb_locked(job, JOB_VERB_RESUME, errp)) {
return;
}
if (job->driver->user_resume) {
job_unlock();
job->driver->user_resume(job);
job_lock();
}
job->user_paused = false;
job_resume_locked(job);
}
/* Called with job_mutex held, but releases it temporarily. */
static void job_do_dismiss_locked(Job *job)
{
assert(job);
job->busy = false;
job->paused = false;
job->deferred_to_main_loop = true;
job_txn_del_job_locked(job);
job_state_transition_locked(job, JOB_STATUS_NULL);
job_unref_locked(job);
}
void job_dismiss_locked(Job **jobptr, Error **errp)
{
Job *job = *jobptr;
/* similarly to _complete, this is QMP-interface only. */
assert(job->id);
if (job_apply_verb_locked(job, JOB_VERB_DISMISS, errp)) {
return;
}
job_do_dismiss_locked(job);
*jobptr = NULL;
}
void job_early_fail(Job *job)
{
JOB_LOCK_GUARD();
assert(job->status == JOB_STATUS_CREATED);
job_do_dismiss_locked(job);
}
/* Called with job_mutex held. */
static void job_conclude_locked(Job *job)
{
job_state_transition_locked(job, JOB_STATUS_CONCLUDED);
if (job->auto_dismiss || !job_started_locked(job)) {
job_do_dismiss_locked(job);
}
}
/* Called with job_mutex held. */
static void job_update_rc_locked(Job *job)
{
if (!job->ret && job_is_cancelled_locked(job)) {
job->ret = -ECANCELED;
}
if (job->ret) {
if (!job->err) {
error_setg(&job->err, "%s", strerror(-job->ret));
}
job_state_transition_locked(job, JOB_STATUS_ABORTING);
}
}
static void job_commit(Job *job)
{
assert(!job->ret);
GLOBAL_STATE_CODE();
if (job->driver->commit) {
job->driver->commit(job);
}
}
static void job_abort(Job *job)
{
assert(job->ret);
GLOBAL_STATE_CODE();
if (job->driver->abort) {
job->driver->abort(job);
}
}
static void job_clean(Job *job)
{
GLOBAL_STATE_CODE();
if (job->driver->clean) {
job->driver->clean(job);
}
}
/*
* Called with job_mutex held, but releases it temporarily.
* Takes AioContext lock internally to invoke a job->driver callback.
*/
static int job_finalize_single_locked(Job *job)
{
int job_ret;
AioContext *ctx = job->aio_context;
assert(job_is_completed_locked(job));
/* Ensure abort is called for late-transactional failures */
job_update_rc_locked(job);
job_ret = job->ret;
job_unlock();
aio_context_acquire(ctx);
if (!job_ret) {
job_commit(job);
} else {
job_abort(job);
}
job_clean(job);
if (job->cb) {
job->cb(job->opaque, job_ret);
}
aio_context_release(ctx);
job_lock();
/* Emit events only if we actually started */
if (job_started_locked(job)) {
if (job_is_cancelled_locked(job)) {
job_event_cancelled_locked(job);
} else {
job_event_completed_locked(job);
}
}
job_txn_del_job_locked(job);
job_conclude_locked(job);
return 0;
}
/*
* Called with job_mutex held, but releases it temporarily.
* Takes AioContext lock internally to invoke a job->driver callback.
