Better names for struct disk elements: d_maxsize, d_stripeoffset

and d_stripesisze;

Introduce si_stripesize and si_stripeoffset in struct cdev so we
can make the visible to clustering code.

Add stripesize and stripeoffset to providers.

DTRT with stripesize and stripeoffset in various places in GEOM.

9 files changed, 50 insertions, 6 deletions

diff --git a/sys/geom/bde/g_bde.c b/sys/geom/bde/g_bde.c
index 1f6f9cf..13fb280 100644
--- a/sys/geom/bde/g_bde.c
+++ b/sys/geom/bde/g_bde.c
@@ -248,6 +248,8 @@ g_bde_config(struct g_configargs *ga)
 		g_topology_lock();
 		pp = g_new_providerf(gp, gp->name);
 		pp->flags |= G_PF_CANDELETE;
+		pp->slicesize = kp->zone_cont;
+		pp->sliceoffset = 0;
 		pp->mediasize = sc->mediasize;
 		pp->sectorsize = sc->sectorsize;
 		g_error_provider(pp, 0);
diff --git a/sys/geom/geom.h b/sys/geom/geom.h
index 7bd3ca7..9047f071 100644
--- a/sys/geom/geom.h
+++ b/sys/geom/geom.h
@@ -170,6 +170,8 @@ struct g_provider {
 	u_int			index;
 	off_t			mediasize;
 	u_int			sectorsize;
+	u_int			stripesize;
+	u_int			stripeoffset;
 	struct g_stat		*stat;
 	u_int			flags;
 #define G_PF_CANDELETE		0x1
diff --git a/sys/geom/geom_dev.c b/sys/geom/geom_dev.c
index c626f14..fe3ad37 100644
--- a/sys/geom/geom_dev.c
+++ b/sys/geom/geom_dev.c
@@ -176,6 +176,8 @@ g_dev_taste(struct g_class *mp, struct g_provider *pp, int insist __unused)
 	mtx_unlock(&Giant);
 	g_topology_lock();
 
