[PATCH 0/4] xfs: fixes and new tests for bulkstat v5

[PATCH 0/4] xfs: fixes and new tests for bulkstat v5

[Darrick J. Wong]

Hi all,

Fix some problems introduced by the creation of the V5 bulkstat ioctl,
and then add some new tests to make sure the new libxfrog bulkstat
wrappers work fine with both the new v5 ioctl and emulating it with the
old v1 ioctl.

If you're going to start using this mess, you probably ought to just
pull from my git trees, which are linked below.

This is an extraordinary way to destroy everything.  Enjoy!
Comments and questions are, as always, welcome.

--D

kernel git tree:
https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=bulkstat-v5

xfsprogs git tree:
https://git.kernel.org/cgit/linux/kernel/git/djwong/xfsprogs-dev.git/log/?h=bulkstat-v5

fstests git tree:
https://git.kernel.org/cgit/linux/kernel/git/djwong/xfstests-dev.git/log/?h=bulkstat-v5

[PATCH 1/4] xfs/122: ignore inode geometry structure

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Ignore new in-core structure.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 tests/xfs/122 |    1 +
 1 file changed, 1 insertion(+)


diff --git a/tests/xfs/122 b/tests/xfs/122
index b66b78a6..89a39a23 100755
--- a/tests/xfs/122
+++ b/tests/xfs/122
@@ -177,6 +177,7 @@ xfs_dirent_t
 xfs_fsop_getparents_handlereq_t
 xfs_dinode_core_t
 struct xfs_iext_cursor
+struct xfs_ino_geometry
 EOF
 
 echo 'int main(int argc, char *argv[]) {' >>$cprog

Re: [PATCH 1/4] xfs/122: ignore inode geometry structure

[Brian Foster]

On Tue, Jul 23, 2019 at 09:13:20PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Ignore new in-core structure.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---

Reviewed-by: Brian Foster <bfoster@redhat.com>

>  tests/xfs/122 |    1 +
>  1 file changed, 1 insertion(+)
> 
> 
> diff --git a/tests/xfs/122 b/tests/xfs/122
> index b66b78a6..89a39a23 100755
> --- a/tests/xfs/122
> +++ b/tests/xfs/122
> @@ -177,6 +177,7 @@ xfs_dirent_t
>  xfs_fsop_getparents_handlereq_t
>  xfs_dinode_core_t
>  struct xfs_iext_cursor
> +struct xfs_ino_geometry
>  EOF
>  
>  echo 'int main(int argc, char *argv[]) {' >>$cprog
> 

[PATCH 2/4] xfs/122: mask wonky ioctls

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Don't check the structure size of the inogrp/bstat/fsop_bulkreq
structures because they're incorrectly padded.  When we remove the
old typdefs the old filter stops working.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 tests/xfs/122 |    3 +++
 1 file changed, 3 insertions(+)


diff --git a/tests/xfs/122 b/tests/xfs/122
index 89a39a23..64b63cb1 100755
--- a/tests/xfs/122
+++ b/tests/xfs/122
@@ -148,12 +148,15 @@ xfs_growfs_data_t
 xfs_growfs_rt_t
 xfs_bstime_t
 xfs_bstat_t
+struct xfs_bstat
 xfs_fsop_bulkreq_t
+struct xfs_fsop_bulkreq
 xfs_icsb_cnts_t
 xfs_icdinode_t
 xfs_ictimestamp_t
 xfs_inobt_rec_incore_t
 xfs_inogrp_t
+struct xfs_inogrp
 xfs_fid2_t
 xfs_fsop_handlereq_t
 xfs_fsop_setdm_handlereq_t

Re: [PATCH 2/4] xfs/122: mask wonky ioctls

[Brian Foster]

On Tue, Jul 23, 2019 at 09:13:26PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Don't check the structure size of the inogrp/bstat/fsop_bulkreq
> structures because they're incorrectly padded.  When we remove the
> old typdefs the old filter stops working.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---

Reviewed-by: Brian Foster <bfoster@redhat.com>

>  tests/xfs/122 |    3 +++
>  1 file changed, 3 insertions(+)
> 
> 
> diff --git a/tests/xfs/122 b/tests/xfs/122
> index 89a39a23..64b63cb1 100755
> --- a/tests/xfs/122
> +++ b/tests/xfs/122
> @@ -148,12 +148,15 @@ xfs_growfs_data_t
>  xfs_growfs_rt_t
>  xfs_bstime_t
>  xfs_bstat_t
> +struct xfs_bstat
>  xfs_fsop_bulkreq_t
> +struct xfs_fsop_bulkreq
>  xfs_icsb_cnts_t
>  xfs_icdinode_t
>  xfs_ictimestamp_t
>  xfs_inobt_rec_incore_t
>  xfs_inogrp_t
> +struct xfs_inogrp
>  xfs_fid2_t
>  xfs_fsop_handlereq_t
>  xfs_fsop_setdm_handlereq_t
> 

[PATCH 3/4] xfs/122: add the new v5 bulkstat/inumbers ioctl structures

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

The new v5 bulkstat and inumbers structures are correctly padded so that
no format changes are necessary across platforms, so add them to the
output.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 tests/xfs/122.out |    7 +++++++
 1 file changed, 7 insertions(+)


diff --git a/tests/xfs/122.out b/tests/xfs/122.out
index cf9ac9e2..e2f346eb 100644
--- a/tests/xfs/122.out
+++ b/tests/xfs/122.out
@@ -66,6 +66,10 @@ sizeof(struct xfs_btree_block_lhdr) = 64
 sizeof(struct xfs_btree_block_shdr) = 48
 sizeof(struct xfs_bud_log_format) = 16
 sizeof(struct xfs_bui_log_format) = 16
+sizeof(struct xfs_bulk_ireq) = 64
+sizeof(struct xfs_bulkstat) = 192
+sizeof(struct xfs_bulkstat_req) = 64
+sizeof(struct xfs_bulkstat_single_req) = 224
 sizeof(struct xfs_clone_args) = 32
 sizeof(struct xfs_cud_log_format) = 16
 sizeof(struct xfs_cui_log_format) = 16
@@ -89,6 +93,9 @@ sizeof(struct xfs_fsop_geom_v4) = 112
 sizeof(struct xfs_icreate_log) = 28
 sizeof(struct xfs_inode_log_format) = 56
 sizeof(struct xfs_inode_log_format_32) = 52
+sizeof(struct xfs_inumbers) = 24
+sizeof(struct xfs_inumbers_req) = 64
+sizeof(struct xfs_ireq) = 32
 sizeof(struct xfs_log_dinode) = 176
 sizeof(struct xfs_map_extent) = 32
 sizeof(struct xfs_phys_extent) = 16

Re: [PATCH 3/4] xfs/122: add the new v5 bulkstat/inumbers ioctl structures

[Brian Foster]

On Tue, Jul 23, 2019 at 09:13:32PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> The new v5 bulkstat and inumbers structures are correctly padded so that
> no format changes are necessary across platforms, so add them to the
> output.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  tests/xfs/122.out |    7 +++++++
>  1 file changed, 7 insertions(+)
> 
> 
> diff --git a/tests/xfs/122.out b/tests/xfs/122.out
> index cf9ac9e2..e2f346eb 100644
> --- a/tests/xfs/122.out
> +++ b/tests/xfs/122.out
> @@ -66,6 +66,10 @@ sizeof(struct xfs_btree_block_lhdr) = 64
>  sizeof(struct xfs_btree_block_shdr) = 48
>  sizeof(struct xfs_bud_log_format) = 16
>  sizeof(struct xfs_bui_log_format) = 16
> +sizeof(struct xfs_bulk_ireq) = 64
> +sizeof(struct xfs_bulkstat) = 192
> +sizeof(struct xfs_bulkstat_req) = 64
> +sizeof(struct xfs_bulkstat_single_req) = 224
>  sizeof(struct xfs_clone_args) = 32
>  sizeof(struct xfs_cud_log_format) = 16
>  sizeof(struct xfs_cui_log_format) = 16
> @@ -89,6 +93,9 @@ sizeof(struct xfs_fsop_geom_v4) = 112
>  sizeof(struct xfs_icreate_log) = 28
>  sizeof(struct xfs_inode_log_format) = 56
>  sizeof(struct xfs_inode_log_format_32) = 52
> +sizeof(struct xfs_inumbers) = 24
> +sizeof(struct xfs_inumbers_req) = 64
> +sizeof(struct xfs_ireq) = 32

I don't see xfs_bulkstat_single_req or xfs_ireq in the latest kernel
headers. Do we still have those? Otherwise looks fine.

