[linux-2.6-block.git] / drivers / lightnvm / pblk-rl.c

// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (C) 2016 CNEX Labs
 * Initial release: Javier Gonzalez <javier@cnexlabs.com>
 *                  Matias Bjorling <matias@cnexlabs.com>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License version
 * 2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * pblk-rl.c - pblk's rate limiter for user I/O
 *
 */

#include "pblk.h"

static void pblk_rl_kick_u_timer(struct pblk_rl *rl)
{
	mod_timer(&rl->u_timer, jiffies + msecs_to_jiffies(5000));
}

int pblk_rl_is_limit(struct pblk_rl *rl)
{
	int rb_space;

	rb_space = atomic_read(&rl->rb_space);

	return (rb_space == 0);
}

int pblk_rl_user_may_insert(struct pblk_rl *rl, int nr_entries)
{
	int rb_user_cnt = atomic_read(&rl->rb_user_cnt);
	int rb_space = atomic_read(&rl->rb_space);

	if (unlikely(rb_space >= 0) && (rb_space - nr_entries < 0))
		return NVM_IO_ERR;

	if (rb_user_cnt >= rl->rb_user_max)
		return NVM_IO_REQUEUE;

	return NVM_IO_OK;
}

void pblk_rl_inserted(struct pblk_rl *rl, int nr_entries)
{
	int rb_space = atomic_read(&rl->rb_space);

	if (unlikely(rb_space >= 0))
		atomic_sub(nr_entries, &rl->rb_space);
}

int pblk_rl_gc_may_insert(struct pblk_rl *rl, int nr_entries)
{
	int rb_gc_cnt = atomic_read(&rl->rb_gc_cnt);
	int rb_user_active;

	/* If there is no user I/O let GC take over space on the write buffer */
	rb_user_active = READ_ONCE(rl->rb_user_active);
	return (!(rb_gc_cnt >= rl->rb_gc_max && rb_user_active));
}

void pblk_rl_user_in(struct pblk_rl *rl, int nr_entries)
{
	atomic_add(nr_entries, &rl->rb_user_cnt);

	/* Release user I/O state. Protect from GC */
	smp_store_release(&rl->rb_user_active, 1);
	pblk_rl_kick_u_timer(rl);
}

void pblk_rl_werr_line_in(struct pblk_rl *rl)
{
	atomic_inc(&rl->werr_lines);
}

void pblk_rl_werr_line_out(struct pblk_rl *rl)
{
	atomic_dec(&rl->werr_lines);
}

void pblk_rl_gc_in(struct pblk_rl *rl, int nr_entries)
{
	atomic_add(nr_entries, &rl->rb_gc_cnt);
}

void pblk_rl_out(struct pblk_rl *rl, int nr_user, int nr_gc)
{
	atomic_sub(nr_user, &rl->rb_user_cnt);
	atomic_sub(nr_gc, &rl->rb_gc_cnt);
}

unsigned long pblk_rl_nr_free_blks(struct pblk_rl *rl)
{
	return atomic_read(&rl->free_blocks);
}

unsigned long pblk_rl_nr_user_free_blks(struct pblk_rl *rl)
{
	return atomic_read(&rl->free_user_blocks);
}

static void __pblk_rl_update_rates(struct pblk_rl *rl,
				   unsigned long free_blocks)
{
	struct pblk *pblk = container_of(rl, struct pblk, rl);
	int max = rl->rb_budget;
	int werr_gc_needed = atomic_read(&rl->werr_lines);

	if (free_blocks >= rl->high) {
		if (werr_gc_needed) {
			/* Allocate a small budget for recovering
			 * lines with write errors
			 */
			rl->rb_gc_max = 1 << rl->rb_windows_pw;
			rl->rb_user_max = max - rl->rb_gc_max;
			rl->rb_state = PBLK_RL_WERR;
		} else {
			rl->rb_user_max = max;
			rl->rb_gc_max = 0;
			rl->rb_state = PBLK_RL_OFF;
		}
	} else if (free_blocks < rl->high) {
		int shift = rl->high_pw - rl->rb_windows_pw;
		int user_windows = free_blocks >> shift;
		int user_max = user_windows << ilog2(NVM_MAX_VLBA);

		rl->rb_user_max = user_max;
		rl->rb_gc_max = max - user_max;

		if (free_blocks <= rl->rsv_blocks) {
			rl->rb_user_max = 0;
			rl->rb_gc_max = max;
		}

