From: Mark Fasheh <mfasheh@suse.de>
To: linux-btrfs@vger.kernel.org
Cc: David Sterba <dsterba@suse.com>, Chris Mason <clm@fb.com>,
Mark Fasheh <mfasheh@suse.de>
Subject: [PATCH 2/3] btrfs-progs: add 'du' command
Date: Wed, 20 Jan 2016 13:49:26 -0800 [thread overview]
Message-ID: <1453326567-20454-3-git-send-email-mfasheh@suse.de> (raw)
In-Reply-To: <1453326567-20454-1-git-send-email-mfasheh@suse.de>
'btrfs du' differs from regular du in that it will work to resolve which
blocks are shared between files in its list. This gives the user a more
accurate bytecount from which they can make decisions regarding management
of their file space.
We still print a total number of bytes counted (like regular du), but also
print the number of bytes which were found to have been shared amongst the
file set provided. From there it becomes trivial to calculate how much space
is exclusively owned.
Signed-off-by: Mark Fasheh <mfasheh@suse.de>
---
Makefile.in | 2 +-
cmds-du.c | 368 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
cmds-du.h | 7 ++
cmds-filesystem.c | 3 +-
4 files changed, 378 insertions(+), 2 deletions(-)
create mode 100644 cmds-du.c
create mode 100644 cmds-du.h
diff --git a/Makefile.in b/Makefile.in
index 8e24808..901c9e8 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -74,7 +74,7 @@ cmds_objects = cmds-subvolume.o cmds-filesystem.o cmds-device.o cmds-scrub.o \
cmds-inspect.o cmds-balance.o cmds-send.o cmds-receive.o \
cmds-quota.o cmds-qgroup.o cmds-replace.o cmds-check.o \
cmds-restore.o cmds-rescue.o chunk-recover.o super-recover.o \
- cmds-property.o cmds-fi-usage.o
+ cmds-property.o cmds-fi-usage.o cmds-du.o
libbtrfs_objects = send-stream.o send-utils.o rbtree.o btrfs-list.o crc32c.o \
uuid-tree.o utils-lib.o rbtree-utils.o
libbtrfs_headers = send-stream.h send-utils.h send.h rbtree.h btrfs-list.h \
diff --git a/cmds-du.c b/cmds-du.c
new file mode 100644
index 0000000..57758d2
--- /dev/null
+++ b/cmds-du.c
@@ -0,0 +1,368 @@
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdarg.h>
+#include <getopt.h>
+#include <fcntl.h>
+#include <dirent.h>
+
+#include <sys/ioctl.h>
+#include <linux/fs.h>
+#include <linux/fiemap.h>
+
+#include "utils.h"
+#include "commands.h"
+#include "kerncompat.h"
+#include "rbtree.h"
+
+static int summarize = 0;
+static unsigned unit_mode = UNITS_RAW;
+static char path[PATH_MAX] = { 0, };
+static char *pathp = path;
+static char *path_max = &path[PATH_MAX - 1];
+
+/* Track which inodes we've seen for the purposes of hardlink detection. */
+struct seen_inode {
+ struct rb_node i_node;
+ u64 i_ino;
+ u64 i_subvol;
+};
+static struct rb_root seen_inodes = RB_ROOT;
+
+static int cmp_si(struct seen_inode *si, u64 ino, u64 subvol)
+{
+ if (ino < si->i_ino)
+ return -1;
+ else if (ino > si->i_ino)
+ return 1;
+ if (subvol < si->i_subvol)
+ return -1;
+ else if (subvol > si->i_subvol)