Brian

>  sizeof(struct xfs_log_dinode) = 176
>  sizeof(struct xfs_map_extent) = 32
>  sizeof(struct xfs_phys_extent) = 16
> 

Re: [PATCH 3/4] xfs/122: add the new v5 bulkstat/inumbers ioctl structures

[Darrick J. Wong]

On Wed, Jul 31, 2019 at 07:40:20AM -0400, Brian Foster wrote:
> On Tue, Jul 23, 2019 at 09:13:32PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > The new v5 bulkstat and inumbers structures are correctly padded so that
> > no format changes are necessary across platforms, so add them to the
> > output.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> >  tests/xfs/122.out |    7 +++++++
> >  1 file changed, 7 insertions(+)
> > 
> > 
> > diff --git a/tests/xfs/122.out b/tests/xfs/122.out
> > index cf9ac9e2..e2f346eb 100644
> > --- a/tests/xfs/122.out
> > +++ b/tests/xfs/122.out
> > @@ -66,6 +66,10 @@ sizeof(struct xfs_btree_block_lhdr) = 64
> >  sizeof(struct xfs_btree_block_shdr) = 48
> >  sizeof(struct xfs_bud_log_format) = 16
> >  sizeof(struct xfs_bui_log_format) = 16
> > +sizeof(struct xfs_bulk_ireq) = 64
> > +sizeof(struct xfs_bulkstat) = 192
> > +sizeof(struct xfs_bulkstat_req) = 64
> > +sizeof(struct xfs_bulkstat_single_req) = 224
> >  sizeof(struct xfs_clone_args) = 32
> >  sizeof(struct xfs_cud_log_format) = 16
> >  sizeof(struct xfs_cui_log_format) = 16
> > @@ -89,6 +93,9 @@ sizeof(struct xfs_fsop_geom_v4) = 112
> >  sizeof(struct xfs_icreate_log) = 28
> >  sizeof(struct xfs_inode_log_format) = 56
> >  sizeof(struct xfs_inode_log_format_32) = 52
> > +sizeof(struct xfs_inumbers) = 24
> > +sizeof(struct xfs_inumbers_req) = 64
> > +sizeof(struct xfs_ireq) = 32
> 
> I don't see xfs_bulkstat_single_req or xfs_ireq in the latest kernel
> headers. Do we still have those? Otherwise looks fine.

Ooops, I forgot to remove those, will fix.

--D

> Brian
> 
> >  sizeof(struct xfs_log_dinode) = 176
> >  sizeof(struct xfs_map_extent) = 32
> >  sizeof(struct xfs_phys_extent) = 16
> > 

[PATCH 4/4] xfs: test new v5 bulkstat commands

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Check that the new v5 bulkstat commands do everything the old one do,
and then make sure the new functionality actually works.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 common/xfs                 |    6 +
 src/Makefile               |    2 
 src/bulkstat_null_ocount.c |   61 +++++++++
 tests/xfs/085              |    2 
 tests/xfs/086              |    2 
 tests/xfs/087              |    2 
 tests/xfs/088              |    2 
 tests/xfs/089              |    2 
 tests/xfs/091              |    2 
 tests/xfs/093              |    2 
 tests/xfs/097              |    2 
 tests/xfs/130              |    2 
 tests/xfs/235              |    2 
 tests/xfs/271              |    2 
 tests/xfs/744              |  212 +++++++++++++++++++++++++++++++
 tests/xfs/744.out          |  297 ++++++++++++++++++++++++++++++++++++++++++++
 tests/xfs/745              |   44 +++++++
 tests/xfs/745.out          |    2 
 tests/xfs/group            |    2 
 19 files changed, 636 insertions(+), 12 deletions(-)
 create mode 100644 src/bulkstat_null_ocount.c
 create mode 100755 tests/xfs/744
 create mode 100644 tests/xfs/744.out
 create mode 100755 tests/xfs/745
 create mode 100644 tests/xfs/745.out


diff --git a/common/xfs b/common/xfs
index 2b38e94b..1bce3c18 100644
--- a/common/xfs
+++ b/common/xfs
@@ -878,3 +878,9 @@ _force_xfsv4_mount_options()
 	fi
 	echo "MOUNT_OPTIONS = $MOUNT_OPTIONS" >>$seqres.full
 }
+
+# Find AG count of mounted filesystem
+_xfs_mount_agcount()
+{
+	$XFS_INFO_PROG "$1" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g'
+}
diff --git a/src/Makefile b/src/Makefile
index 9d3d2529..c4fcf370 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -28,7 +28,7 @@ LINUX_TARGETS = xfsctl bstat t_mtab getdevicesize preallo_rw_pattern_reader \
 	attr-list-by-handle-cursor-test listxattr dio-interleaved t_dir_type \
 	dio-invalidate-cache stat_test t_encrypted_d_revalidate \
 	attr_replace_test swapon mkswap t_attr_corruption t_open_tmpfiles \
-	fscrypt-crypt-util
+	fscrypt-crypt-util bulkstat_null_ocount
 
 SUBDIRS = log-writes perf
 
diff --git a/src/bulkstat_null_ocount.c b/src/bulkstat_null_ocount.c
new file mode 100644
index 00000000..b8f8fd39
--- /dev/null
+++ b/src/bulkstat_null_ocount.c
@@ -0,0 +1,61 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2019 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * Ensure the kernel returns the new lastip even when ocount is null.
+ */
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/ioctl.h>
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <fcntl.h>
+#include <xfs/xfs.h>
+
+static void die(const char *tag)
+{
+	perror(tag);
+	exit(1);
+}
+
+int main(int argc, char *argv[])
+{
+	struct xfs_bstat	bstat;
+	__u64			last;
+	struct xfs_fsop_bulkreq bulkreq = {
+		.lastip		= &last,
+		.icount		= 1,
+		.ubuffer	= &bstat,
+		.ocount		= NULL,
+	};
+	int ret;
+	int fd;
+
+	fd = open(argv[1], O_RDONLY);
+	if (fd < 0)
+		die(argv[1]);
+
+	last = 0;
+	ret = ioctl(fd, XFS_IOC_FSINUMBERS, &bulkreq);
+	if (ret)
+		die("inumbers");
+
+	if (last == 0)
+		printf("inumbers last = %llu\n", (unsigned long long)last);
+
+	last = 0;
+	ret = ioctl(fd, XFS_IOC_FSBULKSTAT, &bulkreq);
+	if (ret)
+		die("bulkstat");
+
+	if (last == 0)
+		printf("bulkstat last = %llu\n", (unsigned long long)last);
+
+	ret = close(fd);
+	if (ret)
+		die("close");
+
+	return 0;
+}
diff --git a/tests/xfs/085 b/tests/xfs/085
index 23095413..18ddeff8 100755
--- a/tests/xfs/085
+++ b/tests/xfs/085
@@ -63,7 +63,7 @@ for x in `seq 2 64`; do
 	touch "${TESTFILE}.${x}"
 done
 inode="$(stat -c '%i' "${TESTFILE}.1")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 umount "${SCRATCH_MNT}"
 
 echo "+ check fs"
diff --git a/tests/xfs/086 b/tests/xfs/086
index 8602a565..7429d39d 100755
--- a/tests/xfs/086
+++ b/tests/xfs/086
@@ -64,7 +64,7 @@ for x in `seq 2 64`; do
 	touch "${TESTFILE}.${x}"
 done
 inode="$(stat -c '%i' "${TESTFILE}.1")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 test "${agcount}" -gt 1 || _notrun "Single-AG XFS not supported"
 umount "${SCRATCH_MNT}"
 