		/* In the worst case, we will need to GC lines in the low list
		 * (high valid sector count). If there are lines to GC on high
		 * or mid lists, these will be prioritized
		 */
		rl->rb_state = PBLK_RL_LOW;
	}

	if (rl->rb_state != PBLK_RL_OFF)
		pblk_gc_should_start(pblk);
	else
		pblk_gc_should_stop(pblk);
}

void pblk_rl_update_rates(struct pblk_rl *rl)
{
	__pblk_rl_update_rates(rl, pblk_rl_nr_user_free_blks(rl));
}

void pblk_rl_free_lines_inc(struct pblk_rl *rl, struct pblk_line *line)
{
	int blk_in_line = atomic_read(&line->blk_in_line);
	int free_blocks;

	atomic_add(blk_in_line, &rl->free_blocks);
	free_blocks = atomic_add_return(blk_in_line, &rl->free_user_blocks);

	__pblk_rl_update_rates(rl, free_blocks);
}

void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line,
			    bool used)
{
	int blk_in_line = atomic_read(&line->blk_in_line);
	int free_blocks;

	atomic_sub(blk_in_line, &rl->free_blocks);

	if (used)
		free_blocks = atomic_sub_return(blk_in_line,
							&rl->free_user_blocks);
	else
		free_blocks = atomic_read(&rl->free_user_blocks);

	__pblk_rl_update_rates(rl, free_blocks);
}

int pblk_rl_high_thrs(struct pblk_rl *rl)
{
	return rl->high;
}

int pblk_rl_max_io(struct pblk_rl *rl)
{
	return rl->rb_max_io;
}

static void pblk_rl_u_timer(struct timer_list *t)
{
	struct pblk_rl *rl = from_timer(rl, t, u_timer);

	/* Release user I/O state. Protect from GC */
	smp_store_release(&rl->rb_user_active, 0);
}

void pblk_rl_free(struct pblk_rl *rl)
{
	del_timer(&rl->u_timer);
}

void pblk_rl_init(struct pblk_rl *rl, int budget)
{
	struct pblk *pblk = container_of(rl, struct pblk, rl);
	struct nvm_tgt_dev *dev = pblk->dev;
	struct nvm_geo *geo = &dev->geo;
	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
	struct pblk_line_meta *lm = &pblk->lm;
	int min_blocks = lm->blk_per_line * PBLK_GC_RSV_LINE;
	int sec_meta, blk_meta;

	unsigned int rb_windows;

	/* Consider sectors used for metadata */
	sec_meta = (lm->smeta_sec + lm->emeta_sec[0]) * l_mg->nr_free_lines;
	blk_meta = DIV_ROUND_UP(sec_meta, geo->clba);

	rl->high = pblk->op_blks - blk_meta - lm->blk_per_line;
	rl->high_pw = get_count_order(rl->high);

	rl->rsv_blocks = min_blocks;

	/* This will always be a power-of-2 */
	rb_windows = budget / NVM_MAX_VLBA;
	rl->rb_windows_pw = get_count_order(rb_windows);