+ return 1;
+ return 0;
+}
+
+static int mark_inode_seen(u64 ino, u64 subvol)
+{
+ int cmp;
+ struct rb_node **p = &seen_inodes.rb_node;
+ struct rb_node *parent = NULL;
+ struct seen_inode *si;
+
+ while (*p) {
+ parent = *p;
+
+ si = rb_entry(parent, struct seen_inode, i_node);
+ cmp = cmp_si(si, ino, subvol);
+ if (cmp < 0)
+ p = &(*p)->rb_left;
+ else if (cmp > 0)
+ p = &(*p)->rb_right;
+ else
+ BUG();
+ }
+
+ si = calloc(1, sizeof(*si));
+ if (!si)
+ return ENOMEM;
+
+ si->i_ino = ino;
+ si->i_subvol = subvol;
+
+ rb_link_node(&si->i_node, parent, p);
+ rb_insert_color(&si->i_node, &seen_inodes);
+
+ return 0;
+}
+
+static int inode_seen(u64 ino, u64 subvol)
+{
+ int cmp;
+ struct rb_node *n = seen_inodes.rb_node;
+ struct seen_inode *si;
+
+ while (n) {
+ si = rb_entry(n, struct seen_inode, i_node);
+
+ cmp = cmp_si(si, ino, subvol);
+ if (cmp < 0)
+ n = n->rb_left;
+ else if (cmp > 0)
+ n = n->rb_right;
+ else
+ return EEXIST;
+ }
+ return 0;
+
+}
+
+const char * const cmd_filesystem_du_usage[] = {
+ "btrfs filesystem du [options] <path> [<path>..]",
+ "Summarize disk usage of each file.",
+ "-h|--human-readable",
+ " human friendly numbers, base 1024 (default)",
+ "-s display only a total for each argument",
+ NULL
+};
+
+/*
+ * Inline extents are skipped because they do not take data space,
+ * delalloc and unknown are skipped because we do not know how much
+ * space they will use yet.
+ */
+#define SKIP_FLAGS (FIEMAP_EXTENT_UNKNOWN|FIEMAP_EXTENT_DELALLOC|FIEMAP_EXTENT_DATA_INLINE)
+static int du_calc_file_space(int dirfd, const char *filename,
+ uint64_t *ret_total, uint64_t *ret_shared)
+{
+ char buf[16384];
+ struct fiemap *fiemap = (struct fiemap *)buf;
+ struct fiemap_extent *fm_ext = &fiemap->fm_extents[0];
+ int count = (sizeof(buf) - sizeof(*fiemap)) /
+ sizeof(struct fiemap_extent);
+ unsigned int i, ret;
+ int last = 0;
+ int rc;
+ u64 ext_len;
+ int fd;
+ u64 file_total = 0;
+ u64 file_shared = 0;
+ u32 flags;
+
+ memset(fiemap, 0, sizeof(struct fiemap));
+
+ fd = openat(dirfd, filename, O_RDONLY);
+ if (fd == -1) {
+ ret = errno;
+ fprintf(stderr, "ERROR: can't access '%s': %s\n",
+ filename, strerror(ret));
+ return ret;
+ }
+
+ do {
+ fiemap->fm_length = ~0ULL;
+ fiemap->fm_extent_count = count;
+ rc = ioctl(fd, FS_IOC_FIEMAP, (unsigned long) fiemap);
+ if (rc < 0) {
+ ret = errno;
+ goto out_close;
+ }
+
+ /* If 0 extents are returned, then more ioctls are not needed */
+ if (fiemap->fm_mapped_extents == 0)
+ break;
+
+ for (i = 0; i < fiemap->fm_mapped_extents; i++) {
+ ext_len = fm_ext[i].fe_length;
+ flags = fm_ext[i].fe_flags;
+
+ if (flags & FIEMAP_EXTENT_LAST)
+ last = 1;
+
+ if (flags & SKIP_FLAGS)
+ continue;
+
+ file_total += ext_len;
+ if (flags & FIEMAP_EXTENT_SHARED)
+ file_shared += ext_len;
+ }
+
+ fiemap->fm_start = (fm_ext[i - 1].fe_logical +
+ fm_ext[i - 1].fe_length);
+ } while (last == 0);
