From: Jie Liu <jeff.liu@xxxxxxxxxx>
Add source files for xfs_log_rlimit.c The new file is used for log
size calculations and validation shared with userspace.
[dchinner: xfs_log_calc_max_attrsetm_res() does not modify the
tr_attrsetm reservation, just calculates the maximum. ]
[dchinner: rework loop in xfs_log_get_max_trans_res() ]
[dchinner: implement xfs_log_calc_unit_res() in util.c to give mkfs
a worse case calculation of the log size needed. ]
Signed-off-by: Jie Liu <jeff.liu@xxxxxxxxxx>
Signed-off-by: Dave Chinner <dchinner@xxxxxxxxxx>
---
include/xfs_log_format.h | 12 ++++-
libxfs/Makefile | 1 +
libxfs/util.c | 107 ++++++++++++++++++++++++++++++++++++
libxfs/xfs_log_rlimit.c | 135 ++++++++++++++++++++++++++++++++++++++++++++++
mkfs/maxtrres.c | 60 +++++----------------
mkfs/xfs_mkfs.c | 89 +++++++++++++++---------------
mkfs/xfs_mkfs.h | 4 +-
7 files changed, 313 insertions(+), 95 deletions(-)
create mode 100644 libxfs/xfs_log_rlimit.c
diff --git a/include/xfs_log_format.h b/include/xfs_log_format.h
index 9f9aeb6..8f46b6a 100644
--- a/include/xfs_log_format.h
+++ b/include/xfs_log_format.h
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Jie Liu.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -18,6 +19,9 @@
#ifndef __XFS_LOG_FORMAT_H__
#define __XFS_LOG_FORMAT_H__
+struct xfs_mount;
+struct xfs_trans_res;
+
typedef __uint32_t xlog_tid_t;
#define XLOG_MIN_ICLOGS 2
@@ -39,6 +43,9 @@ typedef __uint32_t xlog_tid_t;
#define XLOG_HEADER_SIZE 512
+/* Minimum number of transactions that must fit in the log (defined by mkfs) */
+#define XFS_MIN_LOG_FACTOR 3
+
#define XLOG_REC_SHIFT(log) \
BTOBB(1 << (xfs_sb_version_haslogv2(&log->l_mp->m_sb) ? \
XLOG_MAX_RECORD_BSHIFT : XLOG_BIG_RECORD_BSHIFT))
@@ -123,7 +130,6 @@ typedef struct xlog_op_header {
__u16 oh_res2; /* 32 bit align : 2 b */
} xlog_op_header_t;
-
/* valid values for h_fmt */
#define XLOG_FMT_UNKNOWN 0
#define XLOG_FMT_LINUX_LE 1
@@ -175,4 +181,8 @@ typedef struct xfs_log_iovec {
uint i_type; /* type of region */
} xfs_log_iovec_t;
+int xfs_log_calc_unit_res(struct xfs_mount *mp, int unit_bytes);
+int xfs_log_calc_minimum_size(struct xfs_mount *);
+
+
#endif /* __XFS_LOG_FORMAT_H__ */
diff --git a/libxfs/Makefile b/libxfs/Makefile
index df0a5eb..21520a5 100644
--- a/libxfs/Makefile
+++ b/libxfs/Makefile
@@ -19,6 +19,7 @@ CFILES = cache.c init.c kmem.c logitem.c radix-tree.c rdwr.c trans.c util.c \
xfs_ialloc_btree.c xfs_bmap_btree.c xfs_da_btree.c \
xfs_dir2.c xfs_dir2_leaf.c xfs_attr_leaf.c xfs_dir2_block.c \
xfs_dir2_node.c xfs_dir2_data.c xfs_dir2_sf.c xfs_bmap.c \
+ xfs_log_rlimit.c \
xfs_sb.c xfs_rtalloc.c xfs_attr.c xfs_attr_remote.c \
xfs_symlink_remote.c \
xfs_trans_resv.c \
diff --git a/libxfs/util.c b/libxfs/util.c
index d7459e0..460fcb3 100644
--- a/libxfs/util.c
+++ b/libxfs/util.c
@@ -22,6 +22,113 @@
#include <stdarg.h>
/*
+ * Calculate the worst case log unit reservation for a given superblock
+ * configuration. Copied and munged from the kernel code, and assumes a
+ * worse case header usage (maximum log buffer sizes)
+ */
+int
+xfs_log_calc_unit_res(
+ struct xfs_mount *mp,
+ int unit_bytes)
+{
+ int iclog_space;
+ int iclog_header_size;
+ int iclog_size;
+ uint num_headers;
+
+ if (xfs_sb_version_haslogv2(&mp->m_sb)) {
+ iclog_size = XLOG_MAX_RECORD_BSIZE;
+ iclog_header_size = BTOBB(iclog_size / XLOG_HEADER_CYCLE_SIZE);
+ } else {
