Merged revisions 6930-6933,6960-6961 via svnmerge from

svn://svn.code.sf.net/p/scst/svn/trunk

........
  r6930 | vlnb | 2016-08-02 19:49:31 -0700 (Tue, 02 Aug 2016) | 3 lines
  
  scst_vdisk: don't report unmap alignment, if it is equal to unmap granularity
........
  r6931 | bvassche | 2016-08-08 13:52:00 -0700 (Mon, 08 Aug 2016) | 1 line
  
  scst/README.dlm: Update startup sequence information
........
  r6932 | bvassche | 2016-08-09 13:18:03 -0700 (Tue, 09 Aug 2016) | 5 lines
  
  srpt/Testing.txt: Leave out --norandommap
  
  --verify and --norandommap are mutually exclusive. Hence leave out
  --norandommap.
........
  r6933 | bvassche | 2016-08-11 11:36:48 -0700 (Thu, 11 Aug 2016) | 1 line
  
  scst/README.dlm: Correct instructions for waiting for logout
........
  r6960 | vlnb | 2016-08-19 18:24:03 -0700 (Fri, 19 Aug 2016) | 3 lines
  
  scst: fix compilation on older kernels
........
  r6961 | vlnb | 2016-08-19 18:33:47 -0700 (Fri, 19 Aug 2016) | 5 lines
  
  scst README: add section about VMware and Ceph RBD space reclaim
........


git-svn-id: http://svn.code.sf.net/p/scst/svn/branches/3.2.x@6964 d57e44dd-8a1f-0410-8b47-8ef2f437770f

scst/README

@@ -1952,6 +1952,20 @@ context switch is natural for such potentially long operation as
 EXTENDED COPY.
 
 
+VMware and Ceph RBD space reclaim
+---------------------------------
+
+VMware with VMFS5 filesystem ignores UNMAP alignment, so if you use 4MB
+Ceph RBD objects and VMFS5, only some discards will reclaim RBD space
+due to 1MB discard not often hitting the tail of objects.
+
+Thus, to have efficient ESXi space reclamation with RBD and VMFS5, you are
+recommended to use 1 MB object size in Ceph.
+
+See https://sourceforge.net/p/scst/mailman/message/35287598 thread for
+details.
+
+
 Caching
 -------
 

scst/README.dlm

@@ -85,8 +85,9 @@ Startup and Shutdown
 --------------------
 
 The startup sequence is as follows:
-* Load and configure SCST with cluster_mode = 0 and with all target ports
-  disabled.
+* Load the DLM kernel module. If not loaded explicitly, "modprobe scst" will
+  load the DLM kernel module implicitly.
+* Load and configure SCST with all target ports disabled.
 * Enable cluster mode for all SCST devices that can be accessed through more
   than one cluster node:
     for x in /sys/kernel/scst_tgt/handlers/*/*/; do
@@ -103,10 +104,17 @@ The startup sequence is as follows:
 	crmadmin -S "$dc" 2>/dev/null |
 	sed 's/^Status of crmd@[^[:blank:]]*:[[:blank:]]\([^[:blank:]]*\).*/\1/'
     }
-    for ((i=0;i<300;i++)); do
-	[ "$(pacemaker_dc_status)" = "S_IDLE" ] && break
+    timeout=300
+    for ((i=0;i<timeout;i++)); do
+	if [ "$(pacemaker_dc_status)" = "S_IDLE" ]; then
+	    echo "Pacemaker reached idle state after $i s"
+	    break
+	fi
 	sleep 1
     done
+    if [ "$i" = "$timeout" ]; then
+	echo "Pacemaker did not reach the IDLE state in $i s"
+    fi
 * Enable SCST target ports.
 * If no DLM resource has been configured in Pacemaker, start dlm_controld.pcmk
   explicitly.
@@ -118,7 +126,8 @@ The proper shutdown order is as follows:
         echo 0 > $x &
     done
     wait
-    while ls -Ad /sys/kernel/scst_tgt/targets/*/*/sessions/* >/dev/null 2>&1; do
+    while ls -Ad /sys/kernel/scst_tgt/targets/*/*/sessions/* 2>&1 |
+          grep -vE '/sys/kernel/scst_tgt/targets/(copy_manager|scst_local)/'; do
         sleep 1
     done
 * Tell SCST to release the DLM lockspaces:

scst/README_in-tree

@@ -1805,6 +1805,20 @@ context switch is natural for such potentially long operation as
 EXTENDED COPY.
 
