Add to GEOM RAID class module for reading non-degraded RAID5 volumes and

some environment to differentiate 4 possible RAID5 on-disk layouts.

Tested with Intel and AMD RAID BIOSes.

MFC after:	2 weeks

11 files changed, 473 insertions, 32 deletions

diff --git a/sbin/geom/class/raid/graid.8 b/sbin/geom/class/raid/graid.8
index 4ed3296..e9fd5a4 100644
--- a/sbin/geom/class/raid/graid.8
+++ b/sbin/geom/class/raid/graid.8
@@ -24,7 +24,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd October 26, 2011
+.Dd April 19, 2012
 .Dt GRAID 8
 .Os
 .Sh NAME
@@ -242,7 +242,8 @@ own risk: RAID1 (3+ disks), RAID10 (6+ disks).
 The GEOM RAID class follows a modular design, allowing different RAID levels
 to be used.
 Support for the following RAID levels is currently implemented: RAID0, RAID1,
-RAID1E, RAID10, SINGLE, CONCAT.
+RAID1E, RAID5, RAID10, SINGLE, CONCAT.
+RAID5 support is read-only and only for volumes in optimal state.
 .Sh RAID LEVEL MIGRATION
 The GEOM RAID class has no support for RAID level migration, allowed by some
 metadata formats.
diff --git a/sys/conf/files b/sys/conf/files
index 4aaacfa..ecabaa8 100644
--- a/sys/conf/files
+++ b/sys/conf/files
@@ -2405,6 +2405,7 @@ geom/raid/tr_concat.c		optional geom_raid
 geom/raid/tr_raid0.c		optional geom_raid
 geom/raid/tr_raid1.c		optional geom_raid
 geom/raid/tr_raid1e.c		optional geom_raid
+geom/raid/tr_raid5.c		optional geom_raid
 geom/raid3/g_raid3.c		optional geom_raid3
 geom/raid3/g_raid3_ctl.c	optional geom_raid3
 geom/shsec/g_shsec.c		optional geom_shsec
diff --git a/sys/geom/raid/g_raid.c b/sys/geom/raid/g_raid.c
index 3c4a3f8..4a509e8 100644
--- a/sys/geom/raid/g_raid.c
+++ b/sys/geom/raid/g_raid.c
@@ -281,6 +281,14 @@ g_raid_volume_level2str(int level, int qual)
 	case G_RAID_VOLUME_RL_RAID4:
 		return ("RAID4");
 	case G_RAID_VOLUME_RL_RAID5:
+		if (qual == G_RAID_VOLUME_RLQ_R5RA)
+			return ("RAID5RA");
+		if (qual == G_RAID_VOLUME_RLQ_R5RS)
+			return ("RAID5RS");
+		if (qual == G_RAID_VOLUME_RLQ_R5LA)
+			return ("RAID5LA");
+		if (qual == G_RAID_VOLUME_RLQ_R5LS)
+			return ("RAID5LS");
 		return ("RAID5");
 	case G_RAID_VOLUME_RL_RAID6:
 		return ("RAID6");
@@ -313,9 +321,20 @@ g_raid_volume_str2level(const char *str, int *level, int *qual)
 		*level = G_RAID_VOLUME_RL_RAID3;
 	else if (strcasecmp(str, "RAID4") == 0)
 		*level = G_RAID_VOLUME_RL_RAID4;
-	else if (strcasecmp(str, "RAID5") == 0)
+	else if (strcasecmp(str, "RAID5RA") == 0) {
 		*level = G_RAID_VOLUME_RL_RAID5;
-	else if (strcasecmp(str, "RAID6") == 0)
+		*qual = G_RAID_VOLUME_RLQ_R5RA;
+	} else if (strcasecmp(str, "RAID5RS") == 0) {
+		*level = G_RAID_VOLUME_RL_RAID5;
+		*qual = G_RAID_VOLUME_RLQ_R5RS;
+	} else if (strcasecmp(str, "RAID5") == 0 ||
+		   strcasecmp(str, "RAID5LA") == 0) {
+		*level = G_RAID_VOLUME_RL_RAID5;
+		*qual = G_RAID_VOLUME_RLQ_R5LA;
+	} else if (strcasecmp(str, "RAID5LS") == 0) {
+		*level = G_RAID_VOLUME_RL_RAID5;
+		*qual = G_RAID_VOLUME_RLQ_R5LS;
+	} else if (strcasecmp(str, "RAID6") == 0)
 		*level = G_RAID_VOLUME_RL_RAID6;
 	else if (strcasecmp(str, "RAID10") == 0 ||
 		 strcasecmp(str, "RAID1E") == 0)
diff --git a/sys/geom/raid/g_raid.h b/sys/geom/raid/g_raid.h
index 1c14ad6..6b67e8c 100644
--- a/sys/geom/raid/g_raid.h
+++ b/sys/geom/raid/g_raid.h
@@ -227,6 +227,10 @@ struct g_raid_subdisk {
 #define G_RAID_VOLUME_RL_UNKNOWN	0xff
 
 #define G_RAID_VOLUME_RLQ_NONE		0x00
+#define G_RAID_VOLUME_RLQ_R5RA		0x00
+#define G_RAID_VOLUME_RLQ_R5RS		0x01
+#define G_RAID_VOLUME_RLQ_R5LA		0x02
+#define G_RAID_VOLUME_RLQ_R5LS		0x03
 #define G_RAID_VOLUME_RLQ_UNKNOWN	0xff
 
