The following changes since commit f5fe7c17ac4e309e47e78f0f9761aebc8d2f2c81:

Merge tag 'pull-tcg-20230823-2' of https://gitlab.com/rth7680/qemu into staging (2023-08-28 16:07:04 -0400)

are available in the Git repository at:

https://gitlab.com/hreitz/qemu.git tags/pull-block-2023-09-01

for you to fetch changes up to 380448464dd89291cf7fd7434be6c225482a334d:

tests/file-io-error: New test (2023-08-29 13:01:24 +0200)

Block patches

- Fix for file-posix's zoning code crashing on I/O errors

- Throttling refactoring

Zhenwei Pi (9):

throttle: introduce enum ThrottleDirection

test-throttle: use enum ThrottleDirection

throttle: support read-only and write-only

test-throttle: test read only and write only

cryptodev: use NULL throttle timer cb for read direction

throttle: use enum ThrottleDirection instead of bool is_write

throttle: use THROTTLE_MAX/ARRAY_SIZE for hard code

fsdev: Use ThrottleDirection instread of bool is_write

block/throttle-groups: Use ThrottleDirection instread of bool is_write

fsdev/qemu-fsdev-throttle.h | 4 +-

include/block/throttle-groups.h | 6 +-

include/qemu/throttle.h | 16 +-

backends/cryptodev.c | 12 +-

block/block-backend.c | 4 +-

block/file-posix.c | 42 +++---

block/throttle-groups.c | 163 +++++++++++----------

block/throttle.c | 8 +-

fsdev/qemu-fsdev-throttle.c | 18 ++-

hw/9pfs/cofile.c | 4 +-

tests/unit/test-throttle.c | 76 +++++++++-

util/throttle.c | 84 +++++++----

tests/qemu-iotests/tests/file-io-error | 119 +++++++++++++++

tests/qemu-iotests/tests/file-io-error.out | 33 +++++

14 files changed, 418 insertions(+), 171 deletions(-)

create mode 100755 tests/qemu-iotests/tests/file-io-error

create mode 100644 tests/qemu-iotests/tests/file-io-error.out

2.41.0

From: zhenwei pi <pizhenwei@bytedance.com>

Operations on a cryptodev are considered as *write* only, the callback

of read direction is never invoked. Use NULL instead of an unreachable

path(cryptodev_backend_throttle_timer_cb on read direction).

The dummy read timer(never invoked) is already removed here, it means

that the 'FIXME' tag is no longer needed.

Reviewed-by: Alberto Garcia <berto@igalia.com>

Reviewed-by: Hanna Czenczek <hreitz@redhat.com>

Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>

Message-Id: <20230728022006.1098509-6-pizhenwei@bytedance.com>

Signed-off-by: Hanna Czenczek <hreitz@redhat.com>

backends/cryptodev.c | 3 +--

diff --git a/backends/cryptodev.c b/backends/cryptodev.c

--- a/backends/cryptodev.c

+++ b/backends/cryptodev.c

@@ -XXX,XX +XXX,XX @@ static void cryptodev_backend_set_throttle(CryptoDevBackend *backend, int field,

if (!enabled) {

throttle_init(&backend->ts);

throttle_timers_init(&backend->tt, qemu_get_aio_context(),

- QEMU_CLOCK_REALTIME,

- cryptodev_backend_throttle_timer_cb, /* FIXME */

+ QEMU_CLOCK_REALTIME, NULL,

cryptodev_backend_throttle_timer_cb, backend);

}

2.41.0

From: zhenwei pi <pizhenwei@bytedance.com>

'bool is_write' style is obsolete from throttle framework, adapt

block throttle groups to the new style:

- use ThrottleDirection instead of 'bool is_write'. Ex,

schedule_next_request(ThrottleGroupMember *tgm, bool is_write)

-> schedule_next_request(ThrottleGroupMember *tgm, ThrottleDirection direction)

- use THROTTLE_MAX instead of hard code. Ex, ThrottleGroupMember *tokens[2]

-> ThrottleGroupMember *tokens[THROTTLE_MAX]

- use ThrottleDirection instead of hard code on iteration. Ex, (i = 0; i < 2; i++)

-> for (dir = THROTTLE_READ; dir < THROTTLE_MAX; dir++)

Use a simple python script to test the new style:

#!/usr/bin/python3

import subprocess

import random

import time

commands = ['virsh blkdeviotune jammy vda --write-bytes-sec ', \

'virsh blkdeviotune jammy vda --write-iops-sec ', \

'virsh blkdeviotune jammy vda --read-bytes-sec ', \

'virsh blkdeviotune jammy vda --read-iops-sec ']

for loop in range(1, 1000):

time.sleep(random.randrange(3, 5))

command = commands[random.randrange(0, 3)] + str(random.randrange(0, 1000000))

subprocess.run(command, shell=True, check=True)

This works fine.

Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>

Message-Id: <20230728022006.1098509-10-pizhenwei@bytedance.com>

Reviewed-by: Hanna Czenczek <hreitz@redhat.com>

Signed-off-by: Hanna Czenczek <hreitz@redhat.com>

---

include/block/throttle-groups.h | 6 +-

block/block-backend.c | 4 +-

block/throttle-groups.c | 161 ++++++++++++++++----------------

block/throttle.c | 8 +-

4 files changed, 90 insertions(+), 89 deletions(-)

diff --git a/include/block/throttle-groups.h b/include/block/throttle-groups.h

index XXXXXXX..XXXXXXX 100644

--- a/include/block/throttle-groups.h

+++ b/include/block/throttle-groups.h

@@ -XXX,XX +XXX,XX @@ typedef struct ThrottleGroupMember {

AioContext *aio_context;

/* throttled_reqs_lock protects the CoQueues for throttled requests. */

CoMutex throttled_reqs_lock;

- CoQueue throttled_reqs[2];

+ CoQueue throttled_reqs[THROTTLE_MAX];

/* Nonzero if the I/O limits are currently being ignored; generally

* it is zero. Accessed with atomic operations.

@@ -XXX,XX +XXX,XX @@ typedef struct ThrottleGroupMember {

* throttle_state tells us if I/O limits are configured. */

ThrottleState *throttle_state;

ThrottleTimers throttle_timers;

- unsigned pending_reqs[2];

+ unsigned pending_reqs[THROTTLE_MAX];

QLIST_ENTRY(ThrottleGroupMember) round_robin;

} ThrottleGroupMember;

@@ -XXX,XX +XXX,XX @@ void throttle_group_restart_tgm(ThrottleGroupMember *tgm);

void coroutine_fn throttle_group_co_io_limits_intercept(ThrottleGroupMember *tgm,

int64_t bytes,

- bool is_write);

+ ThrottleDirection direction);

void throttle_group_attach_aio_context(ThrottleGroupMember *tgm,

AioContext *new_context);

void throttle_group_detach_aio_context(ThrottleGroupMember *tgm);

diff --git a/block/block-backend.c b/block/block-backend.c

index XXXXXXX..XXXXXXX 100644

--- a/block/block-backend.c

+++ b/block/block-backend.c

@@ -XXX,XX +XXX,XX @@ blk_co_do_preadv_part(BlockBackend *blk, int64_t offset, int64_t bytes,

/* throttling disk I/O */

if (blk->public.throttle_group_member.throttle_state) {

throttle_group_co_io_limits_intercept(&blk->public.throttle_group_member,

- bytes, false);

+ bytes, THROTTLE_READ);

}

ret = bdrv_co_preadv_part(blk->root, offset, bytes, qiov, qiov_offset,

@@ -XXX,XX +XXX,XX @@ blk_co_do_pwritev_part(BlockBackend *blk, int64_t offset, int64_t bytes,

/* throttling disk I/O */

if (blk->public.throttle_group_member.throttle_state) {

throttle_group_co_io_limits_intercept(&blk->public.throttle_group_member,

- bytes, true);

+ bytes, THROTTLE_WRITE);

}

if (!blk->enable_write_cache) {

diff --git a/block/throttle-groups.c b/block/throttle-groups.c

index XXXXXXX..XXXXXXX 100644

--- a/block/throttle-groups.c

+++ b/block/throttle-groups.c

@@ -XXX,XX +XXX,XX @@

static void throttle_group_obj_init(Object *obj);

static void throttle_group_obj_complete(UserCreatable *obj, Error **errp);

-static void timer_cb(ThrottleGroupMember *tgm, bool is_write);

+static void timer_cb(ThrottleGroupMember *tgm, ThrottleDirection direction);