+
+ *ret_total = file_total;
+ *ret_shared = file_shared;
+
+ ret = 0;
+out_close:
+ close(fd);
+ return ret;
+}
+
+struct du_dir_ctxt {
+ uint64_t bytes_total;
+ uint64_t bytes_shared;
+};
+
+static int du_add_file(const char *filename, int dirfd, uint64_t *ret_total,
+ uint64_t *ret_shared, int top_level);
+
+static int du_walk_dir(struct du_dir_ctxt *ctxt)
+{
+ int fd, ret, type;
+ DIR *dirstream = NULL;
+ struct dirent *entry;
+
+ fd = open_file_or_dir(path, &dirstream);
+ if (fd < 0)
+ return fd;
+
+ ret = 0;
+ do {
+ uint64_t tot, shr;
+
+ errno = 0;
+ entry = readdir(dirstream);
+ if (entry) {
+ if (strcmp(entry->d_name, ".") == 0
+ || strcmp(entry->d_name, "..") == 0)
+ continue;
+
+ type = entry->d_type;
+ if (type == DT_REG || type == DT_DIR) {
+ tot = shr = 0;
+
+ ret = du_add_file(entry->d_name,
+ dirfd(dirstream), &tot,
+ &shr, 0);
+ if (ret)
+ break;
+
+ ctxt->bytes_total += tot;
+ ctxt->bytes_shared += shr;
+ }
+ }
+ } while (entry != NULL);
+
+ close_file_or_dir(fd, dirstream);
+ return ret;
+}
+
+static int du_add_file(const char *filename, int dirfd, uint64_t *ret_total,
+ uint64_t *ret_shared, int top_level)
+{
+ int ret, len = strlen(filename);
+ char *pathtmp;
+ struct stat st;
+ struct du_dir_ctxt dir;
+ uint64_t file_total = 0;
+ uint64_t file_shared = 0;
+ u64 subvol;
+ int fd;
+ DIR *dirstream = NULL;
+
+ ret = fstatat(dirfd, filename, &st, 0);
+ if (ret) {
+ ret = errno;
+ return ret;
+ }
+
+ if (!S_ISREG(st.st_mode) && !S_ISDIR(st.st_mode))
+ return 0;
+
+ if (len > (path_max - pathp)) {
+ fprintf(stderr, "ERROR: Path max exceeded: %s %s\n", path,
+ filename);
+ return ENAMETOOLONG;
+ }
+
+ pathtmp = pathp;
+ if (pathp == path)
+ ret = sprintf(pathp, "%s", filename);
+ else
+ ret = sprintf(pathp, "/%s", filename);
+ pathp += ret;
+
+ fd = open_file_or_dir(path, &dirstream);
+ if (fd < 0) {
+ ret = fd;
+ goto out;
+ }
+
+ ret = lookup_ino_rootid(fd, &subvol);
+ if (ret)
+ goto out_close;
+
+ if (inode_seen(st.st_ino, subvol))
+ goto out_close;
+
+ ret = mark_inode_seen(st.st_ino, subvol);
+ if (ret)
+ goto out_close;
+
+ if (S_ISREG(st.st_mode)) {
+ ret = du_calc_file_space(dirfd, filename, &file_total,
+ &file_shared);
+ if (ret)
+ goto out_close;
+ } else if (S_ISDIR(st.st_mode)) {
+ memset(&dir, 0, sizeof(dir));
+
+ ret = du_walk_dir(&dir);
+ *pathp = '\0';
+ if (ret)
+ goto out_close;
+
+ file_total = dir.bytes_total;
+ file_shared = dir.bytes_shared;
+ }
+
+ if (!summarize || top_level) {
+ printf("%s\t%s\t%s\n", pretty_size_mode(file_total, unit_mode),
+ pretty_size_mode((file_total - file_shared), unit_mode),
+ path);
+ }
+
+ if (ret_total)
+ *ret_total = file_total;
+ if (ret_shared)
+ *ret_shared = file_shared;
+
+out_close:
+ close_file_or_dir(fd, dirstream);
+out:
+ /* reset path to just before this element */
+ pathp = pathtmp;
+
+ return ret;
+}
+
+int cmd_filesystem_du(int argc, char **argv)