 struct g_raid_volume;
diff --git a/sys/geom/raid/md_intel.c b/sys/geom/raid/md_intel.c
index 0935223..13f6dfd 100644
--- a/sys/geom/raid/md_intel.c
+++ b/sys/geom/raid/md_intel.c
@@ -682,11 +682,13 @@ g_raid_md_intel_supported(int level, int qual, int disks, int force)
 			return (0);
 		if (!force && disks > 6)
 			return (0);
+		if (qual != G_RAID_VOLUME_RLQ_R5LA)
+			return (0);
 		break;
 	default:
 		return (0);
 	}
-	if (qual != G_RAID_VOLUME_RLQ_NONE)
+	if (level != G_RAID_VOLUME_RL_RAID5 && qual != G_RAID_VOLUME_RLQ_NONE)
 		return (0);
 	return (1);
 }
@@ -1029,6 +1031,7 @@ g_raid_md_intel_start(struct g_raid_softc *sc)
 		mmap = intel_get_map(mvol, 0);
 		vol = g_raid_create_volume(sc, mvol->name, -1);
 		vol->v_md_data = (void *)(intptr_t)i;
+		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
 		if (mmap->type == INTEL_T_RAID0)
 			vol->v_raid_level = G_RAID_VOLUME_RL_RAID0;
 		else if (mmap->type == INTEL_T_RAID1 &&
@@ -1045,11 +1048,11 @@ g_raid_md_intel_start(struct g_raid_softc *sc)
 				vol->v_raid_level = G_RAID_VOLUME_RL_RAID1;
 			else
 				vol->v_raid_level = G_RAID_VOLUME_RL_RAID1E;
-		} else if (mmap->type == INTEL_T_RAID5)
+		} else if (mmap->type == INTEL_T_RAID5) {
 			vol->v_raid_level = G_RAID_VOLUME_RL_RAID5;
-		else
+			vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_R5LA;
+		} else
 			vol->v_raid_level = G_RAID_VOLUME_RL_UNKNOWN;
-		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
 		vol->v_strip_size = (u_int)mmap->strip_sectors * 512; //ZZZ
 		vol->v_disks_count = mmap->total_disks;
 		vol->v_mediasize = (off_t)mvol->total_sectors * 512; //ZZZ
@@ -1485,6 +1488,8 @@ g_raid_md_ctl_intel(struct g_raid_md_object *md,
 			gctl_error(req, "No RAID level.");
 			return (-3);
 		}
+		if (strcasecmp(levelname, "RAID5") == 0)
+			levelname = "RAID5LA";
 		if (g_raid_volume_str2level(levelname, &level, &qual)) {
 			gctl_error(req, "Unknown RAID level '%s'.", levelname);
 			return (-4);
@@ -1631,7 +1636,7 @@ g_raid_md_ctl_intel(struct g_raid_md_object *md,
 		vol = g_raid_create_volume(sc, volname, -1);
 		vol->v_md_data = (void *)(intptr_t)0;
 		vol->v_raid_level = level;
-		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
+		vol->v_raid_level_qualifier = qual;
 		vol->v_strip_size = strip;
 		vol->v_disks_count = numdisks;
 		if (level == G_RAID_VOLUME_RL_RAID0)
@@ -1658,8 +1663,12 @@ g_raid_md_ctl_intel(struct g_raid_md_object *md,
 			if (sd->sd_disk->d_consumer != NULL) {
 				g_raid_change_disk_state(disk,
 				    G_RAID_DISK_S_ACTIVE);
-				g_raid_change_subdisk_state(sd,
-				    G_RAID_SUBDISK_S_ACTIVE);
+				if (level == G_RAID_VOLUME_RL_RAID5)
+					g_raid_change_subdisk_state(sd,
+					    G_RAID_SUBDISK_S_UNINITIALIZED);
+				else
+					g_raid_change_subdisk_state(sd,
+					    G_RAID_SUBDISK_S_ACTIVE);
 				g_raid_event_send(sd, G_RAID_SUBDISK_E_NEW,
 				    G_RAID_EVENT_SUBDISK);
 			} else {
@@ -1694,6 +1703,8 @@ g_raid_md_ctl_intel(struct g_raid_md_object *md,
 			gctl_error(req, "No RAID level.");
 			return (-3);
 		}
+		if (strcasecmp(levelname, "RAID5") == 0)
+			levelname = "RAID5LA";
 		if (g_raid_volume_str2level(levelname, &level, &qual)) {
 			gctl_error(req, "Unknown RAID level '%s'.", levelname);
 			return (-4);
@@ -1818,7 +1829,7 @@ g_raid_md_ctl_intel(struct g_raid_md_object *md,
 		vol = g_raid_create_volume(sc, volname, -1);
 		vol->v_md_data = (void *)(intptr_t)i;
 		vol->v_raid_level = level;
-		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
+		vol->v_raid_level_qualifier = qual;
 		vol->v_strip_size = strip;