/* The ThrottleGroup structure (with its ThrottleState) is shared

* among different ThrottleGroupMembers and it's independent from

@@ -XXX,XX +XXX,XX @@ struct ThrottleGroup {

QemuMutex lock; /* This lock protects the following four fields */

ThrottleState ts;

QLIST_HEAD(, ThrottleGroupMember) head;

- ThrottleGroupMember *tokens[2];

- bool any_timer_armed[2];

+ ThrottleGroupMember *tokens[THROTTLE_MAX];

+ bool any_timer_armed[THROTTLE_MAX];

QEMUClockType clock_type;

/* This field is protected by the global QEMU mutex */

@@ -XXX,XX +XXX,XX @@ static ThrottleGroupMember *throttle_group_next_tgm(ThrottleGroupMember *tgm)

* This assumes that tg->lock is held.

*

* @tgm: the ThrottleGroupMember

- * @is_write: the type of operation (read/write)

+ * @direction: the ThrottleDirection

* @ret: whether the ThrottleGroupMember has pending requests.

*/

static inline bool tgm_has_pending_reqs(ThrottleGroupMember *tgm,

- bool is_write)

+ ThrottleDirection direction)

{

- return tgm->pending_reqs[is_write];

+ return tgm->pending_reqs[direction];

}

/* Return the next ThrottleGroupMember in the round-robin sequence with pending

@@ -XXX,XX +XXX,XX @@ static inline bool tgm_has_pending_reqs(ThrottleGroupMember *tgm,

* This assumes that tg->lock is held.

*

* @tgm: the current ThrottleGroupMember

- * @is_write: the type of operation (read/write)

+ * @direction: the ThrottleDirection

* @ret: the next ThrottleGroupMember with pending requests, or tgm if

* there is none.

*/

static ThrottleGroupMember *next_throttle_token(ThrottleGroupMember *tgm,

- bool is_write)

+ ThrottleDirection direction)

{

ThrottleState *ts = tgm->throttle_state;

ThrottleGroup *tg = container_of(ts, ThrottleGroup, ts);

@@ -XXX,XX +XXX,XX @@ static ThrottleGroupMember *next_throttle_token(ThrottleGroupMember *tgm,

* it's being drained. Skip the round-robin search and return tgm

* immediately if it has pending requests. Otherwise we could be

* forcing it to wait for other member's throttled requests. */

- if (tgm_has_pending_reqs(tgm, is_write) &&

+ if (tgm_has_pending_reqs(tgm, direction) &&

qatomic_read(&tgm->io_limits_disabled)) {

return tgm;

}

- start = token = tg->tokens[is_write];

+ start = token = tg->tokens[direction];

/* get next bs round in round robin style */

token = throttle_group_next_tgm(token);

- while (token != start && !tgm_has_pending_reqs(token, is_write)) {

+ while (token != start && !tgm_has_pending_reqs(token, direction)) {

token = throttle_group_next_tgm(token);

}

@@ -XXX,XX +XXX,XX @@ static ThrottleGroupMember *next_throttle_token(ThrottleGroupMember *tgm,

* then decide the token is the current tgm because chances are

* the current tgm got the current request queued.

*/

- if (token == start && !tgm_has_pending_reqs(token, is_write)) {

+ if (token == start && !tgm_has_pending_reqs(token, direction)) {

token = tgm;

}

/* Either we return the original TGM, or one with pending requests */

- assert(token == tgm || tgm_has_pending_reqs(token, is_write));

+ assert(token == tgm || tgm_has_pending_reqs(token, direction));

return token;

}

@@ -XXX,XX +XXX,XX @@ static ThrottleGroupMember *next_throttle_token(ThrottleGroupMember *tgm,

* This assumes that tg->lock is held.

*

* @tgm: the current ThrottleGroupMember

- * @is_write: the type of operation (read/write)

+ * @direction: the ThrottleDirection

* @ret: whether the I/O request needs to be throttled or not

*/

static bool throttle_group_schedule_timer(ThrottleGroupMember *tgm,

- bool is_write)

+ ThrottleDirection direction)

{

ThrottleState *ts = tgm->throttle_state;

ThrottleGroup *tg = container_of(ts, ThrottleGroup, ts);

ThrottleTimers *tt = &tgm->throttle_timers;

- ThrottleDirection direction = is_write ? THROTTLE_WRITE : THROTTLE_READ;

bool must_wait;

if (qatomic_read(&tgm->io_limits_disabled)) {

@@ -XXX,XX +XXX,XX @@ static bool throttle_group_schedule_timer(ThrottleGroupMember *tgm,