*/
static void job_cancel_async_locked(Job *job, bool force)
{
AioContext *ctx = job->aio_context;
GLOBAL_STATE_CODE();
if (job->driver->cancel) {
job_unlock();
aio_context_acquire(ctx);
jobs: Give Job.force_cancel more meaning We largely have two cancel modes for jobs: First, there is actual cancelling. The job is terminated as soon as possible, without trying to reach a consistent result. Second, we have mirror in the READY state. Technically, the job is not really cancelled, but it just is a different completion mode. The job can still run for an indefinite amount of time while it tries to reach a consistent result. We want to be able to clearly distinguish which cancel mode a job is in (when it has been cancelled). We can use Job.force_cancel for this, but right now it only reflects cancel requests from the user with force=true, but clearly, jobs that do not even distinguish between force=false and force=true are effectively always force-cancelled. So this patch has Job.force_cancel signify whether the job will terminate as soon as possible (force_cancel=true) or whether it will effectively remain running despite being "cancelled" (force_cancel=false). To this end, we let jobs that provide JobDriver.cancel() tell the generic job code whether they will terminate as soon as possible or not, and for jobs that do not provide that method we assume they will. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211006151940.214590-7-hreitz@redhat.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2021-10-06 23:19:33 +08:00
force = job->driver->cancel(job, force);
aio_context_release(ctx);
job_lock();
jobs: Give Job.force_cancel more meaning We largely have two cancel modes for jobs: First, there is actual cancelling. The job is terminated as soon as possible, without trying to reach a consistent result. Second, we have mirror in the READY state. Technically, the job is not really cancelled, but it just is a different completion mode. The job can still run for an indefinite amount of time while it tries to reach a consistent result. We want to be able to clearly distinguish which cancel mode a job is in (when it has been cancelled). We can use Job.force_cancel for this, but right now it only reflects cancel requests from the user with force=true, but clearly, jobs that do not even distinguish between force=false and force=true are effectively always force-cancelled. So this patch has Job.force_cancel signify whether the job will terminate as soon as possible (force_cancel=true) or whether it will effectively remain running despite being "cancelled" (force_cancel=false). To this end, we let jobs that provide JobDriver.cancel() tell the generic job code whether they will terminate as soon as possible or not, and for jobs that do not provide that method we assume they will. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211006151940.214590-7-hreitz@redhat.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2021-10-06 23:19:33 +08:00
} else {
/* No .cancel() means the job will behave as if force-cancelled */
force = true;
}
jobs: Give Job.force_cancel more meaning We largely have two cancel modes for jobs: First, there is actual cancelling. The job is terminated as soon as possible, without trying to reach a consistent result. Second, we have mirror in the READY state. Technically, the job is not really cancelled, but it just is a different completion mode. The job can still run for an indefinite amount of time while it tries to reach a consistent result. We want to be able to clearly distinguish which cancel mode a job is in (when it has been cancelled). We can use Job.force_cancel for this, but right now it only reflects cancel requests from the user with force=true, but clearly, jobs that do not even distinguish between force=false and force=true are effectively always force-cancelled. So this patch has Job.force_cancel signify whether the job will terminate as soon as possible (force_cancel=true) or whether it will effectively remain running despite being "cancelled" (force_cancel=false). To this end, we let jobs that provide JobDriver.cancel() tell the generic job code whether they will terminate as soon as possible or not, and for jobs that do not provide that method we assume they will. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211006151940.214590-7-hreitz@redhat.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2021-10-06 23:19:33 +08:00
if (job->user_paused) {
/* Do not call job_enter here, the caller will handle it. */
if (job->driver->user_resume) {
job_unlock();
job->driver->user_resume(job);
job_lock();
}
job->user_paused = false;
assert(job->pause_count > 0);
job->pause_count--;
}
/*
* Ignore soft cancel requests after the job is already done
* (We will still invoke job->driver->cancel() above, but if the
* job driver supports soft cancelling and the job is done, that
* should be a no-op, too. We still call it so it can override
* @force.)
*/
if (force || !job->deferred_to_main_loop) {
job->cancelled = true;
/* To prevent 'force == false' overriding a previous 'force == true' */
job->force_cancel |= force;
}
}
/*
* Called with job_mutex held, but releases it temporarily.
* Takes AioContext lock internally to invoke a job->driver callback.
*/
static void job_completed_txn_abort_locked(Job *job)
{
JobTxn *txn = job->txn;
Job *other_job;
if (txn->aborting) {
/*
* We are cancelled by another job, which will handle everything.
*/
return;
}
txn->aborting = true;
job_txn_ref_locked(txn);
job_ref_locked(job);
/* Other jobs are effectively cancelled by us, set the status for
* them; this job, however, may or may not be cancelled, depending
* on the caller, so leave it. */
QLIST_FOREACH(other_job, &txn->jobs, txn_list) {
if (other_job != job) {
/*
* This is a transaction: If one job failed, no result will matter.