 		vol->v_disks_count = numdisks;
 		if (level == G_RAID_VOLUME_RL_RAID0)
@@ -1843,8 +1854,12 @@ g_raid_md_ctl_intel(struct g_raid_md_object *md,
 			sd->sd_size = size;
 			TAILQ_INSERT_TAIL(&disk->d_subdisks, sd, sd_next);
 			if (disk->d_state == G_RAID_DISK_S_ACTIVE) {
-				g_raid_change_subdisk_state(sd,
-				    G_RAID_SUBDISK_S_ACTIVE);
+				if (level == G_RAID_VOLUME_RL_RAID5)
+					g_raid_change_subdisk_state(sd,
+					    G_RAID_SUBDISK_S_UNINITIALIZED);
+				else
+					g_raid_change_subdisk_state(sd,
+					    G_RAID_SUBDISK_S_ACTIVE);
 				g_raid_event_send(sd, G_RAID_SUBDISK_E_NEW,
 				    G_RAID_EVENT_SUBDISK);
 			}
@@ -2245,6 +2260,9 @@ g_raid_md_write_intel(struct g_raid_md_object *md, struct g_raid_volume *tvol,
 			mmap0->status = INTEL_S_FAILURE;
 		else if (vol->v_state == G_RAID_VOLUME_S_DEGRADED)
 			mmap0->status = INTEL_S_DEGRADED;
+		else if (g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_UNINITIALIZED)
+		    == g_raid_nsubdisks(vol, -1))
+			mmap0->status = INTEL_S_UNINITIALIZED;
 		else
 			mmap0->status = INTEL_S_READY;
 		if (vol->v_raid_level == G_RAID_VOLUME_RL_RAID0)
@@ -2288,7 +2306,8 @@ g_raid_md_write_intel(struct g_raid_md_object *md, struct g_raid_volume *tvol,
 			    sd->sd_state == G_RAID_SUBDISK_S_RESYNC) {
 				mmap1->disk_idx[sdi] |= INTEL_DI_RBLD;
 			} else if (sd->sd_state != G_RAID_SUBDISK_S_ACTIVE &&
-			    sd->sd_state != G_RAID_SUBDISK_S_STALE) {
+			    sd->sd_state != G_RAID_SUBDISK_S_STALE &&
+			    sd->sd_state != G_RAID_SUBDISK_S_UNINITIALIZED) {
 				mmap0->disk_idx[sdi] |= INTEL_DI_RBLD;
 				if (mvol->migr_state)
 					mmap1->disk_idx[sdi] |= INTEL_DI_RBLD;
diff --git a/sys/geom/raid/md_jmicron.c b/sys/geom/raid/md_jmicron.c
index b07a378..937aff7 100644
--- a/sys/geom/raid/md_jmicron.c
+++ b/sys/geom/raid/md_jmicron.c
@@ -400,13 +400,15 @@ g_raid_md_jmicron_supported(int level, int qual, int disks, int force)
 	case G_RAID_VOLUME_RL_RAID5:
 		if (disks < 3)
 			return (0);
+		if (qual != G_RAID_VOLUME_RLQ_R5LA)
+			return (0);
 		if (!force)
 			return (0);
 		break;
 	default:
 		return (0);
 	}
-	if (qual != G_RAID_VOLUME_RLQ_NONE)
+	if (level != G_RAID_VOLUME_RL_RAID5 && qual != G_RAID_VOLUME_RLQ_NONE)
 		return (0);
 	return (1);
 }
@@ -657,6 +659,7 @@ g_raid_md_jmicron_start(struct g_raid_softc *sc)
 	vol = g_raid_create_volume(sc, buf, -1);
 	size = ((off_t)meta->disk_sectors_high << 16) + meta->disk_sectors_low;
 	size *= 512; //ZZZ
+	vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
 	if (meta->type == JMICRON_T_RAID0) {
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID0;
 		vol->v_mediasize = size * mdi->mdio_total_disks;
@@ -674,12 +677,12 @@ g_raid_md_jmicron_start(struct g_raid_softc *sc)
 		vol->v_mediasize = 0;
 	} else if (meta->type == JMICRON_T_RAID5) {
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID5;
+		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_R5LA;
 		vol->v_mediasize = size * (mdi->mdio_total_disks - 1);
 	} else {
 		vol->v_raid_level = G_RAID_VOLUME_RL_UNKNOWN;
 		vol->v_mediasize = 0;
 	}
-	vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
 	vol->v_strip_size = 1024 << meta->stripe_shift; //ZZZ
 	vol->v_disks_count = mdi->mdio_total_disks;
 	vol->v_sectorsize = 512; //ZZZ
@@ -1057,6 +1060,8 @@ g_raid_md_ctl_jmicron(struct g_raid_md_object *md,
 			gctl_error(req, "No RAID level.");
 			return (-3);
 		}
+		if (strcasecmp(levelname, "RAID5") == 0)
+			levelname = "RAID5LA";
 		if (g_raid_volume_str2level(levelname, &level, &qual)) {
 			gctl_error(req, "Unknown RAID level '%s'.", levelname);
