| Message ID | 1342797098-6493-1-git-send-email-jbacik@fusionio.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
> +static int is_ssd(const char *file) > +{ > + char *dev = strrchr(file, '/'); [ ... ] > + snprintf(path, PATH_MAX, "/sys/block/%s/queue/rotational", dev); > + fd = open(path, O_RDONLY); Hmm, this doesn't seem right. The last path component can have nothing to do with the underlying device name. And /sys/block doesn't have entries for partition devices. After some poking around (and hints from Eric), it looks like the thing to do is stat the file to find that it's a block device and then use blkid_devno_to_wholedisk(). That parses the link from /sys/dev/block/$maj:$min to find the containing device. Then it'll work for partitions and nutty udev vanity symlinks. - 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
On Fri, Jul 20, 2012 at 12:18:59PM -0600, Zach Brown wrote: > > +static int is_ssd(const char *file) > > +{ > > + char *dev = strrchr(file, '/'); > > [ ... ] > > > + snprintf(path, PATH_MAX, "/sys/block/%s/queue/rotational", dev); > > + fd = open(path, O_RDONLY); > > Hmm, this doesn't seem right. The last path component can have nothing > to do with the underlying device name. And /sys/block doesn't have > entries for partition devices. > > After some poking around (and hints from Eric), it looks like the thing > to do is stat the file to find that it's a block device and then use > blkid_devno_to_wholedisk(). That parses the link from > /sys/dev/block/$maj:$min to find the containing device. > > Then it'll work for partitions and nutty udev vanity symlinks. > Partitions are for losers. 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 Fri, Jul 20, 2012 at 02:20:56PM -0400, Josef Bacik wrote: > > Partitions are for losers. *fist bump* - 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
On Fri, Jul 20, 2012 at 12:20:56PM -0600, Josef Bacik wrote: > On Fri, Jul 20, 2012 at 12:18:59PM -0600, Zach Brown wrote: > > > +static int is_ssd(const char *file) > > > +{ > > > + char *dev = strrchr(file, '/'); > > > > [ ... ] > > > > > + snprintf(path, PATH_MAX, "/sys/block/%s/queue/rotational", dev); > > > + fd = open(path, O_RDONLY); > > > > Hmm, this doesn't seem right. The last path component can have nothing > > to do with the underlying device name. And /sys/block doesn't have > > entries for partition devices. > > > > After some poking around (and hints from Eric), it looks like the thing > > to do is stat the file to find that it's a block device and then use > > blkid_devno_to_wholedisk(). That parses the link from > > /sys/dev/block/$maj:$min to find the containing device. > > > > Then it'll work for partitions and nutty udev vanity symlinks. > > > > Partitions are for losers. Winners make a separate /boot, even when using syslinux to boot directly off btrfs (syslinux is awesome everyone, switch now). -chris -- 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/mkfs.c b/mkfs.c index dff5eb8..858ba21 100644 --- a/mkfs.c +++ b/mkfs.c @@ -234,7 +234,7 @@ static int create_one_raid_group(struct btrfs_trans_handle *trans, static int create_raid_groups(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 data_profile, int data_profile_opt, u64 metadata_profile, - int metadata_profile_opt, int mixed) + int metadata_profile_opt, int mixed, int ssd) { u64 num_devices = btrfs_super_num_devices(&root->fs_info->super_copy); u64 allowed; @@ -246,7 +246,7 @@ static int create_raid_groups(struct btrfs_trans_handle *trans, */ if (!metadata_profile_opt && !mixed) { metadata_profile = (num_devices > 1) ? - BTRFS_BLOCK_GROUP_RAID1 : BTRFS_BLOCK_GROUP_DUP; + BTRFS_BLOCK_GROUP_RAID1 : (ssd) ? 0: BTRFS_BLOCK_GROUP_DUP; } if (!data_profile_opt && !mixed) { data_profile = (num_devices > 1) ? @@ -1201,6 +1201,36 @@ static int zero_output_file(int out_fd, u64 size, u32 sectorsize) return ret; } +static int is_ssd(const char *file) +{ + char *dev = strrchr(file, '/'); + char path[PATH_MAX]; + int fd; + char rotational; + + if (!dev) + return 0; + + /* Move ahead one to the actual device */ + dev++; + if (*dev == '\0') + return 0; + + snprintf(path, PATH_MAX, "/sys/block/%s/queue/rotational", dev); + fd = open(path, O_RDONLY); + if (fd < 0) { + return 0; + } + + if (read(fd, &rotational, sizeof(char)) < sizeof(char)) { + close(fd); + return 0; + } + close(fd); + + return !atoi((const char *)&rotational); +} + int main(int ac, char **av) { char *file; @@ -1227,6 +1257,7 @@ int main(int ac, char **av) int data_profile_opt = 0; int metadata_profile_opt = 0; int nodiscard = 0; + int ssd = 0; char *source_dir = NULL; int source_dir_set = 0; @@ -1352,6 +1383,9 @@ int main(int ac, char **av) exit(1); } } + + ssd = is_ssd(file); + if (mixed) { if (metadata_profile != data_profile) { fprintf(stderr, "With mixed block groups data and metadata " @@ -1438,7 +1472,7 @@ raid_groups: if (!source_dir_set) { ret = create_raid_groups(trans, root, data_profile, data_profile_opt, metadata_profile, - metadata_profile_opt, mixed); + metadata_profile_opt, mixed, ssd); BUG_ON(ret); }
SSD's do not gain anything by having metadata DUP turned on. The underlying file system that is a part of all SSD's could easily map duplicate metadat blocks into the same erase block which effectively eliminates the benefit of duplicating the metadata on disk. So detect if we are formatting a single SSD drive and if we are do not use DUP. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> --- mkfs.c | 40 +++++++++++++++++++++++++++++++++++++--- 1 files changed, 37 insertions(+), 3 deletions(-)