target/vhost-scsi: Add per-cpu ida tag pre-allocation for v3.12

index 7172d00..64ceb51 100644
@@ -232,6 +232,35 @@ struct se_session *transport_init_session(void)
 }
 EXPORT_SYMBOL(transport_init_session);
 
+struct se_session *transport_init_session_tags(unsigned int tag_num,
+					       unsigned int tag_size)
+{
+	struct se_session *se_sess;
+	int rc;
+
+	se_sess = transport_init_session();
+	if (IS_ERR(se_sess))
+		return se_sess;
+
+	se_sess->sess_cmd_map = kzalloc(tag_num * tag_size, GFP_KERNEL);
+	if (!se_sess->sess_cmd_map) {
+		pr_err("Unable to allocate se_sess->sess_cmd_map\n");
+		transport_free_session(se_sess);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	rc = percpu_ida_init(&se_sess->sess_tag_pool, tag_num);
+	if (rc < 0) {
+		pr_err("Unable to init se_sess->sess_tag_pool,"
+			" tag_num: %u\n", tag_num);
+		transport_free_session(se_sess);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	return se_sess;
+}
+EXPORT_SYMBOL(transport_init_session_tags);
+
 /*
  * Called with spin_lock_irqsave(&struct se_portal_group->session_lock called.
  */

@@ -367,6 +396,10 @@ EXPORT_SYMBOL(transport_deregister_session_configfs);
 
 void transport_free_session(struct se_session *se_sess)
 {
+	if (se_sess->sess_cmd_map) {
+		percpu_ida_destroy(&se_sess->sess_tag_pool);
+		kfree(se_sess->sess_cmd_map);
+	}
 	kmem_cache_free(se_sess_cache, se_sess);
 }
 EXPORT_SYMBOL(transport_free_session);

index d92ec67..a842884 100644
@@ -5,6 +5,7 @@
 #include <linux/configfs.h>
 #include <linux/dma-mapping.h>
 #include <linux/blkdev.h>
+#include <linux/idr.h>
 #include <scsi/scsi_cmnd.h>
 #include <net/sock.h>
 #include <net/tcp.h>

@@ -418,6 +419,8 @@ struct se_cmd {
 	enum dma_data_direction	data_direction;
 	/* For SAM Task Attribute */
 	int			sam_task_attr;
+	/* Used for se_sess->sess_tag_pool */
+	unsigned int		map_tag;
 	/* Transport protocol dependent state, see transport_state_table */
 	enum transport_state_table t_state;
 	unsigned		cmd_wait_set:1;

@@ -539,6 +542,8 @@ struct se_session {
 	struct list_head	sess_wait_list;
 	spinlock_t		sess_cmd_lock;
 	struct kref		sess_kref;
+	void			*sess_cmd_map;
+	struct percpu_ida	sess_tag_pool;
 };
 
 struct se_device;

index 1dcce9c..d34f2ce 100644
@@ -84,6 +84,7 @@ struct target_core_fabric_ops {
 };
 
 struct se_session *transport_init_session(void);
+struct se_session *transport_init_session_tags(unsigned int, unsigned int);
 void	__transport_register_session(struct se_portal_group *,
 		struct se_node_acl *, struct se_session *, void *);
 void	transport_register_session(struct se_portal_group *,

index 1e5e820..6224434 100644
@@ -48,6 +48,7 @@
 #include <linux/virtio_scsi.h>
 #include <linux/llist.h>
 #include <linux/bitmap.h>
+#include <linux/idr.h>
 
 #include "vhost.c"
 #include "vhost.h"

@@ -55,6 +56,7 @@
 #define TCM_VHOST_VERSION  "v0.1"
 #define TCM_VHOST_NAMELEN 256
 #define TCM_VHOST_MAX_CDB_SIZE 32
+#define TCM_VHOST_DEFAULT_TAGS 256
 
 struct vhost_scsi_inflight {
 	/* Wait for the flush operation to finish */

@@ -448,6 +450,7 @@ static void tcm_vhost_release_cmd(struct se_cmd *se_cmd)
 {
 	struct tcm_vhost_cmd *tv_cmd = container_of(se_cmd,
 				struct tcm_vhost_cmd, tvc_se_cmd);
+	struct se_session *se_sess = se_cmd->se_sess;
 
 	if (tv_cmd->tvc_sgl_count) {
 		u32 i;