 			return (-4);
@@ -1192,7 +1197,7 @@ g_raid_md_ctl_jmicron(struct g_raid_md_object *md,
 		vol = g_raid_create_volume(sc, volname, -1);
 		vol->v_md_data = (void *)(intptr_t)0;
 		vol->v_raid_level = level;
-		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
+		vol->v_raid_level_qualifier = qual;
 		vol->v_strip_size = strip;
 		vol->v_disks_count = numdisks;
 		if (level == G_RAID_VOLUME_RL_RAID0 ||
diff --git a/sys/geom/raid/md_nvidia.c b/sys/geom/raid/md_nvidia.c
index 0cbab4b..cd79d00 100644
--- a/sys/geom/raid/md_nvidia.c
+++ b/sys/geom/raid/md_nvidia.c
@@ -407,11 +407,14 @@ g_raid_md_nvidia_supported(int level, int qual, int disks, int force)
 	case G_RAID_VOLUME_RL_RAID5:
 		if (disks < 3)
 			return (0);
+		if (qual != G_RAID_VOLUME_RLQ_R5LA &&
+		    qual != G_RAID_VOLUME_RLQ_R5LS)
+			return (0);
 		break;
 	default:
 		return (0);
 	}
-	if (qual != G_RAID_VOLUME_RLQ_NONE)
+	if (level != G_RAID_VOLUME_RL_RAID5 && qual != G_RAID_VOLUME_RLQ_NONE)
 		return (0);
 	return (1);
 }
@@ -679,10 +682,11 @@ g_raid_md_nvidia_start(struct g_raid_softc *sc)
 		size = 0;
 	} else if (meta->type == NVIDIA_T_RAID5) {
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID5;
+		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_R5LA;
 		size = vol->v_mediasize / (mdi->mdio_total_disks - 1);
 	} else if (meta->type == NVIDIA_T_RAID5_SYM) {
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID5;
-//		vol->v_raid_level_qualifier = 0x03;
+		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_R5LS;
 		size = vol->v_mediasize / (mdi->mdio_total_disks - 1);
 	} else {
 		vol->v_raid_level = G_RAID_VOLUME_RL_UNKNOWN;
@@ -1059,6 +1063,8 @@ g_raid_md_ctl_nvidia(struct g_raid_md_object *md,
 			gctl_error(req, "No RAID level.");
 			return (-3);
 		}
+		if (strcasecmp(levelname, "RAID5") == 0)
+			levelname = "RAID5LS";
 		if (g_raid_volume_str2level(levelname, &level, &qual)) {
 			gctl_error(req, "Unknown RAID level '%s'.", levelname);
 			return (-4);
@@ -1206,7 +1212,7 @@ g_raid_md_ctl_nvidia(struct g_raid_md_object *md,
 		vol = g_raid_create_volume(sc, volname, -1);
 		vol->v_md_data = (void *)(intptr_t)0;
 		vol->v_raid_level = level;
-		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
+		vol->v_raid_level_qualifier = qual;
 		vol->v_strip_size = strip;
 		vol->v_disks_count = numdisks;
 		vol->v_mediasize = volsize;
@@ -1454,8 +1460,8 @@ g_raid_md_write_nvidia(struct g_raid_md_object *md, struct g_raid_volume *tvol,
 	else if (vol->v_raid_level == G_RAID_VOLUME_RL_CONCAT ||
 	    vol->v_raid_level == G_RAID_VOLUME_RL_SINGLE)
 		meta->type = NVIDIA_T_CONCAT;
-//	else if (vol->v_raid_level_qualifier == 0)
-//		meta->type = NVIDIA_T_RAID5;
+	else if (vol->v_raid_level_qualifier == G_RAID_VOLUME_RLQ_R5LA)
+		meta->type = NVIDIA_T_RAID5;
 	else
 		meta->type = NVIDIA_T_RAID5_SYM;
 	meta->strip_sectors = vol->v_strip_size / vol->v_sectorsize;
diff --git a/sys/geom/raid/md_promise.c b/sys/geom/raid/md_promise.c
index 55b4b21..df31ce6 100644
--- a/sys/geom/raid/md_promise.c
+++ b/sys/geom/raid/md_promise.c
@@ -595,11 +595,13 @@ g_raid_md_promise_supported(int level, int qual, int disks, int force)
 	case G_RAID_VOLUME_RL_RAID5:
 		if (disks < 3)
 			return (0);
+		if (qual != G_RAID_VOLUME_RLQ_R5LA)
+			return (0);
 		break;
 	default:
 		return (0);
 	}
-	if (qual != G_RAID_VOLUME_RLQ_NONE)
+	if (level != G_RAID_VOLUME_RL_RAID5 && qual != G_RAID_VOLUME_RLQ_NONE)
 		return (0);
 	return (1);
 }
@@ -848,6 +850,7 @@ g_raid_md_promise_start(struct g_raid_volume *vol)
 	pv = vol->v_md_data;
