1// SPDX-License-Identifier: GPL-2.0
   2/*
   3 * Copyright (c) 2010 Red Hat, Inc. All Rights Reserved.
 
 
 
 
 
 
 
 
 
 
 
 
 
   4 */
   5
   6#include "xfs.h"
   7#include "xfs_fs.h"
   8#include "xfs_format.h"
   9#include "xfs_log_format.h"
  10#include "xfs_shared.h"
  11#include "xfs_trans_resv.h"
  12#include "xfs_mount.h"
 
 
  13#include "xfs_extent_busy.h"
 
  14#include "xfs_trans.h"
  15#include "xfs_trans_priv.h"
  16#include "xfs_log.h"
  17#include "xfs_log_priv.h"
  18#include "xfs_trace.h"
  19
  20struct workqueue_struct *xfs_discard_wq;
  21
  22/*
  23 * Allocate a new ticket. Failing to get a new ticket makes it really hard to
  24 * recover, so we don't allow failure here. Also, we allocate in a context that
  25 * we don't want to be issuing transactions from, so we need to tell the
  26 * allocation code this as well.
  27 *
  28 * We don't reserve any space for the ticket - we are going to steal whatever
  29 * space we require from transactions as they commit. To ensure we reserve all
  30 * the space required, we need to set the current reservation of the ticket to
  31 * zero so that we know to steal the initial transaction overhead from the
  32 * first transaction commit.
  33 */
  34static struct xlog_ticket *
  35xlog_cil_ticket_alloc(
  36	struct xlog	*log)
  37{
  38	struct xlog_ticket *tic;
  39
  40	tic = xlog_ticket_alloc(log, 0, 1, 0);
 
 
  41
  42	/*
  43	 * set the current reservation to zero so we know to steal the basic
  44	 * transaction overhead reservation from the first transaction commit.
  45	 */
  46	tic->t_curr_res = 0;
  47	tic->t_iclog_hdrs = 0;
  48	return tic;
  49}
  50
  51static inline void
  52xlog_cil_set_iclog_hdr_count(struct xfs_cil *cil)
  53{
  54	struct xlog	*log = cil->xc_log;
  55
  56	atomic_set(&cil->xc_iclog_hdrs,
  57		   (XLOG_CIL_BLOCKING_SPACE_LIMIT(log) /
  58			(log->l_iclog_size - log->l_iclog_hsize)));
  59}
  60
  61/*
  62 * Check if the current log item was first committed in this sequence.
  63 * We can't rely on just the log item being in the CIL, we have to check
  64 * the recorded commit sequence number.
  65 *
  66 * Note: for this to be used in a non-racy manner, it has to be called with
  67 * CIL flushing locked out. As a result, it should only be used during the
  68 * transaction commit process when deciding what to format into the item.
  69 */
  70static bool
  71xlog_item_in_current_chkpt(
  72	struct xfs_cil		*cil,
  73	struct xfs_log_item	*lip)
  74{
  75	if (test_bit(XLOG_CIL_EMPTY, &cil->xc_flags))
  76		return false;
  77
  78	/*
  79	 * li_seq is written on the first commit of a log item to record the
  80	 * first checkpoint it is written to. Hence if it is different to the
  81	 * current sequence, we're in a new checkpoint.
  82	 */
  83	return lip->li_seq == READ_ONCE(cil->xc_current_sequence);
  84}
  85
  86bool
  87xfs_log_item_in_current_chkpt(
  88	struct xfs_log_item *lip)
  89{
  90	return xlog_item_in_current_chkpt(lip->li_log->l_cilp, lip);
  91}
  92
  93/*
  94 * Unavoidable forward declaration - xlog_cil_push_work() calls
  95 * xlog_cil_ctx_alloc() itself.
  96 */
  97static void xlog_cil_push_work(struct work_struct *work);
  98
  99static struct xfs_cil_ctx *
 100xlog_cil_ctx_alloc(void)
 101{
 102	struct xfs_cil_ctx	*ctx;
 103
 104	ctx = kmem_zalloc(sizeof(*ctx), KM_NOFS);
 105	INIT_LIST_HEAD(&ctx->committing);
 106	INIT_LIST_HEAD(&ctx->busy_extents);
 107	INIT_LIST_HEAD(&ctx->log_items);
 108	INIT_LIST_HEAD(&ctx->lv_chain);
 109	INIT_WORK(&ctx->push_work, xlog_cil_push_work);
 110	return ctx;
 111}
 112
 113/*
 114 * Aggregate the CIL per cpu structures into global counts, lists, etc and
 115 * clear the percpu state ready for the next context to use. This is called
 116 * from the push code with the context lock held exclusively, hence nothing else
 117 * will be accessing or modifying the per-cpu counters.
 118 */
 119static void
 120xlog_cil_push_pcp_aggregate(
 121	struct xfs_cil		*cil,
 122	struct xfs_cil_ctx	*ctx)
 123{
 124	struct xlog_cil_pcp	*cilpcp;
 125	int			cpu;
 126
 127	for_each_online_cpu(cpu) {
 128		cilpcp = per_cpu_ptr(cil->xc_pcp, cpu);
 129
 130		ctx->ticket->t_curr_res += cilpcp->space_reserved;
 131		cilpcp->space_reserved = 0;
 132
 133		if (!list_empty(&cilpcp->busy_extents)) {
 134			list_splice_init(&cilpcp->busy_extents,
 135					&ctx->busy_extents);
 136		}
 137		if (!list_empty(&cilpcp->log_items))
 138			list_splice_init(&cilpcp->log_items, &ctx->log_items);
 139
 140		/*
 141		 * We're in the middle of switching cil contexts.  Reset the
 142		 * counter we use to detect when the current context is nearing
 143		 * full.
 144		 */
 145		cilpcp->space_used = 0;
 146	}
 147}
 148
 149/*
 150 * Aggregate the CIL per-cpu space used counters into the global atomic value.
 151 * This is called when the per-cpu counter aggregation will first pass the soft
 152 * limit threshold so we can switch to atomic counter aggregation for accurate
 153 * detection of hard limit traversal.
 154 */
 155static void
 156xlog_cil_insert_pcp_aggregate(
 157	struct xfs_cil		*cil,
 158	struct xfs_cil_ctx	*ctx)
 159{
 160	struct xlog_cil_pcp	*cilpcp;
 161	int			cpu;
 162	int			count = 0;
 163
 164	/* Trigger atomic updates then aggregate only for the first caller */
 165	if (!test_and_clear_bit(XLOG_CIL_PCP_SPACE, &cil->xc_flags))
 166		return;
 167
 168	for_each_online_cpu(cpu) {
 169		int	old, prev;
 170
 171		cilpcp = per_cpu_ptr(cil->xc_pcp, cpu);
 172		do {
 173			old = cilpcp->space_used;
 174			prev = cmpxchg(&cilpcp->space_used, old, 0);
 175		} while (old != prev);
 176		count += old;
 177	}
 178	atomic_add(count, &ctx->space_used);
 179}
 180
 181static void
 182xlog_cil_ctx_switch(
 183	struct xfs_cil		*cil,
 184	struct xfs_cil_ctx	*ctx)
 185{
 186	xlog_cil_set_iclog_hdr_count(cil);
 187	set_bit(XLOG_CIL_EMPTY, &cil->xc_flags);
 188	set_bit(XLOG_CIL_PCP_SPACE, &cil->xc_flags);
 189	ctx->sequence = ++cil->xc_current_sequence;
 190	ctx->cil = cil;
 191	cil->xc_ctx = ctx;
 192}
 193
 194/*
 195 * After the first stage of log recovery is done, we know where the head and
 196 * tail of the log are. We need this log initialisation done before we can
 197 * initialise the first CIL checkpoint context.
 198 *
 199 * Here we allocate a log ticket to track space usage during a CIL push.  This
 200 * ticket is passed to xlog_write() directly so that we don't slowly leak log
 201 * space by failing to account for space used by log headers and additional
 202 * region headers for split regions.
 203 */
 204void
 205xlog_cil_init_post_recovery(
 206	struct xlog	*log)
 207{
 208	log->l_cilp->xc_ctx->ticket = xlog_cil_ticket_alloc(log);
 209	log->l_cilp->xc_ctx->sequence = 1;
 210	xlog_cil_set_iclog_hdr_count(log->l_cilp);
 211}
 212
 213static inline int
 214xlog_cil_iovec_space(
 215	uint	niovecs)
 216{
 217	return round_up((sizeof(struct xfs_log_vec) +
 218					niovecs * sizeof(struct xfs_log_iovec)),
 219			sizeof(uint64_t));
 220}
 221
 222/*
 223 * Allocate or pin log vector buffers for CIL insertion.
 224 *
 225 * The CIL currently uses disposable buffers for copying a snapshot of the
 226 * modified items into the log during a push. The biggest problem with this is
 227 * the requirement to allocate the disposable buffer during the commit if:
 228 *	a) does not exist; or
 229 *	b) it is too small
 230 *
 231 * If we do this allocation within xlog_cil_insert_format_items(), it is done
 232 * under the xc_ctx_lock, which means that a CIL push cannot occur during
 233 * the memory allocation. This means that we have a potential deadlock situation
 234 * under low memory conditions when we have lots of dirty metadata pinned in
 235 * the CIL and we need a CIL commit to occur to free memory.
 236 *
 237 * To avoid this, we need to move the memory allocation outside the
 238 * xc_ctx_lock, but because the log vector buffers are disposable, that opens
 239 * up a TOCTOU race condition w.r.t. the CIL committing and removing the log
 240 * vector buffers between the check and the formatting of the item into the
 241 * log vector buffer within the xc_ctx_lock.
 242 *
 243 * Because the log vector buffer needs to be unchanged during the CIL push
 244 * process, we cannot share the buffer between the transaction commit (which
 245 * modifies the buffer) and the CIL push context that is writing the changes
 246 * into the log. This means skipping preallocation of buffer space is
 247 * unreliable, but we most definitely do not want to be allocating and freeing
 248 * buffers unnecessarily during commits when overwrites can be done safely.
 249 *
 250 * The simplest solution to this problem is to allocate a shadow buffer when a
 251 * log item is committed for the second time, and then to only use this buffer
 252 * if necessary. The buffer can remain attached to the log item until such time
 253 * it is needed, and this is the buffer that is reallocated to match the size of
 254 * the incoming modification. Then during the formatting of the item we can swap
 255 * the active buffer with the new one if we can't reuse the existing buffer. We
 256 * don't free the old buffer as it may be reused on the next modification if
 257 * it's size is right, otherwise we'll free and reallocate it at that point.
 258 *
 259 * This function builds a vector for the changes in each log item in the
 260 * transaction. It then works out the length of the buffer needed for each log
 261 * item, allocates them and attaches the vector to the log item in preparation
 262 * for the formatting step which occurs under the xc_ctx_lock.
 263 *
 264 * While this means the memory footprint goes up, it avoids the repeated
 265 * alloc/free pattern that repeated modifications of an item would otherwise
 266 * cause, and hence minimises the CPU overhead of such behaviour.
 267 */
 268static void
 269xlog_cil_alloc_shadow_bufs(
 270	struct xlog		*log,
 271	struct xfs_trans	*tp)
 272{
 273	struct xfs_log_item	*lip;
 274
 275	list_for_each_entry(lip, &tp->t_items, li_trans) {
 276		struct xfs_log_vec *lv;
 277		int	niovecs = 0;
 278		int	nbytes = 0;
 279		int	buf_size;
 280		bool	ordered = false;
 281
 282		/* Skip items which aren't dirty in this transaction. */
 283		if (!test_bit(XFS_LI_DIRTY, &lip->li_flags))
 284			continue;
 285
 286		/* get number of vecs and size of data to be stored */
 287		lip->li_ops->iop_size(lip, &niovecs, &nbytes);
 288
 289		/*
 290		 * Ordered items need to be tracked but we do not wish to write
 291		 * them. We need a logvec to track the object, but we do not
 292		 * need an iovec or buffer to be allocated for copying data.
 293		 */
 294		if (niovecs == XFS_LOG_VEC_ORDERED) {
 295			ordered = true;
 296			niovecs = 0;
 297			nbytes = 0;
 298		}
 299
 300		/*
 301		 * We 64-bit align the length of each iovec so that the start of
 302		 * the next one is naturally aligned.  We'll need to account for
 303		 * that slack space here.
 304		 *
 305		 * We also add the xlog_op_header to each region when
 306		 * formatting, but that's not accounted to the size of the item
 307		 * at this point. Hence we'll need an addition number of bytes
 308		 * for each vector to hold an opheader.
 309		 *
 310		 * Then round nbytes up to 64-bit alignment so that the initial
 311		 * buffer alignment is easy to calculate and verify.
 312		 */
 313		nbytes += niovecs *
 314			(sizeof(uint64_t) + sizeof(struct xlog_op_header));
 315		nbytes = round_up(nbytes, sizeof(uint64_t));
 316
 317		/*
 318		 * The data buffer needs to start 64-bit aligned, so round up
 319		 * that space to ensure we can align it appropriately and not
 320		 * overrun the buffer.
 321		 */
 322		buf_size = nbytes + xlog_cil_iovec_space(niovecs);
 323
 324		/*
 325		 * if we have no shadow buffer, or it is too small, we need to
 326		 * reallocate it.
 327		 */
 328		if (!lip->li_lv_shadow ||
 329		    buf_size > lip->li_lv_shadow->lv_size) {
 330			/*
 331			 * We free and allocate here as a realloc would copy
 332			 * unnecessary data. We don't use kvzalloc() for the
 333			 * same reason - we don't need to zero the data area in
 334			 * the buffer, only the log vector header and the iovec
 335			 * storage.
 336			 */
 337			kmem_free(lip->li_lv_shadow);
 338			lv = xlog_kvmalloc(buf_size);
 339
 340			memset(lv, 0, xlog_cil_iovec_space(niovecs));
 341
 342			INIT_LIST_HEAD(&lv->lv_list);
 343			lv->lv_item = lip;
 344			lv->lv_size = buf_size;
 345			if (ordered)
 346				lv->lv_buf_len = XFS_LOG_VEC_ORDERED;
 347			else
 348				lv->lv_iovecp = (struct xfs_log_iovec *)&lv[1];
 349			lip->li_lv_shadow = lv;
 350		} else {
 351			/* same or smaller, optimise common overwrite case */
 352			lv = lip->li_lv_shadow;
 353			if (ordered)
 354				lv->lv_buf_len = XFS_LOG_VEC_ORDERED;
 355			else
 356				lv->lv_buf_len = 0;
 357			lv->lv_bytes = 0;
 358		}
 359
 360		/* Ensure the lv is set up according to ->iop_size */
 361		lv->lv_niovecs = niovecs;
 362
 363		/* The allocated data region lies beyond the iovec region */
 364		lv->lv_buf = (char *)lv + xlog_cil_iovec_space(niovecs);
 365	}
 366
 367}
 368
 369/*
 370 * Prepare the log item for insertion into the CIL. Calculate the difference in
 371 * log space it will consume, and if it is a new item pin it as well.
 
