Message ID | 1381203700-17558-1-git-send-email-anand.jain@oracle.com (mailing list archive) |
---|---|
State | Accepted, archived |
Headers | show |
On Tue, Oct 08, 2013 at 11:41:38AM +0800, Anand Jain wrote: > As of now btrfs filesystem show reads directly from > disks. So sometimes output can be stale, mainly when > user wants to cross verify their operation like, > label or device delete or add... etc. so this > patch will read from the kernel ioctl if it finds > that disk is mounted. Sorry for holding this patch back for so long, I wanted to find out why the output is different before and after. The reason is simple, for my convenience I've added the user to the 'disk' group so I can access the block devices without root. This patch stops to read the block devs directly and uses the ioctl BTRFS_IOC_FS_INFO which gives me a silent EPERM. If I run show with sudo, then all filesystems show up as expected. david -- 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
> The reason is simple, for my > convenience I've added the user to the 'disk' group so I can access the > block devices without root. This patch stops to read the block devs > directly and uses the ioctl BTRFS_IOC_FS_INFO which gives me a silent > EPERM. > If I run show with sudo, then all filesystems show up as > expected. Ah. good catch. but what do you suggest. ? as of now user on disk-group is able to do things like, check --repair. but won't be able to list the subvol / manage the snapshots, and is able to run btrfs fi show -d but won't be able to run btrfs fi show / -m. what do we do in this case ? OR when /sys interface is ready that should help to manage based on the user-groups.? 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
On Wed, Oct 16, 2013 at 05:25:37PM +0800, Anand Jain wrote: > Ah. good catch. but what do you suggest. ? > as of now user on disk-group is able to do things like, > check --repair. but won't be able to list the subvol / > manage the snapshots, and is able to run btrfs fi show -d > but won't be able to run btrfs fi show / -m. what do we > do in this case ? We can drop the CAP_SYS_ADMIN check from btrfs_ioctl_fs_info, it does not do or revealy sensitive information. > OR when /sys interface is ready that should help to manage > based on the user-groups.? This could be decided case by case if a sysfs file should not be publicly readable, otherwise let them be readable to everyone. david -- 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
On Tue, Oct 15, 2013 at 07:13:33PM +0200, David Sterba wrote: > On Tue, Oct 08, 2013 at 11:41:38AM +0800, Anand Jain wrote: > > As of now btrfs filesystem show reads directly from > > disks. So sometimes output can be stale, mainly when > > user wants to cross verify their operation like, > > label or device delete or add... etc. so this > > patch will read from the kernel ioctl if it finds > > that disk is mounted. > > Sorry for holding this patch back for so long, I wanted to find out why > the output is different before and after. The reason is simple, for my > convenience I've added the user to the 'disk' group so I can access the > block devices without root. This patch stops to read the block devs > directly and uses the ioctl BTRFS_IOC_FS_INFO which gives me a silent > EPERM. If I run show with sudo, then all filesystems show up as > expected. > This patch needs to be dropped (I imagine the whole series too) as it breaks xfstests btrfs/003. If I do btrfs fi show on a mounted fs I get this [root@destiny btrfs-progs]# btrfs fi show /dev/sdc Btrfs v0.20-rc1-483-ge0173f6 [root@destiny btrfs-progs]# Chris this is in integration, so maybe a revert? Thanks, Josef -- 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