diff --git a/tests/xfs/087 b/tests/xfs/087
index ede8e447..b3d3bca9 100755
--- a/tests/xfs/087
+++ b/tests/xfs/087
@@ -64,7 +64,7 @@ for x in `seq 2 64`; do
 	touch "${TESTFILE}.${x}"
 done
 inode="$(stat -c '%i' "${TESTFILE}.1")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 umount "${SCRATCH_MNT}"
 
 echo "+ check fs"
diff --git a/tests/xfs/088 b/tests/xfs/088
index 6f36efab..74b45163 100755
--- a/tests/xfs/088
+++ b/tests/xfs/088
@@ -64,7 +64,7 @@ for x in `seq 2 64`; do
 	touch "${TESTFILE}.${x}"
 done
 inode="$(stat -c '%i' "${TESTFILE}.1")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 umount "${SCRATCH_MNT}"
 
 echo "+ check fs"
diff --git a/tests/xfs/089 b/tests/xfs/089
index 5c398299..bcbc6363 100755
--- a/tests/xfs/089
+++ b/tests/xfs/089
@@ -64,7 +64,7 @@ for x in `seq 2 64`; do
 	touch "${TESTFILE}.${x}"
 done
 inode="$(stat -c '%i' "${TESTFILE}.1")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 umount "${SCRATCH_MNT}"
 
 echo "+ check fs"
diff --git a/tests/xfs/091 b/tests/xfs/091
index 5d6cd363..be56d8ae 100755
--- a/tests/xfs/091
+++ b/tests/xfs/091
@@ -64,7 +64,7 @@ for x in `seq 2 64`; do
 	touch "${TESTFILE}.${x}"
 done
 inode="$(stat -c '%i' "${TESTFILE}.1")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 umount "${SCRATCH_MNT}"
 
 echo "+ check fs"
diff --git a/tests/xfs/093 b/tests/xfs/093
index e09e8499..4c4fbdc4 100755
--- a/tests/xfs/093
+++ b/tests/xfs/093
@@ -64,7 +64,7 @@ for x in `seq 2 64`; do
 	touch "${TESTFILE}.${x}"
 done
 inode="$(stat -c '%i' "${TESTFILE}.1")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 umount "${SCRATCH_MNT}"
 
 echo "+ check fs"
diff --git a/tests/xfs/097 b/tests/xfs/097
index db355de6..68eae1d4 100755
--- a/tests/xfs/097
+++ b/tests/xfs/097
@@ -67,7 +67,7 @@ for x in `seq 2 64`; do
 	touch "${TESTFILE}.${x}"
 done
 inode="$(stat -c '%i' "${TESTFILE}.1")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 umount "${SCRATCH_MNT}"
 
 echo "+ check fs"
diff --git a/tests/xfs/130 b/tests/xfs/130
index 71c1181f..f15366a6 100755
--- a/tests/xfs/130
+++ b/tests/xfs/130
@@ -43,7 +43,7 @@ _scratch_mkfs_xfs > /dev/null
 echo "+ mount fs image"
 _scratch_mount
 blksz="$(stat -f -c '%s' "${SCRATCH_MNT}")"
-agcount="$($XFS_INFO_PROG "${SCRATCH_MNT}" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')"
+agcount="$(_xfs_mount_agcount $SCRATCH_MNT)"
 
 echo "+ make some files"
 _pwrite_byte 0x62 0 $((blksz * 64)) "${SCRATCH_MNT}/file0" >> "$seqres.full"
diff --git a/tests/xfs/235 b/tests/xfs/235
index 669f58b0..64b0a0b5 100755
--- a/tests/xfs/235
+++ b/tests/xfs/235
@@ -41,7 +41,7 @@ _scratch_mkfs_xfs > /dev/null
 echo "+ mount fs image"
 _scratch_mount
 blksz=$(stat -f -c '%s' ${SCRATCH_MNT})
-agcount=$($XFS_INFO_PROG ${SCRATCH_MNT} | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')
+agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
 
 echo "+ make some files"
 _pwrite_byte 0x62 0 $((blksz * 64)) ${SCRATCH_MNT}/file0 >> $seqres.full
diff --git a/tests/xfs/271 b/tests/xfs/271
index db14bfec..38844246 100755
--- a/tests/xfs/271
+++ b/tests/xfs/271
@@ -37,7 +37,7 @@ echo "Format and mount"
 _scratch_mkfs > "$seqres.full" 2>&1
 _scratch_mount
 
-agcount=$($XFS_INFO_PROG $SCRATCH_MNT | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g')
+agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
 