 372 */
 373STATIC void
 374xfs_cil_prepare_item(
 375	struct xlog		*log,
 376	struct xfs_log_vec	*lv,
 377	struct xfs_log_vec	*old_lv,
 378	int			*diff_len)
 
 379{
 380	/* Account for the new LV being passed in */
 381	if (lv->lv_buf_len != XFS_LOG_VEC_ORDERED)
 382		*diff_len += lv->lv_bytes;
 
 
 383
 384	/*
 385	 * If there is no old LV, this is the first time we've seen the item in
 386	 * this CIL context and so we need to pin it. If we are replacing the
 387	 * old_lv, then remove the space it accounts for and make it the shadow
 388	 * buffer for later freeing. In both cases we are now switching to the
 389	 * shadow buffer, so update the pointer to it appropriately.
 390	 */
 391	if (!old_lv) {
 392		if (lv->lv_item->li_ops->iop_pin)
 393			lv->lv_item->li_ops->iop_pin(lv->lv_item);
 394		lv->lv_item->li_lv_shadow = NULL;
 395	} else if (old_lv != lv) {
 396		ASSERT(lv->lv_buf_len != XFS_LOG_VEC_ORDERED);
 397
 398		*diff_len -= old_lv->lv_bytes;
 399		lv->lv_item->li_lv_shadow = old_lv;
 
 400	}
 401
 402	/* attach new log vector to log item */
 403	lv->lv_item->li_lv = lv;
 404
 405	/*
 406	 * If this is the first time the item is being committed to the
 407	 * CIL, store the sequence number on the log item so we can
 408	 * tell in future commits whether this is the first checkpoint
 409	 * the item is being committed into.
 410	 */
 411	if (!lv->lv_item->li_seq)
 412		lv->lv_item->li_seq = log->l_cilp->xc_ctx->sequence;
 413}
 414
 415/*
 416 * Format log item into a flat buffers
 417 *
 418 * For delayed logging, we need to hold a formatted buffer containing all the
 419 * changes on the log item. This enables us to relog the item in memory and
 420 * write it out asynchronously without needing to relock the object that was
 421 * modified at the time it gets written into the iclog.
 422 *
 423 * This function takes the prepared log vectors attached to each log item, and
 424 * formats the changes into the log vector buffer. The buffer it uses is
 425 * dependent on the current state of the vector in the CIL - the shadow lv is
 426 * guaranteed to be large enough for the current modification, but we will only
 427 * use that if we can't reuse the existing lv. If we can't reuse the existing
 428 * lv, then simple swap it out for the shadow lv. We don't free it - that is
 429 * done lazily either by th enext modification or the freeing of the log item.
 430 *
 431 * We don't set up region headers during this process; we simply copy the
 432 * regions into the flat buffer. We can do this because we still have to do a
 433 * formatting step to write the regions into the iclog buffer.  Writing the
 434 * ophdrs during the iclog write means that we can support splitting large
 435 * regions across iclog boundares without needing a change in the format of the
 436 * item/region encapsulation.
 437 *
 438 * Hence what we need to do now is change the rewrite the vector array to point
 439 * to the copied region inside the buffer we just allocated. This allows us to
 440 * format the regions into the iclog as though they are being formatted
 441 * directly out of the objects themselves.
 442 */
 443static void
 444xlog_cil_insert_format_items(
 445	struct xlog		*log,
 446	struct xfs_trans	*tp,
 447	int			*diff_len)
 
 448{
 449	struct xfs_log_item	*lip;
 
 450
 451	/* Bail out if we didn't find a log item.  */
 452	if (list_empty(&tp->t_items)) {
 453		ASSERT(0);
 454		return;
 455	}
 456
 457	list_for_each_entry(lip, &tp->t_items, li_trans) {
 
 458		struct xfs_log_vec *lv;
 459		struct xfs_log_vec *old_lv = NULL;
 460		struct xfs_log_vec *shadow;
 
 
 461		bool	ordered = false;
 462
 463		/* Skip items which aren't dirty in this transaction. */
 464		if (!test_bit(XFS_LI_DIRTY, &lip->li_flags))
 
 
 
 
 
 
 
 465			continue;
 466
 467		/*
 468		 * The formatting size information is already attached to
 469		 * the shadow lv on the log item.
 