 	meta = pv->pv_meta;
 
+	vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
 	if (meta->type == PROMISE_T_RAID0)
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID0;
 	else if (meta->type == PROMISE_T_RAID1) {
@@ -857,15 +860,15 @@ g_raid_md_promise_start(struct g_raid_volume *vol)
 			vol->v_raid_level = G_RAID_VOLUME_RL_RAID1E;
 	} else if (meta->type == PROMISE_T_RAID3)
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID3;
-	else if (meta->type == PROMISE_T_RAID5)
+	else if (meta->type == PROMISE_T_RAID5) {
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID5;
-	else if (meta->type == PROMISE_T_SPAN)
+		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_R5LA;
+	} else if (meta->type == PROMISE_T_SPAN)
 		vol->v_raid_level = G_RAID_VOLUME_RL_CONCAT;
 	else if (meta->type == PROMISE_T_JBOD)
 		vol->v_raid_level = G_RAID_VOLUME_RL_SINGLE;
 	else
 		vol->v_raid_level = G_RAID_VOLUME_RL_UNKNOWN;
-	vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
 	vol->v_strip_size = 512 << meta->stripe_shift; //ZZZ
 	vol->v_disks_count = meta->total_disks;
 	vol->v_mediasize = (off_t)meta->total_sectors * 512; //ZZZ
@@ -1241,6 +1244,8 @@ g_raid_md_ctl_promise(struct g_raid_md_object *md,
 			gctl_error(req, "No RAID level.");
 			return (-3);
 		}
+		if (strcasecmp(levelname, "RAID5") == 0)
+			levelname = "RAID5LA";
 		if (g_raid_volume_str2level(levelname, &level, &qual)) {
 			gctl_error(req, "Unknown RAID level '%s'.", levelname);
 			return (-4);
@@ -1415,7 +1420,7 @@ g_raid_md_ctl_promise(struct g_raid_md_object *md,
 		vol = g_raid_create_volume(sc, volname, -1);
 		vol->v_md_data = pv;
 		vol->v_raid_level = level;
-		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
+		vol->v_raid_level_qualifier = qual;
 		vol->v_strip_size = strip;
 		vol->v_disks_count = numdisks;
 		if (level == G_RAID_VOLUME_RL_RAID0 ||
diff --git a/sys/geom/raid/md_sii.c b/sys/geom/raid/md_sii.c
index 92f2dec..840ae24 100644
--- a/sys/geom/raid/md_sii.c
+++ b/sys/geom/raid/md_sii.c
@@ -456,11 +456,13 @@ g_raid_md_sii_supported(int level, int qual, int disks, int force)
 	case G_RAID_VOLUME_RL_RAID5:
 		if (disks < 3)
 			return (0);
+		if (qual != G_RAID_VOLUME_RLQ_R5LS)
+			return (0);
 		break;
 	default:
 		return (0);
 	}
-	if (qual != G_RAID_VOLUME_RLQ_NONE)
+	if (level != G_RAID_VOLUME_RL_RAID5 && qual != G_RAID_VOLUME_RLQ_NONE)
 		return (0);
 	return (1);
 }
@@ -719,6 +721,7 @@ g_raid_md_sii_start(struct g_raid_softc *sc)
 	sii_meta_get_name(meta, buf);
 	vol = g_raid_create_volume(sc, buf, -1);
 	vol->v_mediasize = (off_t)meta->total_sectors * 512;
+	vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
 	if (meta->type == SII_T_RAID0) {
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID0;
 		size = vol->v_mediasize / mdi->mdio_total_disks;
@@ -736,6 +739,7 @@ g_raid_md_sii_start(struct g_raid_softc *sc)
 		size = 0;
 	} else if (meta->type == SII_T_RAID5) {
 		vol->v_raid_level = G_RAID_VOLUME_RL_RAID5;
+		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_R5LS;
 		size = vol->v_mediasize / (mdi->mdio_total_disks - 1);
 	} else if (meta->type == SII_T_JBOD) {
 		vol->v_raid_level = G_RAID_VOLUME_RL_SINGLE;
@@ -744,7 +748,6 @@ g_raid_md_sii_start(struct g_raid_softc *sc)
 		vol->v_raid_level = G_RAID_VOLUME_RL_UNKNOWN;
 		size = 0;
 	}
-	vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
 	vol->v_strip_size = meta->strip_sectors * 512; //ZZZ
 	vol->v_disks_count = mdi->mdio_total_disks;
 	vol->v_sectorsize = 512; //ZZZ
@@ -1144,6 +1147,8 @@ g_raid_md_ctl_sii(struct g_raid_md_object *md,