@@ -458,7 +461,7 @@ static void tcm_vhost_release_cmd(struct se_cmd *se_cmd)
         }
 
 	tcm_vhost_put_inflight(tv_cmd->inflight);
-	kfree(tv_cmd);
+	percpu_ida_free(&se_sess->sess_tag_pool, se_cmd->map_tag);
 }
 
 static int tcm_vhost_shutdown_session(struct se_session *se_sess)

@@ -700,7 +703,7 @@ static void vhost_scsi_complete_cmd_work(struct vhost_work *work)
 		vhost_signal(&vs->dev, &vs->vqs[vq].vq);
 }
 
-static struct tcm_vhost_cmd *vhost_scsi_allocate_cmd(
+static struct tcm_vhost_cmd *vhost_scsi_get_tag(
 	struct vhost_virtqueue *vq,
 	struct tcm_vhost_tpg *tv_tpg,
 	struct virtio_scsi_cmd_req *v_req,

@@ -709,18 +712,21 @@ static struct tcm_vhost_cmd *vhost_scsi_allocate_cmd(
 {
 	struct tcm_vhost_cmd *tv_cmd;
 	struct tcm_vhost_nexus *tv_nexus;
+	struct se_session *se_sess;
+	int tag;
 
 	tv_nexus = tv_tpg->tpg_nexus;
 	if (!tv_nexus) {
 		pr_err("Unable to locate active struct tcm_vhost_nexus\n");
 		return ERR_PTR(-EIO);
 	}
+	se_sess = tv_nexus->tvn_se_sess;
 
-	tv_cmd = kzalloc(sizeof(struct tcm_vhost_cmd), GFP_ATOMIC);
-	if (!tv_cmd) {
-		pr_err("Unable to allocate struct tcm_vhost_cmd\n");
-		return ERR_PTR(-ENOMEM);
-	}
+	tag = percpu_ida_alloc(&se_sess->sess_tag_pool, GFP_KERNEL);
+	tv_cmd = &((struct tcm_vhost_cmd *)se_sess->sess_cmd_map)[tag];
+	memset(tv_cmd, 0, sizeof(struct tcm_vhost_cmd));
+
+	tv_cmd->tvc_se_cmd.map_tag = tag;
 	tv_cmd->tvc_tag = v_req->tag;
 	tv_cmd->tvc_task_attr = v_req->task_attr;
 	tv_cmd->tvc_exp_data_len = exp_data_len;

@@ -982,10 +988,10 @@ static void vhost_scsi_handle_vq(struct vhost_scsi *vs,
 		for (i = 0; i < data_num; i++)
 			exp_data_len += vq->iov[data_first + i].iov_len;
 
-		tv_cmd = vhost_scsi_allocate_cmd(vq, tv_tpg, &v_req,
-					exp_data_len, data_direction);
+		tv_cmd = vhost_scsi_get_tag(vq, tv_tpg, &v_req,
+					    exp_data_len, data_direction);
 		if (IS_ERR(tv_cmd)) {
-			vq_err(vq, "vhost_scsi_allocate_cmd failed %ld\n",
+			vq_err(vq, "vhost_scsi_get_tag failed %ld\n",
 					PTR_ERR(tv_cmd));
 			goto err_cmd;
 		}