 470		 */
 471		shadow = lip->li_lv_shadow;
 472		if (shadow->lv_buf_len == XFS_LOG_VEC_ORDERED)
 473			ordered = true;
 
 
 
 474
 475		/* Skip items that do not have any vectors for writing */
 476		if (!shadow->lv_niovecs && !ordered)
 477			continue;
 
 
 
 
 
 
 478
 479		/* compare to existing item size */
 480		old_lv = lip->li_lv;
 481		if (lip->li_lv && shadow->lv_size <= lip->li_lv->lv_size) {
 
 
 
 
 
 
 
 
 
 
 
 
 482			/* same or smaller, optimise common overwrite case */
 483			lv = lip->li_lv;
 
 484
 485			if (ordered)
 486				goto insert;
 487
 488			/*
 489			 * set the item up as though it is a new insertion so
 490			 * that the space reservation accounting is correct.
 491			 */
 
 492			*diff_len -= lv->lv_bytes;
 493
 494			/* Ensure the lv is set up according to ->iop_size */
 495			lv->lv_niovecs = shadow->lv_niovecs;
 496
 497			/* reset the lv buffer information for new formatting */
 498			lv->lv_buf_len = 0;
 499			lv->lv_bytes = 0;
 500			lv->lv_buf = (char *)lv +
 501					xlog_cil_iovec_space(lv->lv_niovecs);
 502		} else {
 503			/* switch to shadow buffer! */
 504			lv = shadow;
 505			lv->lv_item = lip;
 
 506			if (ordered) {
 507				/* track as an ordered logvec */
 508				ASSERT(lip->li_lv == NULL);
 
 509				goto insert;
 510			}
 
 511		}
 512
 
 
 
 
 
 
 
 513		ASSERT(IS_ALIGNED((unsigned long)lv->lv_buf, sizeof(uint64_t)));
 
 514		lip->li_ops->iop_format(lip, lv);
 515insert:
 516		xfs_cil_prepare_item(log, lv, old_lv, diff_len);
 
 517	}
 518}
 519
 520/*
 521 * The use of lockless waitqueue_active() requires that the caller has
 522 * serialised itself against the wakeup call in xlog_cil_push_work(). That
 523 * can be done by either holding the push lock or the context lock.
 524 */
 525static inline bool
 526xlog_cil_over_hard_limit(
 527	struct xlog	*log,
 528	int32_t		space_used)
 529{
 530	if (waitqueue_active(&log->l_cilp->xc_push_wait))
 531		return true;
 532	if (space_used >= XLOG_CIL_BLOCKING_SPACE_LIMIT(log))
 533		return true;
 534	return false;
 535}
 536
 537/*
 538 * Insert the log items into the CIL and calculate the difference in space
 539 * consumed by the item. Add the space to the checkpoint ticket and calculate
 540 * if the change requires additional log metadata. If it does, take that space
 541 * as well. Remove the amount of space we added to the checkpoint ticket from
 542 * the current transaction ticket so that the accounting works out correctly.
 543 */
 544static void
 545xlog_cil_insert_items(
 546	struct xlog		*log,
 547	struct xfs_trans	*tp,
 548	uint32_t		released_space)
 549{
 550	struct xfs_cil		*cil = log->l_cilp;
 551	struct xfs_cil_ctx	*ctx = cil->xc_ctx;
 552	struct xfs_log_item	*lip;
 553	int			len = 0;
 554	int			iovhdr_res = 0, split_res = 0, ctx_res = 0;
 555	int			space_used;
 556	int			order;
 557	struct xlog_cil_pcp	*cilpcp;
 558
 559	ASSERT(tp);
 560
 561	/*
 562	 * We can do this safely because the context can't checkpoint until we
 563	 * are done so it doesn't matter exactly how we update the CIL.
 564	 */
 565	xlog_cil_insert_format_items(log, tp, &len);
 566
 567	/*
 568	 * Subtract the space released by intent cancelation from the space we
 569	 * consumed so that we remove it from the CIL space and add it back to
 570	 * the current transaction reservation context.
 571	 */
 572	len -= released_space;
 573
 574	/*
 575	 * Grab the per-cpu pointer for the CIL before we start any accounting.
 576	 * That ensures that we are running with pre-emption disabled and so we
 577	 * can't be scheduled away between split sample/update operations that
 578	 * are done without outside locking to serialise them.
 579	 */
 580	cilpcp = get_cpu_ptr(cil->xc_pcp);
 581
 582	/*
 583	 * We need to take the CIL checkpoint unit reservation on the first
 584	 * commit into the CIL. Test the XLOG_CIL_EMPTY bit first so we don't
 585	 * unnecessarily do an atomic op in the fast path here. We can clear the
 586	 * XLOG_CIL_EMPTY bit as we are under the xc_ctx_lock here and that
 587	 * needs to be held exclusively to reset the XLOG_CIL_EMPTY bit.
 588	 */
 589	if (test_bit(XLOG_CIL_EMPTY, &cil->xc_flags) &&
 590	    test_and_clear_bit(XLOG_CIL_EMPTY, &cil->xc_flags))
 591		ctx_res = ctx->ticket->t_unit_res;
 592
 593	/*
 594	 * Check if we need to steal iclog headers. atomic_read() is not a
 595	 * locked atomic operation, so we can check the value before we do any
 596	 * real atomic ops in the fast path. If we've already taken the CIL unit
 597	 * reservation from this commit, we've already got one iclog header
 598	 * space reserved so we have to account for that otherwise we risk
 599	 * overrunning the reservation on this ticket.
 600	 *
 601	 * If the CIL is already at the hard limit, we might need more header
 602	 * space that originally reserved. So steal more header space from every
 603	 * commit that occurs once we are over the hard limit to ensure the CIL
 604	 * push won't run out of reservation space.
 605	 *
 606	 * This can steal more than we need, but that's OK.
 607	 *
 608	 * The cil->xc_ctx_lock provides the serialisation necessary for safely
 609	 * calling xlog_cil_over_hard_limit() in this context.
 610	 */
 611	space_used = atomic_read(&ctx->space_used) + cilpcp->space_used + len;
 612	if (atomic_read(&cil->xc_iclog_hdrs) > 0 ||
 613	    xlog_cil_over_hard_limit(log, space_used)) {
 614		split_res = log->l_iclog_hsize +
 615					sizeof(struct xlog_op_header);
 616		if (ctx_res)
 617			ctx_res += split_res * (tp->t_ticket->t_iclog_hdrs - 1);
 618		else
 619			ctx_res = split_res * tp->t_ticket->t_iclog_hdrs;
 620		atomic_sub(tp->t_ticket->t_iclog_hdrs, &cil->xc_iclog_hdrs);
 621	}
 622	cilpcp->space_reserved += ctx_res;
 623
 624	/*
 625	 * Accurately account when over the soft limit, otherwise fold the
 626	 * percpu count into the global count if over the per-cpu threshold.
 627	 */
 628	if (!test_bit(XLOG_CIL_PCP_SPACE, &cil->xc_flags)) {
 629		atomic_add(len, &ctx->space_used);
 630	} else if (cilpcp->space_used + len >
 631			(XLOG_CIL_SPACE_LIMIT(log) / num_online_cpus())) {
 632		space_used = atomic_add_return(cilpcp->space_used + len,
 633						&ctx->space_used);
 634		cilpcp->space_used = 0;
 635
 636		/*
 637		 * If we just transitioned over the soft limit, we need to
 638		 * transition to the global atomic counter.
 
 639		 */
 640		if (space_used >= XLOG_CIL_SPACE_LIMIT(log))
 641			xlog_cil_insert_pcp_aggregate(cil, ctx);
 642	} else {
 643		cilpcp->space_used += len;
 644	}
 
 
 
 
 
 645	/* attach the transaction to the CIL if it has any busy extents */
 646	if (!list_empty(&tp->t_busy))
 647		list_splice_init(&tp->t_busy, &cilpcp->busy_extents);
 648
 649	/*
 650	 * Now update the order of everything modified in the transaction
 651	 * and insert items into the CIL if they aren't already there.
 652	 * We do this here so we only need to take the CIL lock once during
 653	 * the transaction commit.
 654	 */
 655	order = atomic_inc_return(&ctx->order_id);
 656	list_for_each_entry(lip, &tp->t_items, li_trans) {
 657		/* Skip items which aren't dirty in this transaction. */
 658		if (!test_bit(XFS_LI_DIRTY, &lip->li_flags))
 659			continue;
 660
 661		lip->li_order_id = order;
 662		if (!list_empty(&lip->li_cil))
 663			continue;
 664		list_add_tail(&lip->li_cil, &cilpcp->log_items);
 
 
 
 
 
 
 
 
 
 665	}
 666	put_cpu_ptr(cilpcp);
 
 667
 668	/*
 669	 * If we've overrun the reservation, dump the tx details before we move
 670	 * the log items. Shutdown is imminent...
 671	 */
 672	tp->t_ticket->t_curr_res -= ctx_res + len;
 673	if (WARN_ON(tp->t_ticket->t_curr_res < 0)) {
 674		xfs_warn(log->l_mp, "Transaction log reservation overrun:");
 675		xfs_warn(log->l_mp,
 676			 "  log items: %d bytes (iov hdrs: %d bytes)",
 677			 len, iovhdr_res);
 678		xfs_warn(log->l_mp, "  split region headers: %d bytes",
 679			 split_res);
 680		xfs_warn(log->l_mp, "  ctx ticket: %d bytes", ctx_res);
 681		xlog_print_trans(tp);
 682		xlog_force_shutdown(log, SHUTDOWN_LOG_IO_ERROR);
 683	}
 684}
 685
 686static void
 687xlog_cil_free_logvec(
 688	struct list_head	*lv_chain)
 689{
 690	struct xfs_log_vec	*lv;
 691
 692	while (!list_empty(lv_chain)) {
 693		lv = list_first_entry(lv_chain, struct xfs_log_vec, lv_list);
 694		list_del_init(&lv->lv_list);
 695		kmem_free(lv);
 