 echo "Get fsmap" | tee -a $seqres.full
 $XFS_IO_PROG -c 'fsmap -v' $SCRATCH_MNT > $TEST_DIR/fsmap
diff --git a/tests/xfs/744 b/tests/xfs/744
new file mode 100755
index 00000000..ef605301
--- /dev/null
+++ b/tests/xfs/744
@@ -0,0 +1,212 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2014 Red Hat, Inc.  All Rights Reserved.
+# Copyright (c) 2019 Oracle, Inc.  All Rights Reserved.
+#
+# FS QA Test No. 744
+#
+# Use the xfs_io bulkstat utility to verify bulkstat finds all inodes in a
+# filesystem.  Test under various inode counts, inobt record layouts and
+# bulkstat batch sizes.  Test v1 and v5 ioctls explicitly, as well as the
+# ioctl version autodetection code in libfrog.
+#
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+	cd /
+	rm -f $tmp.*
+}
+
+bstat_versions()
+{
+	echo "default"
+	echo "v1 -v1"
+	if [ -n "$has_v5" ]; then
+		echo "v5 -v5"
+	else
+		echo "v5"
+	fi
+}
+
+# print the number of inodes counted by bulkstat
+bstat_count()
+{
+	local batchsize="$1"
+	local tag="$2"
+
+	bstat_versions | while read v_tag v_flag; do
+		echo "$tag($v_tag): passing \"$v_flag\" to bulkstat" >> $seqres.full
+		echo -n "bulkstat $tag($v_tag): "
+		$XFS_IO_PROG -c "bulkstat -n $batchsize $vflag" $SCRATCH_MNT | grep ino | wc -l
+	done
+}
+
+# print the number of inodes counted by per-ag bulkstat
+bstat_perag_count()
+{
+	local batchsize="$1"
+	local tag="$2"
+
+	local agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
+
+	bstat_versions | while read v_tag v_flag; do
+		echo -n "bulkstat $tag($v_tag): "
+		seq 0 $((agcount - 1)) | while read ag; do
+			$XFS_IO_PROG -c "bulkstat -a $ag -n $batchsize $v_flag" $SCRATCH_MNT
+		done | grep ino | wc -l
+	done
+}
+
+# Sum the number of allocated inodes in each AG in a fs.
+inumbers_ag()
+{
+	local agcount="$1"
+	local batchsize="$2"
+	local mount="$3"
+	local v_flag="$4"
+
+	seq 0 $((agcount - 1)) | while read ag; do
+		$XFS_IO_PROG -c "inumbers -a $ag -n $batchsize $v_flag" $mount
+	done | grep alloccount | awk '{x += $3} END { print(x) }'
+}
+
+# Sum the number of allocated inodes in the whole fs all at once.
+inumbers_fs()
+{
+	local dir="$1"
+	local v_flag="$2"
+
+	$XFS_IO_PROG -c "inumbers $v_flag" "$dir" | grep alloccount | \
+		awk '{x += $3} END { print(x) }'
+}
+
+# print the number of inodes counted by inumbers
+inumbers_count()
+{
+	local expect="$1"
+
+	# There probably aren't more than 10 hidden inodes, right?
+	local tolerance=10
+
+	# Force all background inode cleanup
+	_scratch_cycle_mount
+
+	bstat_versions | while read v_tag v_flag; do
+		echo -n "inumbers all($v_tag): "
+		nr=$(inumbers_fs $SCRATCH_MNT $v_flag)
+		_within_tolerance "inumbers" $nr $expect $tolerance -v
+
+		local agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
+		for batchsize in 64 2 1; do
+			echo -n "inumbers $batchsize($v_tag): "
+			nr=$(inumbers_ag $agcount $batchsize $SCRATCH_MNT $v_flag)
+			_within_tolerance "inumbers" $nr $expect $tolerance -v
+		done
+	done
+}
+
+# compare the src/bstat output against the xfs_io bstat output
+bstat_compare()
+{
+	bstat_versions | while read v_tag v_flag; do
+		diff -u <(./src/bstat $SCRATCH_MNT | grep ino | awk '{print $2}') \
+			<($XFS_IO_PROG -c "bulkstat $v_flag" $SCRATCH_MNT | grep ino | awk '{print $3}')
+	done
+}
+
+# print bulkstat counts using varied batch sizes
+bstat_test()
+{
+	expect=`find $SCRATCH_MNT -print | wc -l`
+	echo
+	echo "expect $expect"
+
+	for sz in 4096 71 32 1; do
+		bstat_count $sz "$sz all"
+		bstat_perag_count $sz "$sz perag"
+		bstat_compare
+		inumbers_count $expect
+	done
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+_require_scratch
+_require_xfs_io_command bulkstat
+_require_xfs_io_command bulkstat_single
+_require_xfs_io_command inumbers
+
+# real QA test starts here
+
+_supported_fs xfs
+_supported_os Linux
+
+rm -f $seqres.full
+
+DIRCOUNT=8
+INOCOUNT=$((2048 / DIRCOUNT))
+
+_scratch_mkfs "-d agcount=$DIRCOUNT" >> $seqres.full 2>&1 || _fail "mkfs failed"
+_scratch_mount
+
+# Figure out if we have v5 bulkstat/inumbers ioctls.
+has_v5=
+bs_root_out="$($XFS_IO_PROG -c 'bulkstat_single root' $SCRATCH_MNT 2>>$seqres.full)"
+test -n "$bs_root_out" && has_v5=1
+
+echo "this will be 1 if we have v5 bulkstat: $has_v5" >> $seqres.full
+
+# If v5 bulkstat is present, query the root inode and compare it to the stat
+# output of $SCRATCH_MNT to make sure it gave us the correct number
+if [ -n "$has_v5" ]; then
+	bs_root=$(echo "$bs_root_out" | grep ino | awk '{print $3}')
+	stat_root=$(stat -c '%i' $SCRATCH_MNT)
+	if [ "$stat_root" -ne "$bs_root" ]; then
+		echo "stat says root is $stat_root but bulkstat says $bs_root"
+	fi
+fi
+
+# create a set of directories and fill each with a fixed number of files
+for dir in $(seq 1 $DIRCOUNT); do
+	mkdir -p $SCRATCH_MNT/$dir
+	for i in $(seq 1 $INOCOUNT); do
+		touch $SCRATCH_MNT/$dir/$i
+	done
+done
+bstat_test
+
+# remove every other file from each dir
+for dir in $(seq 1 $DIRCOUNT); do
+	for i in $(seq 2 2 $INOCOUNT); do
+		rm -f $SCRATCH_MNT/$dir/$i
+	done
+done
+bstat_test
+
+# remove the entire second half of files
+for dir in $(seq 1 $DIRCOUNT); do
+	for i in $(seq $((INOCOUNT / 2)) $INOCOUNT); do
+		rm -f $SCRATCH_MNT/$dir/$i
+	done
+done
+bstat_test
+
+# remove all regular files
+for dir in $(seq 1 $DIRCOUNT); do
+	rm -f $SCRATCH_MNT/$dir/*
+done
+bstat_test
+
+# success, all done
+status=0
+exit
diff --git a/tests/xfs/744.out b/tests/xfs/744.out
new file mode 100644
index 00000000..de89a8ff
--- /dev/null
+++ b/tests/xfs/744.out
@@ -0,0 +1,297 @@
+QA output created by 744
+
+expect 2057
+bulkstat 4096 all(default): 2057
+bulkstat 4096 all(v1): 2057
+bulkstat 4096 all(v5): 2057
+bulkstat 4096 perag(default): 2057
+bulkstat 4096 perag(v1): 2057
+bulkstat 4096 perag(v5): 2057
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 71 all(default): 2057
+bulkstat 71 all(v1): 2057
+bulkstat 71 all(v5): 2057
+bulkstat 71 perag(default): 2057
+bulkstat 71 perag(v1): 2057
+bulkstat 71 perag(v5): 2057
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 32 all(default): 2057
+bulkstat 32 all(v1): 2057
+bulkstat 32 all(v5): 2057
+bulkstat 32 perag(default): 2057
+bulkstat 32 perag(v1): 2057
+bulkstat 32 perag(v5): 2057
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 1 all(default): 2057
+bulkstat 1 all(v1): 2057
+bulkstat 1 all(v5): 2057
+bulkstat 1 perag(default): 2057
+bulkstat 1 perag(v1): 2057
+bulkstat 1 perag(v5): 2057
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+
+expect 1033
+bulkstat 4096 all(default): 1033
+bulkstat 4096 all(v1): 1033
+bulkstat 4096 all(v5): 1033
+bulkstat 4096 perag(default): 1033
+bulkstat 4096 perag(v1): 1033
+bulkstat 4096 perag(v5): 1033
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 71 all(default): 1033
+bulkstat 71 all(v1): 1033
+bulkstat 71 all(v5): 1033
+bulkstat 71 perag(default): 1033
+bulkstat 71 perag(v1): 1033
+bulkstat 71 perag(v5): 1033
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 32 all(default): 1033
+bulkstat 32 all(v1): 1033
+bulkstat 32 all(v5): 1033
+bulkstat 32 perag(default): 1033
+bulkstat 32 perag(v1): 1033
+bulkstat 32 perag(v5): 1033
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 1 all(default): 1033
+bulkstat 1 all(v1): 1033
+bulkstat 1 all(v5): 1033
+bulkstat 1 perag(default): 1033
+bulkstat 1 perag(v1): 1033
+bulkstat 1 perag(v5): 1033
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+
+expect 521
+bulkstat 4096 all(default): 521
+bulkstat 4096 all(v1): 521
+bulkstat 4096 all(v5): 521
+bulkstat 4096 perag(default): 521
+bulkstat 4096 perag(v1): 521
+bulkstat 4096 perag(v5): 521
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 71 all(default): 521
+bulkstat 71 all(v1): 521
+bulkstat 71 all(v5): 521
+bulkstat 71 perag(default): 521
+bulkstat 71 perag(v1): 521
+bulkstat 71 perag(v5): 521
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 32 all(default): 521
+bulkstat 32 all(v1): 521
+bulkstat 32 all(v5): 521
+bulkstat 32 perag(default): 521
+bulkstat 32 perag(v1): 521
+bulkstat 32 perag(v5): 521
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 1 all(default): 521
+bulkstat 1 all(v1): 521
+bulkstat 1 all(v5): 521
+bulkstat 1 perag(default): 521
+bulkstat 1 perag(v1): 521
+bulkstat 1 perag(v5): 521
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+
+expect 9
+bulkstat 4096 all(default): 9
+bulkstat 4096 all(v1): 9
+bulkstat 4096 all(v5): 9
+bulkstat 4096 perag(default): 9
+bulkstat 4096 perag(v1): 9
+bulkstat 4096 perag(v5): 9
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 71 all(default): 9
+bulkstat 71 all(v1): 9
+bulkstat 71 all(v5): 9
+bulkstat 71 perag(default): 9
+bulkstat 71 perag(v1): 9
+bulkstat 71 perag(v5): 9
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 32 all(default): 9
+bulkstat 32 all(v1): 9
+bulkstat 32 all(v5): 9
+bulkstat 32 perag(default): 9
+bulkstat 32 perag(v1): 9
+bulkstat 32 perag(v5): 9
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
+bulkstat 1 all(default): 9
+bulkstat 1 all(v1): 9
+bulkstat 1 all(v5): 9
+bulkstat 1 perag(default): 9
+bulkstat 1 perag(v1): 9
+bulkstat 1 perag(v5): 9
+inumbers all(default): inumbers is in range
+inumbers 64(default): inumbers is in range
+inumbers 2(default): inumbers is in range
+inumbers 1(default): inumbers is in range
+inumbers all(v1): inumbers is in range
+inumbers 64(v1): inumbers is in range
+inumbers 2(v1): inumbers is in range
+inumbers 1(v1): inumbers is in range
+inumbers all(v5): inumbers is in range
+inumbers 64(v5): inumbers is in range
+inumbers 2(v5): inumbers is in range
+inumbers 1(v5): inumbers is in range
diff --git a/tests/xfs/745 b/tests/xfs/745
new file mode 100755
index 00000000..6931d46b
--- /dev/null
+++ b/tests/xfs/745
@@ -0,0 +1,44 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0+
+# Copyright (c) 2019 Oracle, Inc.  All Rights Reserved.
+#
+# FS QA Test No. 745
+#
+# Regression test for a long-standing bug in BULKSTAT and INUMBERS where
+# the kernel fails to write thew new @lastip value back to userspace if
+# @ocount is NULL.
+#
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+    cd /
+    rm -f $tmp.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+_require_test_program "bulkstat_null_ocount"
+
+# real QA test starts here
+
+_supported_fs xfs
+_supported_os Linux
+
+rm -f $seqres.full
+
+echo "Silence is golden."
+src/bulkstat_null_ocount $TEST_DIR
+
+# success, all done
+status=0
+exit
diff --git a/tests/xfs/745.out b/tests/xfs/745.out
new file mode 100644
index 00000000..ce947de2
--- /dev/null
+++ b/tests/xfs/745.out
@@ -0,0 +1,2 @@
+QA output created by 745
+Silence is golden.
diff --git a/tests/xfs/group b/tests/xfs/group
index 270d82ff..ef0cf92c 100644
--- a/tests/xfs/group
+++ b/tests/xfs/group
@@ -506,3 +506,5 @@
 506 auto quick health
 507 clone
 508 auto quick quota
+744 auto ioctl quick
+745 auto ioctl quick