+ iclog_size = XLOG_BIG_RECORD_BSIZE;
+ iclog_header_size = BBSIZE;
+ }
+
+ /*
+ * Permanent reservations have up to 'cnt'-1 active log operations
+ * in the log. A unit in this case is the amount of space for one
+ * of these log operations. Normal reservations have a cnt of 1
+ * and their unit amount is the total amount of space required.
+ *
+ * The following lines of code account for non-transaction data
+ * which occupy space in the on-disk log.
+ *
+ * Normal form of a transaction is:
+ * <oph><trans-hdr><start-oph><reg1-oph><reg1><reg2-oph>...<commit-oph>
+ * and then there are LR hdrs, split-recs and roundoff at end of syncs.
+ *
+ * We need to account for all the leadup data and trailer data
+ * around the transaction data.
+ * And then we need to account for the worst case in terms of using
+ * more space.
+ * The worst case will happen if:
+ * - the placement of the transaction happens to be such that the
+ * roundoff is at its maximum
+ * - the transaction data is synced before the commit record is synced
+ * i.e. <transaction-data><roundoff> | <commit-rec><roundoff>
+ * Therefore the commit record is in its own Log Record.
+ * This can happen as the commit record is called with its
+ * own region to xlog_write().
+ * This then means that in the worst case, roundoff can happen for
+ * the commit-rec as well.
+ * The commit-rec is smaller than padding in this scenario and so it is
+ * not added separately.
+ */
+
+ /* for trans header */
+ unit_bytes += sizeof(xlog_op_header_t);
+ unit_bytes += sizeof(xfs_trans_header_t);
+
+ /* for start-rec */
+ unit_bytes += sizeof(xlog_op_header_t);
+
+ /*
+ * for LR headers - the space for data in an iclog is the size minus
+ * the space used for the headers. If we use the iclog size, then we
+ * undercalculate the number of headers required.
+ *
+ * Furthermore - the addition of op headers for split-recs might
+ * increase the space required enough to require more log and op
+ * headers, so take that into account too.
+ *
+ * IMPORTANT: This reservation makes the assumption that if this
+ * transaction is the first in an iclog and hence has the LR headers
+ * accounted to it, then the remaining space in the iclog is
+ * exclusively for this transaction. i.e. if the transaction is larger
+ * than the iclog, it will be the only thing in that iclog.
+ * Fundamentally, this means we must pass the entire log vector to
+ * xlog_write to guarantee this.
+ */
+ iclog_space = iclog_size - iclog_header_size;
+ num_headers = howmany(unit_bytes, iclog_space);
+
+ /* for split-recs - ophdrs added when data split over LRs */
+ unit_bytes += sizeof(xlog_op_header_t) * num_headers;
+
+ /* add extra header reservations if we overrun */
+ while (!num_headers ||
+ howmany(unit_bytes, iclog_space) > num_headers) {
+ unit_bytes += sizeof(xlog_op_header_t);
+ num_headers++;
+ }
+ unit_bytes += iclog_header_size * num_headers;
+
+ /* for commit-rec LR header - note: padding will subsume the ophdr */
+ unit_bytes += iclog_header_size;
+
+ /* for roundoff padding for transaction data and one for commit record */
+ if (xfs_sb_version_haslogv2(&mp->m_sb) && mp->m_sb.sb_logsunit > 1) {
+ /* log su roundoff */
+ unit_bytes += 2 * mp->m_sb.sb_logsunit;
+ } else {
+ /* BB roundoff */
+ unit_bytes += 2 * BBSIZE;
+ }
+
+ return unit_bytes;
+}
+
+/*
* Change the requested timestamp in the given inode.