 696	}
 697}
 698
 699static void
 700xlog_discard_endio_work(
 701	struct work_struct	*work)
 702{
 703	struct xfs_cil_ctx	*ctx =
 704		container_of(work, struct xfs_cil_ctx, discard_endio_work);
 705	struct xfs_mount	*mp = ctx->cil->xc_log->l_mp;
 706
 707	xfs_extent_busy_clear(mp, &ctx->busy_extents, false);
 708	kmem_free(ctx);
 709}
 710
 711/*
 712 * Queue up the actual completion to a thread to avoid IRQ-safe locking for
 713 * pagb_lock.  Note that we need a unbounded workqueue, otherwise we might
 714 * get the execution delayed up to 30 seconds for weird reasons.
 715 */
 716static void
 717xlog_discard_endio(
 718	struct bio		*bio)
 719{
 720	struct xfs_cil_ctx	*ctx = bio->bi_private;
 721
 722	INIT_WORK(&ctx->discard_endio_work, xlog_discard_endio_work);
 723	queue_work(xfs_discard_wq, &ctx->discard_endio_work);
 724	bio_put(bio);
 725}
 726
 727static void
 728xlog_discard_busy_extents(
 729	struct xfs_mount	*mp,
 730	struct xfs_cil_ctx	*ctx)
 731{
 732	struct list_head	*list = &ctx->busy_extents;
 733	struct xfs_extent_busy	*busyp;
 734	struct bio		*bio = NULL;
 735	struct blk_plug		plug;
 736	int			error = 0;
 737
 738	ASSERT(xfs_has_discard(mp));
 739
 740	blk_start_plug(&plug);
 741	list_for_each_entry(busyp, list, list) {
 742		trace_xfs_discard_extent(mp, busyp->agno, busyp->bno,
 743					 busyp->length);
 744
 745		error = __blkdev_issue_discard(mp->m_ddev_targp->bt_bdev,
 746				XFS_AGB_TO_DADDR(mp, busyp->agno, busyp->bno),
 747				XFS_FSB_TO_BB(mp, busyp->length),
 748				GFP_NOFS, &bio);
 749		if (error && error != -EOPNOTSUPP) {
 750			xfs_info(mp,
 751	 "discard failed for extent [0x%llx,%u], error %d",
 752				 (unsigned long long)busyp->bno,
 753				 busyp->length,
 754				 error);
 755			break;
 756		}
 757	}
 758
 759	if (bio) {
 760		bio->bi_private = ctx;
 761		bio->bi_end_io = xlog_discard_endio;
 762		submit_bio(bio);
 763	} else {
 764		xlog_discard_endio_work(&ctx->discard_endio_work);
 765	}
 766	blk_finish_plug(&plug);
 767}
 768
 769/*
 770 * Mark all items committed and clear busy extents. We free the log vector
 771 * chains in a separate pass so that we unpin the log items as quickly as
 772 * possible.
 773 */
 774static void
 775xlog_cil_committed(
 776	struct xfs_cil_ctx	*ctx)
 
 777{
 
 778	struct xfs_mount	*mp = ctx->cil->xc_log->l_mp;
 779	bool			abort = xlog_is_shutdown(ctx->cil->xc_log);
 780
 781	/*
 782	 * If the I/O failed, we're aborting the commit and already shutdown.
 783	 * Wake any commit waiters before aborting the log items so we don't
 784	 * block async log pushers on callbacks. Async log pushers explicitly do
 785	 * not wait on log force completion because they may be holding locks
 786	 * required to unpin items.
 787	 */
 788	if (abort) {
 789		spin_lock(&ctx->cil->xc_push_lock);
 790		wake_up_all(&ctx->cil->xc_start_wait);
 791		wake_up_all(&ctx->cil->xc_commit_wait);
 792		spin_unlock(&ctx->cil->xc_push_lock);
 793	}
 794
 795	xfs_trans_committed_bulk(ctx->cil->xc_log->l_ailp, &ctx->lv_chain,
 796					ctx->start_lsn, abort);
 797
 798	xfs_extent_busy_sort(&ctx->busy_extents);
 799	xfs_extent_busy_clear(mp, &ctx->busy_extents,
 800			      xfs_has_discard(mp) && !abort);
 801
 
 
 
 
 
 
 802	spin_lock(&ctx->cil->xc_push_lock);
 