* Therefore, pass force=true to terminate all other jobs as quickly
* as possible.
*/
job_cancel_async_locked(other_job, true);
}
}
while (!QLIST_EMPTY(&txn->jobs)) {
other_job = QLIST_FIRST(&txn->jobs);
if (!job_is_completed_locked(other_job)) {
assert(job_cancel_requested_locked(other_job));
job_finish_sync_locked(other_job, NULL, NULL);
}
job_finalize_single_locked(other_job);
}
job_unref_locked(job);
job_txn_unref_locked(txn);
}
/* Called with job_mutex held, but releases it temporarily */
static int job_prepare_locked(Job *job)
{
int ret;
AioContext *ctx = job->aio_context;
GLOBAL_STATE_CODE();
if (job->ret == 0 && job->driver->prepare) {
job_unlock();
aio_context_acquire(ctx);
ret = job->driver->prepare(job);
aio_context_release(ctx);
job_lock();
job->ret = ret;
job_update_rc_locked(job);
}
return job->ret;
}
/* Called with job_mutex held */
static int job_needs_finalize_locked(Job *job)
{
return !job->auto_finalize;
}
/* Called with job_mutex held */
static void job_do_finalize_locked(Job *job)
{
int rc;
assert(job && job->txn);
/* prepare the transaction to complete */
rc = job_txn_apply_locked(job, job_prepare_locked);
if (rc) {
job_completed_txn_abort_locked(job);
} else {
job_txn_apply_locked(job, job_finalize_single_locked);
}
}
void job_finalize_locked(Job *job, Error **errp)
{
assert(job && job->id);
if (job_apply_verb_locked(job, JOB_VERB_FINALIZE, errp)) {
return;
}
job_do_finalize_locked(job);
}
/* Called with job_mutex held. */
static int job_transition_to_pending_locked(Job *job)
{
job_state_transition_locked(job, JOB_STATUS_PENDING);
if (!job->auto_finalize) {
job_event_pending_locked(job);
}
return 0;
}
void job_transition_to_ready(Job *job)
{
JOB_LOCK_GUARD();
job_state_transition_locked(job, JOB_STATUS_READY);
job_event_ready_locked(job);
}
/* Called with job_mutex held. */
static void job_completed_txn_success_locked(Job *job)
{
JobTxn *txn = job->txn;
Job *other_job;
job_state_transition_locked(job, JOB_STATUS_WAITING);
/*
* Successful completion, see if there are other running jobs in this
* txn.
*/
QLIST_FOREACH(other_job, &txn->jobs, txn_list) {
if (!job_is_completed_locked(other_job)) {
return;
}
assert(other_job->ret == 0);
}
job_txn_apply_locked(job, job_transition_to_pending_locked);
/* If no jobs need manual finalization, automatically do so */
if (job_txn_apply_locked(job, job_needs_finalize_locked) == 0) {
job_do_finalize_locked(job);
}
}
/* Called with job_mutex held. */
static void job_completed_locked(Job *job)
{
assert(job && job->txn && !job_is_completed_locked(job));
job_update_rc_locked(job);
trace_job_completed(job, job->ret);
if (job->ret) {
job_completed_txn_abort_locked(job);
} else {
job_completed_txn_success_locked(job);
}
}
/**
* Useful only as a type shim for aio_bh_schedule_oneshot.
* Called with job_mutex *not* held.
*/
static void job_exit(void *opaque)
{
Job *job = (Job *)opaque;
JOB_LOCK_GUARD();
job_ref_locked(job);
/* This is a lie, we're not quiescent, but still doing the completion
* callbacks. However, completion callbacks tend to involve operations that
* drain block nodes, and if .drained_poll still returned true, we would
* deadlock. */
job->busy = false;
job_event_idle_locked(job);
job_completed_locked(job);
job_unref_locked(job);
}
/**
* All jobs must allow a pause point before entering their job proper. This
* ensures that jobs can be paused prior to being started, then resumed later.
