diff mbox

[2/2] btrfs-progs: introduce btrfs filesystem show --kernel

Message ID 1375949227-23729-3-git-send-email-anand.jain@oracle.com (mailing list archive)
State Under Review, archived
Headers show

Commit Message

Anand Jain Aug. 8, 2013, 8:07 a.m. UTC
As of now btrfs filesystem show reads directly from
disks. So sometimes output can be stale, mainly when
user want to verify their last operation like,
labeling or device delete or add... etc.

This patch adds --kernel option to the 'filesystem show'
subcli, which will read from the kernel instead of
the disks directly.

also this path adds the group profile info to the
output

eg:
-----------------
btrfs fi show --kernel
Label: none  uuid: 39f55f14-e5ca-4a01-899d-915fd35bde05 mounted: /btrfs
        Group profile: metadata: RAID1  data: RAID1
        Total devices 2 FS bytes used 7.40GB
        devid    1 size 48.23GB used 11.04GB path /dev/dm-5
        devid    2 size 44.99GB used 11.03GB path /dev/mapper/mpathe

Label: none  uuid: a0beeb78-0019-4bdf-8002-0900a123ee07 mounted: /btrfs1
        Group profile: mixed: single
        Total devices 1 FS bytes used 7.40GB
        devid    1 size 15.00GB used 9.01GB path /dev/mapper/mpathbp1

btrfs fi show --kernel /btrfs2
Label: none  uuid: 9d6a347e-e8a0-44fe-9d2a-d28ee45ef33f mounted: /btrfs2
        Group profile: metadata: DUP  data: single
        Total devices 1 FS bytes used 2.22MB
        devid    1 size 15.00GB used 1.32GB path /dev/mapper/mpathcp1

btrfs fi show --kernel 9d6a347e-e8a0-44fe-9d2a-d28ee45ef33f
Label: none  uuid: 9d6a347e-e8a0-44fe-9d2a-d28ee45ef33f mounted: /btrfs2
        Group profile: metadata: DUP  data: single
        Total devices 1 FS bytes used 2.22MB
        devid    1 size 15.00GB used 1.32GB path /dev/mapper/mpathcp1
------------

v3->v4:
        dropped the dependence of used_bytes from the ioctl
        kernel, Instead used the get_df to calculate the
        used space.
        dropped the function device_list_add_from_kernel
        to update the original device_list_add instead
        I have my own print and device filters, this way I
        can add the group profile information in the show
        output.
v2->v3:
        Do the stuffs without adding new ioctl
        new dependencies: this patch also depends on
        path 9/13 to 12/13 also sent here.
v1->v2:
        code optimized to remove redundancy

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
 cmds-filesystem.c | 165 +++++++++++++++++++++++++++++++++++++++++++++++++++---
 man/btrfs.8.in    |   5 +-
 2 files changed, 159 insertions(+), 11 deletions(-)

Comments

Zach Brown Aug. 8, 2013, 6:08 p.m. UTC | #1
On Thu, Aug 08, 2013 at 04:07:07PM +0800, Anand Jain wrote:
> As of now btrfs filesystem show reads directly from
> disks. So sometimes output can be stale, mainly when
> user want to verify their last operation like,
> labeling or device delete or add... etc.
> 
> This patch adds --kernel option to the 'filesystem show'
> subcli, which will read from the kernel instead of
> the disks directly.

Why should this be an option?

When mounted, the kernel cache is authoritative.  It was always a bug to
read stale data from disk.

The kernel should be read first, and if that isn't available it can fall
back to offering unreliable data from disk with a giant wraning.

Right?

- z
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Anand Jain Aug. 9, 2013, 10:57 a.m. UTC | #2
On 09/08/2013 02:08, Zach Brown wrote:
> On Thu, Aug 08, 2013 at 04:07:07PM +0800, Anand Jain wrote:
>> As of now btrfs filesystem show reads directly from
>> disks. So sometimes output can be stale, mainly when
>> user want to verify their last operation like,
>> labeling or device delete or add... etc.
>>
>> This patch adds --kernel option to the 'filesystem show'
>> subcli, which will read from the kernel instead of
>> the disks directly.
>
> Why should this be an option?
>
> When mounted, the kernel cache is authoritative.  It was always a bug to
> read stale data from disk.
>
> The kernel should be read first, and if that isn't available it can fall
> back to offering unreliable data from disk with a giant wraning.
>
> Right?

  Yes !
  For the no option - what I like is more an intelligent way
  of listing all btrfs - mounted and unmounted by appropriately
  reading from the kernel and disks respectively.