 
 803	list_del(&ctx->committing);
 804	spin_unlock(&ctx->cil->xc_push_lock);
 805
 806	xlog_cil_free_logvec(&ctx->lv_chain);
 807
 808	if (!list_empty(&ctx->busy_extents))
 809		xlog_discard_busy_extents(mp, ctx);
 810	else
 811		kmem_free(ctx);
 812}
 813
 814void
 815xlog_cil_process_committed(
 816	struct list_head	*list)
 817{
 818	struct xfs_cil_ctx	*ctx;
 819
 820	while ((ctx = list_first_entry_or_null(list,
 821			struct xfs_cil_ctx, iclog_entry))) {
 822		list_del(&ctx->iclog_entry);
 823		xlog_cil_committed(ctx);
 824	}
 825}
 826
 827/*
 828* Record the LSN of the iclog we were just granted space to start writing into.
 829* If the context doesn't have a start_lsn recorded, then this iclog will
 830* contain the start record for the checkpoint. Otherwise this write contains
 831* the commit record for the checkpoint.
 832*/
 833void
 834xlog_cil_set_ctx_write_state(
 835	struct xfs_cil_ctx	*ctx,
 836	struct xlog_in_core	*iclog)
 837{
 838	struct xfs_cil		*cil = ctx->cil;
 839	xfs_lsn_t		lsn = be64_to_cpu(iclog->ic_header.h_lsn);
 840
 841	ASSERT(!ctx->commit_lsn);
 842	if (!ctx->start_lsn) {
 843		spin_lock(&cil->xc_push_lock);
 844		/*
 845		 * The LSN we need to pass to the log items on transaction
 846		 * commit is the LSN reported by the first log vector write, not
 847		 * the commit lsn. If we use the commit record lsn then we can
 848		 * move the grant write head beyond the tail LSN and overwrite
 849		 * it.
 850		 */
 851		ctx->start_lsn = lsn;
 852		wake_up_all(&cil->xc_start_wait);
 853		spin_unlock(&cil->xc_push_lock);
 854
 855		/*
 856		 * Make sure the metadata we are about to overwrite in the log
 857		 * has been flushed to stable storage before this iclog is
 858		 * issued.
 859		 */
 860		spin_lock(&cil->xc_log->l_icloglock);
 861		iclog->ic_flags |= XLOG_ICL_NEED_FLUSH;
 862		spin_unlock(&cil->xc_log->l_icloglock);
 863		return;
 864	}
 865
 866	/*
 867	 * Take a reference to the iclog for the context so that we still hold
 868	 * it when xlog_write is done and has released it. This means the
 869	 * context controls when the iclog is released for IO.
 870	 */
 871	atomic_inc(&iclog->ic_refcnt);
 872
 873	/*
 874	 * xlog_state_get_iclog_space() guarantees there is enough space in the
 875	 * iclog for an entire commit record, so we can attach the context
 876	 * callbacks now.  This needs to be done before we make the commit_lsn
 877	 * visible to waiters so that checkpoints with commit records in the
 878	 * same iclog order their IO completion callbacks in the same order that
 879	 * the commit records appear in the iclog.
 880	 */
 881	spin_lock(&cil->xc_log->l_icloglock);
 882	list_add_tail(&ctx->iclog_entry, &iclog->ic_callbacks);
 883	spin_unlock(&cil->xc_log->l_icloglock);
 884
 885	/*
 886	 * Now we can record the commit LSN and wake anyone waiting for this
 887	 * sequence to have the ordered commit record assigned to a physical
 888	 * location in the log.
 889	 */
 890	spin_lock(&cil->xc_push_lock);
 891	ctx->commit_iclog = iclog;
 892	ctx->commit_lsn = lsn;
 893	wake_up_all(&cil->xc_commit_wait);
 894	spin_unlock(&cil->xc_push_lock);
 895}
 896
 897
 898/*
 899 * Ensure that the order of log writes follows checkpoint sequence order. This
 900 * relies on the context LSN being zero until the log write has guaranteed the
 901 * LSN that the log write will start at via xlog_state_get_iclog_space().
 902 */
 903enum _record_type {
 904	_START_RECORD,
 905	_COMMIT_RECORD,
 906};
 907
 908static int
 909xlog_cil_order_write(
 910	struct xfs_cil		*cil,
 911	xfs_csn_t		sequence,
 912	enum _record_type	record)
 913{
 914	struct xfs_cil_ctx	*ctx;
 915
 916restart:
 917	spin_lock(&cil->xc_push_lock);
 918	list_for_each_entry(ctx, &cil->xc_committing, committing) {
 919		/*
 920		 * Avoid getting stuck in this loop because we were woken by the
 921		 * shutdown, but then went back to sleep once already in the
 922		 * shutdown state.
 923		 */
 924		if (xlog_is_shutdown(cil->xc_log)) {
 925			spin_unlock(&cil->xc_push_lock);
 926			return -EIO;
 927		}
 928
 929		/*
 930		 * Higher sequences will wait for this one so skip them.
 931		 * Don't wait for our own sequence, either.
 932		 */
 933		if (ctx->sequence >= sequence)
 934			continue;
 935
 936		/* Wait until the LSN for the record has been recorded. */
 937		switch (record) {
 938		case _START_RECORD:
 939			if (!ctx->start_lsn) {
 940				xlog_wait(&cil->xc_start_wait, &cil->xc_push_lock);
 941				goto restart;
 942			}
 943			break;
 944		case _COMMIT_RECORD:
 945			if (!ctx->commit_lsn) {
 946				xlog_wait(&cil->xc_commit_wait, &cil->xc_push_lock);
 947				goto restart;
 948			}
 949			break;
 950		}
 951	}
 952	spin_unlock(&cil->xc_push_lock);
 953	return 0;
 954}
 955
 956/*
 957 * Write out the log vector change now attached to the CIL context. This will
 958 * write a start record that needs to be strictly ordered in ascending CIL
 959 * sequence order so that log recovery will always use in-order start LSNs when
 960 * replaying checkpoints.
 961 */
 962static int
 963xlog_cil_write_chain(
 964	struct xfs_cil_ctx	*ctx,
 965	uint32_t		chain_len)
 966{
 967	struct xlog		*log = ctx->cil->xc_log;
 968	int			error;
 969
 970	error = xlog_cil_order_write(ctx->cil, ctx->sequence, _START_RECORD);
 971	if (error)
 972		return error;
 973	return xlog_write(log, ctx, &ctx->lv_chain, ctx->ticket, chain_len);
 974}
 975
 976/*
 977 * Write out the commit record of a checkpoint transaction to close off a
 978 * running log write. These commit records are strictly ordered in ascending CIL
 979 * sequence order so that log recovery will always replay the checkpoints in the
 980 * correct order.
 981 */
 982static int
 983xlog_cil_write_commit_record(
 984	struct xfs_cil_ctx	*ctx)
 985{
 986	struct xlog		*log = ctx->cil->xc_log;
 987	struct xlog_op_header	ophdr = {
 988		.oh_clientid = XFS_TRANSACTION,
 989		.oh_tid = cpu_to_be32(ctx->ticket->t_tid),
 990		.oh_flags = XLOG_COMMIT_TRANS,
 991	};
 992	struct xfs_log_iovec	reg = {
 993		.i_addr = &ophdr,
 994		.i_len = sizeof(struct xlog_op_header),
 995		.i_type = XLOG_REG_TYPE_COMMIT,
 996	};
 997	struct xfs_log_vec	vec = {
 998		.lv_niovecs = 1,
 999		.lv_iovecp = &reg,
1000	};
1001	int			error;
1002	LIST_HEAD(lv_chain);
1003	list_add(&vec.lv_list, &lv_chain);
1004
1005	if (xlog_is_shutdown(log))
1006		return -EIO;
1007
1008	error = xlog_cil_order_write(ctx->cil, ctx->sequence, _COMMIT_RECORD);
1009	if (error)
1010		return error;
1011
1012	/* account for space used by record data */
1013	ctx->ticket->t_curr_res -= reg.i_len;
1014	error = xlog_write(log, ctx, &lv_chain, ctx->ticket, reg.i_len);
1015	if (error)
1016		xlog_force_shutdown(log, SHUTDOWN_LOG_IO_ERROR);
1017	return error;
1018}
1019
1020struct xlog_cil_trans_hdr {
1021	struct xlog_op_header	oph[2];
1022	struct xfs_trans_header	thdr;
1023	struct xfs_log_iovec	lhdr[2];
1024};
1025
1026/*
1027 * Build a checkpoint transaction header to begin the journal transaction.  We
1028 * need to account for the space used by the transaction header here as it is
1029 * not accounted for in xlog_write().
1030 *
1031 * This is the only place we write a transaction header, so we also build the
1032 * log opheaders that indicate the start of a log transaction and wrap the
1033 * transaction header. We keep the start record in it's own log vector rather
1034 * than compacting them into a single region as this ends up making the logic
1035 * in xlog_write() for handling empty opheaders for start, commit and unmount
1036 * records much simpler.
1037 */
1038static void
1039xlog_cil_build_trans_hdr(
1040	struct xfs_cil_ctx	*ctx,
1041	struct xlog_cil_trans_hdr *hdr,
1042	struct xfs_log_vec	*lvhdr,
1043	int			num_iovecs)
1044{
1045	struct xlog_ticket	*tic = ctx->ticket;
1046	__be32			tid = cpu_to_be32(tic->t_tid);
1047
1048	memset(hdr, 0, sizeof(*hdr));
1049
1050	/* Log start record */
1051	hdr->oph[0].oh_tid = tid;
1052	hdr->oph[0].oh_clientid = XFS_TRANSACTION;
1053	hdr->oph[0].oh_flags = XLOG_START_TRANS;
1054
1055	/* log iovec region pointer */
1056	hdr->lhdr[0].i_addr = &hdr->oph[0];
1057	hdr->lhdr[0].i_len = sizeof(struct xlog_op_header);
1058	hdr->lhdr[0].i_type = XLOG_REG_TYPE_LRHEADER;
1059
1060	/* log opheader */
1061	hdr->oph[1].oh_tid = tid;
1062	hdr->oph[1].oh_clientid = XFS_TRANSACTION;
1063	hdr->oph[1].oh_len = cpu_to_be32(sizeof(struct xfs_trans_header));
1064
1065	/* transaction header in host byte order format */
1066	hdr->thdr.th_magic = XFS_TRANS_HEADER_MAGIC;
1067	hdr->thdr.th_type = XFS_TRANS_CHECKPOINT;
1068	hdr->thdr.th_tid = tic->t_tid;
1069	hdr->thdr.th_num_items = num_iovecs;
1070
1071	/* log iovec region pointer */
1072	hdr->lhdr[1].i_addr = &hdr->oph[1];
1073	hdr->lhdr[1].i_len = sizeof(struct xlog_op_header) +
1074				sizeof(struct xfs_trans_header);
1075	hdr->lhdr[1].i_type = XLOG_REG_TYPE_TRANSHDR;
1076
1077	lvhdr->lv_niovecs = 2;
1078	lvhdr->lv_iovecp = &hdr->lhdr[0];
1079	lvhdr->lv_bytes = hdr->lhdr[0].i_len + hdr->lhdr[1].i_len;
1080
1081	tic->t_curr_res -= lvhdr->lv_bytes;
1082}
1083
1084/*
1085 * CIL item reordering compare function. We want to order in ascending ID order,
1086 * but we want to leave items with the same ID in the order they were added to
1087 * the list. This is important for operations like reflink where we log 4 order
1088 * dependent intents in a single transaction when we overwrite an existing
1089 * shared extent with a new shared extent. i.e. BUI(unmap), CUI(drop),
1090 * CUI (inc), BUI(remap)...
1091 */
1092static int
1093xlog_cil_order_cmp(
1094	void			*priv,
1095	const struct list_head	*a,
1096	const struct list_head	*b)
1097{
1098	struct xfs_log_vec	*l1 = container_of(a, struct xfs_log_vec, lv_list);
1099	struct xfs_log_vec	*l2 = container_of(b, struct xfs_log_vec, lv_list);
1100
1101	return l1->lv_order_id > l2->lv_order_id;
1102}
1103
1104/*
1105 * Pull all the log vectors off the items in the CIL, and remove the items from
1106 * the CIL. We don't need the CIL lock here because it's only needed on the
1107 * transaction commit side which is currently locked out by the flush lock.
1108 *
1109 * If a log item is marked with a whiteout, we do not need to write it to the
1110 * journal and so we just move them to the whiteout list for the caller to
1111 * dispose of appropriately.
1112 */
1113static void
1114xlog_cil_build_lv_chain(
1115	struct xfs_cil_ctx	*ctx,
1116	struct list_head	*whiteouts,
1117	uint32_t		*num_iovecs,
1118	uint32_t		*num_bytes)
1119{
1120	while (!list_empty(&ctx->log_items)) {
1121		struct xfs_log_item	*item;
1122		struct xfs_log_vec	*lv;
1123
1124		item = list_first_entry(&ctx->log_items,
1125					struct xfs_log_item, li_cil);
1126
1127		if (test_bit(XFS_LI_WHITEOUT, &item->li_flags)) {
1128			list_move(&item->li_cil, whiteouts);
1129			trace_xfs_cil_whiteout_skip(item);
1130			continue;
1131		}
1132
1133		lv = item->li_lv;
1134		lv->lv_order_id = item->li_order_id;
1135
1136		/* we don't write ordered log vectors */
1137		if (lv->lv_buf_len != XFS_LOG_VEC_ORDERED)
1138			*num_bytes += lv->lv_bytes;
1139		*num_iovecs += lv->lv_niovecs;
1140		list_add_tail(&lv->lv_list, &ctx->lv_chain);
1141
1142		list_del_init(&item->li_cil);
1143		item->li_order_id = 0;
1144		item->li_lv = NULL;
1145	}
1146}
1147
1148static void
1149xlog_cil_cleanup_whiteouts(
1150	struct list_head	*whiteouts)
1151{
1152	while (!list_empty(whiteouts)) {
1153		struct xfs_log_item *item = list_first_entry(whiteouts,
1154						struct xfs_log_item, li_cil);
1155		list_del_init(&item->li_cil);
1156		trace_xfs_cil_whiteout_unpin(item);
1157		item->li_ops->iop_unpin(item, 1);
1158	}
1159}
1160
1161/*
1162 * Push the Committed Item List to the log.
1163 *
1164 * If the current sequence is the same as xc_push_seq we need to do a flush. If
1165 * xc_push_seq is less than the current sequence, then it has already been
1166 * flushed and we don't need to do anything - the caller will wait for it to
1167 * complete if necessary.
1168 *
1169 * xc_push_seq is checked unlocked against the sequence number for a match.
1170 * Hence we can allow log forces to run racily and not issue pushes for the
1171 * same sequence twice.  If we get a race between multiple pushes for the same
1172 * sequence they will block on the first one and then abort, hence avoiding
1173 * needless pushes.
1174 */
1175static void
1176xlog_cil_push_work(
1177	struct work_struct	*work)
1178{
1179	struct xfs_cil_ctx	*ctx =
1180		container_of(work, struct xfs_cil_ctx, push_work);
1181	struct xfs_cil		*cil = ctx->cil;
1182	struct xlog		*log = cil->xc_log;
1183	struct xfs_cil_ctx	*new_ctx;
1184	int			num_iovecs = 0;
1185	int			num_bytes = 0;
 
1186	int			error = 0;
1187	struct xlog_cil_trans_hdr thdr;
1188	struct xfs_log_vec	lvhdr = {};
1189	xfs_csn_t		push_seq;
1190	bool			push_commit_stable;
1191	LIST_HEAD		(whiteouts);
1192	struct xlog_ticket	*ticket;
 
 
1193
1194	new_ctx = xlog_cil_ctx_alloc();
1195	new_ctx->ticket = xlog_cil_ticket_alloc(log);
1196
1197	down_write(&cil->xc_ctx_lock);
 
