buf0flu.c

这是linux下运行的mysql软件包,可用于linux 下安装 php + mysql + apach 的网络配置

				|| flush_type == BUF_FLUSH_SINGLE_PAGE);

	mutex_enter(&(buf_pool->mutex));

	block = buf_page_hash_get(space, offset);

	ut_a(!block || block->state == BUF_BLOCK_FILE_PAGE);

	if (flush_type == BUF_FLUSH_LIST
	    && block && buf_flush_ready_for_flush(block, flush_type)) {
	
		block->io_fix = BUF_IO_WRITE;

		/* If AWE is enabled and the page is not mapped to a frame,
		then map it */

		if (block->frame == NULL) {
			ut_a(srv_use_awe);

			/* We set second parameter TRUE because the block is
			in the LRU list and we must put it to
			awe_LRU_free_mapped list once mapped to a frame */
		
			buf_awe_map_page_to_frame(block, TRUE);
		}

		block->flush_type = flush_type;

		if (buf_pool->n_flush[flush_type] == 0) {

			os_event_reset(buf_pool->no_flush[flush_type]);
		}

		(buf_pool->n_flush[flush_type])++;

		locked = FALSE;
		
		/* If the simulated aio thread is not running, we must
		not wait for any latch, as we may end up in a deadlock:
		if buf_fix_count == 0, then we know we need not wait */

		if (block->buf_fix_count == 0) {
			rw_lock_s_lock_gen(&(block->lock), BUF_IO_WRITE);

			locked = TRUE;
		}

		mutex_exit(&(buf_pool->mutex));

		if (!locked) {
			buf_flush_buffered_writes();

			rw_lock_s_lock_gen(&(block->lock), BUF_IO_WRITE);
		}

#ifdef UNIV_DEBUG
		if (buf_debug_prints) {
			fprintf(stderr,
				"Flushing page space %lu, page no %lu \n",
				(ulong) block->space, (ulong) block->offset);
		}
#endif /* UNIV_DEBUG */

		buf_flush_write_block_low(block);
		
		return(1);

	} else if (flush_type == BUF_FLUSH_LRU && block
			&& buf_flush_ready_for_flush(block, flush_type)) {

		/* VERY IMPORTANT:
		Because any thread may call the LRU flush, even when owning
		locks on pages, to avoid deadlocks, we must make sure that the
		s-lock is acquired on the page without waiting: this is
		accomplished because in the if-condition above we require
		the page not to be bufferfixed (in function
		..._ready_for_flush). */

		block->io_fix = BUF_IO_WRITE;

		/* If AWE is enabled and the page is not mapped to a frame,
		then map it */

		if (block->frame == NULL) {
			ut_a(srv_use_awe);

			/* We set second parameter TRUE because the block is
			in the LRU list and we must put it to
			awe_LRU_free_mapped list once mapped to a frame */
		
			buf_awe_map_page_to_frame(block, TRUE);
		}

		block->flush_type = flush_type;

		if (buf_pool->n_flush[flush_type] == 0) {

			os_event_reset(buf_pool->no_flush[flush_type]);
		}

		(buf_pool->n_flush[flush_type])++;

		rw_lock_s_lock_gen(&(block->lock), BUF_IO_WRITE);

		/* Note that the s-latch is acquired before releasing the
		buf_pool mutex: this ensures that the latch is acquired
		immediately. */
		
		mutex_exit(&(buf_pool->mutex));

		buf_flush_write_block_low(block);

		return(1);

	} else if (flush_type == BUF_FLUSH_SINGLE_PAGE && block
			&& buf_flush_ready_for_flush(block, flush_type)) {
	
		block->io_fix = BUF_IO_WRITE;

		/* If AWE is enabled and the page is not mapped to a frame,
		then map it */

		if (block->frame == NULL) {
			ut_a(srv_use_awe);