	/* To start with, all buffer is available to user I/O writers */
	rl->rb_budget = budget;
	rl->rb_user_max = budget;
	rl->rb_max_io = budget >> 1;
	rl->rb_gc_max = 0;
	rl->rb_state = PBLK_RL_HIGH;

	atomic_set(&rl->rb_user_cnt, 0);
	atomic_set(&rl->rb_gc_cnt, 0);
	atomic_set(&rl->rb_space, -1);
	atomic_set(&rl->werr_lines, 0);

	timer_setup(&rl->u_timer, pblk_rl_u_timer, 0);

	rl->rb_user_active = 0;
	rl->rb_gc_active = 0;
}
Commit	Line	Data
02a1520d	1	// SPDX-License-Identifier: GPL-2.0
a4bd217b JG	2	/*
	3	* Copyright (C) 2016 CNEX Labs
	4	* Initial release: Javier Gonzalez <javier@cnexlabs.com>
	5	* Matias Bjorling <matias@cnexlabs.com>
	6	*
	7	* This program is free software; you can redistribute it and/or
	8	* modify it under the terms of the GNU General Public License version
	9	* 2 as published by the Free Software Foundation.
	10	*
	11	* This program is distributed in the hope that it will be useful, but
	12	* WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	14	* General Public License for more details.
	15	*
	16	* pblk-rl.c - pblk's rate limiter for user I/O
	17	*
	18	*/
	19
	20	#include "pblk.h"
	21
	22	static void pblk_rl_kick_u_timer(struct pblk_rl *rl)
	23	{
	24	mod_timer(&rl->u_timer, jiffies + msecs_to_jiffies(5000));
	25	}
	26
588726d3 JG	27	int pblk_rl_is_limit(struct pblk_rl *rl)
	28	{
	29	int rb_space;
	30
	31	rb_space = atomic_read(&rl->rb_space);
	32
	33	return (rb_space == 0);
	34	}
	35
a4bd217b JG	36	int pblk_rl_user_may_insert(struct pblk_rl *rl, int nr_entries)
	37	{
	38	int rb_user_cnt = atomic_read(&rl->rb_user_cnt);
588726d3 JG	39	int rb_space = atomic_read(&rl->rb_space);
	40
	41	if (unlikely(rb_space >= 0) && (rb_space - nr_entries < 0))
	42	return NVM_IO_ERR;
	43
	44	if (rb_user_cnt >= rl->rb_user_max)
	45	return NVM_IO_REQUEUE;
a4bd217b	46
588726d3 JG	47	return NVM_IO_OK;
	48	}
	49
	50	void pblk_rl_inserted(struct pblk_rl *rl, int nr_entries)
	51	{
	52	int rb_space = atomic_read(&rl->rb_space);
	53
	54	if (unlikely(rb_space >= 0))
	55	atomic_sub(nr_entries, &rl->rb_space);
a4bd217b JG	56	}
	57
	58	int pblk_rl_gc_may_insert(struct pblk_rl *rl, int nr_entries)
	59	{
	60	int rb_gc_cnt = atomic_read(&rl->rb_gc_cnt);
	61	int rb_user_active;
	62
	63	/* If there is no user I/O let GC take over space on the write buffer */
	64	rb_user_active = READ_ONCE(rl->rb_user_active);
b20ba1bc	65	return (!(rb_gc_cnt >= rl->rb_gc_max && rb_user_active));
a4bd217b JG	66	}
	67
	68	void pblk_rl_user_in(struct pblk_rl *rl, int nr_entries)
	69	{
	70	atomic_add(nr_entries, &rl->rb_user_cnt);
	71
	72	/* Release user I/O state. Protect from GC */
	73	smp_store_release(&rl->rb_user_active, 1);
	74	pblk_rl_kick_u_timer(rl);
	75	}
	76
48b8d208 HH	77	void pblk_rl_werr_line_in(struct pblk_rl *rl)
	78	{
	79	atomic_inc(&rl->werr_lines);
	80	}
	81