1198
1199	spin_lock(&cil->xc_push_lock);
1200	push_seq = cil->xc_push_seq;
1201	ASSERT(push_seq <= ctx->sequence);
1202	push_commit_stable = cil->xc_push_commit_stable;
1203	cil->xc_push_commit_stable = false;
1204
1205	/*
1206	 * As we are about to switch to a new, empty CIL context, we no longer
1207	 * need to throttle tasks on CIL space overruns. Wake any waiters that
1208	 * the hard push throttle may have caught so they can start committing
1209	 * to the new context. The ctx->xc_push_lock provides the serialisation
1210	 * necessary for safely using the lockless waitqueue_active() check in
1211	 * this context.
1212	 */
1213	if (waitqueue_active(&cil->xc_push_wait))
1214		wake_up_all(&cil->xc_push_wait);
1215
1216	xlog_cil_push_pcp_aggregate(cil, ctx);
1217
1218	/*
1219	 * Check if we've anything to push. If there is nothing, then we don't
1220	 * move on to a new sequence number and so we have to be able to push
1221	 * this sequence again later.
1222	 */
1223	if (test_bit(XLOG_CIL_EMPTY, &cil->xc_flags)) {
1224		cil->xc_push_seq = 0;
1225		spin_unlock(&cil->xc_push_lock);
1226		goto out_skip;
1227	}
1228
1229
1230	/* check for a previously pushed sequence */
1231	if (push_seq < ctx->sequence) {
1232		spin_unlock(&cil->xc_push_lock);
1233		goto out_skip;
1234	}
1235
1236	/*
1237	 * We are now going to push this context, so add it to the committing
1238	 * list before we do anything else. This ensures that anyone waiting on
1239	 * this push can easily detect the difference between a "push in
1240	 * progress" and "CIL is empty, nothing to do".
1241	 *
1242	 * IOWs, a wait loop can now check for:
1243	 *	the current sequence not being found on the committing list;
1244	 *	an empty CIL; and
1245	 *	an unchanged sequence number
1246	 * to detect a push that had nothing to do and therefore does not need
1247	 * waiting on. If the CIL is not empty, we get put on the committing
1248	 * list before emptying the CIL and bumping the sequence number. Hence
1249	 * an empty CIL and an unchanged sequence number means we jumped out
1250	 * above after doing nothing.
1251	 *
1252	 * Hence the waiter will either find the commit sequence on the
1253	 * committing list or the sequence number will be unchanged and the CIL
1254	 * still dirty. In that latter case, the push has not yet started, and
1255	 * so the waiter will have to continue trying to check the CIL
1256	 * committing list until it is found. In extreme cases of delay, the
1257	 * sequence may fully commit between the attempts the wait makes to wait
1258	 * on the commit sequence.
1259	 */
1260	list_add(&ctx->committing, &cil->xc_committing);
1261	spin_unlock(&cil->xc_push_lock);
1262
1263	xlog_cil_build_lv_chain(ctx, &whiteouts, &num_iovecs, &num_bytes);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1264
1265	/*
1266	 * Switch the contexts so we can drop the context lock and move out
 
 
 
 
 
 
 
 
 
 
 
 
1267	 * of a shared context. We can't just go straight to the commit record,
1268	 * though - we need to synchronise with previous and future commits so
1269	 * that the commit records are correctly ordered in the log to ensure
1270	 * that we process items during log IO completion in the correct order.
1271	 *
1272	 * For example, if we get an EFI in one checkpoint and the EFD in the
1273	 * next (e.g. due to log forces), we do not want the checkpoint with
1274	 * the EFD to be committed before the checkpoint with the EFI.  Hence
1275	 * we must strictly order the commit records of the checkpoints so
1276	 * that: a) the checkpoint callbacks are attached to the iclogs in the
1277	 * correct order; and b) the checkpoints are replayed in correct order
1278	 * in log recovery.
1279	 *
1280	 * Hence we need to add this context to the committing context list so
1281	 * that higher sequences will wait for us to write out a commit record
1282	 * before they do.
1283	 *
1284	 * xfs_log_force_seq requires us to mirror the new sequence into the cil
1285	 * structure atomically with the addition of this sequence to the
1286	 * committing list. This also ensures that we can do unlocked checks
1287	 * against the current sequence in log forces without risking
1288	 * deferencing a freed context pointer.
1289	 */
1290	spin_lock(&cil->xc_push_lock);
1291	xlog_cil_ctx_switch(cil, new_ctx);
1292	spin_unlock(&cil->xc_push_lock);
1293	up_write(&cil->xc_ctx_lock);
1294
1295	/*
1296	 * Sort the log vector chain before we add the transaction headers.
1297	 * This ensures we always have the transaction headers at the start
1298	 * of the chain.
1299	 */
1300	list_sort(NULL, &ctx->lv_chain, xlog_cil_order_cmp);
1301
1302	/*
1303	 * Build a checkpoint transaction header and write it to the log to
1304	 * begin the transaction. We need to account for the space used by the
1305	 * transaction header here as it is not accounted for in xlog_write().
1306	 * Add the lvhdr to the head of the lv chain we pass to xlog_write() so
1307	 * it gets written into the iclog first.
1308	 */
1309	xlog_cil_build_trans_hdr(ctx, &thdr, &lvhdr, num_iovecs);
1310	num_bytes += lvhdr.lv_bytes;
1311	list_add(&lvhdr.lv_list, &ctx->lv_chain);
1312
1313	/*
1314	 * Take the lvhdr back off the lv_chain immediately after calling
1315	 * xlog_cil_write_chain() as it should not be passed to log IO
1316	 * completion.
1317	 */
1318	error = xlog_cil_write_chain(ctx, num_bytes);
1319	list_del(&lvhdr.lv_list);
1320	if (error)
1321		goto out_abort_free_ticket;
 
 
1322
1323	error = xlog_cil_write_commit_record(ctx);
1324	if (error)
1325		goto out_abort_free_ticket;
1326
1327	/*
1328	 * Grab the ticket from the ctx so we can ungrant it after releasing the
1329	 * commit_iclog. The ctx may be freed by the time we return from
1330	 * releasing the commit_iclog (i.e. checkpoint has been completed and
1331	 * callback run) so we can't reference the ctx after the call to
1332	 * xlog_state_release_iclog().
1333	 */
1334	ticket = ctx->ticket;
1335
1336	/*
1337	 * If the checkpoint spans multiple iclogs, wait for all previous iclogs
1338	 * to complete before we submit the commit_iclog. We can't use state
1339	 * checks for this - ACTIVE can be either a past completed iclog or a
1340	 * future iclog being filled, while WANT_SYNC through SYNC_DONE can be a
1341	 * past or future iclog awaiting IO or ordered IO completion to be run.
1342	 * In the latter case, if it's a future iclog and we wait on it, the we
1343	 * will hang because it won't get processed through to ic_force_wait
1344	 * wakeup until this commit_iclog is written to disk.  Hence we use the
1345	 * iclog header lsn and compare it to the commit lsn to determine if we
1346	 * need to wait on iclogs or not.
1347	 */
1348	spin_lock(&log->l_icloglock);
1349	if (ctx->start_lsn != ctx->commit_lsn) {
1350		xfs_lsn_t	plsn;
1351
1352		plsn = be64_to_cpu(ctx->commit_iclog->ic_prev->ic_header.h_lsn);
1353		if (plsn && XFS_LSN_CMP(plsn, ctx->commit_lsn) < 0) {
1354			/*
1355			 * Waiting on ic_force_wait orders the completion of
1356			 * iclogs older than ic_prev. Hence we only need to wait
1357			 * on the most recent older iclog here.
1358			 */
1359			xlog_wait_on_iclog(ctx->commit_iclog->ic_prev);
1360			spin_lock(&log->l_icloglock);
1361		}
1362
1363		/*
1364		 * We need to issue a pre-flush so that the ordering for this
1365		 * checkpoint is correctly preserved down to stable storage.
1366		 */
1367		ctx->commit_iclog->ic_flags |= XLOG_ICL_NEED_FLUSH;
 
 
 
 
 
 
 
 
 
1368	}
 
 
 
 
 
 
 
 
 
 
 
 
 