 			gctl_error(req, "No RAID level.");
 			return (-3);
 		}
+		if (strcasecmp(levelname, "RAID5") == 0)
+			levelname = "RAID5LS";
 		if (g_raid_volume_str2level(levelname, &level, &qual)) {
 			gctl_error(req, "Unknown RAID level '%s'.", levelname);
 			return (-4);
@@ -1278,7 +1283,7 @@ g_raid_md_ctl_sii(struct g_raid_md_object *md,
 		vol = g_raid_create_volume(sc, volname, -1);
 		vol->v_md_data = (void *)(intptr_t)0;
 		vol->v_raid_level = level;
-		vol->v_raid_level_qualifier = G_RAID_VOLUME_RLQ_NONE;
+		vol->v_raid_level_qualifier = qual;
 		vol->v_strip_size = strip;
 		vol->v_disks_count = numdisks;
 		if (level == G_RAID_VOLUME_RL_RAID0 ||
diff --git a/sys/geom/raid/tr_raid5.c b/sys/geom/raid/tr_raid5.c
new file mode 100644
index 0000000..eb628fd
--- /dev/null
+++ b/sys/geom/raid/tr_raid5.c
@@ -0,0 +1,376 @@
+/*-
+ * Copyright (c) 2012 Alexander Motin <mav@FreeBSD.org>
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
+
+#include <sys/cdefs.h>
+__FBSDID("$FreeBSD$");
+
+#include <sys/param.h>
+#include <sys/bio.h>
+#include <sys/endian.h>
+#include <sys/kernel.h>
+#include <sys/kobj.h>
+#include <sys/limits.h>
+#include <sys/lock.h>
+#include <sys/malloc.h>
+#include <sys/mutex.h>
+#include <sys/sysctl.h>
+#include <sys/systm.h>
+#include <geom/geom.h>
+#include "geom/raid/g_raid.h"
+#include "g_raid_tr_if.h"
+
+SYSCTL_DECL(_kern_geom_raid);
+
+static MALLOC_DEFINE(M_TR_RAID5, "tr_raid5_data", "GEOM_RAID RAID5 data");
+
+#define TR_RAID5_NONE 0
+#define TR_RAID5_REBUILD 1
+#define TR_RAID5_RESYNC 2
+
+#define TR_RAID5_F_DOING_SOME	0x1
+#define TR_RAID5_F_LOCKED	0x2
+#define TR_RAID5_F_ABORT	0x4
+
+struct g_raid_tr_raid5_object {
+	struct g_raid_tr_object	 trso_base;
+	int			 trso_starting;
+	int			 trso_stopping;
+	int			 trso_type;
+	int			 trso_recover_slabs; /* slabs before rest */
+	int			 trso_fair_io;
+	int			 trso_meta_update;
+	int			 trso_flags;
+	struct g_raid_subdisk	*trso_failed_sd; /* like per volume */
+	void			*trso_buffer;	 /* Buffer space */
+	struct bio		 trso_bio;
+};
+
+static g_raid_tr_taste_t g_raid_tr_taste_raid5;
+static g_raid_tr_event_t g_raid_tr_event_raid5;
+static g_raid_tr_start_t g_raid_tr_start_raid5;
+static g_raid_tr_stop_t g_raid_tr_stop_raid5;
+static g_raid_tr_iostart_t g_raid_tr_iostart_raid5;
+static g_raid_tr_iodone_t g_raid_tr_iodone_raid5;
+static g_raid_tr_kerneldump_t g_raid_tr_kerneldump_raid5;
+static g_raid_tr_locked_t g_raid_tr_locked_raid5;
+static g_raid_tr_free_t g_raid_tr_free_raid5;
+
+static kobj_method_t g_raid_tr_raid5_methods[] = {
+	KOBJMETHOD(g_raid_tr_taste,	g_raid_tr_taste_raid5),
+	KOBJMETHOD(g_raid_tr_event,	g_raid_tr_event_raid5),
+	KOBJMETHOD(g_raid_tr_start,	g_raid_tr_start_raid5),
+	KOBJMETHOD(g_raid_tr_stop,	g_raid_tr_stop_raid5),
+	KOBJMETHOD(g_raid_tr_iostart,	g_raid_tr_iostart_raid5),
+	KOBJMETHOD(g_raid_tr_iodone,	g_raid_tr_iodone_raid5),
+	KOBJMETHOD(g_raid_tr_kerneldump, g_raid_tr_kerneldump_raid5),
+	KOBJMETHOD(g_raid_tr_locked,	g_raid_tr_locked_raid5),
+	KOBJMETHOD(g_raid_tr_free,	g_raid_tr_free_raid5),
+	{ 0, 0 }
+};
+
+static struct g_raid_tr_class g_raid_tr_raid5_class = {
+	"RAID5",
+	g_raid_tr_raid5_methods,
+	sizeof(struct g_raid_tr_raid5_object),
+	.trc_priority = 100
+};
+
+static int
+g_raid_tr_taste_raid5(struct g_raid_tr_object *tr, struct g_raid_volume *vol)
+{
+	struct g_raid_tr_raid5_object *trs;
+	u_int qual;
+
+	trs = (struct g_raid_tr_raid5_object *)tr;
+	qual = tr->tro_volume->v_raid_level_qualifier;
+	if (tr->tro_volume->v_raid_level == G_RAID_VOLUME_RL_RAID5 &&
+	    qual >= 0 && qual <= 3) {
+		/* RAID5 */
+	} else
