xfs_buf_item.c

linux 内核源代码

	bip->bli_flags |= XFS_BLI_DIRTY;

	/*
	 * Convert byte offsets to bit numbers.
	 */
	first_bit = first >> XFS_BLI_SHIFT;
	last_bit = last >> XFS_BLI_SHIFT;

	/*
	 * Calculate the total number of bits to be set.
	 */
	bits_to_set = last_bit - first_bit + 1;

	/*
	 * Get a pointer to the first word in the bitmap
	 * to set a bit in.
	 */
	word_num = first_bit >> BIT_TO_WORD_SHIFT;
	wordp = &(bip->bli_format.blf_data_map[word_num]);

	/*
	 * Calculate the starting bit in the first word.
	 */
	bit = first_bit & (uint)(NBWORD - 1);

	/*
	 * First set any bits in the first word of our range.
	 * If it starts at bit 0 of the word, it will be
	 * set below rather than here.  That is what the variable
	 * bit tells us. The variable bits_set tracks the number
	 * of bits that have been set so far.  End_bit is the number
	 * of the last bit to be set in this word plus one.
	 */
	if (bit) {
		end_bit = MIN(bit + bits_to_set, (uint)NBWORD);
		mask = ((1 << (end_bit - bit)) - 1) << bit;
		*wordp |= mask;
		wordp++;
		bits_set = end_bit - bit;
	} else {
		bits_set = 0;
	}

	/*
	 * Now set bits a whole word at a time that are between
	 * first_bit and last_bit.
	 */
	while ((bits_to_set - bits_set) >= NBWORD) {
		*wordp |= 0xffffffff;
		bits_set += NBWORD;
		wordp++;
	}

	/*
	 * Finally, set any bits left to be set in one last partial word.
	 */
	end_bit = bits_to_set - bits_set;
	if (end_bit) {
		mask = (1 << end_bit) - 1;
		*wordp |= mask;
	}

	xfs_buf_item_log_debug(bip, first, last);
}


/*
 * Return 1 if the buffer has some data that has been logged (at any
 * point, not just the current transaction) and 0 if not.
 */
uint
xfs_buf_item_dirty(
	xfs_buf_log_item_t	*bip)
{
	return (bip->bli_flags & XFS_BLI_DIRTY);
}

/*
 * This is called when the buf log item is no longer needed.  It should
 * free the buf log item associated with the given buffer and clear
 * the buffer's pointer to the buf log item.  If there are no more
 * items in the list, clear the b_iodone field of the buffer (see
 * xfs_buf_attach_iodone() below).
 */
void
xfs_buf_item_relse(
	xfs_buf_t	*bp)
{
	xfs_buf_log_item_t	*bip;

	xfs_buftrace("XFS_RELSE", bp);
	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t*);
	XFS_BUF_SET_FSPRIVATE(bp, bip->bli_item.li_bio_list);
	if ((XFS_BUF_FSPRIVATE(bp, void *) == NULL) &&
	    (XFS_BUF_IODONE_FUNC(bp) != NULL)) {
		XFS_BUF_CLR_IODONE_FUNC(bp);
	}

#ifdef XFS_TRANS_DEBUG
	kmem_free(bip->bli_orig, XFS_BUF_COUNT(bp));
	bip->bli_orig = NULL;
	kmem_free(bip->bli_logged, XFS_BUF_COUNT(bp) / NBBY);
	bip->bli_logged = NULL;
#endif /* XFS_TRANS_DEBUG */

#ifdef XFS_BLI_TRACE
	ktrace_free(bip->bli_trace);
#endif
	kmem_zone_free(xfs_buf_item_zone, bip);
}


/*
 * Add the given log item with its callback to the list of callbacks
 * to be called when the buffer's I/O completes.  If it is not set
 * already, set the buffer's b_iodone() routine to be
 * xfs_buf_iodone_callbacks() and link the log item into the list of
 * items rooted at b_fsprivate.  Items are always added as the second
 * entry in the list if there is a first, because the buf item code
 * assumes that the buf log item is first.
 */
void
xfs_buf_attach_iodone(
	xfs_buf_t	*bp,
	void		(*cb)(xfs_buf_t *, xfs_log_item_t *),
	xfs_log_item_t	*lip)
{
	xfs_log_item_t	*head_lip;

	ASSERT(XFS_BUF_ISBUSY(bp));
	ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);

	lip->li_cb = cb;
	if (XFS_BUF_FSPRIVATE(bp, void *) != NULL) {
		head_lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
		lip->li_bio_list = head_lip->li_bio_list;
		head_lip->li_bio_list = lip;
	} else {
		XFS_BUF_SET_FSPRIVATE(bp, lip);
	}