+	dev->si_stripesize = pp->stripesize;
+	dev->si_stripeoffset = pp->stripeoffset;
 	gp->softc = dev;
 	dev->si_drv1 = gp;
 	dev->si_drv2 = cp;
diff --git a/sys/geom/geom_disk.c b/sys/geom/geom_disk.c
index d763bfe..43d9155 100644
--- a/sys/geom/geom_disk.c
+++ b/sys/geom/geom_disk.c
@@ -290,6 +290,8 @@ g_disk_create(void *arg)
 	pp->sectorsize = dp->d_sectorsize;
 	if (dp->d_flags & DISKFLAG_CANDELETE)
 		pp->flags |= G_PF_CANDELETE;
+	pp->stripeoffset = dp->d_stripeoffset;
+	pp->stripesize = dp->d_stripesize;
 	g_error_provider(pp, 0);
 	if (bootverbose)
 		printf("GEOM: new disk %s\n", gp->name);
diff --git a/sys/geom/geom_slice.c b/sys/geom/geom_slice.c
index e0632b3..11f1380 100644
--- a/sys/geom/geom_slice.c
+++ b/sys/geom/geom_slice.c
@@ -291,7 +291,7 @@ g_slice_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct
 int
 g_slice_config(struct g_geom *gp, u_int idx, int how, off_t offset, off_t length, u_int sectorsize, const char *fmt, ...)
 {
-	struct g_provider *pp;
+	struct g_provider *pp, *pp2;
 	struct g_slicer *gsp;
 	struct g_slice *gsl;
 	va_list ap;
@@ -346,8 +346,10 @@ g_slice_config(struct g_geom *gp, u_int idx, int how, off_t offset, off_t length
 	sbuf_vprintf(sb, fmt, ap);
 	sbuf_finish(sb);
 	pp = g_new_providerf(gp, sbuf_data(sb));
-	pp->flags =
-		LIST_FIRST(&gp->consumer)->provider->flags & G_PF_CANDELETE;
+	pp2 = LIST_FIRST(&gp->consumer)->provider;
+	pp->flags = pp2->flags & G_PF_CANDELETE;
+	pp->stripesize = pp2->stripesize;
+	pp->stripeoffset = (pp2->stripeoffset + offset) % pp->stripesize;
 	if (bootverbose)
 		printf("GEOM: Configure %s, start %jd length %jd end %jd\n",
 		    pp->name, (intmax_t)offset, (intmax_t)length,
diff --git a/sys/geom/notes b/sys/geom/notes
index eff24c5..9e674f7 100644
--- a/sys/geom/notes
+++ b/sys/geom/notes
@@ -86,3 +86,33 @@ allows userland to mmap(2) the pages containing the statistics data.
 In order to indicate to userland when the data in a statstics
 structure might be inconsistent, g_io_deliver() atomically sets a
 flag "updating" and resets it when the structure is again consistent.
+-----------------------------------------------------------------------
+maxsize, stripesize and stripeoffset
+
+maxsize is the biggest request we are willing to handle.  If not
+set there is no upper bound on the size of a request and the code
+is responsible for chopping it up.  Only hardware methods should
+set an upper bound in this field.  Geom_disk will inherit the upper
+bound set by the device driver.
+
+stripesize is the width of any natural request boundaries for the
+device.  This would be the width of a stripe on a raid-5 unit or
+one zone in GBDE.  The idea with this field is to hint to clustering
+type code to not trivially overrun these boundaries.
+
+stripeoffset is the amount of the first stripe which lies before the
+devices beginning.
+
+If we have a device with 64k stripes:
+	[0...64k[
+	[64k...128k[
+	[128k..192k[
+Then it will have stripesize = 64k and stripeoffset = 0.
+
+If we put a MBR on this device, where slice#1 starts on sector#63,
+then this slice will have: stripesize = 64k, stripeoffset = 63 * sectorsize.
+
+If the clustering code wants to widen a request which writes to
+sector#53 of the slice, it can calculate how many bytes till the end of
+the stripe as:
+	stripewith - (53 * sectorsize + stripeoffset) % stripewidth.
diff --git a/sys/sys/conf.h b/sys/sys/conf.h
index 20085b2..adb95df 100644
--- a/sys/sys/conf.h
+++ b/sys/sys/conf.h
@@ -78,6 +78,8 @@ struct cdev {
 	void		*si_drv1, *si_drv2;
 	struct cdevsw	*si_devsw;
 	int		si_iosize_max;	/* maximum I/O size (for physio &al) */
+	u_int		si_stripesize;
+	u_int		si_stripeoffset;
 	uid_t		si_uid;
 	gid_t		si_gid;
 	mode_t		si_mode;
diff --git a/sys/sys/disk.h b/sys/sys/disk.h
index e84c427..c9b1b85 100644
--- a/sys/sys/disk.h
+++ b/sys/sys/disk.h
@@ -58,9 +58,9 @@ struct disk {
 	off_t			d_mediasize;
 	u_int			d_fwsectors;
 	u_int			d_fwheads;
-	u_int			d_stripe_offset;
-	u_int			d_stripe_width;
-	u_int			d_max_request;
+	u_int			d_maxsize;
+	u_int			d_stripeoffset;
+	u_int			d_stripesize;
 
 	/* Fields private to the driver */
 	void			*d_drv1;
diff --git a/sys/sys/linedisc.h b/sys/sys/linedisc.h
index 20085b2..adb95df 100644
--- a/sys/sys/linedisc.h
+++ b/sys/sys/linedisc.h
@@ -78,6 +78,8 @@ struct cdev {
 	void		*si_drv1, *si_drv2;
 	struct cdevsw	*si_devsw;
 	int		si_iosize_max;	/* maximum I/O size (for physio &al) */
+	u_int		si_stripesize;
+	u_int		si_stripeoffset;
 	uid_t		si_uid;
 	gid_t		si_gid;
 	mode_t		si_mode;