https://github.com/torvalds/linux
Revision 4fca50d440cc5d4dc570ad5484cc0b70b381bc2a authored by Jan Kara on 08 September 2022, 09:21:24 UTC, committed by Theodore Ts'o on 22 September 2022, 02:11:34 UTC
One of the side-effects of mb_optimize_scan was that the optimized functions to select next group to try were called even before we tried the goal group. As a result we no longer allocate files close to corresponding inodes as well as we don't try to expand currently allocated extent in the same group. This results in reaim regression with workfile.disk workload of upto 8% with many clients on my test machine: baseline mb_optimize_scan Hmean disk-1 2114.16 ( 0.00%) 2099.37 ( -0.70%) Hmean disk-41 87794.43 ( 0.00%) 83787.47 * -4.56%* Hmean disk-81 148170.73 ( 0.00%) 135527.05 * -8.53%* Hmean disk-121 177506.11 ( 0.00%) 166284.93 * -6.32%* Hmean disk-161 220951.51 ( 0.00%) 207563.39 * -6.06%* Hmean disk-201 208722.74 ( 0.00%) 203235.59 ( -2.63%) Hmean disk-241 222051.60 ( 0.00%) 217705.51 ( -1.96%) Hmean disk-281 252244.17 ( 0.00%) 241132.72 * -4.41%* Hmean disk-321 255844.84 ( 0.00%) 245412.84 * -4.08%* Also this is causing huge regression (time increased by a factor of 5 or so) when untarring archive with lots of small files on some eMMC storage cards. Fix the problem by making sure we try goal group first. Fixes: 196e402adf2e ("ext4: improve cr 0 / cr 1 group scanning") CC: stable@kernel.org Reported-and-tested-by: Stefan Wahren <stefan.wahren@i2se.com> Tested-by: Ojaswin Mujoo <ojaswin@linux.ibm.com> Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com> Link: https://lore.kernel.org/all/20220727105123.ckwrhbilzrxqpt24@quack3/ Link: https://lore.kernel.org/all/0d81a7c2-46b7-6010-62a4-3e6cfc1628d6@i2se.com/ Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20220908092136.11770-1-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
1 parent 7e18e42
Tip revision: 4fca50d440cc5d4dc570ad5484cc0b70b381bc2a authored by Jan Kara on 08 September 2022, 09:21:24 UTC
ext4: make mballoc try target group first even with mb_optimize_scan
ext4: make mballoc try target group first even with mb_optimize_scan
Tip revision: 4fca50d
algboss.c
// SPDX-License-Identifier: GPL-2.0-or-later
/*
* Create default crypto algorithm instances.
*
* Copyright (c) 2006 Herbert Xu <herbert@gondor.apana.org.au>
*/
#include <crypto/internal/aead.h>
#include <linux/completion.h>
#include <linux/ctype.h>
#include <linux/err.h>
#include <linux/init.h>
#include <linux/kthread.h>
#include <linux/module.h>
#include <linux/notifier.h>
#include <linux/rtnetlink.h>
#include <linux/sched/signal.h>
#include <linux/slab.h>
#include <linux/string.h>
#include "internal.h"
struct cryptomgr_param {
struct rtattr *tb[CRYPTO_MAX_ATTRS + 2];
struct {
struct rtattr attr;
struct crypto_attr_type data;
} type;
struct {
struct rtattr attr;
struct crypto_attr_alg data;
} attrs[CRYPTO_MAX_ATTRS];
char template[CRYPTO_MAX_ALG_NAME];
struct crypto_larval *larval;
u32 otype;
u32 omask;
};
struct crypto_test_param {
char driver[CRYPTO_MAX_ALG_NAME];
char alg[CRYPTO_MAX_ALG_NAME];
u32 type;
};
static int cryptomgr_probe(void *data)
{
struct cryptomgr_param *param = data;
struct crypto_template *tmpl;
int err;
tmpl = crypto_lookup_template(param->template);
if (!tmpl)
goto out;
do {
err = tmpl->create(tmpl, param->tb);
} while (err == -EAGAIN && !signal_pending(current));
crypto_tmpl_put(tmpl);
out:
complete_all(¶m->larval->completion);
crypto_alg_put(¶m->larval->alg);
kfree(param);
module_put_and_kthread_exit(0);
}
static int cryptomgr_schedule_probe(struct crypto_larval *larval)
{
struct task_struct *thread;
struct cryptomgr_param *param;
const char *name = larval->alg.cra_name;
const char *p;
unsigned int len;
int i;