*/
static void coroutine_fn job_co_entry(void *opaque)
{
Job *job = opaque;
int ret;
assert(job && job->driver && job->driver->run);
WITH_JOB_LOCK_GUARD() {
assert(job->aio_context == qemu_get_current_aio_context());
job_pause_point_locked(job);
}
ret = job->driver->run(job, &job->err);
WITH_JOB_LOCK_GUARD() {
job->ret = ret;
job->deferred_to_main_loop = true;
job->busy = true;
}
aio_bh_schedule_oneshot(qemu_get_aio_context(), job_exit, job);
}
void job_start(Job *job)
{
assert(qemu_in_main_thread());
WITH_JOB_LOCK_GUARD() {
assert(job && !job_started_locked(job) && job->paused &&
job->driver && job->driver->run);
job->co = qemu_coroutine_create(job_co_entry, job);
job->pause_count--;
job->busy = true;
job->paused = false;
job_state_transition_locked(job, JOB_STATUS_RUNNING);
}
aio_co_enter(job->aio_context, job->co);
}
void job_cancel_locked(Job *job, bool force)
{
if (job->status == JOB_STATUS_CONCLUDED) {
job_do_dismiss_locked(job);
return;
}
job_cancel_async_locked(job, force);
if (!job_started_locked(job)) {
job_completed_locked(job);
} else if (job->deferred_to_main_loop) {
/*
* job_cancel_async() ignores soft-cancel requests for jobs
* that are already done (i.e. deferred to the main loop). We
* have to check again whether the job is really cancelled.
job: Add job_cancel_requested() Most callers of job_is_cancelled() actually want to know whether the job is on its way to immediate termination. For example, we refuse to pause jobs that are cancelled; but this only makes sense for jobs that are really actually cancelled. A mirror job that is cancelled during READY with force=false should absolutely be allowed to pause. This "cancellation" (which is actually a kind of completion) may take an indefinite amount of time, and so should behave like any job during normal operation. For example, with on-target-error=stop, the job should stop on write errors. (In contrast, force-cancelled jobs should not get write errors, as they should just terminate and not do further I/O.) Therefore, redefine job_is_cancelled() to only return true for jobs that are force-cancelled (which as of HEAD^ means any job that interprets the cancellation request as a request for immediate termination), and add job_cancel_requested() as the general variant, which returns true for any jobs which have been requested to be cancelled, whether it be immediately or after an arbitrarily long completion phase. Finally, here is a justification for how different job_is_cancelled() invocations are treated by this patch: - block/mirror.c (mirror_run()): - The first invocation is a while loop that should loop until the job has been cancelled or scheduled for completion. What kind of cancel does not matter, only the fact that the job is supposed to end. - The second invocation wants to know whether the job has been soft-cancelled. Calling job_cancel_requested() is a bit too broad, but if the job were force-cancelled, we should leave the main loop as soon as possible anyway, so this should not matter here. - The last two invocations already check force_cancel, so they should continue to use job_is_cancelled(). - block/backup.c, block/commit.c, block/stream.c, anything in tests/: These jobs know only force-cancel, so there is no difference between job_is_cancelled() and job_cancel_requested(). We can continue using job_is_cancelled(). - job.c: - job_pause_point(), job_yield(), job_sleep_ns(): Only force-cancelled jobs should be prevented from being paused. Continue using job_is_cancelled(). - job_update_rc(), job_finalize_single(), job_finish_sync(): These functions are all called after the job has left its main loop. The mirror job (the only job that can be soft-cancelled) will clear .cancelled before leaving the main loop if it has been soft-cancelled. Therefore, these functions will observe .cancelled to be true only if the job has been force-cancelled. We can continue to use job_is_cancelled(). (Furthermore, conceptually, a soft-cancelled mirror job should not report to have been cancelled. It should report completion (see also the block-job-cancel QAPI documentation). Therefore, it makes sense for these functions not to distinguish between a soft-cancelled mirror job and a job that has