	ASSERT((XFS_BUF_IODONE_FUNC(bp) == xfs_buf_iodone_callbacks) ||
	       (XFS_BUF_IODONE_FUNC(bp) == NULL));
	XFS_BUF_SET_IODONE_FUNC(bp, xfs_buf_iodone_callbacks);
}

STATIC void
xfs_buf_do_callbacks(
	xfs_buf_t	*bp,
	xfs_log_item_t	*lip)
{
	xfs_log_item_t	*nlip;

	while (lip != NULL) {
		nlip = lip->li_bio_list;
		ASSERT(lip->li_cb != NULL);
		/*
		 * Clear the next pointer so we don't have any
		 * confusion if the item is added to another buf.
		 * Don't touch the log item after calling its
		 * callback, because it could have freed itself.
		 */
		lip->li_bio_list = NULL;
		lip->li_cb(bp, lip);
		lip = nlip;
	}
}

/*
 * This is the iodone() function for buffers which have had callbacks
 * attached to them by xfs_buf_attach_iodone().  It should remove each
 * log item from the buffer's list and call the callback of each in turn.
 * When done, the buffer's fsprivate field is set to NULL and the buffer
 * is unlocked with a call to iodone().
 */
void
xfs_buf_iodone_callbacks(
	xfs_buf_t	*bp)
{
	xfs_log_item_t	*lip;
	static ulong	lasttime;
	static xfs_buftarg_t *lasttarg;
	xfs_mount_t	*mp;

	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
	lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);

	if (XFS_BUF_GETERROR(bp) != 0) {
		/*
		 * If we've already decided to shutdown the filesystem
		 * because of IO errors, there's no point in giving this
		 * a retry.
		 */
		mp = lip->li_mountp;
		if (XFS_FORCED_SHUTDOWN(mp)) {
			ASSERT(XFS_BUF_TARGET(bp) == mp->m_ddev_targp);
			XFS_BUF_SUPER_STALE(bp);
			xfs_buftrace("BUF_IODONE_CB", bp);
			xfs_buf_do_callbacks(bp, lip);
			XFS_BUF_SET_FSPRIVATE(bp, NULL);
			XFS_BUF_CLR_IODONE_FUNC(bp);

			/*
			 * XFS_SHUT flag gets set when we go thru the
			 * entire buffer cache and deliberately start
			 * throwing away delayed write buffers.
			 * Since there's no biowait done on those,
			 * we should just brelse them.
			 */
			if (XFS_BUF_ISSHUT(bp)) {
			    XFS_BUF_UNSHUT(bp);
				xfs_buf_relse(bp);
			} else {
				xfs_biodone(bp);
			}

			return;
		}

		if ((XFS_BUF_TARGET(bp) != lasttarg) ||
		    (time_after(jiffies, (lasttime + 5*HZ)))) {
			lasttime = jiffies;
			cmn_err(CE_ALERT, "Device %s, XFS metadata write error"
					" block 0x%llx in %s",
				XFS_BUFTARG_NAME(XFS_BUF_TARGET(bp)),
			      (__uint64_t)XFS_BUF_ADDR(bp), mp->m_fsname);
		}
		lasttarg = XFS_BUF_TARGET(bp);

		if (XFS_BUF_ISASYNC(bp)) {
			/*
			 * If the write was asynchronous then noone will be
			 * looking for the error.  Clear the error state
			 * and write the buffer out again delayed write.
			 *
			 * XXXsup This is OK, so long as we catch these
			 * before we start the umount; we don't want these
			 * DELWRI metadata bufs to be hanging around.
			 */
			XFS_BUF_ERROR(bp,0); /* errno of 0 unsets the flag */

			if (!(XFS_BUF_ISSTALE(bp))) {
				XFS_BUF_DELAYWRITE(bp);
				XFS_BUF_DONE(bp);
				XFS_BUF_SET_START(bp);
			}
			ASSERT(XFS_BUF_IODONE_FUNC(bp));
			xfs_buftrace("BUF_IODONE ASYNC", bp);
			xfs_buf_relse(bp);
		} else {
			/*
			 * If the write of the buffer was not asynchronous,
			 * then we want to make sure to return the error
			 * to the caller of bwrite().  Because of this we
			 * cannot clear the B_ERROR state at this point.
			 * Instead we install a callback function that
			 * will be called when the buffer is released, and
			 * that routine will clear the error state and
			 * set the buffer to be written out again after
			 * some delay.
			 */
			/* We actually overwrite the existing b-relse
			   function at times, but we're gonna be shutting down
			   anyway. */
			XFS_BUF_SET_BRELSE_FUNC(bp,xfs_buf_error_relse);
			XFS_BUF_DONE(bp);
			XFS_BUF_V_IODONESEMA(bp);
		}
		return;
	}
#ifdef XFSERRORDEBUG
	xfs_buftrace("XFS BUFCB NOERR", bp);
#endif
	xfs_buf_do_callbacks(bp, lip);
	XFS_BUF_SET_FSPRIVATE(bp, NULL);
	XFS_BUF_CLR_IODONE_FUNC(bp);
	xfs_biodone(bp);
}