Re: [PATCH 4/4] xfs: test new v5 bulkstat commands

[Brian Foster]

On Tue, Jul 23, 2019 at 09:13:38PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Check that the new v5 bulkstat commands do everything the old one do,
> and then make sure the new functionality actually works.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  common/xfs                 |    6 +
>  src/Makefile               |    2 
>  src/bulkstat_null_ocount.c |   61 +++++++++
>  tests/xfs/085              |    2 
>  tests/xfs/086              |    2 
>  tests/xfs/087              |    2 
>  tests/xfs/088              |    2 
>  tests/xfs/089              |    2 
>  tests/xfs/091              |    2 
>  tests/xfs/093              |    2 
>  tests/xfs/097              |    2 
>  tests/xfs/130              |    2 
>  tests/xfs/235              |    2 
>  tests/xfs/271              |    2 
>  tests/xfs/744              |  212 +++++++++++++++++++++++++++++++
>  tests/xfs/744.out          |  297 ++++++++++++++++++++++++++++++++++++++++++++
>  tests/xfs/745              |   44 +++++++
>  tests/xfs/745.out          |    2 
>  tests/xfs/group            |    2 
>  19 files changed, 636 insertions(+), 12 deletions(-)
>  create mode 100644 src/bulkstat_null_ocount.c
>  create mode 100755 tests/xfs/744
>  create mode 100644 tests/xfs/744.out
>  create mode 100755 tests/xfs/745
>  create mode 100644 tests/xfs/745.out
> 
> 
> diff --git a/common/xfs b/common/xfs
> index 2b38e94b..1bce3c18 100644
> --- a/common/xfs
> +++ b/common/xfs
> @@ -878,3 +878,9 @@ _force_xfsv4_mount_options()
>  	fi
>  	echo "MOUNT_OPTIONS = $MOUNT_OPTIONS" >>$seqres.full
>  }
> +
> +# Find AG count of mounted filesystem
> +_xfs_mount_agcount()
> +{
> +	$XFS_INFO_PROG "$1" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g'
> +}

This and the associated changes to existing tests should probably be a
separate patch.

...
> diff --git a/tests/xfs/744 b/tests/xfs/744
> new file mode 100755
> index 00000000..ef605301
> --- /dev/null
> +++ b/tests/xfs/744
> @@ -0,0 +1,212 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2014 Red Hat, Inc.  All Rights Reserved.
> +# Copyright (c) 2019 Oracle, Inc.  All Rights Reserved.
> +#
> +# FS QA Test No. 744
> +#
> +# Use the xfs_io bulkstat utility to verify bulkstat finds all inodes in a
> +# filesystem.  Test under various inode counts, inobt record layouts and
> +# bulkstat batch sizes.  Test v1 and v5 ioctls explicitly, as well as the
> +# ioctl version autodetection code in libfrog.
> +#

Apparently I don't have xfs_io bulkstat support. Is that posted
somewhere? At a glance the test looks mostly fine..

> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=1	# failure is the default!
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +_cleanup()
> +{
> +	cd /
> +	rm -f $tmp.*
> +}
> +
> +bstat_versions()
> +{
> +	echo "default"
> +	echo "v1 -v1"
> +	if [ -n "$has_v5" ]; then
> +		echo "v5 -v5"
> +	else
> +		echo "v5"
> +	fi
> +}
> +
> +# print the number of inodes counted by bulkstat
> +bstat_count()
> +{
> +	local batchsize="$1"
> +	local tag="$2"
> +
> +	bstat_versions | while read v_tag v_flag; do
> +		echo "$tag($v_tag): passing \"$v_flag\" to bulkstat" >> $seqres.full
> +		echo -n "bulkstat $tag($v_tag): "
> +		$XFS_IO_PROG -c "bulkstat -n $batchsize $vflag" $SCRATCH_MNT | grep ino | wc -l

s/vflag/v_flag/ ?

> +	done
> +}
> +
> +# print the number of inodes counted by per-ag bulkstat
> +bstat_perag_count()
> +{
> +	local batchsize="$1"
> +	local tag="$2"
> +
> +	local agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
> +
> +	bstat_versions | while read v_tag v_flag; do
> +		echo -n "bulkstat $tag($v_tag): "
> +		seq 0 $((agcount - 1)) | while read ag; do
> +			$XFS_IO_PROG -c "bulkstat -a $ag -n $batchsize $v_flag" $SCRATCH_MNT
> +		done | grep ino | wc -l
> +	done
> +}
> +
> +# Sum the number of allocated inodes in each AG in a fs.
> +inumbers_ag()
> +{
> +	local agcount="$1"
> +	local batchsize="$2"
> +	local mount="$3"
> +	local v_flag="$4"
> +
> +	seq 0 $((agcount - 1)) | while read ag; do
> +		$XFS_IO_PROG -c "inumbers -a $ag -n $batchsize $v_flag" $mount
> +	done | grep alloccount | awk '{x += $3} END { print(x) }'
> +}
> +
> +# Sum the number of allocated inodes in the whole fs all at once.
> +inumbers_fs()
> +{
> +	local dir="$1"
> +	local v_flag="$2"
> +
> +	$XFS_IO_PROG -c "inumbers $v_flag" "$dir" | grep alloccount | \
> +		awk '{x += $3} END { print(x) }'
> +}
> +
> +# print the number of inodes counted by inumbers
> +inumbers_count()
> +{
> +	local expect="$1"
> +
> +	# There probably aren't more than 10 hidden inodes, right?
> +	local tolerance=10
> +
> +	# Force all background inode cleanup

Comment took me a second to grok. This refers to unlinked inode cleanup,
right?

> +	_scratch_cycle_mount
> +
> +	bstat_versions | while read v_tag v_flag; do
> +		echo -n "inumbers all($v_tag): "
> +		nr=$(inumbers_fs $SCRATCH_MNT $v_flag)
> +		_within_tolerance "inumbers" $nr $expect $tolerance -v
> +
> +		local agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
> +		for batchsize in 64 2 1; do

Perhaps we should stuff a value > than the per-record inode count in
here as well.