*
* This was once shared with the kernel, but has diverged to the point
diff --git a/libxfs/xfs_log_rlimit.c b/libxfs/xfs_log_rlimit.c
new file mode 100644
index 0000000..5b5edc5
--- /dev/null
+++ b/libxfs/xfs_log_rlimit.c
@@ -0,0 +1,135 @@
+/*
+ * Copyright (c) 2013 Jie Liu.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+#include <xfs.h>
+
+/*
+ * Calculate the maximum length in bytes that would be required for a local
+ * attribute value as large attributes out of line are not logged.
+ */
+STATIC int
+xfs_log_calc_max_attrsetm_res(
+ struct xfs_mount *mp)
+{
+ int size;
+ int nblks;
+
+ size = xfs_attr_leaf_entsize_local_max(mp->m_sb.sb_blocksize) -
+ MAXNAMELEN - 1;
+ nblks = XFS_DAENTER_SPACE_RES(mp, XFS_ATTR_FORK);
+ nblks += XFS_B_TO_FSB(mp, size);
+ nblks += XFS_NEXTENTADD_SPACE_RES(mp, size, XFS_ATTR_FORK);
+
+ return M_RES(mp)->tr_attrsetm.tr_logres +
+ M_RES(mp)->tr_attrsetrt.tr_logres * nblks;
+}
+
+/*
+ * Iterate over the log space reservation table to figure out and return
+ * the maximum one in terms of the pre-calculated values which were done
+ * at mount time.
+ */
+STATIC void
+xfs_log_get_max_trans_res(
+ struct xfs_mount *mp,
+ struct xfs_trans_res *max_resp)
+{
+ struct xfs_trans_res *resp;
+ struct xfs_trans_res *end_resp;
+ int log_space = 0;
+ int attr_space;
+
+ attr_space = xfs_log_calc_max_attrsetm_res(mp);
+
+ resp = (struct xfs_trans_res *)M_RES(mp);
+ end_resp = (struct xfs_trans_res *)(M_RES(mp) + 1);
+ for (; resp < end_resp; resp++) {
+ int tmp = resp->tr_logcount > 1 ?
+ resp->tr_logres * resp->tr_logcount :
+ resp->tr_logres;
+ if (log_space < tmp) {
+ log_space = tmp;
+ *max_resp = *resp; /* struct copy */
+ }
+ }
+
+ if (attr_space > log_space) {
+ *max_resp = M_RES(mp)->tr_attrsetm; /* struct copy */
+ max_resp->tr_logres = attr_space;
+ }
+}
+
+
+/*
+ * Calculate the minimum valid log size for the given superblock configuration.
+ * Used to calculate the minimum log size at mkfs time, and to determine if
+ * the log is large enough or not at mount time. Returns the minimum size in
+ * filesystem block size units.
+ */
+int
+xfs_log_calc_minimum_size(
+ struct xfs_mount *mp)
+{
+ struct xfs_trans_res tres = {0};
+ int max_logres;
+ int min_logblks = 0;
+ int lsunit = 0;
+
+ xfs_log_get_max_trans_res(mp, &tres);
+
+ max_logres = xfs_log_calc_unit_res(mp, tres.tr_logres);
+ if (tres.tr_logcount > 1)
+ max_logres *= tres.tr_logcount;
+
+ if (xfs_sb_version_haslogv2(&mp->m_sb) && mp->m_sb.sb_logsunit > 1)
+ lsunit = BTOBB(mp->m_sb.sb_logsunit);
+
+ /*
+ * Two factors should be taken into account for calculating the minimum
+ * log space.