@@ -1662,9 +1668,12 @@ static int tcm_vhost_make_nexus(struct tcm_vhost_tpg *tv_tpg,
 		return -ENOMEM;
 	}
 	/*
-	 *  Initialize the struct se_session pointer
+	 *  Initialize the struct se_session pointer and setup tagpool
+	 *  for struct tcm_vhost_cmd descriptors
 	 */
-	tv_nexus->tvn_se_sess = transport_init_session();
+	tv_nexus->tvn_se_sess = transport_init_session_tags(
+						TCM_VHOST_DEFAULT_TAGS,
+						sizeof(struct tcm_vhost_cmd));
 	if (IS_ERR(tv_nexus->tvn_se_sess)) {
 		mutex_unlock(&tv_tpg->tv_tpg_mutex);
 		kfree(tv_nexus);

index 1e5e820..6224434 100644
@@ -48,6 +48,7 @@
 #include <linux/virtio_scsi.h>
 #include <linux/llist.h>
 #include <linux/bitmap.h>
+#include <linux/idr.h>
 
 #include "vhost.c"
 #include "vhost.h"

@@ -55,6 +56,7 @@
 #define TCM_VHOST_VERSION  "v0.1"
 #define TCM_VHOST_NAMELEN 256
 #define TCM_VHOST_MAX_CDB_SIZE 32
+#define TCM_VHOST_DEFAULT_TAGS 256
 
 struct vhost_scsi_inflight {
 	/* Wait for the flush operation to finish */

@@ -448,6 +450,7 @@ static void tcm_vhost_release_cmd(struct se_cmd *se_cmd)
 {
 	struct tcm_vhost_cmd *tv_cmd = container_of(se_cmd,
 				struct tcm_vhost_cmd, tvc_se_cmd);
+	struct se_session *se_sess = se_cmd->se_sess;
 
 	if (tv_cmd->tvc_sgl_count) {
 		u32 i;

@@ -458,7 +461,7 @@ static void tcm_vhost_release_cmd(struct se_cmd *se_cmd)
         }
 
 	tcm_vhost_put_inflight(tv_cmd->inflight);
-	kfree(tv_cmd);
+	percpu_ida_free(&se_sess->sess_tag_pool, se_cmd->map_tag);
 }
 
 static int tcm_vhost_shutdown_session(struct se_session *se_sess)

@@ -700,7 +703,7 @@ static void vhost_scsi_complete_cmd_work(struct vhost_work *work)
 		vhost_signal(&vs->dev, &vs->vqs[vq].vq);
 }
 
-static struct tcm_vhost_cmd *vhost_scsi_allocate_cmd(
+static struct tcm_vhost_cmd *vhost_scsi_get_tag(
 	struct vhost_virtqueue *vq,
 	struct tcm_vhost_tpg *tv_tpg,
 	struct virtio_scsi_cmd_req *v_req,

@@ -709,18 +712,21 @@ static struct tcm_vhost_cmd *vhost_scsi_allocate_cmd(
 {
 	struct tcm_vhost_cmd *tv_cmd;
 	struct tcm_vhost_nexus *tv_nexus;
+	struct se_session *se_sess;
+	int tag;
 
 	tv_nexus = tv_tpg->tpg_nexus;
 	if (!tv_nexus) {
 		pr_err("Unable to locate active struct tcm_vhost_nexus\n");
 		return ERR_PTR(-EIO);
 	}
+	se_sess = tv_nexus->tvn_se_sess;
 
-	tv_cmd = kzalloc(sizeof(struct tcm_vhost_cmd), GFP_ATOMIC);
-	if (!tv_cmd) {
-		pr_err("Unable to allocate struct tcm_vhost_cmd\n");
-		return ERR_PTR(-ENOMEM);
-	}
+	tag = percpu_ida_alloc(&se_sess->sess_tag_pool, GFP_KERNEL);
+	tv_cmd = &((struct tcm_vhost_cmd *)se_sess->sess_cmd_map)[tag];
+	memset(tv_cmd, 0, sizeof(struct tcm_vhost_cmd));
+
+	tv_cmd->tvc_se_cmd.map_tag = tag;
 	tv_cmd->tvc_tag = v_req->tag;
 	tv_cmd->tvc_task_attr = v_req->task_attr;
 	tv_cmd->tvc_exp_data_len = exp_data_len;

@@ -982,10 +988,10 @@ static void vhost_scsi_handle_vq(struct vhost_scsi *vs,
 		for (i = 0; i < data_num; i++)
 			exp_data_len += vq->iov[data_first + i].iov_len;
 