}

/* Check if any of the timers in this group is already armed */

- if (tg->any_timer_armed[is_write]) {

+ if (tg->any_timer_armed[direction]) {

return true;

}

@@ -XXX,XX +XXX,XX @@ static bool throttle_group_schedule_timer(ThrottleGroupMember *tgm,

/* If a timer just got armed, set tgm as the current token */

if (must_wait) {

- tg->tokens[is_write] = tgm;

- tg->any_timer_armed[is_write] = true;

+ tg->tokens[direction] = tgm;

+ tg->any_timer_armed[direction] = true;

}

return must_wait;

@@ -XXX,XX +XXX,XX @@ static bool throttle_group_schedule_timer(ThrottleGroupMember *tgm,

* any request was actually pending.

*

* @tgm: the current ThrottleGroupMember

- * @is_write: the type of operation (read/write)

+ * @direction: the ThrottleDirection

*/

static bool coroutine_fn throttle_group_co_restart_queue(ThrottleGroupMember *tgm,

- bool is_write)

+ ThrottleDirection direction)

{

bool ret;

qemu_co_mutex_lock(&tgm->throttled_reqs_lock);

- ret = qemu_co_queue_next(&tgm->throttled_reqs[is_write]);

+ ret = qemu_co_queue_next(&tgm->throttled_reqs[direction]);

qemu_co_mutex_unlock(&tgm->throttled_reqs_lock);

return ret;

@@ -XXX,XX +XXX,XX @@ static bool coroutine_fn throttle_group_co_restart_queue(ThrottleGroupMember *tg

* This assumes that tg->lock is held.

*

* @tgm: the current ThrottleGroupMember

- * @is_write: the type of operation (read/write)

+ * @direction: the ThrottleDirection

*/

-static void schedule_next_request(ThrottleGroupMember *tgm, bool is_write)

+static void schedule_next_request(ThrottleGroupMember *tgm,

+ ThrottleDirection direction)

{

ThrottleState *ts = tgm->throttle_state;

ThrottleGroup *tg = container_of(ts, ThrottleGroup, ts);

@@ -XXX,XX +XXX,XX @@ static void schedule_next_request(ThrottleGroupMember *tgm, bool is_write)

ThrottleGroupMember *token;

/* Check if there's any pending request to schedule next */

- token = next_throttle_token(tgm, is_write);

- if (!tgm_has_pending_reqs(token, is_write)) {

+ token = next_throttle_token(tgm, direction);

+ if (!tgm_has_pending_reqs(token, direction)) {

return;

}

/* Set a timer for the request if it needs to be throttled */

- must_wait = throttle_group_schedule_timer(token, is_write);

+ must_wait = throttle_group_schedule_timer(token, direction);

/* If it doesn't have to wait, queue it for immediate execution */

if (!must_wait) {

/* Give preference to requests from the current tgm */

if (qemu_in_coroutine() &&

- throttle_group_co_restart_queue(tgm, is_write)) {

+ throttle_group_co_restart_queue(tgm, direction)) {

token = tgm;

} else {

ThrottleTimers *tt = &token->throttle_timers;

int64_t now = qemu_clock_get_ns(tg->clock_type);

- timer_mod(tt->timers[is_write], now);

- tg->any_timer_armed[is_write] = true;

+ timer_mod(tt->timers[direction], now);

+ tg->any_timer_armed[direction] = true;

}

- tg->tokens[is_write] = token;

+ tg->tokens[direction] = token;

}

}

@@ -XXX,XX +XXX,XX @@ static void schedule_next_request(ThrottleGroupMember *tgm, bool is_write)

*

* @tgm: the current ThrottleGroupMember

* @bytes: the number of bytes for this I/O

- * @is_write: the type of operation (read/write)

+ * @direction: the ThrottleDirection

*/

void coroutine_fn throttle_group_co_io_limits_intercept(ThrottleGroupMember *tgm,

int64_t bytes,

- bool is_write)

+ ThrottleDirection direction)

{

bool must_wait;

ThrottleGroupMember *token;

ThrottleGroup *tg = container_of(tgm->throttle_state, ThrottleGroup, ts);

- ThrottleDirection direction = is_write ? THROTTLE_WRITE : THROTTLE_READ;

assert(bytes >= 0);