1369
1370	/*
1371	 * The commit iclog must be written to stable storage to guarantee
1372	 * journal IO vs metadata writeback IO is correctly ordered on stable
1373	 * storage.
1374	 *
1375	 * If the push caller needs the commit to be immediately stable and the
1376	 * commit_iclog is not yet marked as XLOG_STATE_WANT_SYNC to indicate it
1377	 * will be written when released, switch it's state to WANT_SYNC right
1378	 * now.
1379	 */
1380	ctx->commit_iclog->ic_flags |= XLOG_ICL_NEED_FUA;
1381	if (push_commit_stable &&
1382	    ctx->commit_iclog->ic_state == XLOG_STATE_ACTIVE)
1383		xlog_state_switch_iclogs(log, ctx->commit_iclog, 0);
1384	ticket = ctx->ticket;
1385	xlog_state_release_iclog(log, ctx->commit_iclog, ticket);
1386
1387	/* Not safe to reference ctx now! */
1388
1389	spin_unlock(&log->l_icloglock);
1390	xlog_cil_cleanup_whiteouts(&whiteouts);
1391	xfs_log_ticket_ungrant(log, ticket);
1392	return;
1393
1394out_skip:
1395	up_write(&cil->xc_ctx_lock);
1396	xfs_log_ticket_put(new_ctx->ticket);
1397	kmem_free(new_ctx);
1398	return;
1399
1400out_abort_free_ticket:
1401	ASSERT(xlog_is_shutdown(log));
1402	xlog_cil_cleanup_whiteouts(&whiteouts);
1403	if (!ctx->commit_iclog) {
1404		xfs_log_ticket_ungrant(log, ctx->ticket);
1405		xlog_cil_committed(ctx);
1406		return;
1407	}
1408	spin_lock(&log->l_icloglock);
1409	ticket = ctx->ticket;
1410	xlog_state_release_iclog(log, ctx->commit_iclog, ticket);
1411	/* Not safe to reference ctx now! */
1412	spin_unlock(&log->l_icloglock);
1413	xfs_log_ticket_ungrant(log, ticket);
1414}
1415
1416/*
1417 * We need to push CIL every so often so we don't cache more than we can fit in
1418 * the log. The limit really is that a checkpoint can't be more than half the
1419 * log (the current checkpoint is not allowed to overwrite the previous
1420 * checkpoint), but commit latency and memory usage limit this to a smaller
1421 * size.
1422 */
1423static void
1424xlog_cil_push_background(
1425	struct xlog	*log) __releases(cil->xc_ctx_lock)
1426{
1427	struct xfs_cil	*cil = log->l_cilp;
1428	int		space_used = atomic_read(&cil->xc_ctx->space_used);
1429
1430	/*
1431	 * The cil won't be empty because we are called while holding the
1432	 * context lock so whatever we added to the CIL will still be there.
1433	 */
1434	ASSERT(!test_bit(XLOG_CIL_EMPTY, &cil->xc_flags));
1435
1436	/*
1437	 * We are done if:
1438	 * - we haven't used up all the space available yet; or
1439	 * - we've already queued up a push; and
1440	 * - we're not over the hard limit; and
1441	 * - nothing has been over the hard limit.
1442	 *
1443	 * If so, we don't need to take the push lock as there's nothing to do.
1444	 */
1445	if (space_used < XLOG_CIL_SPACE_LIMIT(log) ||
1446	    (cil->xc_push_seq == cil->xc_current_sequence &&
1447	     space_used < XLOG_CIL_BLOCKING_SPACE_LIMIT(log) &&
1448	     !waitqueue_active(&cil->xc_push_wait))) {
1449		up_read(&cil->xc_ctx_lock);
1450		return;
1451	}
1452
1453	spin_lock(&cil->xc_push_lock);
1454	if (cil->xc_push_seq < cil->xc_current_sequence) {
1455		cil->xc_push_seq = cil->xc_current_sequence;
1456		queue_work(cil->xc_push_wq, &cil->xc_ctx->push_work);
1457	}
1458
1459	/*
1460	 * Drop the context lock now, we can't hold that if we need to sleep
1461	 * because we are over the blocking threshold. The push_lock is still
1462	 * held, so blocking threshold sleep/wakeup is still correctly
1463	 * serialised here.
1464	 */
1465	up_read(&cil->xc_ctx_lock);
1466
1467	/*
1468	 * If we are well over the space limit, throttle the work that is being
1469	 * done until the push work on this context has begun. Enforce the hard
1470	 * throttle on all transaction commits once it has been activated, even
1471	 * if the committing transactions have resulted in the space usage
1472	 * dipping back down under the hard limit.
1473	 *
1474	 * The ctx->xc_push_lock provides the serialisation necessary for safely
1475	 * calling xlog_cil_over_hard_limit() in this context.
1476	 */
1477	if (xlog_cil_over_hard_limit(log, space_used)) {
1478		trace_xfs_log_cil_wait(log, cil->xc_ctx->ticket);
1479		ASSERT(space_used < log->l_logsize);
1480		xlog_wait(&cil->xc_push_wait, &cil->xc_push_lock);
1481		return;
1482	}
1483
1484	spin_unlock(&cil->xc_push_lock);
1485
1486}
1487
1488/*
1489 * xlog_cil_push_now() is used to trigger an immediate CIL push to the sequence
1490 * number that is passed. When it returns, the work will be queued for
1491 * @push_seq, but it won't be completed.
1492 *
1493 * If the caller is performing a synchronous force, we will flush the workqueue
1494 * to get previously queued work moving to minimise the wait time they will
1495 * undergo waiting for all outstanding pushes to complete. The caller is
1496 * expected to do the required waiting for push_seq to complete.
1497 *
1498 * If the caller is performing an async push, we need to ensure that the
1499 * checkpoint is fully flushed out of the iclogs when we finish the push. If we
1500 * don't do this, then the commit record may remain sitting in memory in an
1501 * ACTIVE iclog. This then requires another full log force to push to disk,
1502 * which defeats the purpose of having an async, non-blocking CIL force
1503 * mechanism. Hence in this case we need to pass a flag to the push work to
1504 * indicate it needs to flush the commit record itself.
1505 */
1506static void
1507xlog_cil_push_now(
1508	struct xlog	*log,
1509	xfs_lsn_t	push_seq,
1510	bool		async)
1511{
1512	struct xfs_cil	*cil = log->l_cilp;
1513
1514	if (!cil)
1515		return;
1516
1517	ASSERT(push_seq && push_seq <= cil->xc_current_sequence);
1518
1519	/* start on any pending background push to minimise wait time on it */
1520	if (!async)
1521		flush_workqueue(cil->xc_push_wq);
1522
1523	spin_lock(&cil->xc_push_lock);
1524
1525	/*
1526	 * If this is an async flush request, we always need to set the
1527	 * xc_push_commit_stable flag even if something else has already queued
1528	 * a push. The flush caller is asking for the CIL to be on stable
1529	 * storage when the next push completes, so regardless of who has queued
1530	 * the push, the flush requires stable semantics from it.
1531	 */
1532	cil->xc_push_commit_stable = async;
1533
1534	/*
1535	 * If the CIL is empty or we've already pushed the sequence then
1536	 * there's no more work that we need to do.
1537	 */
1538	if (test_bit(XLOG_CIL_EMPTY, &cil->xc_flags) ||
1539	    push_seq <= cil->xc_push_seq) {
1540		spin_unlock(&cil->xc_push_lock);
1541		return;
1542	}
1543
1544	cil->xc_push_seq = push_seq;
1545	queue_work(cil->xc_push_wq, &cil->xc_ctx->push_work);
1546	spin_unlock(&cil->xc_push_lock);
1547}
1548
1549bool
1550xlog_cil_empty(
1551	struct xlog	*log)
1552{
1553	struct xfs_cil	*cil = log->l_cilp;
1554	bool		empty = false;
1555
1556	spin_lock(&cil->xc_push_lock);
1557	if (test_bit(XLOG_CIL_EMPTY, &cil->xc_flags))
1558		empty = true;
1559	spin_unlock(&cil->xc_push_lock);
1560	return empty;
1561}
1562
1563/*
1564 * If there are intent done items in this transaction and the related intent was
1565 * committed in the current (same) CIL checkpoint, we don't need to write either
1566 * the intent or intent done item to the journal as the change will be
1567 * journalled atomically within this checkpoint. As we cannot remove items from
1568 * the CIL here, mark the related intent with a whiteout so that the CIL push
1569 * can remove it rather than writing it to the journal. Then remove the intent
1570 * done item from the current transaction and release it so it doesn't get put
1571 * into the CIL at all.
1572 */
1573static uint32_t
1574xlog_cil_process_intents(
1575	struct xfs_cil		*cil,
1576	struct xfs_trans	*tp)
1577{
1578	struct xfs_log_item	*lip, *ilip, *next;
1579	uint32_t		len = 0;
1580
1581	list_for_each_entry_safe(lip, next, &tp->t_items, li_trans) {
1582		if (!(lip->li_ops->flags & XFS_ITEM_INTENT_DONE))
1583			continue;
1584
1585		ilip = lip->li_ops->iop_intent(lip);
1586		if (!ilip || !xlog_item_in_current_chkpt(cil, ilip))
1587			continue;
1588		set_bit(XFS_LI_WHITEOUT, &ilip->li_flags);
1589		trace_xfs_cil_whiteout_mark(ilip);
1590		len += ilip->li_lv->lv_bytes;
1591		kmem_free(ilip->li_lv);
1592		ilip->li_lv = NULL;
1593
1594		xfs_trans_del_item(lip);
1595		lip->li_ops->iop_release(lip);
1596	}
1597	return len;
1598}
1599
1600/*
1601 * Commit a transaction with the given vector to the Committed Item List.
1602 *
1603 * To do this, we need to format the item, pin it in memory if required and
1604 * account for the space used by the transaction. Once we have done that we
1605 * need to release the unused reservation for the transaction, attach the
1606 * transaction to the checkpoint context so we carry the busy extents through
1607 * to checkpoint completion, and then unlock all the items in the transaction.
1608 *
1609 * Called with the context lock already held in read mode to lock out
1610 * background commit, returns without it held once background commits are
1611 * allowed again.
1612 */
1613void
1614xlog_cil_commit(
1615	struct xlog		*log,
1616	struct xfs_trans	*tp,
1617	xfs_csn_t		*commit_seq,
1618	bool			regrant)
1619{
 
1620	struct xfs_cil		*cil = log->l_cilp;
1621	struct xfs_log_item	*lip, *next;
1622	uint32_t		released_space = 0;
1623
1624	/*
1625	 * Do all necessary memory allocation before we lock the CIL.
1626	 * This ensures the allocation does not deadlock with a CIL
1627	 * push in memory reclaim (e.g. from kswapd).
1628	 */
1629	xlog_cil_alloc_shadow_bufs(log, tp);
1630
1631	/* lock out background commit */
1632	down_read(&cil->xc_ctx_lock);
1633
1634	if (tp->t_flags & XFS_TRANS_HAS_INTENT_DONE)
1635		released_space = xlog_cil_process_intents(cil, tp);
 
 
 
1636
1637	xlog_cil_insert_items(log, tp, released_space);
 