-		tv_cmd = vhost_scsi_allocate_cmd(vq, tv_tpg, &v_req,
-					exp_data_len, data_direction);
+		tv_cmd = vhost_scsi_get_tag(vq, tv_tpg, &v_req,
+					    exp_data_len, data_direction);
 		if (IS_ERR(tv_cmd)) {
-			vq_err(vq, "vhost_scsi_allocate_cmd failed %ld\n",
+			vq_err(vq, "vhost_scsi_get_tag failed %ld\n",
 					PTR_ERR(tv_cmd));
 			goto err_cmd;
 		}

@@ -1662,9 +1668,12 @@ static int tcm_vhost_make_nexus(struct tcm_vhost_tpg *tv_tpg,
 		return -ENOMEM;
 	}
 	/*
-	 *  Initialize the struct se_session pointer
+	 *  Initialize the struct se_session pointer and setup tagpool
+	 *  for struct tcm_vhost_cmd descriptors
 	 */
-	tv_nexus->tvn_se_sess = transport_init_session();
+	tv_nexus->tvn_se_sess = transport_init_session_tags(
+						TCM_VHOST_DEFAULT_TAGS,
+						sizeof(struct tcm_vhost_cmd));
 	if (IS_ERR(tv_nexus->tvn_se_sess)) {
 		mutex_unlock(&tv_tpg->tv_tpg_mutex);
 		kfree(tv_nexus);

index 6224434..f200b17 100644
@@ -57,6 +57,8 @@
 #define TCM_VHOST_NAMELEN 256
 #define TCM_VHOST_MAX_CDB_SIZE 32
 #define TCM_VHOST_DEFAULT_TAGS 256
+#define TCM_VHOST_PREALLOC_SGLS 2048
+#define TCM_VHOST_PREALLOC_PAGES 2048
 
 struct vhost_scsi_inflight {
 	/* Wait for the flush operation to finish */

@@ -82,6 +84,7 @@ struct tcm_vhost_cmd {
 	u32 tvc_lun;
 	/* Pointer to the SGL formatted memory from virtio-scsi */
 	struct scatterlist *tvc_sgl;
+	struct page **tvc_upages;
 	/* Pointer to response */
 	struct virtio_scsi_cmd_resp __user *tvc_resp;
 	/* Pointer to vhost_scsi for our device */

@@ -456,8 +459,6 @@ static void tcm_vhost_release_cmd(struct se_cmd *se_cmd)
 		u32 i;
 		for (i = 0; i < tv_cmd->tvc_sgl_count; i++)
 			put_page(sg_page(&tv_cmd->tvc_sgl[i]));
-
-		kfree(tv_cmd->tvc_sgl);
         }
 
 	tcm_vhost_put_inflight(tv_cmd->inflight);

@@ -713,6 +714,8 @@ static struct tcm_vhost_cmd *vhost_scsi_get_tag(
 	struct tcm_vhost_cmd *tv_cmd;
 	struct tcm_vhost_nexus *tv_nexus;
 	struct se_session *se_sess;
+	struct scatterlist *sg;
+	struct page **pages;
 	int tag;
 
 	tv_nexus = tv_tpg->tpg_nexus;

@@ -724,8 +727,12 @@ static struct tcm_vhost_cmd *vhost_scsi_get_tag(
 
 	tag = percpu_ida_alloc(&se_sess->sess_tag_pool, GFP_KERNEL);
 	tv_cmd = &((struct tcm_vhost_cmd *)se_sess->sess_cmd_map)[tag];
+	sg = tv_cmd->tvc_sgl;
+	pages = tv_cmd->tvc_upages;
 	memset(tv_cmd, 0, sizeof(struct tcm_vhost_cmd));
 
+	tv_cmd->tvc_sgl = sg;
+	tv_cmd->tvc_upages = pages;
 	tv_cmd->tvc_se_cmd.map_tag = tag;
 	tv_cmd->tvc_tag = v_req->tag;
 	tv_cmd->tvc_task_attr = v_req->task_attr;