			/* We set second parameter TRUE because the block is
			in the LRU list and we must put it to
			awe_LRU_free_mapped list once mapped to a frame */
		
			buf_awe_map_page_to_frame(block, TRUE);
		}

		block->flush_type = flush_type;

		if (buf_pool->n_flush[block->flush_type] == 0) {

			os_event_reset(buf_pool->no_flush[block->flush_type]);
		}

		(buf_pool->n_flush[flush_type])++;

		mutex_exit(&(buf_pool->mutex));

		rw_lock_s_lock_gen(&(block->lock), BUF_IO_WRITE);

#ifdef UNIV_DEBUG
		if (buf_debug_prints) {
			fprintf(stderr,
			"Flushing single page space %lu, page no %lu \n",
						(ulong) block->space,
			                        (ulong) block->offset);
		}
#endif /* UNIV_DEBUG */

		buf_flush_write_block_low(block);
		
		return(1);
	} else {
		mutex_exit(&(buf_pool->mutex));

		return(0);
	}		
}

/***************************************************************
Flushes to disk all flushable pages within the flush area. */
static
ulint
buf_flush_try_neighbors(
/*====================*/
				/* out: number of pages flushed */
	ulint	space,		/* in: space id */
	ulint	offset,		/* in: page offset */
	ulint	flush_type)	/* in: BUF_FLUSH_LRU or BUF_FLUSH_LIST */
{
	buf_block_t*	block;
	ulint		low, high;
	ulint		count		= 0;
	ulint		i;

	ut_ad(flush_type == BUF_FLUSH_LRU || flush_type == BUF_FLUSH_LIST);

	low = (offset / BUF_FLUSH_AREA) * BUF_FLUSH_AREA;
	high = (offset / BUF_FLUSH_AREA + 1) * BUF_FLUSH_AREA;

	if (UT_LIST_GET_LEN(buf_pool->LRU) < BUF_LRU_OLD_MIN_LEN) {
		/* If there is little space, it is better not to flush any
		block except from the end of the LRU list */
	
		low = offset;
		high = offset + 1;
	}

	/* fprintf(stderr, "Flush area: low %lu high %lu\n", low, high); */
	
	if (high > fil_space_get_size(space)) {
		high = fil_space_get_size(space);
	}

	mutex_enter(&(buf_pool->mutex));

	for (i = low; i < high; i++) {

		block = buf_page_hash_get(space, i);
		ut_a(!block || block->state == BUF_BLOCK_FILE_PAGE);

		if (block && flush_type == BUF_FLUSH_LRU && i != offset
		    && !block->old) {

		        /* We avoid flushing 'non-old' blocks in an LRU flush,
		        because the flushed blocks are soon freed */

		        continue;
		}

		if (block && buf_flush_ready_for_flush(block, flush_type)
		   && (i == offset || block->buf_fix_count == 0)) {
			/* We only try to flush those neighbors != offset
			where the buf fix count is zero, as we then know that
			we probably can latch the page without a semaphore
			wait. Semaphore waits are expensive because we must
			flush the doublewrite buffer before we start
			waiting. */

			mutex_exit(&(buf_pool->mutex));

			/* Note: as we release the buf_pool mutex above, in
			buf_flush_try_page we cannot be sure the page is still
			in a flushable state: therefore we check it again
			inside that function. */

			count += buf_flush_try_page(space, i, flush_type);

			mutex_enter(&(buf_pool->mutex));
		}
	}
				
	mutex_exit(&(buf_pool->mutex));

	return(count);
}

/***********************************************************************
This utility flushes dirty blocks from the end of the LRU list or flush_list.
NOTE 1: in the case of an LRU flush the calling thread may own latches to
pages: to avoid deadlocks, this function must be written so that it cannot
end up waiting for these latches! NOTE 2: in the case of a flush list flush,
the calling thread is not allowed to own any latches on pages! */