 
1638
1639	if (regrant && !xlog_is_shutdown(log))
1640		xfs_log_ticket_regrant(log, tp->t_ticket);
1641	else
1642		xfs_log_ticket_ungrant(log, tp->t_ticket);
1643	tp->t_ticket = NULL;
1644	xfs_trans_unreserve_and_mod_sb(tp);
1645
1646	/*
1647	 * Once all the items of the transaction have been copied to the CIL,
1648	 * the items can be unlocked and possibly freed.
1649	 *
1650	 * This needs to be done before we drop the CIL context lock because we
1651	 * have to update state in the log items and unlock them before they go
1652	 * to disk. If we don't, then the CIL checkpoint can race with us and
1653	 * we can run checkpoint completion before we've updated and unlocked
1654	 * the log items. This affects (at least) processing of stale buffers,
1655	 * inodes and EFIs.
1656	 */
1657	trace_xfs_trans_commit_items(tp, _RET_IP_);
1658	list_for_each_entry_safe(lip, next, &tp->t_items, li_trans) {
1659		xfs_trans_del_item(lip);
1660		if (lip->li_ops->iop_committing)
1661			lip->li_ops->iop_committing(lip, cil->xc_ctx->sequence);
1662	}
1663	if (commit_seq)
1664		*commit_seq = cil->xc_ctx->sequence;
1665
1666	/* xlog_cil_push_background() releases cil->xc_ctx_lock */
1667	xlog_cil_push_background(log);
1668}
1669
1670/*
1671 * Flush the CIL to stable storage but don't wait for it to complete. This
1672 * requires the CIL push to ensure the commit record for the push hits the disk,
1673 * but otherwise is no different to a push done from a log force.
1674 */
1675void
1676xlog_cil_flush(
1677	struct xlog	*log)
1678{
1679	xfs_csn_t	seq = log->l_cilp->xc_current_sequence;
1680
1681	trace_xfs_log_force(log->l_mp, seq, _RET_IP_);
1682	xlog_cil_push_now(log, seq, true);
1683
1684	/*
1685	 * If the CIL is empty, make sure that any previous checkpoint that may
1686	 * still be in an active iclog is pushed to stable storage.
1687	 */
1688	if (test_bit(XLOG_CIL_EMPTY, &log->l_cilp->xc_flags))
1689		xfs_log_force(log->l_mp, 0);
1690}
1691
1692/*
1693 * Conditionally push the CIL based on the sequence passed in.
1694 *
1695 * We only need to push if we haven't already pushed the sequence number given.
1696 * Hence the only time we will trigger a push here is if the push sequence is
1697 * the same as the current context.
1698 *
1699 * We return the current commit lsn to allow the callers to determine if a
1700 * iclog flush is necessary following this call.
1701 */
1702xfs_lsn_t
1703xlog_cil_force_seq(
1704	struct xlog	*log,
1705	xfs_csn_t	sequence)
1706{
1707	struct xfs_cil		*cil = log->l_cilp;
1708	struct xfs_cil_ctx	*ctx;
1709	xfs_lsn_t		commit_lsn = NULLCOMMITLSN;
1710
1711	ASSERT(sequence <= cil->xc_current_sequence);
1712
1713	if (!sequence)
1714		sequence = cil->xc_current_sequence;
1715	trace_xfs_log_force(log->l_mp, sequence, _RET_IP_);
1716
1717	/*
1718	 * check to see if we need to force out the current context.
1719	 * xlog_cil_push() handles racing pushes for the same sequence,
1720	 * so no need to deal with it here.
1721	 */
1722restart:
1723	xlog_cil_push_now(log, sequence, false);
1724
1725	/*
1726	 * See if we can find a previous sequence still committing.
1727	 * We need to wait for all previous sequence commits to complete
1728	 * before allowing the force of push_seq to go ahead. Hence block
1729	 * on commits for those as well.
1730	 */
1731	spin_lock(&cil->xc_push_lock);
1732	list_for_each_entry(ctx, &cil->xc_committing, committing) {
1733		/*
1734		 * Avoid getting stuck in this loop because we were woken by the
1735		 * shutdown, but then went back to sleep once already in the
1736		 * shutdown state.
1737		 */
1738		if (xlog_is_shutdown(log))
1739			goto out_shutdown;
1740		if (ctx->sequence > sequence)
1741			continue;
1742		if (!ctx->commit_lsn) {
1743			/*
1744			 * It is still being pushed! Wait for the push to
1745			 * complete, then start again from the beginning.
1746			 */
1747			XFS_STATS_INC(log->l_mp, xs_log_force_sleep);
1748			xlog_wait(&cil->xc_commit_wait, &cil->xc_push_lock);
1749			goto restart;
1750		}
1751		if (ctx->sequence != sequence)
1752			continue;
1753		/* found it! */
1754		commit_lsn = ctx->commit_lsn;
1755	}
1756
1757	/*
1758	 * The call to xlog_cil_push_now() executes the push in the background.
1759	 * Hence by the time we have got here it our sequence may not have been
1760	 * pushed yet. This is true if the current sequence still matches the
1761	 * push sequence after the above wait loop and the CIL still contains
1762	 * dirty objects. This is guaranteed by the push code first adding the
1763	 * context to the committing list before emptying the CIL.
1764	 *
1765	 * Hence if we don't find the context in the committing list and the
1766	 * current sequence number is unchanged then the CIL contents are
1767	 * significant.  If the CIL is empty, if means there was nothing to push
1768	 * and that means there is nothing to wait for. If the CIL is not empty,
1769	 * it means we haven't yet started the push, because if it had started
1770	 * we would have found the context on the committing list.
1771	 */
1772	if (sequence == cil->xc_current_sequence &&
1773	    !test_bit(XLOG_CIL_EMPTY, &cil->xc_flags)) {
1774		spin_unlock(&cil->xc_push_lock);
1775		goto restart;
1776	}
1777
1778	spin_unlock(&cil->xc_push_lock);
1779	return commit_lsn;
1780
1781	/*
1782	 * We detected a shutdown in progress. We need to trigger the log force
1783	 * to pass through it's iclog state machine error handling, even though
1784	 * we are already in a shutdown state. Hence we can't return
1785	 * NULLCOMMITLSN here as that has special meaning to log forces (i.e.
1786	 * LSN is already stable), so we return a zero LSN instead.
1787	 */
1788out_shutdown:
1789	spin_unlock(&cil->xc_push_lock);
1790	return 0;
1791}
1792
1793/*
1794 * Move dead percpu state to the relevant CIL context structures.
 
 
1795 *
1796 * We have to lock the CIL context here to ensure that nothing is modifying
1797 * the percpu state, either addition or removal. Both of these are done under
1798 * the CIL context lock, so grabbing that exclusively here will ensure we can
1799 * safely drain the cilpcp for the CPU that is dying.
1800 */
1801void
1802xlog_cil_pcp_dead(
1803	struct xlog		*log,
1804	unsigned int		cpu)
1805{
1806	struct xfs_cil		*cil = log->l_cilp;
1807	struct xlog_cil_pcp	*cilpcp = per_cpu_ptr(cil->xc_pcp, cpu);
1808	struct xfs_cil_ctx	*ctx;
1809
1810	down_write(&cil->xc_ctx_lock);
1811	ctx = cil->xc_ctx;
1812	if (ctx->ticket)
1813		ctx->ticket->t_curr_res += cilpcp->space_reserved;
1814	cilpcp->space_reserved = 0;
1815
1816	if (!list_empty(&cilpcp->log_items))
1817		list_splice_init(&cilpcp->log_items, &ctx->log_items);
1818	if (!list_empty(&cilpcp->busy_extents))
1819		list_splice_init(&cilpcp->busy_extents, &ctx->busy_extents);
1820	atomic_add(cilpcp->space_used, &ctx->space_used);
1821	cilpcp->space_used = 0;
1822	up_write(&cil->xc_ctx_lock);
1823}
1824
1825/*
1826 * Perform initial CIL structure initialisation.
1827 */
1828int
1829xlog_cil_init(
1830	struct xlog		*log)
1831{
1832	struct xfs_cil		*cil;
1833	struct xfs_cil_ctx	*ctx;
1834	struct xlog_cil_pcp	*cilpcp;
1835	int			cpu;
1836
1837	cil = kmem_zalloc(sizeof(*cil), KM_MAYFAIL);
1838	if (!cil)
1839		return -ENOMEM;
1840	/*
1841	 * Limit the CIL pipeline depth to 4 concurrent works to bound the
1842	 * concurrency the log spinlocks will be exposed to.
1843	 */
1844	cil->xc_push_wq = alloc_workqueue("xfs-cil/%s",
1845			XFS_WQFLAGS(WQ_FREEZABLE | WQ_MEM_RECLAIM | WQ_UNBOUND),
1846			4, log->l_mp->m_super->s_id);
1847	if (!cil->xc_push_wq)
1848		goto out_destroy_cil;
1849
1850	cil->xc_log = log;
1851	cil->xc_pcp = alloc_percpu(struct xlog_cil_pcp);
1852	if (!cil->xc_pcp)
1853		goto out_destroy_wq;
1854
1855	for_each_possible_cpu(cpu) {
1856		cilpcp = per_cpu_ptr(cil->xc_pcp, cpu);
1857		INIT_LIST_HEAD(&cilpcp->busy_extents);
1858		INIT_LIST_HEAD(&cilpcp->log_items);
1859	}
1860
 
 
1861	INIT_LIST_HEAD(&cil->xc_committing);
 
1862	spin_lock_init(&cil->xc_push_lock);
1863	init_waitqueue_head(&cil->xc_push_wait);
1864	init_rwsem(&cil->xc_ctx_lock);
1865	init_waitqueue_head(&cil->xc_start_wait);
1866	init_waitqueue_head(&cil->xc_commit_wait);
1867	log->l_cilp = cil;
1868
1869	ctx = xlog_cil_ctx_alloc();
1870	xlog_cil_ctx_switch(cil, ctx);
1871	return 0;
 
 
 
1872
1873out_destroy_wq:
1874	destroy_workqueue(cil->xc_push_wq);
1875out_destroy_cil:
1876	kmem_free(cil);
1877	return -ENOMEM;
1878}
1879
1880void
1881xlog_cil_destroy(
1882	struct xlog	*log)
1883{
1884	struct xfs_cil	*cil = log->l_cilp;
1885
1886	if (cil->xc_ctx) {
1887		if (cil->xc_ctx->ticket)
1888			xfs_log_ticket_put(cil->xc_ctx->ticket);
1889		kmem_free(cil->xc_ctx);
1890	}
1891
1892	ASSERT(test_bit(XLOG_CIL_EMPTY, &cil->xc_flags));
1893	free_percpu(cil->xc_pcp);
1894	destroy_workqueue(cil->xc_push_wq);
1895	kmem_free(cil);
1896}
1897