@@ -742,7 +749,7 @@ static struct tcm_vhost_cmd *vhost_scsi_get_tag(
  *
  * Returns the number of scatterlist entries used or -errno on error.
  */
-static int vhost_scsi_map_to_sgl(struct scatterlist *sgl,
+static int vhost_scsi_map_to_sgl(struct tcm_vhost_cmd *tv_cmd, struct scatterlist *sgl,
 	unsigned int sgl_count, struct iovec *iov, int write)
 {
 	unsigned int npages = 0, pages_nr, offset, nbytes;

@@ -752,13 +759,25 @@ static int vhost_scsi_map_to_sgl(struct scatterlist *sgl,
 	struct page **pages;
 	int ret, i;
 
+	if (sgl_count > TCM_VHOST_PREALLOC_SGLS) {
+		pr_err("vhost_scsi_map_to_sgl() psgl_count: %u greater than"
+		       " preallocated TCM_VHOST_PREALLOC_SGLS: %u\n",
+			sgl_count, TCM_VHOST_PREALLOC_SGLS);
+		return -ENOBUFS;
+	}
+
 	pages_nr = iov_num_pages(iov);
 	if (pages_nr > sgl_count)
 		return -ENOBUFS;
 
-	pages = kmalloc(pages_nr * sizeof(struct page *), GFP_KERNEL);
-	if (!pages)
-		return -ENOMEM;
+	if (pages_nr > TCM_VHOST_PREALLOC_PAGES) {
+		pr_err("vhost_scsi_map_to_sgl() pages_nr: %u greater than"
+		       " preallocated TCM_VHOST_PREALLOC_PAGES: %u\n",
+			pages_nr, TCM_VHOST_PREALLOC_PAGES);
+		return -ENOBUFS;
+	}
+
+	pages = tv_cmd->tvc_upages;
 
 	ret = get_user_pages_fast((unsigned long)ptr, pages_nr, write, pages);
 	/* No pages were pinned */

@@ -783,7 +802,6 @@ static int vhost_scsi_map_to_sgl(struct scatterlist *sgl,
 	}
 
 out:
-	kfree(pages);
 	return ret;
 }
 

@@ -803,25 +821,18 @@ static int vhost_scsi_map_iov_to_sgl(struct tcm_vhost_cmd *tv_cmd,
 		sgl_count += iov_num_pages(&iov[i]);
 
 	/* TODO overflow checking */
-
-	sg = kmalloc(sizeof(tv_cmd->tvc_sgl[0]) * sgl_count, GFP_ATOMIC);
-	if (!sg)
-		return -ENOMEM;
-	pr_debug("%s sg %p sgl_count %u is_err %d\n", __func__,
-	       sg, sgl_count, !sg);
+	sg = tv_cmd->tvc_sgl;
 	sg_init_table(sg, sgl_count);
 
-	tv_cmd->tvc_sgl = sg;
 	tv_cmd->tvc_sgl_count = sgl_count;
 
 	pr_debug("Mapping %u iovecs for %u pages\n", niov, sgl_count);
 	for (i = 0; i < niov; i++) {
-		ret = vhost_scsi_map_to_sgl(sg, sgl_count, &iov[i], write);
+		ret = vhost_scsi_map_to_sgl(tv_cmd, sg, sgl_count, &iov[i], write);
 		if (ret < 0) {
 			for (i = 0; i < tv_cmd->tvc_sgl_count; i++)
 				put_page(sg_page(&tv_cmd->tvc_sgl[i]));
-			kfree(tv_cmd->tvc_sgl);
-			tv_cmd->tvc_sgl = NULL;
+
 			tv_cmd->tvc_sgl_count = 0;
 			return ret;
 		}

@@ -1647,11 +1658,31 @@ static void tcm_vhost_drop_nodeacl(struct se_node_acl *se_acl)
 	kfree(nacl);
 }
 