On 10/21/13 10:44 PM, Josef Bacik wrote: > On Tue, Oct 15, 2013 at 07:13:33PM +0200, David Sterba wrote: >> On Tue, Oct 08, 2013 at 11:41:38AM +0800, Anand Jain wrote: >>> As of now btrfs filesystem show reads directly from >>> disks. So sometimes output can be stale, mainly when >>> user wants to cross verify their operation like, >>> label or device delete or add... etc. so this >>> patch will read from the kernel ioctl if it finds >>> that disk is mounted. >> >> Sorry for holding this patch back for so long, I wanted to find out why >> the output is different before and after. The reason is simple, for my >> convenience I've added the user to the 'disk' group so I can access the >> block devices without root. This patch stops to read the block devs >> directly and uses the ioctl BTRFS_IOC_FS_INFO which gives me a silent >> EPERM. If I run show with sudo, then all filesystems show up as >> expected. >> > > This patch needs to be dropped (I imagine the whole series too) as it breaks > xfstests btrfs/003. If I do btrfs fi show on a mounted fs I get this > > [root@destiny btrfs-progs]# btrfs fi show /dev/sdc > Btrfs v0.20-rc1-483-ge0173f6 > [root@destiny btrfs-progs]# > > Chris this is in integration, so maybe a revert? Thanks, That would be less productive approach as a whole. I have sent out the fix for this. Kindly find it. --- [PATCH 1/2] btrfs-progs: make get_btrfs_mount callable [PATCH 2/2] btrfs-progs: filesystem show of specified mounted disk should work --- 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
On Tue, Oct 22, 2013 at 01:53:47PM +0800, Anand Jain wrote: > > > On 10/21/13 10:44 PM, Josef Bacik wrote: > >On Tue, Oct 15, 2013 at 07:13:33PM +0200, David Sterba wrote: > >>On Tue, Oct 08, 2013 at 11:41:38AM +0800, Anand Jain wrote: > >>>As of now btrfs filesystem show reads directly from > >>>disks. So sometimes output can be stale, mainly when > >>>user wants to cross verify their operation like, > >>>label or device delete or add... etc. so this > >>>patch will read from the kernel ioctl if it finds > >>>that disk is mounted. > >> > >>Sorry for holding this patch back for so long, I wanted to find out why > >>the output is different before and after. The reason is simple, for my > >>convenience I've added the user to the 'disk' group so I can access the > >>block devices without root. This patch stops to read the block devs > >>directly and uses the ioctl BTRFS_IOC_FS_INFO which gives me a silent > >>EPERM. If I run show with sudo, then all filesystems show up as > >>expected. > >> > > > >This patch needs to be dropped (I imagine the whole series too) as it breaks > >xfstests btrfs/003. If I do btrfs fi show on a mounted fs I get this > > > >[root@destiny btrfs-progs]# btrfs fi show /dev/sdc > >Btrfs v0.20-rc1-483-ge0173f6 > >[root@destiny btrfs-progs]# > > > >Chris this is in integration, so maybe a revert? Thanks, > > That would be less productive approach as a whole. > I have sent out the fix for this. Kindly find it. > > --- > [PATCH 1/2] btrfs-progs: make get_btrfs_mount callable > [PATCH 2/2] btrfs-progs: filesystem show of specified mounted disk should > work > --- > Did you test these? Because they aren't working for me, so I think a revert is the only solution. Thanks, Josef -- 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
On Tue, Oct 22, 2013 at 09:21:47AM -0400, Josef Bacik wrote: > On Tue, Oct 22, 2013 at 01:53:47PM +0800, Anand Jain wrote: > > That would be less productive approach as a whole. > > I have sent out the fix for this. Kindly find it. > > > > --- > > [PATCH 1/2] btrfs-progs: make get_btrfs_mount callable > > [PATCH 2/2] btrfs-progs: filesystem show of specified mounted disk should > > work > > --- > > > > Did you test these? Because they aren't working for me, so I think a revert is > the only solution. Thanks, Just for the avoidance of doubt -- Josef means a full xfstests run on the patches. It may take a long time, but it does mean that everything(*) gets tested for regressions and helps prevent this to-and-fro of patches trying to fix up earlier bugs. Hugo. (*) OK, not quite everything, but Eric's and Josef's recent work is taking us closer to that goal.
