All of lore.kernel.org
 help / color / mirror / Atom feed
From: Christoph Hellwig <hch@lst.de>
To: linux-xfs@vger.kernel.org
Subject: [PATCH 08/20] xfs: factor out splitting of an iclog from xlog_sync
Date: Mon,  3 Jun 2019 19:29:33 +0200	[thread overview]
Message-ID: <20190603172945.13819-9-hch@lst.de> (raw)
In-Reply-To: <20190603172945.13819-1-hch@lst.de>

Split out a self-contained chunk of code from xlog_sync that calculates
the split offset for an iclog that wraps the log end and bumps the
cycles for the second half.

Use the chance to bring some sanity to the variables used to track the
split in xlog_sync by not changing the count variable, and instead use
split as the offset for the split and use those to calculate the
sizes and offsets for the two write buffers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/xfs/xfs_log.c | 63 +++++++++++++++++++++++++-----------------------
 1 file changed, 33 insertions(+), 30 deletions(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 217941c4105d..0d8805c9570a 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -1804,6 +1804,32 @@ xlog_write_iclog(
 	xfs_buf_submit(bp);
 }
 
+/*
+ * We need to bump cycle number for the part of the iclog that is
+ * written to the start of the log. Watch out for the header magic
+ * number case, though.
+ */
+static unsigned int
+xlog_split_iclog(
+	struct xlog		*log,
+	void			*data,
+	uint64_t		bno,
+	unsigned int		count)
+{
+	unsigned int		split_offset = BBTOB(log->l_logBBsize - bno);
+	unsigned int		i;
+
+	for (i = split_offset; i < count; i += BBSIZE) {
+		uint32_t cycle = get_unaligned_be32(data + i);
+
+		if (++cycle == XLOG_HEADER_MAGIC_NUM)
+			cycle++;
+		put_unaligned_be32(cycle, data + i);
+	}
+
+	return split_offset;
+}
+
 /*
  * Flush out the in-core log (iclog) to the on-disk log in an asynchronous 
  * fashion.  Previously, we should have moved the current iclog
@@ -1832,13 +1858,12 @@ xlog_sync(
 	struct xlog		*log,
 	struct xlog_in_core	*iclog)
 {
-	int		i;
 	uint		count;		/* byte count of bwrite */
 	uint		count_init;	/* initial count before roundup */
 	int		roundoff;       /* roundoff to BB or stripe */
-	int		split = 0;	/* split write into two regions */
 	int		v2 = xfs_sb_version_haslogv2(&log->l_mp->m_sb);
 	uint64_t	bno;
+	unsigned int	split = 0;
 	int		size;
 	bool		need_flush = true;
 
@@ -1881,32 +1906,8 @@ xlog_sync(
 	bno = BLOCK_LSN(be64_to_cpu(iclog->ic_header.h_lsn));
 
 	/* Do we need to split this write into 2 parts? */
-	if (bno + BTOBB(count) > log->l_logBBsize) {
-		char		*dptr;
-
-		split = count - (BBTOB(log->l_logBBsize - bno));
-		count = BBTOB(log->l_logBBsize - bno);
-		iclog->ic_bwritecnt = 2;
-
-		/*
-		 * Bump the cycle numbers at the start of each block in the
-		 * part of the iclog that ends up in the buffer that gets
-		 * written to the start of the log.
-		 *
-		 * Watch out for the header magic number case, though.
-		 */
-		dptr = (char *)&iclog->ic_header + count;
-		for (i = 0; i < split; i += BBSIZE) {
-			uint32_t cycle = be32_to_cpu(*(__be32 *)dptr);
-			if (++cycle == XLOG_HEADER_MAGIC_NUM)
-				cycle++;
-			*(__be32 *)dptr = cpu_to_be32(cycle);
-
-			dptr += BBSIZE;
-		}
-	} else {
-		iclog->ic_bwritecnt = 1;
-	}
+	if (bno + BTOBB(count) > log->l_logBBsize)
+		split = xlog_split_iclog(log, &iclog->ic_header, bno, count);
 
 	/* calculcate the checksum */
 	iclog->ic_header.h_crc = xlog_cksum(log, &iclog->ic_header,
@@ -1939,14 +1940,16 @@ xlog_sync(
 		need_flush = false;
 	}
 
-	iclog->ic_bp->b_io_length = BTOBB(count);
+	iclog->ic_bp->b_io_length = BTOBB(split ? split : count);
+	iclog->ic_bwritecnt = split ? 2 : 1;
 
 	xlog_verify_iclog(log, iclog, count, true);
 	xlog_write_iclog(log, iclog, iclog->ic_bp, bno, need_flush);
 
 	if (split) {
 		xfs_buf_associate_memory(iclog->ic_log->l_xbuf,
-				(char *)&iclog->ic_header + count, split);
+				(char *)&iclog->ic_header + split,
+				count - split);
 		xlog_write_iclog(log, iclog, iclog->ic_log->l_xbuf, 0, false);
 	}
 }
-- 
2.20.1

  parent reply	other threads:[~2019-06-03 17:30 UTC|newest]