+ * 1) The fundamental limitation is that no single transaction can be
+ * larger than half size of the log.
+ *
+ * From mkfs.xfs, this is considered by the XFS_MIN_LOG_FACTOR
+ * define, which is set to 3. That means we can definitely fit
+ * maximally sized 2 transactions in the log. We'll use this same
+ * value here.
+ *
+ * 2) If the lsunit option is specified, a transaction requires 2 LSU
+ * for the reservation because there are two log writes that can
+ * require padding - the transaction data and the commit record which
+ * are written separately and both can require padding to the LSU.
+ * Consider that we can have an active CIL reservation holding 2*LSU,
+ * but the CIL is not over a push threshold, in this case, if we
+ * don't have enough log space for at one new transaction, which
+ * includes another 2*LSU in the reservation, we will run into dead
+ * loop situation in log space grant procedure. i.e.
+ * xlog_grant_head_wait().
+ *
+ * Hence the log size needs to be able to contain two maximally sized
+ * and padded transactions, which is (2 * (2 * LSU + maxlres)).
+ *
+ * Also, the log size should be a multiple of the log stripe unit, round
+ * it up to lsunit boundary if lsunit is specified.
+ */
+ if (lsunit)
+ min_logblks = roundup(BTOBB(max_logres), lsunit) + 2 * lsunit;
+ else
+ min_logblks = BTOBB(max_logres);
+ min_logblks *= XFS_MIN_LOG_FACTOR;
+ return XFS_BB_TO_FSB(mp, min_logblks);
+}
diff --git a/mkfs/maxtrres.c b/mkfs/maxtrres.c
index a45e190..c8cb025 100644
--- a/mkfs/maxtrres.c
+++ b/mkfs/maxtrres.c
@@ -27,46 +27,6 @@
#include <xfs/libxfs.h>
#include "xfs_mkfs.h"
-static void
-max_attrsetm_trans_res_adjust(
- xfs_mount_t *mp)
-{
- int local;
- int size;
- int nblks;
- int res;
-
- /*
- * Determine space the maximal sized attribute will use,
- * to calculate the largest reservation size needed.
- */
- size = libxfs_attr_leaf_newentsize(MAXNAMELEN, 64 * 1024,
- mp->m_sb.sb_blocksize, &local);
- ASSERT(!local);
- nblks = XFS_DAENTER_SPACE_RES(mp, XFS_ATTR_FORK);
- nblks += XFS_B_TO_FSB(mp, size);
- nblks += XFS_NEXTENTADD_SPACE_RES(mp, size, XFS_ATTR_FORK);
- res = M_RES(mp)->tr_attrsetm.tr_logres +
- M_RES(mp)->tr_attrsetrt.tr_logres * nblks;
- M_RES(mp)->tr_attrsetm.tr_logres = res;
-}
-
-static int
-max_trans_res_by_mount(
- xfs_mount_t *mp)
-{
- struct xfs_trans_resv *tr = &mp->m_resv;
- struct xfs_trans_res *p;
- struct xfs_trans_res rval = {0};
-
- for (p = (struct xfs_trans_res *)tr;
- p < (struct xfs_trans_res *)(tr + 1); p++) {
- if (p->tr_logres > rval.tr_logres)
- rval = *p;
- }
- return rval.tr_logres;
-}
-
int
max_trans_res(
int crcs_enabled,
@@ -74,11 +34,12 @@ max_trans_res(
int sectorlog,
int blocklog,