On Tue, Oct 22, 2013 at 09:21:47AM -0400, Josef Bacik wrote: > Did you test these? Because they aren't working for me, so I think a revert is > the only solution. Thanks, The impact of the failing test is imho not that big to justify a full revert from integration branch (were it in master, then yes). I see the bug and I hope Anand will fix it soon. david -- 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
On 10/23/13 12:52 AM, David Sterba wrote: > On Tue, Oct 22, 2013 at 09:21:47AM -0400, Josef Bacik wrote: >> Did you test these? Because they aren't working for me, so I think a revert is >> the only solution. Thanks, > > The impact of the failing test is imho not that big to justify a full > revert from integration branch (were it in master, then yes). I see the > bug and I hope Anand will fix it soon. Sure. but what is the bug in btrfs-progs with this patch-set ? if its about xfstests btrfs/006 apparently it isn't working without this patch-set as well. 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
> On 10/23/13 12:52 AM, David Sterba wrote: >> On Tue, Oct 22, 2013 at 09:21:47AM -0400, Josef Bacik wrote: >>> Did you test these? Because they aren't working for me, so I think a >>> revert is >>> the only solution. Thanks, >> >> The impact of the failing test is imho not that big to justify a full >> revert from integration branch (were it in master, then yes). I see the >> bug and I hope Anand will fix it soon. > > Sure. but what is the bug in btrfs-progs with this patch-set ? > > if its about xfstests btrfs/006 apparently it isn't working > without this patch-set as well. and after btrfs/006 works with out this patch-set, we would need btrfs/006.out to be updated with the new btrfs fi show output/ IMO it should be part of this patch-set I will be sending a patch for xfstests. 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
diff --git a/cmds-filesystem.c b/cmds-filesystem.c index 780baad..9e0c8b9 100644 --- a/cmds-filesystem.c +++ b/cmds-filesystem.c @@ -25,6 +25,9 @@ #include <ctype.h> #include <fcntl.h> #include <ftw.h> +#include <mntent.h> +#include <linux/limits.h> +#include <getopt.h> #include "kerncompat.h" #include "ctree.h" @@ -251,9 +254,131 @@ 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 calc_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 *fs_info, + struct btrfs_ioctl_dev_info_args *dev_info, + struct btrfs_ioctl_space_args *space_info, + char *label, char *path) +{ + int i; + char uuidbuf[37]; + struct btrfs_ioctl_dev_info_args *tmp_dev_info; + + uuid_unparse(fs_info->fsid, uuidbuf); + printf("Label: %s uuid: %s\n", + strlen(label) ? label : "none", uuidbuf); + + printf("\tTotal devices %llu FS bytes used %s\n", + fs_info->num_devices, + pretty_size(calc_used_bytes(space_info))); + + for (i = 0; i < fs_info->num_devices; i++) { + tmp_dev_info = (struct btrfs_ioctl_dev_info_args *)&dev_info[i]; + printf("\tdevid %llu size %s used %s path %s\n", + tmp_dev_info->devid, + pretty_size(tmp_dev_info->total_bytes), + pretty_size(tmp_dev_info->bytes_used), + tmp_dev_info->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) +{ + uuid_t out; + char path[PATH_MAX]; + + if (!input) + return BTRFS_ARG_UNKNOWN; + + if (realpath(input, path)) + return BTRFS_ARG_PATH; + + if (!uuid_parse(input, out)) + return BTRFS_ARG_UUID; + + return BTRFS_ARG_UNKNOWN; +} + +static