ulint
buf_flush_batch(
/*============*/
				/* out: number of blocks for which the write
				request was queued; ULINT_UNDEFINED if there
				was a flush of the same type already running */
	ulint	flush_type,	/* in: BUF_FLUSH_LRU or BUF_FLUSH_LIST; if
				BUF_FLUSH_LIST, then the caller must not own
				any latches on pages */
	ulint	min_n,		/* in: wished minimum mumber of blocks flushed
				(it is not guaranteed that the actual number
				is that big, though) */
	dulint	lsn_limit)	/* in the case BUF_FLUSH_LIST all blocks whose
				oldest_modification is smaller than this
				should be flushed (if their number does not
				exceed min_n), otherwise ignored */
{
	buf_block_t*	block;
	ulint		page_count 	= 0;
	ulint		old_page_count;
	ulint		space;
	ulint		offset;
	ibool		found;
	
	ut_ad((flush_type == BUF_FLUSH_LRU)
					|| (flush_type == BUF_FLUSH_LIST)); 
	ut_ad((flush_type != BUF_FLUSH_LIST)
					|| sync_thread_levels_empty_gen(TRUE));
	mutex_enter(&(buf_pool->mutex));

	if ((buf_pool->n_flush[flush_type] > 0)
	    || (buf_pool->init_flush[flush_type] == TRUE)) {

		/* There is already a flush batch of the same type running */
		
		mutex_exit(&(buf_pool->mutex));

		return(ULINT_UNDEFINED);
	}

	(buf_pool->init_flush)[flush_type] = TRUE;
	
	for (;;) {
		/* If we have flushed enough, leave the loop */
		if (page_count >= min_n) {

			break;
		}
	
		/* Start from the end of the list looking for a suitable
		block to be flushed. */
		
	    	if (flush_type == BUF_FLUSH_LRU) {
			block = UT_LIST_GET_LAST(buf_pool->LRU);
	    	} else {
			ut_ad(flush_type == BUF_FLUSH_LIST);

			block = UT_LIST_GET_LAST(buf_pool->flush_list);
			if (!block
			    || (ut_dulint_cmp(block->oldest_modification,
			    				lsn_limit) >= 0)) {
				/* We have flushed enough */

				break;
			}
	    	}
	    	
	    	found = FALSE;
	
		/* Note that after finding a single flushable page, we try to
		flush also all its neighbors, and after that start from the
		END of the LRU list or flush list again: the list may change
		during the flushing and we cannot safely preserve within this
		function a pointer to a block in the list! */

	    	while ((block != NULL) && !found) {
			ut_a(block->state == BUF_BLOCK_FILE_PAGE);

			if (buf_flush_ready_for_flush(block, flush_type)) {

				found = TRUE;
				space = block->space;
				offset = block->offset;
	    
				mutex_exit(&(buf_pool->mutex));

				old_page_count = page_count;
				
				/* Try to flush also all the neighbors */
				page_count +=
					buf_flush_try_neighbors(space, offset,
								flush_type);
				/* fprintf(stderr,
				"Flush type %lu, page no %lu, neighb %lu\n",
				flush_type, offset,
				page_count - old_page_count); */

				mutex_enter(&(buf_pool->mutex));

			} else if (flush_type == BUF_FLUSH_LRU) {

				block = UT_LIST_GET_PREV(LRU, block);
			} else {
				ut_ad(flush_type == BUF_FLUSH_LIST);

				block = UT_LIST_GET_PREV(flush_list, block);
			}
	    	}

	    	/* If we could not find anything to flush, leave the loop */

	    	if (!found) {
	    		break;
	    	}
	}

	(buf_pool->init_flush)[flush_type] = FALSE;

	if ((buf_pool->n_flush[flush_type] == 0)
	    && (buf_pool->init_flush[flush_type] == FALSE)) {

		/* The running flush batch has ended */

		os_event_set(buf_pool->no_flush[flush_type]);
	}

	mutex_exit(&(buf_pool->mutex));

	buf_flush_buffered_writes();