int inodelog,
- int dirblocklog)
+ int dirblocklog,
+ int log_sunit)
{
xfs_sb_t *sbp;
xfs_mount_t mount;
- int maxres, maxfsb;
+ int maxfsb;
memset(&mount, 0, sizeof(mount));
sbp = &mount.m_sb;
@@ -93,19 +54,22 @@ max_trans_res(
sbp->sb_inodesize = 1 << inodelog;
sbp->sb_inopblock = 1 << (blocklog - inodelog);
sbp->sb_dirblklog = dirblocklog - blocklog;
+
+ log_sunit = (log_sunit == 0) ? 1 : XFS_FSB_TO_B(&mount, log_sunit);
+ sbp->sb_logsunit = log_sunit;
sbp->sb_versionnum =
(crcs_enabled ? XFS_SB_VERSION_5 : XFS_SB_VERSION_4) |
- (dirversion == 2 ? XFS_SB_VERSION_DIRV2BIT : 0);
+ (dirversion == 2 ? XFS_SB_VERSION_DIRV2BIT : 0) |
+ (log_sunit > 1 ? XFS_SB_VERSION_LOGV2BIT : 0);
libxfs_mount(&mount, sbp, 0,0,0,0);
- max_attrsetm_trans_res_adjust(&mount);
- maxres = max_trans_res_by_mount(&mount);
- maxfsb = XFS_B_TO_FSB(&mount, maxres);
+ maxfsb = xfs_log_calc_minimum_size(&mount);
libxfs_umount(&mount);
#if 0
- printf("#define\tMAXTRRES_S%d_B%d_I%d_D%d_V%d\t%lld\n",
- sectorlog, blocklog, inodelog, dirblocklog, dirversion, maxfsb);
+ printf("#define\tMAXTRRES_S%d_B%d_I%d_D%d_V%d_LSU%d\t%d\n",
+ sectorlog, blocklog, inodelog, dirblocklog, dirversion,
+ log_sunit, maxfsb);
#endif
return maxfsb;
diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
index a940150..bcf099f 100644
--- a/mkfs/xfs_mkfs.c
+++ b/mkfs/xfs_mkfs.c
@@ -2111,50 +2111,6 @@ reported by the device (%u).\n"),
sectorsize, xi.rtbsize);
}
- max_tr_res = max_trans_res(crcs_enabled, dirversion,
- sectorlog, blocklog, inodelog, dirblocklog);
- ASSERT(max_tr_res);
- min_logblocks = max_tr_res * XFS_MIN_LOG_FACTOR;
- min_logblocks = MAX(XFS_MIN_LOG_BLOCKS, min_logblocks);
- if (!logsize && dblocks >= (1024*1024*1024) >> blocklog)
- min_logblocks = MAX(min_logblocks, XFS_MIN_LOG_BYTES>>blocklog);
- if (logsize && xi.logBBsize > 0 && logblocks > DTOBT(xi.logBBsize)) {
- fprintf(stderr,
-_("size %s specified for log subvolume is too large, maximum is %lld blocks\n"),
- logsize, (long long)DTOBT(xi.logBBsize));
- usage();
- } else if (!logsize && xi.logBBsize > 0) {
- logblocks = DTOBT(xi.logBBsize);
- } else if (logsize && !xi.logdev && !loginternal) {
- fprintf(stderr,
- _("size specified for non-existent log subvolume\n"));
- usage();
- } else if (loginternal && logsize && logblocks >= dblocks) {
- fprintf(stderr, _("size %lld too large for internal log\n"),
- (long long)logblocks);
- usage();
- } else if (!loginternal && !xi.logdev) {
- logblocks = 0;
- } else if (loginternal && !logsize) {
- /*
- * With a 2GB max log size, default to maximum size
- * at 4TB. This keeps the same ratio from the older
- * max log size of 128M at 256GB fs size. IOWs,
- * the ratio of fs size to log size is 2048:1.