 
+VMware and Ceph RBD space reclaim
+---------------------------------
+
+VMware with VMFS5 filesystem ignores UNMAP alignment, so if you use 4MB
+Ceph RBD objects and VMFS5, only some discards will reclaim RBD space
+due to 1MB discard not often hitting the tail of objects.
+
+Thus, to have efficient ESXi space reclamation with RBD and VMFS5, you are
+recommended to use 1 MB object size in Ceph.
+
+See https://sourceforge.net/p/scst/mailman/message/35287598 thread for
+details.
+
+
 Caching
 -------
 

scst/include/backport.h

@@ -678,4 +678,19 @@ static inline int scsi_bidi_cmnd(struct scsi_cmnd *cmd)
 }
 #endif
 
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0)) && \
+    (LINUX_VERSION_CODE < KERNEL_VERSION(3, 16, 0))
+/**
+ * ktime_before - Compare if a ktime_t value is smaller than another one.
+ * @cmp1:	comparable1
+ * @cmp2:	comparable2
+ *
+ * Return: true if cmp1 happened before cmp2.
+ */
+static inline bool ktime_before(const ktime_t cmp1, const ktime_t cmp2)
+{
+	return ktime_compare(cmp1, cmp2) < 0;
+}
+#endif /* (LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0)) && (LINUX_VERSION_CODE < KERNEL_VERSION(3, 16, 0)) */
+
 #endif /* _SCST_BACKPORT_H_ */

scst/src/dev_handlers/scst_vdisk.c

@@ -1062,6 +1062,8 @@ check:
 	(defined(RHEL_MAJOR) && RHEL_MAJOR -0 >= 6)
 			virt_dev->unmap_opt_gran = q->limits.discard_granularity >> block_shift;
 			virt_dev->unmap_align = q->limits.discard_alignment >> block_shift;
+			if (virt_dev->unmap_opt_gran == virt_dev->unmap_align)
+				virt_dev->unmap_align = 0;
 			virt_dev->unmap_max_lba_cnt = q->limits.max_discard_sectors >> (block_shift - 9);
 			virt_dev->discard_zeroes_data = q->limits.discard_zeroes_data;
 #else

scst/src/scst_main.c

@@ -121,7 +121,9 @@ unsigned long scst_trace_flag;
 
 unsigned long scst_flags;
 
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0)
 unsigned long scst_poll_ns = SCST_DEF_POLL_NS;
+#endif
 
 int scst_max_tasklet_cmd = SCST_DEF_MAX_TASKLET_CMD;
 

scst/src/scst_priv.h

@@ -183,8 +183,10 @@ extern unsigned int scst_setup_id;
 #define SCST_DEF_MAX_TASKLET_CMD 10
 extern int scst_max_tasklet_cmd;
 
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0)
 #define SCST_DEF_POLL_NS 0
 extern unsigned long scst_poll_ns;
+#endif
 
 extern spinlock_t scst_init_lock;
 extern struct list_head scst_init_cmd_list;

scst/src/scst_sysfs.c

@@ -7077,6 +7077,8 @@ static struct kobj_attribute scst_max_tasklet_cmd_attr =
 	__ATTR(max_tasklet_cmd, S_IRUGO | S_IWUSR, scst_max_tasklet_cmd_show,
 	       scst_max_tasklet_cmd_store);
 
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0)
+
 static ssize_t scst_poll_us_show(struct kobject *kobj,
 				  struct kobj_attribute *attr, char *buf)
 {
@@ -7124,6 +7126,8 @@ static struct kobj_attribute scst_poll_us_attr =
 	__ATTR(poll_us, S_IRUGO | S_IWUSR, scst_poll_us_show,
 	       scst_poll_us_store);
 
+#endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0) */
+
 static ssize_t scst_suspend_show(struct kobject *kobj,
 				 struct kobj_attribute *attr, char *buf)
 {
@@ -7398,7 +7402,9 @@ static struct attribute *scst_sysfs_root_default_attrs[] = {
 	&scst_threads_attr.attr,
 	&scst_setup_id_attr.attr,
 	&scst_max_tasklet_cmd_attr.attr,
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0)
 	&scst_poll_us_attr.attr,
+#endif
 	&scst_suspend_attr.attr,
 #if defined(CONFIG_SCST_DEBUG) || defined(CONFIG_SCST_TRACING)
 	&scst_main_trace_level_attr.attr,

scst/src/scst_targ.c

@@ -5620,6 +5620,7 @@ again:
 			thr_locked = false;
 		}
 
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0)
 		if (scst_poll_ns > 0) {
 			struct timespec ts;
 			ktime_t end, kt;
@@ -5652,6 +5653,7 @@ again:
 		}
 
 go:
+#endif
 		spin_lock_irq(&p_cmd_threads->cmd_list_lock);
 		spin_lock(&thr->thr_cmd_list_lock);
 	}

srpt/Testing.txt

@@ -37,8 +37,8 @@ At least the following tests must be run before releasing a new SRPT version:
       rm -rf /mnt/test* && \
       fio --verify=md5 -rw=randwrite --size=10m --bs=4k \
         --loops=1000000 --iodepth=64 --group_reporting --sync=1 --direct=1 \
-        --norandommap --ioengine=aio --directory=/mnt --name=test --thread \
-        --numjobs=80 --runtime=30 && \
+        --ioengine=aio --directory=/mnt --name=test --thread --numjobs=80 \
+	--runtime=30 && \
       fsck -N $dev
     do
       true