> +			echo -n "inumbers $batchsize($v_tag): "
> +			nr=$(inumbers_ag $agcount $batchsize $SCRATCH_MNT $v_flag)
> +			_within_tolerance "inumbers" $nr $expect $tolerance -v
> +		done
> +	done
> +}
> +
> +# compare the src/bstat output against the xfs_io bstat output

This compares actual inode numbers, right? If so, I'd point that out in
the comment.

> +bstat_compare()
> +{
> +	bstat_versions | while read v_tag v_flag; do
> +		diff -u <(./src/bstat $SCRATCH_MNT | grep ino | awk '{print $2}') \
> +			<($XFS_IO_PROG -c "bulkstat $v_flag" $SCRATCH_MNT | grep ino | awk '{print $3}')
> +	done
> +}
> +
...
> diff --git a/tests/xfs/745 b/tests/xfs/745
> new file mode 100755
> index 00000000..6931d46b
> --- /dev/null
> +++ b/tests/xfs/745
> @@ -0,0 +1,44 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0+
> +# Copyright (c) 2019 Oracle, Inc.  All Rights Reserved.
> +#
> +# FS QA Test No. 745
> +#
> +# Regression test for a long-standing bug in BULKSTAT and INUMBERS where
> +# the kernel fails to write thew new @lastip value back to userspace if
> +# @ocount is NULL.
> +#

I think it would be helpful to reference the upstream fix here, which
IIRC is commit f16fe3ecde62 ("xfs: bulkstat should copy lastip whenever
userspace supplies one"). Otherwise this test looks fine to me.

Brian

> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=1	# failure is the default!
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +_cleanup()
> +{
> +    cd /
> +    rm -f $tmp.*
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +
> +_require_test_program "bulkstat_null_ocount"
> +
> +# real QA test starts here
> +
> +_supported_fs xfs
> +_supported_os Linux
> +
> +rm -f $seqres.full
> +
> +echo "Silence is golden."
> +src/bulkstat_null_ocount $TEST_DIR
> +
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/xfs/745.out b/tests/xfs/745.out
> new file mode 100644
> index 00000000..ce947de2
> --- /dev/null
> +++ b/tests/xfs/745.out
> @@ -0,0 +1,2 @@
> +QA output created by 745
> +Silence is golden.
> diff --git a/tests/xfs/group b/tests/xfs/group
> index 270d82ff..ef0cf92c 100644
> --- a/tests/xfs/group
> +++ b/tests/xfs/group
> @@ -506,3 +506,5 @@
>  506 auto quick health
>  507 clone
>  508 auto quick quota
> +744 auto ioctl quick
> +745 auto ioctl quick
> 

Re: [PATCH 4/4] xfs: test new v5 bulkstat commands

[Darrick J. Wong]

On Wed, Jul 31, 2019 at 07:43:07AM -0400, Brian Foster wrote:
> On Tue, Jul 23, 2019 at 09:13:38PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > Check that the new v5 bulkstat commands do everything the old one do,
> > and then make sure the new functionality actually works.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> >  common/xfs                 |    6 +
> >  src/Makefile               |    2 
> >  src/bulkstat_null_ocount.c |   61 +++++++++
> >  tests/xfs/085              |    2 
> >  tests/xfs/086              |    2 
> >  tests/xfs/087              |    2 
> >  tests/xfs/088              |    2 
> >  tests/xfs/089              |    2 
> >  tests/xfs/091              |    2 
> >  tests/xfs/093              |    2 
> >  tests/xfs/097              |    2 
> >  tests/xfs/130              |    2 
> >  tests/xfs/235              |    2 
> >  tests/xfs/271              |    2 
> >  tests/xfs/744              |  212 +++++++++++++++++++++++++++++++
> >  tests/xfs/744.out          |  297 ++++++++++++++++++++++++++++++++++++++++++++
> >  tests/xfs/745              |   44 +++++++
> >  tests/xfs/745.out          |    2 
> >  tests/xfs/group            |    2 
> >  19 files changed, 636 insertions(+), 12 deletions(-)
> >  create mode 100644 src/bulkstat_null_ocount.c
> >  create mode 100755 tests/xfs/744
> >  create mode 100644 tests/xfs/744.out
> >  create mode 100755 tests/xfs/745
> >  create mode 100644 tests/xfs/745.out
> > 
> > 
> > diff --git a/common/xfs b/common/xfs
> > index 2b38e94b..1bce3c18 100644
> > --- a/common/xfs
> > +++ b/common/xfs
> > @@ -878,3 +878,9 @@ _force_xfsv4_mount_options()
> >  	fi
> >  	echo "MOUNT_OPTIONS = $MOUNT_OPTIONS" >>$seqres.full
> >  }
> > +
> > +# Find AG count of mounted filesystem
> > +_xfs_mount_agcount()
> > +{
> > +	$XFS_INFO_PROG "$1" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g'
> > +}
> 
> This and the associated changes to existing tests should probably be a
> separate patch.

Done.

> ...
> > diff --git a/tests/xfs/744 b/tests/xfs/744
> > new file mode 100755
> > index 00000000..ef605301
> > --- /dev/null
> > +++ b/tests/xfs/744
> > @@ -0,0 +1,212 @@
> > +#! /bin/bash
> > +# SPDX-License-Identifier: GPL-2.0
> > +# Copyright (c) 2014 Red Hat, Inc.  All Rights Reserved.
> > +# Copyright (c) 2019 Oracle, Inc.  All Rights Reserved.
> > +#
> > +# FS QA Test No. 744
> > +#
> > +# Use the xfs_io bulkstat utility to verify bulkstat finds all inodes in a
> > +# filesystem.  Test under various inode counts, inobt record layouts and
> > +# bulkstat batch sizes.  Test v1 and v5 ioctls explicitly, as well as the
> > +# ioctl version autodetection code in libfrog.
> > +#
> 
> Apparently I don't have xfs_io bulkstat support. Is that posted
> somewhere? At a glance the test looks mostly fine..

Hm... it's been hanging out in my xfsprogs dev branch for a while, maybe
I haven't sent it yet...?

In any case, it's here:
https://git.kernel.org/pub/scm/linux/kernel/git/djwong/xfsprogs-dev.git/log/?h=bulkstat-v5

> > +seq=`basename $0`
> > +seqres=$RESULT_DIR/$seq
> > +echo "QA output created by $seq"
> > +
> > +here=`pwd`
> > +tmp=/tmp/$$
> > +status=1	# failure is the default!
> > +trap "_cleanup; exit \$status" 0 1 2 3 15
> > +
> > +_cleanup()
> > +{
> > +	cd /
> > +	rm -f $tmp.*
> > +}
> > +
> > +bstat_versions()
> > +{
> > +	echo "default"
> > +	echo "v1 -v1"
> > +	if [ -n "$has_v5" ]; then
> > +		echo "v5 -v5"
> > +	else
> > +		echo "v5"
> > +	fi
> > +}
> > +
> > +# print the number of inodes counted by bulkstat
> > +bstat_count()
> > +{
> > +	local batchsize="$1"
> > +	local tag="$2"
> > +
> > +	bstat_versions | while read v_tag v_flag; do
> > +		echo "$tag($v_tag): passing \"$v_flag\" to bulkstat" >> $seqres.full
> > +		echo -n "bulkstat $tag($v_tag): "
> > +		$XFS_IO_PROG -c "bulkstat -n $batchsize $vflag" $SCRATCH_MNT | grep ino | wc -l
> 
> s/vflag/v_flag/ ?

Good catch!