+ assert(direction < THROTTLE_MAX);

qemu_mutex_lock(&tg->lock);

/* First we check if this I/O has to be throttled. */

- token = next_throttle_token(tgm, is_write);

- must_wait = throttle_group_schedule_timer(token, is_write);

+ token = next_throttle_token(tgm, direction);

+ must_wait = throttle_group_schedule_timer(token, direction);

/* Wait if there's a timer set or queued requests of this type */

- if (must_wait || tgm->pending_reqs[is_write]) {

- tgm->pending_reqs[is_write]++;

+ if (must_wait || tgm->pending_reqs[direction]) {

+ tgm->pending_reqs[direction]++;

qemu_mutex_unlock(&tg->lock);

qemu_co_mutex_lock(&tgm->throttled_reqs_lock);

- qemu_co_queue_wait(&tgm->throttled_reqs[is_write],

+ qemu_co_queue_wait(&tgm->throttled_reqs[direction],

&tgm->throttled_reqs_lock);

qemu_co_mutex_unlock(&tgm->throttled_reqs_lock);

qemu_mutex_lock(&tg->lock);

- tgm->pending_reqs[is_write]--;

+ tgm->pending_reqs[direction]--;

}

/* The I/O will be executed, so do the accounting */

throttle_account(tgm->throttle_state, direction, bytes);

/* Schedule the next request */

- schedule_next_request(tgm, is_write);

+ schedule_next_request(tgm, direction);

qemu_mutex_unlock(&tg->lock);

}

typedef struct {

ThrottleGroupMember *tgm;

- bool is_write;

+ ThrottleDirection direction;

} RestartData;

static void coroutine_fn throttle_group_restart_queue_entry(void *opaque)

@@ -XXX,XX +XXX,XX @@ static void coroutine_fn throttle_group_restart_queue_entry(void *opaque)

ThrottleGroupMember *tgm = data->tgm;

ThrottleState *ts = tgm->throttle_state;

ThrottleGroup *tg = container_of(ts, ThrottleGroup, ts);

- bool is_write = data->is_write;

+ ThrottleDirection direction = data->direction;

bool empty_queue;

- empty_queue = !throttle_group_co_restart_queue(tgm, is_write);

+ empty_queue = !throttle_group_co_restart_queue(tgm, direction);

/* If the request queue was empty then we have to take care of

* scheduling the next one */

if (empty_queue) {

qemu_mutex_lock(&tg->lock);

- schedule_next_request(tgm, is_write);

+ schedule_next_request(tgm, direction);

qemu_mutex_unlock(&tg->lock);

}

@@ -XXX,XX +XXX,XX @@ static void coroutine_fn throttle_group_restart_queue_entry(void *opaque)

aio_wait_kick();

}

-static void throttle_group_restart_queue(ThrottleGroupMember *tgm, bool is_write)

+static void throttle_group_restart_queue(ThrottleGroupMember *tgm,

+ ThrottleDirection direction)

{

Coroutine *co;

RestartData *rd = g_new0(RestartData, 1);

rd->tgm = tgm;

- rd->is_write = is_write;

+ rd->direction = direction;

/* This function is called when a timer is fired or when

* throttle_group_restart_tgm() is called. Either way, there can

* be no timer pending on this tgm at this point */

- assert(!timer_pending(tgm->throttle_timers.timers[is_write]));

+ assert(!timer_pending(tgm->throttle_timers.timers[direction]));

qatomic_inc(&tgm->restart_pending);

@@ -XXX,XX +XXX,XX @@ static void throttle_group_restart_queue(ThrottleGroupMember *tgm, bool is_write

void throttle_group_restart_tgm(ThrottleGroupMember *tgm)

{

- int i;

+ ThrottleDirection dir;

if (tgm->throttle_state) {

- for (i = 0; i < 2; i++) {

- QEMUTimer *t = tgm->throttle_timers.timers[i];

+ for (dir = THROTTLE_READ; dir < THROTTLE_MAX; dir++) {

+ QEMUTimer *t = tgm->throttle_timers.timers[dir];

if (timer_pending(t)) {

/* If there's a pending timer on this tgm, fire it now */

timer_del(t);

- timer_cb(tgm, i);

+ timer_cb(tgm, dir);

} else {

/* Else run the next request from the queue manually */

- throttle_group_restart_queue(tgm, i);

+ throttle_group_restart_queue(tgm, dir);

}

}