- */
- logblocks = (dblocks << blocklog) / 2048;
- logblocks = logblocks >> blocklog;
- logblocks = MAX(min_logblocks, logblocks);
- logblocks = MAX(logblocks,
- MAX(XFS_DFL_LOG_SIZE,
- max_tr_res * XFS_DFL_LOG_FACTOR));
- logblocks = MIN(logblocks, XFS_MAX_LOG_BLOCKS);
- if ((logblocks << blocklog) > XFS_MAX_LOG_BYTES) {
- logblocks = XFS_MAX_LOG_BYTES >> blocklog;
- }
- }
- validate_log_size(logblocks, blocklog, min_logblocks);
-
if (rtsize && xi.rtsize > 0 && rtblocks > DTOBT(xi.rtsize)) {
fprintf(stderr,
_("size %s specified for rt subvolume is too large, "
@@ -2363,6 +2319,51 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
fprintf(stderr, _("log stripe unit adjusted to 32KiB\n"));
}
+ max_tr_res = max_trans_res(crcs_enabled, dirversion,
+ sectorlog, blocklog, inodelog, dirblocklog,
+ lsunit);
+ ASSERT(max_tr_res);
+ min_logblocks = max_tr_res * XFS_MIN_LOG_FACTOR;
+ min_logblocks = MAX(XFS_MIN_LOG_BLOCKS, min_logblocks);
+ if (!logsize && dblocks >= (1024*1024*1024) >> blocklog)
+ min_logblocks = MAX(min_logblocks, XFS_MIN_LOG_BYTES>>blocklog);
+ if (logsize && xi.logBBsize > 0 && logblocks > DTOBT(xi.logBBsize)) {
+ fprintf(stderr,
+_("size %s specified for log subvolume is too large, maximum is %lld blocks\n"),
+ logsize, (long long)DTOBT(xi.logBBsize));
+ usage();
+ } else if (!logsize && xi.logBBsize > 0) {
+ logblocks = DTOBT(xi.logBBsize);
+ } else if (logsize && !xi.logdev && !loginternal) {
+ fprintf(stderr,
+ _("size specified for non-existent log subvolume\n"));
+ usage();
+ } else if (loginternal && logsize && logblocks >= dblocks) {
+ fprintf(stderr, _("size %lld too large for internal log\n"),
+ (long long)logblocks);
+ usage();
+ } else if (!loginternal && !xi.logdev) {
+ logblocks = 0;
+ } else if (loginternal && !logsize) {
+ /*
+ * With a 2GB max log size, default to maximum size
+ * at 4TB. This keeps the same ratio from the older
+ * max log size of 128M at 256GB fs size. IOWs,
+ * the ratio of fs size to log size is 2048:1.
+ */
+ logblocks = (dblocks << blocklog) / 2048;
+ logblocks = logblocks >> blocklog;
+ logblocks = MAX(min_logblocks, logblocks);
+ logblocks = MAX(logblocks,
+ MAX(XFS_DFL_LOG_SIZE,
+ max_tr_res * XFS_DFL_LOG_FACTOR));
+ logblocks = MIN(logblocks, XFS_MAX_LOG_BLOCKS);
+ if ((logblocks << blocklog) > XFS_MAX_LOG_BYTES) {
+ logblocks = XFS_MAX_LOG_BYTES >> blocklog;
+ }
+ }
+ validate_log_size(logblocks, blocklog, min_logblocks);
+
protostring = setup_proto(protofile);
bsize = 1 << (blocklog - BBSHIFT);
mp = &mbuf;
diff --git a/mkfs/xfs_mkfs.h b/mkfs/xfs_mkfs.h
index d10e444..81779aa 100644
--- a/mkfs/xfs_mkfs.h
+++ b/mkfs/xfs_mkfs.h
@@ -54,7 +54,6 @@
#define XFS_MIN_REC_DIRSIZE 12 /* 4096 byte dirblocks (V2) */
#define XFS_DFL_DIR_VERSION 2 /* default directory version */
#define XFS_DFL_LOG_SIZE 1000 /* default log size, blocks */
-#define XFS_MIN_LOG_FACTOR 3 /* min log size factor */
#define XFS_DFL_LOG_FACTOR 16 /* default log size, factor */
/* with max trans reservation */
#define XFS_MAX_INODE_SIG_BITS 32 /* most significant bits in an
@@ -82,6 +81,7 @@ extern void res_failed (int err);
/* maxtrres.c */
extern int max_trans_res (int crcs_enabled, int dirversion,
- int sectorlog, int blocklog, int inodelog, int dirblocklog);
+ int sectorlog, int blocklog, int inodelog, int dirblocklogi,
+ int log_sunit);
#endif /* __XFS_MKFS_H__ */
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