> > +	done
> > +}
> > +
> > +# print the number of inodes counted by per-ag bulkstat
> > +bstat_perag_count()
> > +{
> > +	local batchsize="$1"
> > +	local tag="$2"
> > +
> > +	local agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
> > +
> > +	bstat_versions | while read v_tag v_flag; do
> > +		echo -n "bulkstat $tag($v_tag): "
> > +		seq 0 $((agcount - 1)) | while read ag; do
> > +			$XFS_IO_PROG -c "bulkstat -a $ag -n $batchsize $v_flag" $SCRATCH_MNT
> > +		done | grep ino | wc -l
> > +	done
> > +}
> > +
> > +# Sum the number of allocated inodes in each AG in a fs.
> > +inumbers_ag()
> > +{
> > +	local agcount="$1"
> > +	local batchsize="$2"
> > +	local mount="$3"
> > +	local v_flag="$4"
> > +
> > +	seq 0 $((agcount - 1)) | while read ag; do
> > +		$XFS_IO_PROG -c "inumbers -a $ag -n $batchsize $v_flag" $mount
> > +	done | grep alloccount | awk '{x += $3} END { print(x) }'
> > +}
> > +
> > +# Sum the number of allocated inodes in the whole fs all at once.
> > +inumbers_fs()
> > +{
> > +	local dir="$1"
> > +	local v_flag="$2"
> > +
> > +	$XFS_IO_PROG -c "inumbers $v_flag" "$dir" | grep alloccount | \
> > +		awk '{x += $3} END { print(x) }'
> > +}
> > +
> > +# print the number of inodes counted by inumbers
> > +inumbers_count()
> > +{
> > +	local expect="$1"
> > +
> > +	# There probably aren't more than 10 hidden inodes, right?
> > +	local tolerance=10
> > +
> > +	# Force all background inode cleanup
> 
> Comment took me a second to grok. This refers to unlinked inode cleanup,
> right?

Yes.  I think it's only needed to force deferred inode inactivation to
run... which means that we don't necessarily need it now, but it doesn't
hurt to have it.

> > +	_scratch_cycle_mount
> > +
> > +	bstat_versions | while read v_tag v_flag; do
> > +		echo -n "inumbers all($v_tag): "
> > +		nr=$(inumbers_fs $SCRATCH_MNT $v_flag)
> > +		_within_tolerance "inumbers" $nr $expect $tolerance -v
> > +
> > +		local agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
> > +		for batchsize in 64 2 1; do
> 
> Perhaps we should stuff a value > than the per-record inode count in
> here as well.

I'm confused about this comment -- inumbers returns (cooked) inobt
records, not inode records themselves.  The 64>2>1 sequence here asks
inumbers to return 64 inobt records per call, then 2 per call, then 1.

Maybe this should be 71 or some other prime number...

> > +			echo -n "inumbers $batchsize($v_tag): "
> > +			nr=$(inumbers_ag $agcount $batchsize $SCRATCH_MNT $v_flag)
> > +			_within_tolerance "inumbers" $nr $expect $tolerance -v
> > +		done
> > +	done
> > +}
> > +
> > +# compare the src/bstat output against the xfs_io bstat output
> 
> This compares actual inode numbers, right? If so, I'd point that out in
> the comment.

Ok.

> > +bstat_compare()
> > +{
> > +	bstat_versions | while read v_tag v_flag; do
> > +		diff -u <(./src/bstat $SCRATCH_MNT | grep ino | awk '{print $2}') \
> > +			<($XFS_IO_PROG -c "bulkstat $v_flag" $SCRATCH_MNT | grep ino | awk '{print $3}')
> > +	done
> > +}
> > +
> ...
> > diff --git a/tests/xfs/745 b/tests/xfs/745
> > new file mode 100755
> > index 00000000..6931d46b
> > --- /dev/null
> > +++ b/tests/xfs/745
> > @@ -0,0 +1,44 @@
> > +#! /bin/bash
> > +# SPDX-License-Identifier: GPL-2.0+
> > +# Copyright (c) 2019 Oracle, Inc.  All Rights Reserved.
> > +#
> > +# FS QA Test No. 745
> > +#
> > +# Regression test for a long-standing bug in BULKSTAT and INUMBERS where
> > +# the kernel fails to write thew new @lastip value back to userspace if
> > +# @ocount is NULL.
> > +#
> 
> I think it would be helpful to reference the upstream fix here, which
> IIRC is commit f16fe3ecde62 ("xfs: bulkstat should copy lastip whenever
> userspace supplies one"). Otherwise this test looks fine to me.

Ok, will update.

--D

> Brian
> 
> > +seq=`basename $0`
> > +seqres=$RESULT_DIR/$seq
> > +echo "QA output created by $seq"
> > +
> > +here=`pwd`
> > +tmp=/tmp/$$
> > +status=1	# failure is the default!
> > +trap "_cleanup; exit \$status" 0 1 2 3 15
> > +
> > +_cleanup()
> > +{
> > +    cd /
> > +    rm -f $tmp.*
> > +}
> > +
> > +# get standard environment, filters and checks
> > +. ./common/rc
> > +. ./common/filter
> > +
> > +_require_test_program "bulkstat_null_ocount"
> > +
> > +# real QA test starts here
> > +
> > +_supported_fs xfs
> > +_supported_os Linux
> > +
> > +rm -f $seqres.full
> > +
> > +echo "Silence is golden."
> > +src/bulkstat_null_ocount $TEST_DIR
> > +
> > +# success, all done
> > +status=0
> > +exit
> > diff --git a/tests/xfs/745.out b/tests/xfs/745.out
> > new file mode 100644
> > index 00000000..ce947de2
> > --- /dev/null
> > +++ b/tests/xfs/745.out
> > @@ -0,0 +1,2 @@
> > +QA output created by 745
> > +Silence is golden.
> > diff --git a/tests/xfs/group b/tests/xfs/group
> > index 270d82ff..ef0cf92c 100644
> > --- a/tests/xfs/group
> > +++ b/tests/xfs/group
> > @@ -506,3 +506,5 @@
> >  506 auto quick health
> >  507 clone
> >  508 auto quick quota
> > +744 auto ioctl quick
> > +745 auto ioctl quick
> > 

Re: [PATCH 4/4] xfs: test new v5 bulkstat commands

[Brian Foster]

On Wed, Jul 31, 2019 at 09:44:19AM -0700, Darrick J. Wong wrote:
> On Wed, Jul 31, 2019 at 07:43:07AM -0400, Brian Foster wrote:
> > On Tue, Jul 23, 2019 at 09:13:38PM -0700, Darrick J. Wong wrote:
> > > From: Darrick J. Wong <darrick.wong@oracle.com>
> > > 
> > > Check that the new v5 bulkstat commands do everything the old one do,
> > > and then make sure the new functionality actually works.
> > > 
> > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > > ---
> > >  common/xfs                 |    6 +
> > >  src/Makefile               |    2 
> > >  src/bulkstat_null_ocount.c |   61 +++++++++
> > >  tests/xfs/085              |    2 
> > >  tests/xfs/086              |    2 
> > >  tests/xfs/087              |    2 
> > >  tests/xfs/088              |    2 
> > >  tests/xfs/089              |    2 
> > >  tests/xfs/091              |    2 
> > >  tests/xfs/093              |    2 
> > >  tests/xfs/097              |    2 
> > >  tests/xfs/130              |    2 
> > >  tests/xfs/235              |    2 
> > >  tests/xfs/271              |    2 
> > >  tests/xfs/744              |  212 +++++++++++++++++++++++++++++++
> > >  tests/xfs/744.out          |  297 ++++++++++++++++++++++++++++++++++++++++++++
> > >  tests/xfs/745              |   44 +++++++
> > >  tests/xfs/745.out          |    2 
> > >  tests/xfs/group            |    2 
> > >  19 files changed, 636 insertions(+), 12 deletions(-)
> > >  create mode 100644 src/bulkstat_null_ocount.c
> > >  create mode 100755 tests/xfs/744
> > >  create mode 100644 tests/xfs/744.out
> > >  create mode 100755 tests/xfs/745
> > >  create mode 100644 tests/xfs/745.out
> > > 
> > > 
...
> > > +# print the number of inodes counted by inumbers
> > > +inumbers_count()
> > > +{
> > > +	local expect="$1"
> > > +
> > > +	# There probably aren't more than 10 hidden inodes, right?
> > > +	local tolerance=10
> > > +
> > > +	# Force all background inode cleanup
> > 
> > Comment took me a second to grok. This refers to unlinked inode cleanup,
> > right?
> 
> Yes.  I think it's only needed to force deferred inode inactivation to
> run... which means that we don't necessarily need it now, but it doesn't
> hurt to have it.
> 

Yep, could we just clarify the comment then to refer to cleaning up
unlinked inodes?