completed as normal.) - job_completed_txn_abort(): All jobs other than @job have been force-cancelled. job_is_cancelled() must be true for them. Regarding @job itself: job_completed_txn_abort() is mostly called when the job's return value is not 0. A soft-cancelled mirror has a return value of 0, and so will not end up here then. However, job_cancel() invokes job_completed_txn_abort() if the job has been deferred to the main loop, which is mostly the case for completed jobs (which skip the assertion), but not for sure. To be safe, use job_cancel_requested() in this assertion. - job_complete(): This is function eventually invoked by the user (through qmp_block_job_complete() or qmp_job_complete(), or job_complete_sync(), which comes from qemu-img). The intention here is to prevent a user from invoking job-complete after the job has been cancelled. This should also apply to soft cancelling: After a mirror job has been soft-cancelled, the user should not be able to decide otherwise and have it complete as normal (i.e. pivoting to the target). - job_cancel(): Both functions are equivalent (see comment there), but we want to use job_is_cancelled(), because this shows that we call job_completed_txn_abort() only for force-cancelled jobs. (As explained for job_update_rc(), soft-cancelled jobs should be treated as if they have completed as normal.) Buglink: https://gitlab.com/qemu-project/qemu/-/issues/462 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20211006151940.214590-9-hreitz@redhat.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2021-10-06 23:19:35 +08:00
* (job_cancel_requested() and job_is_cancelled() are equivalent
* here, because job_cancel_async() will make soft-cancel
* requests no-ops when deferred_to_main_loop is true. We
* choose to call job_is_cancelled() to show that we invoke
* job_completed_txn_abort() only for force-cancelled jobs.)
*/
if (job_is_cancelled_locked(job)) {
job_completed_txn_abort_locked(job);
}
} else {
job_enter_cond_locked(job, NULL);
}
}
void job_user_cancel_locked(Job *job, bool force, Error **errp)
{
if (job_apply_verb_locked(job, JOB_VERB_CANCEL, errp)) {
return;
}
job_cancel_locked(job, force);
}
/* A wrapper around job_cancel_locked() taking an Error ** parameter so it may
* be used with job_finish_sync_locked() without the need for (rather nasty)
* function pointer casts there.
*
* Called with job_mutex held.
*/
static void job_cancel_err_locked(Job *job, Error **errp)
{
job_cancel_locked(job, false);
}
/**
* Same as job_cancel_err(), but force-cancel.
* Called with job_mutex held.
*/
static void job_force_cancel_err_locked(Job *job, Error **errp)
{
job_cancel_locked(job, true);
}
int job_cancel_sync_locked(Job *job, bool force)
{
if (force) {
return job_finish_sync_locked(job, &job_force_cancel_err_locked, NULL);
} else {
return job_finish_sync_locked(job, &job_cancel_err_locked, NULL);
}
}
int job_cancel_sync(Job *job, bool force)
{
JOB_LOCK_GUARD();
return job_cancel_sync_locked(job, force);
}
void job_cancel_sync_all(void)
{
Job *job;
JOB_LOCK_GUARD();
while ((job = job_next_locked(NULL))) {
job_cancel_sync_locked(job, true);
}
}
int job_complete_sync_locked(Job *job, Error **errp)
{
return job_finish_sync_locked(job, job_complete_locked, errp);
}
void job_complete_locked(Job *job, Error **errp)
{
/* Should not be reachable via external interface for internal jobs */
assert(job->id);
GLOBAL_STATE_CODE();
if (job_apply_verb_locked(job, JOB_VERB_COMPLETE, errp)) {
return;
}
if (job_cancel_requested_locked(job) || !job->driver->complete) {
error_setg(errp, "The active block job '%s' cannot be completed",
job->id);
return;
}
job_unlock();
job->driver->complete(job, errp);
job_lock();
}
int job_finish_sync_locked(Job *job,
void (*finish)(Job *, Error **errp),
Error **errp)
{
Error *local_err = NULL;
int ret;
GLOBAL_STATE_CODE();
job_ref_locked(job);
if (finish) {
finish(job, &local_err);
}
if (local_err) {
error_propagate(errp, local_err);
job_unref_locked(job);
return -EBUSY;
}
job_unlock();
AIO_WAIT_WHILE_UNLOCKED(job->aio_context,
(job_enter(job), !job_is_completed(job)));
job_lock();
ret = (job_is_cancelled_locked(job) && job->ret == 0)
? -ECANCELED : job->ret;
job_unref_locked(job);
return ret;
}