#ifdef UNIV_DEBUG
	if (buf_debug_prints && page_count > 0) {
		ut_a(flush_type == BUF_FLUSH_LRU
			|| flush_type == BUF_FLUSH_LIST);
		fprintf(stderr, flush_type == BUF_FLUSH_LRU
			? "Flushed %lu pages in LRU flush\n"
			: "Flushed %lu pages in flush list flush\n",
			(ulong) page_count);
	}
#endif /* UNIV_DEBUG */
	
        if (page_count != ULINT_UNDEFINED)
          srv_buf_pool_flushed+= page_count;

	return(page_count);
}

/**********************************************************************
Waits until a flush batch of the given type ends */

void
buf_flush_wait_batch_end(
/*=====================*/
	ulint	type)	/* in: BUF_FLUSH_LRU or BUF_FLUSH_LIST */
{
	ut_ad((type == BUF_FLUSH_LRU) || (type == BUF_FLUSH_LIST));
	
	os_event_wait(buf_pool->no_flush[type]);
}	

/**********************************************************************
Gives a recommendation of how many blocks should be flushed to establish
a big enough margin of replaceable blocks near the end of the LRU list
and in the free list. */
static
ulint
buf_flush_LRU_recommendation(void)
/*==============================*/
			/* out: number of blocks which should be flushed
			from the end of the LRU list */
{
	buf_block_t*	block;
	ulint		n_replaceable;
	ulint		distance	= 0;
	
	mutex_enter(&(buf_pool->mutex));

	n_replaceable = UT_LIST_GET_LEN(buf_pool->free);

	block = UT_LIST_GET_LAST(buf_pool->LRU);

	while ((block != NULL)
	       && (n_replaceable < BUF_FLUSH_FREE_BLOCK_MARGIN
	       				+ BUF_FLUSH_EXTRA_MARGIN)
	       && (distance < BUF_LRU_FREE_SEARCH_LEN)) {

		if (buf_flush_ready_for_replace(block)) {
			n_replaceable++;
		}

		distance++;
			
		block = UT_LIST_GET_PREV(LRU, block);
	}
	
	mutex_exit(&(buf_pool->mutex));

	if (n_replaceable >= BUF_FLUSH_FREE_BLOCK_MARGIN) {

		return(0);
	}
	
	return(BUF_FLUSH_FREE_BLOCK_MARGIN + BUF_FLUSH_EXTRA_MARGIN
							- n_replaceable);
}

/*************************************************************************
Flushes pages from the end of the LRU list if there is too small a margin
of replaceable pages there or in the free list. VERY IMPORTANT: this function
is called also by threads which have locks on pages. To avoid deadlocks, we
flush only pages such that the s-lock required for flushing can be acquired
immediately, without waiting. */ 

void
buf_flush_free_margin(void)
/*=======================*/
{
	ulint	n_to_flush;
	ulint	n_flushed;

	n_to_flush = buf_flush_LRU_recommendation();
	
	if (n_to_flush > 0) {
		n_flushed = buf_flush_batch(BUF_FLUSH_LRU, n_to_flush,
							ut_dulint_zero);
		if (n_flushed == ULINT_UNDEFINED) {
			/* There was an LRU type flush batch already running;
			let us wait for it to end */
		   
		        buf_flush_wait_batch_end(BUF_FLUSH_LRU);
		}
	}
}

/**********************************************************************
Validates the flush list. */
static
ibool
buf_flush_validate_low(void)
/*========================*/
		/* out: TRUE if ok */
{
	buf_block_t*	block;
	dulint		om;
	
	UT_LIST_VALIDATE(flush_list, buf_block_t, buf_pool->flush_list);

	block = UT_LIST_GET_FIRST(buf_pool->flush_list);

	while (block != NULL) {
		om = block->oldest_modification;
		ut_a(block->state == BUF_BLOCK_FILE_PAGE);
		ut_a(ut_dulint_cmp(om, ut_dulint_zero) > 0);
		
		block = UT_LIST_GET_NEXT(flush_list, block);

		if (block) {
			ut_a(ut_dulint_cmp(om, block->oldest_modification)
									>= 0);
		}
	}

	return(TRUE);
}

/**********************************************************************
Validates the flush list. */

ibool
buf_flush_validate(void)
/*====================*/
		/* out: TRUE if ok */
{
	ibool	ret;
	
	mutex_enter(&(buf_pool->mutex));

	ret = buf_flush_validate_low();
	
	mutex_exit(&(buf_pool->mutex));

	return(ret);
}