> > > +	_scratch_cycle_mount
> > > +
> > > +	bstat_versions | while read v_tag v_flag; do
> > > +		echo -n "inumbers all($v_tag): "
> > > +		nr=$(inumbers_fs $SCRATCH_MNT $v_flag)
> > > +		_within_tolerance "inumbers" $nr $expect $tolerance -v
> > > +
> > > +		local agcount=$(_xfs_mount_agcount $SCRATCH_MNT)
> > > +		for batchsize in 64 2 1; do
> > 
> > Perhaps we should stuff a value > than the per-record inode count in
> > here as well.
> 
> I'm confused about this comment -- inumbers returns (cooked) inobt
> records, not inode records themselves.  The 64>2>1 sequence here asks
> inumbers to return 64 inobt records per call, then 2 per call, then 1.
> 
> Maybe this should be 71 or some other prime number...
> 

Er, disregard.. I was thinking in terms of inode count when I read this.

Brian

> > > +			echo -n "inumbers $batchsize($v_tag): "
> > > +			nr=$(inumbers_ag $agcount $batchsize $SCRATCH_MNT $v_flag)
> > > +			_within_tolerance "inumbers" $nr $expect $tolerance -v
> > > +		done
> > > +	done
> > > +}
> > > +
> > > +# compare the src/bstat output against the xfs_io bstat output
> > 
> > This compares actual inode numbers, right? If so, I'd point that out in
> > the comment.
> 
> Ok.
> 
> > > +bstat_compare()
> > > +{
> > > +	bstat_versions | while read v_tag v_flag; do
> > > +		diff -u <(./src/bstat $SCRATCH_MNT | grep ino | awk '{print $2}') \
> > > +			<($XFS_IO_PROG -c "bulkstat $v_flag" $SCRATCH_MNT | grep ino | awk '{print $3}')
> > > +	done
> > > +}
> > > +
> > ...
> > > diff --git a/tests/xfs/745 b/tests/xfs/745
> > > new file mode 100755
> > > index 00000000..6931d46b
> > > --- /dev/null
> > > +++ b/tests/xfs/745
> > > @@ -0,0 +1,44 @@
> > > +#! /bin/bash
> > > +# SPDX-License-Identifier: GPL-2.0+
> > > +# Copyright (c) 2019 Oracle, Inc.  All Rights Reserved.
> > > +#
> > > +# FS QA Test No. 745
> > > +#
> > > +# Regression test for a long-standing bug in BULKSTAT and INUMBERS where
> > > +# the kernel fails to write thew new @lastip value back to userspace if
> > > +# @ocount is NULL.
> > > +#
> > 
> > I think it would be helpful to reference the upstream fix here, which
> > IIRC is commit f16fe3ecde62 ("xfs: bulkstat should copy lastip whenever
> > userspace supplies one"). Otherwise this test looks fine to me.
> 
> Ok, will update.
> 
> --D
> 
> > Brian
> > 
> > > +seq=`basename $0`
> > > +seqres=$RESULT_DIR/$seq
> > > +echo "QA output created by $seq"
> > > +
> > > +here=`pwd`
> > > +tmp=/tmp/$$
> > > +status=1	# failure is the default!
> > > +trap "_cleanup; exit \$status" 0 1 2 3 15
> > > +
> > > +_cleanup()
> > > +{
> > > +    cd /
> > > +    rm -f $tmp.*
> > > +}
> > > +
> > > +# get standard environment, filters and checks
> > > +. ./common/rc
> > > +. ./common/filter
> > > +
> > > +_require_test_program "bulkstat_null_ocount"
> > > +
> > > +# real QA test starts here
> > > +
> > > +_supported_fs xfs
> > > +_supported_os Linux
> > > +
> > > +rm -f $seqres.full
> > > +
> > > +echo "Silence is golden."
> > > +src/bulkstat_null_ocount $TEST_DIR
> > > +
> > > +# success, all done
> > > +status=0
> > > +exit
> > > diff --git a/tests/xfs/745.out b/tests/xfs/745.out
> > > new file mode 100644
> > > index 00000000..ce947de2
> > > --- /dev/null
> > > +++ b/tests/xfs/745.out
> > > @@ -0,0 +1,2 @@
> > > +QA output created by 745
> > > +Silence is golden.
> > > diff --git a/tests/xfs/group b/tests/xfs/group
> > > index 270d82ff..ef0cf92c 100644
> > > --- a/tests/xfs/group
> > > +++ b/tests/xfs/group
> > > @@ -506,3 +506,5 @@
> > >  506 auto quick health
> > >  507 clone
> > >  508 auto quick quota
> > > +744 auto ioctl quick
> > > +745 auto ioctl quick
> > > 

Re: [PATCH 0/4] xfs: fixes and new tests for bulkstat v5

[Eryu Guan]

Hi XFS folks,

On Tue, Jul 23, 2019 at 09:13:14PM -0700, Darrick J. Wong wrote:
> Hi all,
> 
> Fix some problems introduced by the creation of the V5 bulkstat ioctl,
> and then add some new tests to make sure the new libxfrog bulkstat
> wrappers work fine with both the new v5 ioctl and emulating it with the
> old v1 ioctl.

I may need some help on reviewing this patchset, especially the new
bulkstat tests :) Thanks in advance!

But I'd suggest split the last patch into two patches, one introduces &
uses the new helper, the other one adds new tests. Also, it misses a new
entry in .gitignore file.

Thanks,
Eryu

> 
> If you're going to start using this mess, you probably ought to just
> pull from my git trees, which are linked below.
> 
> This is an extraordinary way to destroy everything.  Enjoy!
> Comments and questions are, as always, welcome.
> 
> --D
> 
> kernel git tree:
> https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=bulkstat-v5
> 
> xfsprogs git tree:
> https://git.kernel.org/cgit/linux/kernel/git/djwong/xfsprogs-dev.git/log/?h=bulkstat-v5
> 
> fstests git tree:
> https://git.kernel.org/cgit/linux/kernel/git/djwong/xfstests-dev.git/log/?h=bulkstat-v5

Re: [PATCH 0/4] xfs: fixes and new tests for bulkstat v5

[Darrick J. Wong]

On Sun, Jul 28, 2019 at 07:26:35PM +0800, Eryu Guan wrote:
> Hi XFS folks,
> 
> On Tue, Jul 23, 2019 at 09:13:14PM -0700, Darrick J. Wong wrote:
> > Hi all,
> > 
> > Fix some problems introduced by the creation of the V5 bulkstat ioctl,
> > and then add some new tests to make sure the new libxfrog bulkstat
> > wrappers work fine with both the new v5 ioctl and emulating it with the
> > old v1 ioctl.
> 
> I may need some help on reviewing this patchset, especially the new
> bulkstat tests :) Thanks in advance!

Let me know if you have any questions. :)

> But I'd suggest split the last patch into two patches, one introduces &
> uses the new helper, the other one adds new tests. Also, it misses a new
> entry in .gitignore file.

Oops, will fix.

--D

> Thanks,
> Eryu
> 
> > 
> > If you're going to start using this mess, you probably ought to just
> > pull from my git trees, which are linked below.
> > 
> > This is an extraordinary way to destroy everything.  Enjoy!
> > Comments and questions are, as always, welcome.
> > 
> > --D
> > 
> > kernel git tree:
> > https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=bulkstat-v5
> > 
> > xfsprogs git tree:
> > https://git.kernel.org/cgit/linux/kernel/git/djwong/xfsprogs-dev.git/log/?h=bulkstat-v5
> > 
> > fstests git tree:
> > https://git.kernel.org/cgit/linux/kernel/git/djwong/xfstests-dev.git/log/?h=bulkstat-v5