int btrfs_scan_kernel(void *search) +{ + int ret = 0, fd, type; + FILE *f; + struct mntent *mnt; + struct btrfs_ioctl_fs_info_args fs_info_arg; + struct btrfs_ioctl_dev_info_args *dev_info_arg = NULL; + struct btrfs_ioctl_space_args *space_info_arg; + char label[BTRFS_LABEL_SIZE]; + uuid_t uuid; + + f = setmntent("/proc/self/mounts", "r"); + if (f == NULL) + return 1; + + type = check_arg_type(search); + + while ((mnt = getmntent(f)) != NULL) { + if (strcmp(mnt->mnt_type, "btrfs")) + continue; + ret = get_fs_info(mnt->mnt_dir, &fs_info_arg, + &dev_info_arg); + if (ret) + return ret; + + switch (type) { + case BTRFS_ARG_UUID: + ret = uuid_parse(search, uuid); + if (ret) + return 1; + if (uuid_compare(fs_info_arg.fsid, uuid)) + continue; + break; + case BTRFS_ARG_PATH: + if (strcmp(search, mnt->mnt_dir)) + continue; + break; + default: + break; + } + + fd = open(mnt->mnt_dir, O_RDONLY); + if (fd > 0 && !get_df(fd, &space_info_arg)) { + get_label_mounted(mnt->mnt_dir, label); + print_one_fs(&fs_info_arg, dev_info_arg, + space_info_arg, label, mnt->mnt_dir); + free(space_info_arg); + } + if (fd > 0) + close(fd); + free(dev_info_arg); + } + return ret; +} + static const char * const cmd_show_usage[] = { - "btrfs filesystem show [--all-devices|<uuid>]", + "btrfs filesystem show [options] [<path>|<uuid>]", "Show the structure of a filesystem", + "-d|--all-devices show only disks under /dev containing btrfs filesystem", + "-m|--mounted show only mounted btrfs", "If no argument is given, structure of all present filesystems is shown.", NULL }; @@ -262,38 +387,87 @@ static int cmd_show(int argc, char **argv) { struct list_head *all_uuids; struct btrfs_fs_devices *fs_devices; + struct btrfs_device *device; struct list_head *cur_uuid; char *search = NULL; int ret; int where = BTRFS_SCAN_PROC; - int searchstart = 1; - - if( argc > 1 && !strcmp(argv[1],"--all-devices")){ - where = BTRFS_SCAN_DEV; - searchstart += 1; + int type = 0; + + while (1) { + int long_index; + static struct option long_options[] = { + { "all-devices", no_argument, NULL, 'd'}, + { "mounted", no_argument, NULL, 'm'}, + { NULL, no_argument, NULL, 0 }, + }; + int c = getopt_long(argc, argv, "dm", long_options, + &long_index); + if (c < 0) + break; + switch (c) { + case 'd': + where = BTRFS_SCAN_DEV; + break; + case 'm': + where = BTRFS_SCAN_MOUNTED; + break; + default: + usage(cmd_show_usage); + } } - if (check_argc_max(argc, searchstart + 1)) + if (check_argc_max(argc, optind + 1)) usage(cmd_show_usage); + if (argc > optind) { + search = argv[optind]; + type = check_arg_type(search); + if (type == BTRFS_ARG_UNKNOWN) { + fprintf(stderr, "ERROR: arg type unknown\n"); + usage(cmd_show_usage); + } + } + + if (where == BTRFS_SCAN_DEV) + goto devs_only; - ret = scan_for_btrfs(where, 0); + /* show mounted btrfs */ + btrfs_scan_kernel(search); - if (ret){ - fprintf(stderr, "ERROR: error %d while scanning\n", ret); + /* shows mounted only */ + if (where == BTRFS_SCAN_MOUNTED) + goto out; + +devs_only: + ret = scan_for_btrfs(where, !BTRFS_UPDATE_KERNEL); + + if (ret) { + fprintf(stderr, "ERROR: %d while scanning\n", ret); return 1; } - if(searchstart < argc) - search = argv[searchstart]; - all_uuids = btrfs_scanned_uuids(); list_for_each(cur_uuid, all_uuids) { fs_devices = list_entry(cur_uuid, struct btrfs_fs_devices, list); if (search && uuid_search(fs_devices, search) == 0) continue; + + /* skip mounted as they are already printed by + * btrfs_scan_kernel + */ + /* do it only for the default, no option */ + if (where == BTRFS_SCAN_PROC) { + device = list_entry(fs_devices->devices.next, + struct btrfs_device, dev_list); + ret = check_mounted(device->name); + if (ret) + continue; + } print_one_uuid(fs_devices); } + +out: printf("%s\n", BTRFS_BUILD_VERSION); return 0; } diff --git a/man/btrfs.8.in b/man/btrfs.8.in index 8a0a05d..ff2427e 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--mounted\fP|\fI--all-devices\fP|\fI<uuid>\fP]\fP .PP \fBbtrfs\fP \fBfilesystem sync\fP\fI <path> \fP .PP @@ -51,7 +51,7 @@ btrfs \- control a btrfs filesystem .PP \fBbtrfs\fP \fBdevice delete\fP \fI<device>\fP [\fI<device>...\fP] \fI<path>\fP .PP -\fBbtrfs\fP \fBdevice scan\fP [--all-devices|\fI<device> \fP[\fI<device>...\fP] +\fBbtrfs\fP \fBdevice scan\fP [\fI--all-devices\fP|\fI<device> \P[\fI<device>...\fP] .PP \fBbtrfs\fP \fBdevice ready\fP\fI <device>\fP .PP @@ -261,9 +261,11 @@ 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 -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. +\fBfilesystem show\fR [\fI--mounted\fP|\fI--all-devices\fP|\fI<uuid>\fR]\fR +Show the btrfs filesystem with some additional info. If no option or \fIUUID\fP +is passed, \fBbtrfs\fR shows information of all the btrfs filesystem both mounted +and unmounted. +If \fB--mounted\fP is passed, it would probe btrfs kernel to list mounted btrfs filesystem(s); 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. .TP @@ -410,8 +412,8 @@ Remove device(s) from a filesystem identified by \fI<path>\fR. \fBdevice scan\fR [--all-devices|\fI<device> \fP[\fI<device>...\fP]\fR If one or more devices are passed, these are scanned for a btrfs filesystem. -If no devices are passed, \fBbtrfs\fR scans all the block devices listed -in the /proc/partitions file. +If no devices are passed, \fBbtrfs\fR uses block devices containing btrfs +filesystem as listed by blkid. Finally, if \fB--all-devices\fP is passed, all the devices under /dev are scanned. .TP diff --git a/utils.h b/utils.h index a6d3e5e..0f31db7 100644 --- a/utils.h +++ b/utils.h @@ -25,10 +25,17 @@ #define BTRFS_MKFS_SYSTEM_GROUP_SIZE (4 * 1024 * 1024) -#define BTRFS_SCAN_PROC 1 -#define BTRFS_SCAN_DEV 2 +#define BTRFS_SCAN_PROC (1ULL << 0) +#define BTRFS_SCAN_DEV (1ULL << 1) +#define BTRFS_SCAN_MOUNTED (1ULL << 2) #define BTRFS_SCAN_LBLKID (1ULL << 3) +#define BTRFS_UPDATE_KERNEL 1 + +#define BTRFS_ARG_UNKNOWN 0 +#define BTRFS_ARG_PATH 1 +#define BTRFS_ARG_UUID 2 + int make_btrfs(int fd, const char *device, const char *label, u64 blocks[6], u64 num_bytes, u32 nodesize, u32 leafsize, u32 sectorsize, u32 stripesize, u64 features);
As of now btrfs filesystem show reads directly from disks. So sometimes output can be stale, mainly when user wants to cross verify their operation like, label or device delete or add... etc. so this patch will read from the kernel ioctl if it finds that disk is mounted. v4: rebase on integration-20131001-3 v3: rebase on 20130920 and splits new parameter to new patch v2: accepts David suggested Signed-off-by: Anand Jain <anand.jain@oracle.com> --- cmds-filesystem.c | 200 +++++++++++++++++++++++++++++++++++++++++++++++++---- man/btrfs.8.in | 16 +++-- utils.h | 11 +++- 3 files changed, 205 insertions(+), 22 deletions(-)