	82	void pblk_rl_werr_line_out(struct pblk_rl *rl)
	83	{
	84	atomic_dec(&rl->werr_lines);
	85	}
	86
a4bd217b JG	87	void pblk_rl_gc_in(struct pblk_rl *rl, int nr_entries)
	88	{
	89	atomic_add(nr_entries, &rl->rb_gc_cnt);
	90	}
	91
	92	void pblk_rl_out(struct pblk_rl *rl, int nr_user, int nr_gc)
	93	{
	94	atomic_sub(nr_user, &rl->rb_user_cnt);
	95	atomic_sub(nr_gc, &rl->rb_gc_cnt);
	96	}
	97
	98	unsigned long pblk_rl_nr_free_blks(struct pblk_rl *rl)
	99	{
	100	return atomic_read(&rl->free_blocks);
	101	}
	102
a7689938 JG	103	unsigned long pblk_rl_nr_user_free_blks(struct pblk_rl *rl)
	104	{
	105	return atomic_read(&rl->free_user_blocks);
	106	}
	107
	108	static void __pblk_rl_update_rates(struct pblk_rl *rl,
	109	unsigned long free_blocks)
a4bd217b	110	{
32825ebb	111	struct pblk *pblk = container_of(rl, struct pblk, rl);
32825ebb	112	int max = rl->rb_budget;
48b8d208	113	int werr_gc_needed = atomic_read(&rl->werr_lines);
a4bd217b JG	114
a4bd217b JG	115	if (free_blocks >= rl->high) {
48b8d208 HH	116	if (werr_gc_needed) {
	117	/* Allocate a small budget for recovering
	118	* lines with write errors
	119	*/
	120	rl->rb_gc_max = 1 << rl->rb_windows_pw;
	121	rl->rb_user_max = max - rl->rb_gc_max;
	122	rl->rb_state = PBLK_RL_WERR;
	123	} else {
	124	rl->rb_user_max = max;
	125	rl->rb_gc_max = 0;
	126	rl->rb_state = PBLK_RL_OFF;
	127	}
a4bd217b JG	128	} else if (free_blocks < rl->high) {
	129	int shift = rl->high_pw - rl->rb_windows_pw;
	130	int user_windows = free_blocks >> shift;
afdc23c9	131	int user_max = user_windows << ilog2(NVM_MAX_VLBA);
a4bd217b JG	132
a4bd217b JG	133	rl->rb_user_max = user_max;
b20ba1bc JG	134	rl->rb_gc_max = max - user_max;
	135
	136	if (free_blocks <= rl->rsv_blocks) {
	137	rl->rb_user_max = 0;
	138	rl->rb_gc_max = max;
	139	}
	140
	141	/* In the worst case, we will need to GC lines in the low list
	142	* (high valid sector count). If there are lines to GC on high
	143	* or mid lists, these will be prioritized
	144	*/
	145	rl->rb_state = PBLK_RL_LOW;
a4bd217b JG	146	}
a4bd217b JG	147
48b8d208	148	if (rl->rb_state != PBLK_RL_OFF)
32825ebb RP	149	pblk_gc_should_start(pblk);
	150	else
	151	pblk_gc_should_stop(pblk);
a4bd217b JG	152	}
a4bd217b JG	153
a7689938 JG	154	void pblk_rl_update_rates(struct pblk_rl *rl)
	155	{
	156	__pblk_rl_update_rates(rl, pblk_rl_nr_user_free_blks(rl));
	157	}
	158
a4bd217b JG	159	void pblk_rl_free_lines_inc(struct pblk_rl rl, struct pblk_line line)
a4bd217b JG	160	{
a44f53fa	161	int blk_in_line = atomic_read(&line->blk_in_line);
a7689938	162	int free_blocks;
a4bd217b	163
a44f53fa	164	atomic_add(blk_in_line, &rl->free_blocks);
a7689938 JG	165	free_blocks = atomic_add_return(blk_in_line, &rl->free_user_blocks);
	166
	167	__pblk_rl_update_rates(rl, free_blocks);
a4bd217b JG	168	}
a4bd217b JG	169