+		return (G_RAID_TR_TASTE_FAIL);
+	trs->trso_starting = 1;
+	return (G_RAID_TR_TASTE_SUCCEED);
+}
+
+static int
+g_raid_tr_update_state_raid5(struct g_raid_volume *vol,
+    struct g_raid_subdisk *sd)
+{
+	struct g_raid_tr_raid5_object *trs;
+	struct g_raid_softc *sc;
+	u_int s;
+	int na, ns, nu;
+
+	sc = vol->v_softc;
+	trs = (struct g_raid_tr_raid5_object *)vol->v_tr;
+	if (trs->trso_stopping &&
+	    (trs->trso_flags & TR_RAID5_F_DOING_SOME) == 0)
+		s = G_RAID_VOLUME_S_STOPPED;
+	else if (trs->trso_starting)
+		s = G_RAID_VOLUME_S_STARTING;
+	else {
+		na = g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_ACTIVE);
+		ns = g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_STALE) +
+		     g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_RESYNC);
+		nu = g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_UNINITIALIZED);
+		if (na == vol->v_disks_count)
+			s = G_RAID_VOLUME_S_OPTIMAL;
+		else if (na + ns == vol->v_disks_count ||
+		    na + ns + nu == vol->v_disks_count /* XXX: Temporary. */)
+			s = G_RAID_VOLUME_S_SUBOPTIMAL;
+		else if (na == vol->v_disks_count - 1 ||
+		    na + ns + nu == vol->v_disks_count)
+			s = G_RAID_VOLUME_S_DEGRADED;
+		else
+			s = G_RAID_VOLUME_S_BROKEN;
+	}
+	if (s != vol->v_state) {
+		g_raid_event_send(vol, G_RAID_VOLUME_S_ALIVE(s) ?
+		    G_RAID_VOLUME_E_UP : G_RAID_VOLUME_E_DOWN,
+		    G_RAID_EVENT_VOLUME);
+		g_raid_change_volume_state(vol, s);
+		if (!trs->trso_starting && !trs->trso_stopping)
+			g_raid_write_metadata(sc, vol, NULL, NULL);
+	}
+	return (0);
+}
+
+static int
+g_raid_tr_event_raid5(struct g_raid_tr_object *tr,
+    struct g_raid_subdisk *sd, u_int event)
+{
+
+	g_raid_tr_update_state_raid5(tr->tro_volume, sd);
+	return (0);
+}
+
+static int
+g_raid_tr_start_raid5(struct g_raid_tr_object *tr)
+{
+	struct g_raid_tr_raid5_object *trs;
+	struct g_raid_volume *vol;
+
+	trs = (struct g_raid_tr_raid5_object *)tr;
+	vol = tr->tro_volume;
+	trs->trso_starting = 0;
+	g_raid_tr_update_state_raid5(vol, NULL);
+	return (0);
+}
+
+static int
+g_raid_tr_stop_raid5(struct g_raid_tr_object *tr)
+{
+	struct g_raid_tr_raid5_object *trs;
+	struct g_raid_volume *vol;
+
+	trs = (struct g_raid_tr_raid5_object *)tr;
+	vol = tr->tro_volume;
+	trs->trso_starting = 0;
+	trs->trso_stopping = 1;
+	g_raid_tr_update_state_raid5(vol, NULL);
+	return (0);
+}
+
+static void
+g_raid_tr_iostart_raid5_read(struct g_raid_tr_object *tr, struct bio *bp)
+{
+	struct g_raid_volume *vol;
+	struct g_raid_subdisk *sd;
+	struct bio_queue_head queue;
+	struct bio *cbp;
+	char *addr;
+	off_t offset, start, length, nstripe, remain;
+	int no, pno;
+	u_int strip_size, qual;
+
+	vol = tr->tro_volume;
+	addr = bp->bio_data;
+	strip_size = vol->v_strip_size;
+	qual = tr->tro_volume->v_raid_level_qualifier;
+
+	/* Stripe number. */
+	nstripe = bp->bio_offset / strip_size;
+	/* Start position in stripe. */
+	start = bp->bio_offset % strip_size;
+	/* Parity disk number. */
+	pno = nstripe / (vol->v_disks_count - 1) % vol->v_disks_count;
+	if (qual >= 2)
+		pno = (vol->v_disks_count - 1) - pno;
+	/* Disk number. */
+	no = nstripe % (vol->v_disks_count - 1);
+	if (qual & 1) {
+		no = (pno + no + 1) % vol->v_disks_count;
+	} else if (no >= pno)
+		no++;
+	/* Stripe start position in disk. */
+	offset = (nstripe / (vol->v_disks_count - 1)) * strip_size;
+	/* Length of data to operate. */
+	remain = bp->bio_length;
+
+	bioq_init(&queue);
+	do {
+		length = MIN(strip_size - start, remain);
+		cbp = g_clone_bio(bp);
+		if (cbp == NULL)
+			goto failure;
+		cbp->bio_offset = offset + start;