if (!try_module_get(THIS_MODULE))
goto err;
param = kzalloc(sizeof(*param), GFP_KERNEL);
if (!param)
goto err_put_module;
for (p = name; isalnum(*p) || *p == '-' || *p == '_'; p++)
;
len = p - name;
if (!len || *p != '(')
goto err_free_param;
memcpy(param->template, name, len);
i = 0;
for (;;) {
name = ++p;
for (; isalnum(*p) || *p == '-' || *p == '_'; p++)
;
if (*p == '(') {
int recursion = 0;
for (;;) {
if (!*++p)
goto err_free_param;
if (*p == '(')
recursion++;
else if (*p == ')' && !recursion--)
break;
}
p++;
}
len = p - name;
if (!len)
goto err_free_param;
param->attrs[i].attr.rta_len = sizeof(param->attrs[i]);
param->attrs[i].attr.rta_type = CRYPTOA_ALG;
memcpy(param->attrs[i].data.name, name, len);
param->tb[i + 1] = ¶m->attrs[i].attr;
i++;
if (i >= CRYPTO_MAX_ATTRS)
goto err_free_param;
if (*p == ')')
break;
if (*p != ',')
goto err_free_param;
}
if (!i)
goto err_free_param;
param->tb[i + 1] = NULL;
param->type.attr.rta_len = sizeof(param->type);
param->type.attr.rta_type = CRYPTOA_TYPE;
param->type.data.type = larval->alg.cra_flags & ~CRYPTO_ALG_TESTED;
param->type.data.mask = larval->mask & ~CRYPTO_ALG_TESTED;
param->tb[0] = ¶m->type.attr;
param->otype = larval->alg.cra_flags;
param->omask = larval->mask;
crypto_alg_get(&larval->alg);
param->larval = larval;
thread = kthread_run(cryptomgr_probe, param, "cryptomgr_probe");
if (IS_ERR(thread))
goto err_put_larval;
return NOTIFY_STOP;
err_put_larval:
crypto_alg_put(&larval->alg);
err_free_param:
kfree(param);
err_put_module:
module_put(THIS_MODULE);
err:
return NOTIFY_OK;
}
static int cryptomgr_test(void *data)
{
struct crypto_test_param *param = data;
u32 type = param->type;
int err = 0;
#ifdef CONFIG_CRYPTO_MANAGER_DISABLE_TESTS
goto skiptest;
#endif
if (type & CRYPTO_ALG_TESTED)
goto skiptest;
err = alg_test(param->driver, param->alg, type, CRYPTO_ALG_TESTED);
skiptest:
crypto_alg_tested(param->driver, err);
kfree(param);
module_put_and_kthread_exit(0);
}
static int cryptomgr_schedule_test(struct crypto_alg *alg)
{
struct task_struct *thread;
struct crypto_test_param *param;
u32 type;
if (!try_module_get(THIS_MODULE))
goto err;
param = kzalloc(sizeof(*param), GFP_KERNEL);
if (!param)
goto err_put_module;
memcpy(param->driver, alg->cra_driver_name, sizeof(param->driver));
memcpy(param->alg, alg->cra_name, sizeof(param->alg));
type = alg->cra_flags;
/* Do not test internal algorithms. */
if (type & CRYPTO_ALG_INTERNAL)
type |= CRYPTO_ALG_TESTED;
param->type = type;
thread = kthread_run(cryptomgr_test, param, "cryptomgr_test");
if (IS_ERR(thread))
goto err_free_param;
return NOTIFY_STOP;
err_free_param:
kfree(param);
err_put_module:
module_put(THIS_MODULE);
err:
return NOTIFY_OK;
}
static int cryptomgr_notify(struct notifier_block *this, unsigned long msg,
void *data)
{
switch (msg) {
case CRYPTO_MSG_ALG_REQUEST:
return cryptomgr_schedule_probe(data);
case CRYPTO_MSG_ALG_REGISTER:
return cryptomgr_schedule_test(data);
case CRYPTO_MSG_ALG_LOADED:
break;
}
return NOTIFY_DONE;
}
static struct notifier_block cryptomgr_notifier = {
.notifier_call = cryptomgr_notify,
};
static int __init cryptomgr_init(void)
{
return crypto_register_notifier(&cryptomgr_notifier);
}
static void __exit cryptomgr_exit(void)
{
int err = crypto_unregister_notifier(&cryptomgr_notifier);
BUG_ON(err);
}
/*
* This is arch_initcall() so that the crypto self-tests are run on algorithms
* registered early by subsys_initcall(). subsys_initcall() is needed for
* generic implementations so that they're available for comparison tests when
* other implementations are registered later by module_init().
*/
arch_initcall(cryptomgr_init);
module_exit(cryptomgr_exit);
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("Crypto Algorithm Manager");
Computing file changes ...