+{
+ int ret = 0, error = 0;
+ int i;
+
+ optind = 1;
+ while (1) {
+ int long_index;
+ static const struct option long_options[] = {
+ { "summarize", no_argument, NULL, 's'},
+ { "human-readable", no_argument, NULL, 'h'},
+ { NULL, 0, NULL, 0 }
+ };
+ int c = getopt_long(argc, argv, "sh", long_options,
+ &long_index);
+
+ if (c < 0)
+ break;
+ switch (c) {
+ case 'h':
+ unit_mode = UNITS_HUMAN;
+ break;
+ case 's':
+ summarize = 1;
+ break;
+ default:
+ usage(cmd_filesystem_du_usage);
+ }
+ }
+
+ if (check_argc_min(argc - optind, 1))
+ usage(cmd_filesystem_du_usage);
+
+ printf("total\texclusive\tfilename\n");
+
+ for (i = optind; i < argc; i++) {
+ ret = du_add_file(argv[i], AT_FDCWD, NULL, NULL, 1);
+ if (ret) {
+ fprintf(stderr, "ERROR: can't check space of '%s': %s\n",
+ argv[i], strerror(ret));
+ error = 1;
+ }
+ }
+
+ return error;
+}
diff --git a/cmds-du.h b/cmds-du.h
new file mode 100644
index 0000000..5ffb178
--- /dev/null
+++ b/cmds-du.h
@@ -0,0 +1,7 @@
+#ifndef __CMDS_DU_H__
+#define __CMDS_DU_H__
+
+extern const char * const cmd_filesystem_du_usage[];
+int cmd_filesystem_du(int argc, char **argv);
+
+#endif /* __CMDS_DU_H__ */
diff --git a/cmds-filesystem.c b/cmds-filesystem.c
index 25317fa..4973b10 100644
--- a/cmds-filesystem.c
+++ b/cmds-filesystem.c
@@ -37,7 +37,7 @@
#include "cmds-fi-usage.h"
#include "list_sort.h"
#include "disk-io.h"
-
+#include "cmds-du.h"
/*
* for btrfs fi show, we maintain a hash of fsids we've already printed.
@@ -1287,6 +1287,7 @@ static const char filesystem_cmd_group_info[] =
const struct cmd_group filesystem_cmd_group = {
filesystem_cmd_group_usage, filesystem_cmd_group_info, {
{ "df", cmd_filesystem_df, cmd_filesystem_df_usage, NULL, 0 },
+ { "du", cmd_filesystem_du, cmd_filesystem_du_usage, NULL, 0 },
{ "show", cmd_filesystem_show, cmd_filesystem_show_usage, NULL,
0 },
{ "sync", cmd_filesystem_sync, cmd_filesystem_sync_usage, NULL,
--
2.1.4
next prev parent reply other threads:[~2016-01-20 21:49 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-01-20 21:49 [PATCH] btrfs-progs: add 'du' command Mark Fasheh
2016-01-20 21:49 ` [PATCH 1/3] btrfs-progs: Import interval tree implemenation from Linux v4.0-rc7 Mark Fasheh
2016-01-20 21:49 ` Mark Fasheh [this message]
2016-02-25 8:57 ` [PATCH 2/3] btrfs-progs: add 'du' command Jérôme Poulin
2016-02-25 9:25 ` David Sterba
2016-01-20 21:49 ` [PATCH 3/3] btrfs-progs du: Calculate space shared by each directory arguments file set Mark Fasheh
2016-02-01 23:43 ` [PATCH] btrfs-progs: add 'du' command David Sterba
2016-02-02 20:09 ` Mark Fasheh
2016-02-24 15:28 ` David Sterba
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=1453326567-20454-3-git-send-email-mfasheh@suse.de \
--to=mfasheh@suse.de \
--cc=clm@fb.com \
--cc=dsterba@suse.com \
--cc=linux-btrfs@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).