/*
 * This is a callback routine attached to a buffer which gets an error
 * when being written out synchronously.
 */
STATIC void
xfs_buf_error_relse(
	xfs_buf_t	*bp)
{
	xfs_log_item_t	*lip;
	xfs_mount_t	*mp;

	lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
	mp = (xfs_mount_t *)lip->li_mountp;
	ASSERT(XFS_BUF_TARGET(bp) == mp->m_ddev_targp);

	XFS_BUF_STALE(bp);
	XFS_BUF_DONE(bp);
	XFS_BUF_UNDELAYWRITE(bp);
	XFS_BUF_ERROR(bp,0);
	xfs_buftrace("BUF_ERROR_RELSE", bp);
	if (! XFS_FORCED_SHUTDOWN(mp))
		xfs_force_shutdown(mp, SHUTDOWN_META_IO_ERROR);
	/*
	 * We have to unpin the pinned buffers so do the
	 * callbacks.
	 */
	xfs_buf_do_callbacks(bp, lip);
	XFS_BUF_SET_FSPRIVATE(bp, NULL);
	XFS_BUF_CLR_IODONE_FUNC(bp);
	XFS_BUF_SET_BRELSE_FUNC(bp,NULL);
	xfs_buf_relse(bp);
}


/*
 * This is the iodone() function for buffers which have been
 * logged.  It is called when they are eventually flushed out.
 * It should remove the buf item from the AIL, and free the buf item.
 * It is called by xfs_buf_iodone_callbacks() above which will take
 * care of cleaning up the buffer itself.
 */
/* ARGSUSED */
void
xfs_buf_iodone(
	xfs_buf_t		*bp,
	xfs_buf_log_item_t	*bip)
{
	struct xfs_mount	*mp;
	SPLDECL(s);

	ASSERT(bip->bli_buf == bp);

	mp = bip->bli_item.li_mountp;

	/*
	 * If we are forcibly shutting down, this may well be
	 * off the AIL already. That's because we simulate the
	 * log-committed callbacks to unpin these buffers. Or we may never
	 * have put this item on AIL because of the transaction was
	 * aborted forcibly. xfs_trans_delete_ail() takes care of these.
	 *
	 * Either way, AIL is useless if we're forcing a shutdown.
	 */
	AIL_LOCK(mp,s);
	/*
	 * xfs_trans_delete_ail() drops the AIL lock.
	 */
	xfs_trans_delete_ail(mp, (xfs_log_item_t *)bip, s);

#ifdef XFS_TRANS_DEBUG
	kmem_free(bip->bli_orig, XFS_BUF_COUNT(bp));
	bip->bli_orig = NULL;
	kmem_free(bip->bli_logged, XFS_BUF_COUNT(bp) / NBBY);
	bip->bli_logged = NULL;
#endif /* XFS_TRANS_DEBUG */

#ifdef XFS_BLI_TRACE
	ktrace_free(bip->bli_trace);
#endif
	kmem_zone_free(xfs_buf_item_zone, bip);
}

#if defined(XFS_BLI_TRACE)
void
xfs_buf_item_trace(
	char			*id,
	xfs_buf_log_item_t	*bip)
{
	xfs_buf_t		*bp;
	ASSERT(bip->bli_trace != NULL);

	bp = bip->bli_buf;
	ktrace_enter(bip->bli_trace,
		     (void *)id,
		     (void *)bip->bli_buf,
		     (void *)((unsigned long)bip->bli_flags),
		     (void *)((unsigned long)bip->bli_recur),
		     (void *)((unsigned long)atomic_read(&bip->bli_refcount)),
		     (void *)((unsigned long)
				(0xFFFFFFFF & XFS_BUF_ADDR(bp) >> 32)),
		     (void *)((unsigned long)(0xFFFFFFFF & XFS_BUF_ADDR(bp))),
		     (void *)((unsigned long)XFS_BUF_COUNT(bp)),
		     (void *)((unsigned long)XFS_BUF_BFLAGS(bp)),
		     XFS_BUF_FSPRIVATE(bp, void *),
		     XFS_BUF_FSPRIVATE2(bp, void *),
		     (void *)(unsigned long)XFS_BUF_ISPINNED(bp),
		     (void *)XFS_BUF_IODONE_FUNC(bp),
		     (void *)((unsigned long)(XFS_BUF_VALUSEMA(bp))),
		     (void *)bip->bli_item.li_desc,
		     (void *)((unsigned long)bip->bli_item.li_flags));
}
#endif /* XFS_BLI_TRACE */