  However to do this, we need a better disk list with in btrfs-progs
  which isn't there yet. I am getting those related components ready
  to be there someday.

  So at the moment an interim solution for the bug, which will
  also help the long term design.

Thanks, Anand
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Zach Brown Aug. 9, 2013, 6:03 p.m. UTC | #3
> >Right?
> 
>  Yes !
>  For the no option - what I like is more an intelligent way
>  of listing all btrfs - mounted and unmounted by appropriately
>  reading from the kernel and disks respectively.

Great!

It'll be good to get this fixed.  I stumbled across this behaviour of
btrfs fi show lying to me just yesterday.

- z
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/cmds-filesystem.c b/cmds-filesystem.c
index be8afde..74ad30b 100644
--- a/cmds-filesystem.c
+++ b/cmds-filesystem.c
@@ -22,6 +22,9 @@ 
 #include <errno.h>
 #include <uuid/uuid.h>
 #include <ctype.h>
+#include <mntent.h>
+#include <fcntl.h>
+#include <linux/limits.h>
 
 #include "kerncompat.h"
 #include "ctree.h"
@@ -251,8 +254,124 @@  static void print_one_uuid(struct btrfs_fs_devices *fs_devices)
 	printf("\n");
 }
 
+/* adds up all the used spaces as reported by the space info ioctl
+ */
+static u64 cal_used_bytes(struct btrfs_ioctl_space_args *si)
+{
+	u64 ret = 0;
+	int i;
+	for (i = 0; i < si->total_spaces; i++)
+		ret += si->spaces[i].used_bytes;
+	return ret;
+}
+
+static int print_one_fs(struct btrfs_ioctl_fs_info_args *fi,
+		struct btrfs_ioctl_dev_info_args *di_n,
+		struct btrfs_ioctl_space_args *si_n, char *label, char *path)
+{
+	int i;
+	char uuidbuf[37];
+	struct btrfs_ioctl_dev_info_args *di = di_n;
+	u64 flags;
+
+	uuid_unparse(fi->fsid, uuidbuf);
+	printf("Label: %s  uuid: %s mounted: %s\n",
+		strlen(label)?label:"none", uuidbuf, path);
+	printf("\tGroup profile:");
+	for (i = si_n->total_spaces - 1; i >= 0; i--) {
+		flags = si_n->spaces[i].flags;
+		if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+			continue;
+		printf(" %s: %s", group_type_str(flags),
+					group_profile_str(flags));
+		printf(" ");
+	}
+	printf("\n");
+
+	printf("\tTotal devices %llu FS bytes used %s\n",
+				fi->num_devices,
+			pretty_size(cal_used_bytes(si_n)));
+
+	for (i = 0; i < fi->num_devices; i++) {
+		di = (struct btrfs_ioctl_dev_info_args *)&di_n[i];
+		printf("\tdevid    %llu size %s used %s path %s\n",
+			di->devid,
+			pretty_size(di->total_bytes),
+			pretty_size(di->bytes_used),
+			di->path);
+	}
+
+	printf("\n");
+	return 0;
+}
+
+/* This function checks if the given input parameter is
+ * an uuid or a path
+ * return -1: some error in the given input
+ * return 0: unknow input
+ * return 1: given input is uuid
+ * return 2: given input is path
+ */
+static int check_arg_type(char *input, u8 *processed)
+{
+	int ret = 0;
+	if (!uuid_parse(input, processed))
+		ret = 1;
+	else if (realpath(input, (char *)processed))
+		ret = 2;
+	return ret;
+}
+
+static int btrfs_scan_kernel(void *input, int type)
+{
+	int ret = 0, fd;
+	FILE *f;
+	struct mntent *mnt;
+	struct btrfs_ioctl_fs_info_args fi;
+	struct btrfs_ioctl_dev_info_args *di = NULL;
+	struct btrfs_ioctl_space_args *si;
+	char label[BTRFS_LABEL_SIZE];
+
+	if ((f = setmntent ("/proc/mounts", "r")) == NULL)
+		return -errno;
+
+	while ((mnt = getmntent (f)) != NULL) {
+		if (strcmp(mnt->mnt_type, "btrfs"))
+			continue;
+		ret = get_fs_info(mnt->mnt_dir, &fi, &di);
+		if (ret)
+			return ret;
+
+		switch (type) {
+		case 0:
+			break;
+		case 1:
+			if (uuid_compare(fi.fsid, (u8 *)input))
+				continue;
+			break;
+		case 2:
+			if (strcmp(input, mnt->mnt_dir))
+				continue;
+			break;
+		default:
+			break;
+		}
+
+		fd = open(mnt->mnt_dir, O_RDONLY);
+		if (fd > 0 && !get_df(fd, &si)) {
+			get_label_mounted(mnt->mnt_dir, label);
+			print_one_fs(&fi, di, si, label, mnt->mnt_dir);
+			free(si);
+		}
+		if (fd > 0)
+			close(fd);
+		free(di);
+	}
+	return ret;
+}
+
 static const char * const cmd_show_usage[] = {
-	"btrfs filesystem show [--all-devices|<uuid>]",
+	"btrfs filesystem show [--all-devices|--mapper|--kernel|<uuid>]",
 	"Show the structure of a filesystem",
 	"If no argument is given, structure of all present filesystems is shown.",
 	NULL
@@ -264,23 +383,51 @@  static int cmd_show(int argc, char **argv)
 	struct btrfs_fs_devices *fs_devices;
 	struct list_head *cur_uuid;
 	char *search = 0;
-	int ret;
+	int ret = 0;
 	int where = BTRFS_SCAN_PROC;
 	int searchstart = 1;
+	u8 processed[PATH_MAX];
 