Thread overview: 51+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-06-03 17:29 use bios directly in the log code v2 Christoph Hellwig
2019-06-03 17:29 ` [PATCH 01/20] xfs: remove the no-op spinlock_destroy stub Christoph Hellwig
2019-06-03 17:29 ` [PATCH 02/20] xfs: remove the never used _XBF_COMPOUND flag Christoph Hellwig
2019-06-03 17:29 ` [PATCH 03/20] xfs: renumber XBF_WRITE_FAIL Christoph Hellwig
2019-06-03 17:29 ` [PATCH 04/20] xfs: make mem_to_page available outside of xfs_buf.c Christoph Hellwig
2019-06-03 17:29 ` [PATCH 05/20] xfs: reformat xlog_get_lowest_lsn Christoph Hellwig
2019-06-03 17:29 ` [PATCH 06/20] xfs: don't use REQ_PREFLUSH for split log writes Christoph Hellwig
2019-06-04 16:12   ` Brian Foster
2019-06-04 22:45     ` Dave Chinner
2019-06-05 10:51       ` Brian Foster
2019-06-05 15:14         ` Christoph Hellwig
2019-06-05 15:47           ` Brian Foster
2019-06-03 17:29 ` [PATCH 07/20] xfs: factor out log buffer writing from xlog_sync Christoph Hellwig
2019-06-04  2:20   ` Dave Chinner
2019-06-03 17:29 ` Christoph Hellwig [this message]
2019-06-04  2:21   ` [PATCH 08/20] xfs: factor out splitting of an iclog " Dave Chinner
2019-06-03 17:29 ` [PATCH 09/20] xfs: factor out iclog size calculation " Christoph Hellwig
2019-06-04  2:23   ` Dave Chinner
2019-06-03 17:29 ` [PATCH 10/20] xfs: update both stat counters together in xlog_sync Christoph Hellwig
2019-06-03 17:29 ` [PATCH 11/20] xfs: remove the syncing argument from xlog_verify_iclog Christoph Hellwig
2019-06-03 17:29 ` [PATCH 12/20] xfs: make use of the l_targ field in struct xlog Christoph Hellwig
2019-06-03 17:29 ` [PATCH 13/20] xfs: use bios directly to write log buffers Christoph Hellwig
2019-06-04  5:54   ` Dave Chinner
2019-06-04  6:10     ` Christoph Hellwig
2019-06-03 17:29 ` [PATCH 14/20] xfs: move the log ioend workqueue to struct xlog Christoph Hellwig
2019-06-19 12:19   ` Christoph Hellwig
2019-06-19 22:51     ` Darrick J. Wong
2019-06-20  6:08       ` Christoph Hellwig
2019-06-03 17:29 ` [PATCH 15/20] xfs: return an offset instead of a pointer from xlog_align Christoph Hellwig
2019-06-03 17:29 ` [PATCH 16/20] xfs: use bios directly to read and write the log recovery buffers Christoph Hellwig
2019-06-04  6:13   ` Dave Chinner
2019-06-05 15:09     ` Christoph Hellwig
2019-06-03 17:29 ` [PATCH 17/20] xfs: stop using bp naming for " Christoph Hellwig
2019-06-04  6:19   ` Dave Chinner
2019-06-03 17:29 ` [PATCH 18/20] xfs: remove unused buffer cache APIs Christoph Hellwig
2019-06-04  6:24   ` Dave Chinner
2019-06-05 15:12     ` Christoph Hellwig
2019-06-05 21:24       ` Dave Chinner
2019-06-03 17:29 ` [PATCH 19/20] xfs: properly type the b_log_item field in struct xfs_buf Christoph Hellwig
2019-06-04  6:25   ` Dave Chinner
2019-06-03 17:29 ` [PATCH 20/20] xfs: remove the b_io_length " Christoph Hellwig
2019-06-04  6:27   ` Dave Chinner
2019-06-03 17:35 ` use bios directly in the log code v2 Darrick J. Wong
2019-06-03 17:38   ` Christoph Hellwig
2019-06-04 17:25     ` Darrick J. Wong
2019-06-04 17:54       ` Christoph Hellwig
2019-06-04 18:42         ` Brian Foster
2019-06-04 18:58           ` Christoph Hellwig
  -- strict thread matches above, loose matches on Subject: below --
2019-05-23 17:37 Christoph Hellwig
2019-05-23 17:37 ` [PATCH 08/20] xfs: factor out splitting of an iclog from xlog_sync Christoph Hellwig
2019-05-23 23:17   ` Dave Chinner
2019-05-24  6:16     ` Christoph Hellwig

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190603172945.13819-9-hch@lst.de \
    --to=hch@lst.de \
    --cc=linux-xfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.