+		cbp->bio_data = addr;
+		cbp->bio_length = length;
+		cbp->bio_caller1 = &vol->v_subdisks[no];
+		bioq_insert_tail(&queue, cbp);
+		no++;
+		if (qual & 1) {
+			no %= vol->v_disks_count;
+			if (no == pno) {
+				if (qual < 2) {
+					pno = (pno + 1) % vol->v_disks_count;
+					no = (no + 2) % vol->v_disks_count;
+				} else if (pno == 0)
+					pno = vol->v_disks_count - 1;
+				else
+					pno--;
+				offset += strip_size;
+			}
+		} else {
+			if (no == pno)
+				no++;
+			if (no >= vol->v_disks_count) {
+				no %= vol->v_disks_count;
+				if (qual < 2)
+					pno = (pno + 1) % vol->v_disks_count;
+				else if (pno == 0)
+					pno = vol->v_disks_count - 1;
+				else
+					pno--;
+				offset += strip_size;
+			}
+			if (no == pno)
+				no++;
+		}
+		remain -= length;
+		addr += length;
+		start = 0;
+	} while (remain > 0);
+	for (cbp = bioq_first(&queue); cbp != NULL;
+	    cbp = bioq_first(&queue)) {
+		bioq_remove(&queue, cbp);
+		sd = cbp->bio_caller1;
+		cbp->bio_caller1 = NULL;
+		g_raid_subdisk_iostart(sd, cbp);
+	}
+	return;
+failure:
+	for (cbp = bioq_first(&queue); cbp != NULL;
+	    cbp = bioq_first(&queue)) {
+		bioq_remove(&queue, cbp);
+		g_destroy_bio(cbp);
+	}
+	if (bp->bio_error == 0)
+		bp->bio_error = ENOMEM;
+	g_raid_iodone(bp, bp->bio_error);
+}
+
+static void
+g_raid_tr_iostart_raid5(struct g_raid_tr_object *tr, struct bio *bp)
+{
+	struct g_raid_volume *vol;
+	struct g_raid_tr_raid5_object *trs;
+
+	vol = tr->tro_volume;
+	trs = (struct g_raid_tr_raid5_object *)tr;
+	if (vol->v_state < G_RAID_VOLUME_S_SUBOPTIMAL) {
+		g_raid_iodone(bp, EIO);
+		return;
+	}
+	switch (bp->bio_cmd) {
+	case BIO_READ:
+		g_raid_tr_iostart_raid5_read(tr, bp);
+		break;
+	case BIO_WRITE:
+	case BIO_DELETE:
+	case BIO_FLUSH:
+		g_raid_iodone(bp, ENODEV);
+		break;
+	default:
+		KASSERT(1 == 0, ("Invalid command here: %u (volume=%s)",
+		    bp->bio_cmd, vol->v_name));
+		break;
+	}
+}
+
+static void
+g_raid_tr_iodone_raid5(struct g_raid_tr_object *tr,
+    struct g_raid_subdisk *sd, struct bio *bp)
+{
+	struct bio *pbp;
+	int error;
+
+	pbp = bp->bio_parent;
+	pbp->bio_inbed++;
+	error = bp->bio_error;
+	g_destroy_bio(bp);
+	if (pbp->bio_children == pbp->bio_inbed) {
+		pbp->bio_completed = pbp->bio_length;
+		g_raid_iodone(pbp, error);
+	}
+}
+
+static int
+g_raid_tr_kerneldump_raid5(struct g_raid_tr_object *tr,
+    void *virtual, vm_offset_t physical, off_t offset, size_t length)
+{
+
+	return (ENODEV);
+}
+
+static int
+g_raid_tr_locked_raid5(struct g_raid_tr_object *tr, void *argp)
+{
+	struct bio *bp;
+	struct g_raid_subdisk *sd;
+
+	bp = (struct bio *)argp;
+	sd = (struct g_raid_subdisk *)bp->bio_caller1;
+	g_raid_subdisk_iostart(sd, bp);
+
+	return (0);
+}
+
+static int
+g_raid_tr_free_raid5(struct g_raid_tr_object *tr)
+{
+	struct g_raid_tr_raid5_object *trs;
+
+	trs = (struct g_raid_tr_raid5_object *)tr;
+
+	if (trs->trso_buffer != NULL) {
+		free(trs->trso_buffer, M_TR_RAID5);
+		trs->trso_buffer = NULL;
+	}
+	return (0);
+}
+
+G_RAID_TR_DECLARE(g_raid_tr_raid5);
diff --git a/sys/modules/geom/geom_raid/Makefile b/sys/modules/geom/geom_raid/Makefile
index 4487807..26c3532 100644
--- a/sys/modules/geom/geom_raid/Makefile
+++ b/sys/modules/geom/geom_raid/Makefile
@@ -11,7 +11,7 @@ SRCS+=	g_raid_tr_if.h g_raid_tr_if.c
 
 SRCS+=	md_intel.c md_jmicron.c md_nvidia.c md_promise.c md_sii.c
 
-SRCS+=	tr_concat.c tr_raid0.c tr_raid1.c tr_raid1e.c
+SRCS+=	tr_concat.c tr_raid0.c tr_raid1.c tr_raid1e.c tr_raid5.c
 
 MFILES=	kern/bus_if.m kern/device_if.m
 MFILES+= geom/raid/g_raid_md_if.m geom/raid/g_raid_tr_if.m