-	if( argc > 1 && !strcmp(argv[1],"--all-devices")){
+	if( argc > 1 && !strcmp(argv[1], "--all-devices")){
 		where = BTRFS_SCAN_DEV;
 		searchstart += 1;
+	} else if (argc > 1 && !strcmp(argv[1], "--kernel")) {
+		where = 0;
+		searchstart += 1;
 	}
 
-	if (check_argc_max(argc, searchstart + 1))
-		usage(cmd_show_usage);
+	if (!where) {
+		if (! (searchstart < argc))
+			ret = btrfs_scan_kernel(NULL, 0);
 
-	ret = scan_for_btrfs(where, 0);
+		while (searchstart < argc) {
+			ret = check_arg_type(argv[searchstart], processed);
+			if (ret < 0) {
+				fprintf(stderr, "ERROR at input %s\n",
+							argv[searchstart]);
+				return 1;
+			}
+			if (!ret) {
+				fprintf(stderr, "ERROR unknown %s\n",
+					argv[searchstart]);
+				return 1;
+			}
+
+			ret = btrfs_scan_kernel(processed, ret);
+			if (ret)
+				break;
+			searchstart++;
+		}
+		if (ret)
+			fprintf(stderr, "ERROR: scan kernel failed, %d\n",
+				ret);
+		return ret;
+	}
 
-	if (ret){
-		fprintf(stderr, "ERROR: error %d while scanning\n", ret);
-		return 18;
+	ret = scan_for_btrfs(where, 0);
+	if (ret) {
+		fprintf(stderr, "ERROR: %d while scanning\n", ret);
+		return 1;
 	}
 	
 	if(searchstart < argc)
diff --git a/man/btrfs.8.in b/man/btrfs.8.in
index e9ec1a7..6383469 100644
--- a/man/btrfs.8.in
+++ b/man/btrfs.8.in
@@ -25,7 +25,7 @@  btrfs \- control a btrfs filesystem
 .PP
 \fBbtrfs\fP \fBfilesystem df\fP\fI <path>\fP
 .PP
-\fBbtrfs\fP \fBfilesystem show\fP [--all-devices|\fI<uuid>\fP|\fI<label>\fP]\fP
+\fBbtrfs\fP \fBfilesystem show\fP\fI [--all-devices|--kernel] [\fI<uuid>|<path>]\fP\fP
 .PP
 \fBbtrfs\fP \fBfilesystem sync\fP\fI <path> \fP
 .PP
@@ -254,11 +254,12 @@  Show information of a given subvolume in the \fI<path>\fR.
 Show space usage information for a mount point.
 .TP
 
-\fBfilesystem show\fR [--all-devices|\fI<uuid>\fR|\fI<label>\fR]\fR
+\fBfilesystem show\fR [--all-devices|--kernel] [\fI<uuid>|<path>]\fP\fP
 Show the btrfs filesystem with some additional info. If no \fIUUID\fP or
 \fIlabel\fP is passed, \fBbtrfs\fR show info of all the btrfs filesystem.
 If \fB--all-devices\fP is passed, all the devices under /dev are scanned;
 otherwise the devices list is extracted from the /proc/partitions file.
+The --kernel will scan the btrfs kernel for the mounted btrfs.
 .TP
 
 \fBfilesystem sync\fR\fI <path> \fR