+static void tcm_vhost_free_cmd_map_res(struct tcm_vhost_nexus *nexus,
+				       struct se_session *se_sess)
+{
+	struct tcm_vhost_cmd *tv_cmd;
+	int i;
+
+	if (!se_sess->sess_cmd_map)
+		return;
+
+	for (i = 0; i < TCM_VHOST_DEFAULT_TAGS; i++) {
+		tv_cmd = &((struct tcm_vhost_cmd *)se_sess->sess_cmd_map)[i];
+
+		kfree(tv_cmd->tvc_sgl);
+		kfree(tv_cmd->tvc_upages);
+	}
+}
+
 static int tcm_vhost_make_nexus(struct tcm_vhost_tpg *tv_tpg,
 	const char *name)
 {
 	struct se_portal_group *se_tpg;
+	struct se_session *se_sess;
 	struct tcm_vhost_nexus *tv_nexus;
+	struct tcm_vhost_cmd *tv_cmd;
+	unsigned int i;
 
 	mutex_lock(&tv_tpg->tv_tpg_mutex);
 	if (tv_tpg->tpg_nexus) {

@@ -1679,6 +1710,26 @@ static int tcm_vhost_make_nexus(struct tcm_vhost_tpg *tv_tpg,
 		kfree(tv_nexus);
 		return -ENOMEM;
 	}
+	se_sess = tv_nexus->tvn_se_sess;
+	for (i = 0; i < TCM_VHOST_DEFAULT_TAGS; i++) {
+		tv_cmd = &((struct tcm_vhost_cmd *)se_sess->sess_cmd_map)[i];
+
+		tv_cmd->tvc_sgl = kzalloc(sizeof(struct scatterlist) *
+					TCM_VHOST_PREALLOC_SGLS, GFP_KERNEL);
+		if (!tv_cmd->tvc_sgl) {
+			mutex_unlock(&tv_tpg->tv_tpg_mutex);
+			pr_err("Unable to allocate tv_cmd->tvc_sgl\n");
+			goto out;
+		}
+
+		tv_cmd->tvc_upages = kzalloc(sizeof(struct page *) *
+					TCM_VHOST_PREALLOC_PAGES, GFP_KERNEL);
+		if (!tv_cmd->tvc_upages) {
+			mutex_unlock(&tv_tpg->tv_tpg_mutex);
+			pr_err("Unable to allocate tv_cmd->tvc_upages\n");
+			goto out;
+		}
+	}
 	/*
 	 * Since we are running in 'demo mode' this call with generate a
 	 * struct se_node_acl for the tcm_vhost struct se_portal_group with

@@ -1690,9 +1741,7 @@ static int tcm_vhost_make_nexus(struct tcm_vhost_tpg *tv_tpg,
 		mutex_unlock(&tv_tpg->tv_tpg_mutex);
 		pr_debug("core_tpg_check_initiator_node_acl() failed"
 				" for %s\n", name);
-		transport_free_session(tv_nexus->tvn_se_sess);
-		kfree(tv_nexus);
-		return -ENOMEM;
+		goto out;
 	}
 	/*
 	 * Now register the TCM vhost virtual I_T Nexus as active with the

@@ -1704,6 +1753,12 @@ static int tcm_vhost_make_nexus(struct tcm_vhost_tpg *tv_tpg,
 
 	mutex_unlock(&tv_tpg->tv_tpg_mutex);
 	return 0;
+
+out:
+	tcm_vhost_free_cmd_map_res(tv_nexus, se_sess);
+	transport_free_session(se_sess);
+	kfree(tv_nexus);
+	return -ENOMEM;
 }
 
 static int tcm_vhost_drop_nexus(struct tcm_vhost_tpg *tpg)

@@ -1743,6 +1798,8 @@ static int tcm_vhost_drop_nexus(struct tcm_vhost_tpg *tpg)
 	pr_debug("TCM_vhost_ConfigFS: Removing I_T Nexus to emulated"
 		" %s Initiator Port: %s\n", tcm_vhost_dump_proto_id(tpg->tport),
 		tv_nexus->tvn_se_sess->se_node_acl->initiatorname);
+
+	tcm_vhost_free_cmd_map_res(tv_nexus, se_sess);
 	/*
 	 * Release the SCSI I_T Nexus to the emulated vhost Target Port
 	 */