a7689938 JG	170	void pblk_rl_free_lines_dec(struct pblk_rl rl, struct pblk_line line,
a7689938 JG	171	bool used)
a4bd217b	172	{
a44f53fa	173	int blk_in_line = atomic_read(&line->blk_in_line);
a7689938	174	int free_blocks;
a4bd217b	175
a44f53fa	176	atomic_sub(blk_in_line, &rl->free_blocks);
a7689938 JG	177
	178	if (used)
	179	free_blocks = atomic_sub_return(blk_in_line,
	180	&rl->free_user_blocks);
	181	else
	182	free_blocks = atomic_read(&rl->free_user_blocks);
	183
	184	__pblk_rl_update_rates(rl, free_blocks);
a4bd217b JG	185	}
a4bd217b JG	186
b20ba1bc	187	int pblk_rl_high_thrs(struct pblk_rl *rl)
a4bd217b JG	188	{
	189	return rl->high;
	190	}
	191
da67e68f JG	192	int pblk_rl_max_io(struct pblk_rl *rl)
	193	{
	194	return rl->rb_max_io;
	195	}
	196
87c1d2d3	197	static void pblk_rl_u_timer(struct timer_list *t)
a4bd217b	198	{
87c1d2d3	199	struct pblk_rl *rl = from_timer(rl, t, u_timer);
a4bd217b JG	200
	201	/* Release user I/O state. Protect from GC */
	202	smp_store_release(&rl->rb_user_active, 0);
	203	}
	204
	205	void pblk_rl_free(struct pblk_rl *rl)
	206	{
	207	del_timer(&rl->u_timer);
	208	}
	209
	210	void pblk_rl_init(struct pblk_rl *rl, int budget)
	211	{
b20ba1bc	212	struct pblk *pblk = container_of(rl, struct pblk, rl);
a7689938 JG	213	struct nvm_tgt_dev *dev = pblk->dev;
	214	struct nvm_geo *geo = &dev->geo;
	215	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
b20ba1bc JG	216	struct pblk_line_meta *lm = &pblk->lm;
b20ba1bc JG	217	int min_blocks = lm->blk_per_line * PBLK_GC_RSV_LINE;
a7689938 JG	218	int sec_meta, blk_meta;
a7689938 JG	219
a4bd217b JG	220	unsigned int rb_windows;
a4bd217b JG	221
a7689938 JG	222	/* Consider sectors used for metadata */
a7689938 JG	223	sec_meta = (lm->smeta_sec + lm->emeta_sec[0]) * l_mg->nr_free_lines;
e46f4e48	224	blk_meta = DIV_ROUND_UP(sec_meta, geo->clba);
a4bd217b	225
a7689938 JG	226	rl->high = pblk->op_blks - blk_meta - lm->blk_per_line;
a7689938 JG	227	rl->high_pw = get_count_order(rl->high);
b20ba1bc JG	228
	229	rl->rsv_blocks = min_blocks;
	230
a4bd217b	231	/* This will always be a power-of-2 */
afdc23c9	232	rb_windows = budget / NVM_MAX_VLBA;
b20ba1bc	233	rl->rb_windows_pw = get_count_order(rb_windows);
a4bd217b JG	234
	235	/* To start with, all buffer is available to user I/O writers */
	236	rl->rb_budget = budget;
	237	rl->rb_user_max = budget;
da67e68f	238	rl->rb_max_io = budget >> 1;
a4bd217b JG	239	rl->rb_gc_max = 0;
a4bd217b JG	240	rl->rb_state = PBLK_RL_HIGH;
588726d3 JG	241
588726d3 JG	242	atomic_set(&rl->rb_user_cnt, 0);
a4bd217b	243	atomic_set(&rl->rb_gc_cnt, 0);
588726d3	244	atomic_set(&rl->rb_space, -1);
48b8d208	245	atomic_set(&rl->werr_lines, 0);
a4bd217b	246
87c1d2d3	247	timer_setup(&rl->u_timer, pblk_rl_u_timer, 0);
b20ba1bc	248
a4bd217b	249	rl->rb_user_active = 0;
b20ba1bc	250	rl->rb_gc_active = 0;
a4bd217b	251	}