From 362caa76662fdf7094e39f452d61dd7e7dc2a9a1 Mon Sep 17 00:00:00 2001
From: Mike Christie <michaelc@cs.wisc.edu>
Date: Wed, 29 Jul 2015 04:23:46 -0500
Subject: [PATCH] libceph: add support for write same requests
References: fate#318836
Patch-mainline: Not yet, SES2 clustered LIO/RBD

This adds a new ceph request writesame. Write a buffer of length
writesame.data_length bytes at writesame.offset over writesame.length
bytes.

This command maps to SCSI's WRITE SAME request. In the next patches
rbd and lio will hook in to this support.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: David Disseldorp <ddiss@suse.de>
Acked-by: Luis Henriques <lhenriques@suse.com>
[luis: updated get_num_data_items() introduced by commit 0d9c1ab3be4c
 ("libceph: preallocate message data items")]
[luis: rebased atop 26f887e0a3c4 ("libceph, rbd, ceph: move
 ceph_osdc_alloc_messages() calls")]
---
 include/linux/ceph/osd_client.h |   15 ++++++++++++++
 include/linux/ceph/rados.h      |    6 +++++
 net/ceph/osd_client.c           |   41 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 62 insertions(+)

--- a/include/linux/ceph/osd_client.h
+++ b/include/linux/ceph/osd_client.h
@@ -141,6 +141,12 @@ struct ceph_osd_req_op {
 			u64 expected_object_size;
 			u64 expected_write_size;
 		} alloc_hint;
+		struct {
+			u64 offset;
+			u64 length;
+			u64 data_length;
+			struct ceph_osd_data request_data;
+		} writesame;
 	};
 };
 
@@ -397,6 +403,10 @@ extern void osd_req_op_raw_data_in_pages
 					u32 alignment, bool pages_from_pool,
 					bool own_pages);
 
+extern void osd_req_op_writesame_init(struct ceph_osd_request *osd_req,
+					unsigned int which, u16 opcode,
+					u64 offset, u64 length,
+					u64 data_length);
 extern void osd_req_op_extent_init(struct ceph_osd_request *osd_req,
 					unsigned int which, u16 opcode,
 					u64 offset, u64 length,
@@ -435,6 +445,11 @@ extern void osd_req_op_extent_osd_data_s
 					unsigned int which,
 					struct scatterlist *sgl,
 					unsigned int init_sg_offset,
+					u64 length);
+extern void osd_req_op_writesame_osd_data_sg(struct ceph_osd_request *,
+					unsigned int which,
+					struct scatterlist *sgl,
+					unsigned int init_sg_offset,
 					u64 length);
 extern void osd_req_op_cls_request_data_pagelist(struct ceph_osd_request *,
 					unsigned int which,
--- a/include/linux/ceph/rados.h
+++ b/include/linux/ceph/rados.h
@@ -227,6 +227,7 @@ extern const char *ceph_osd_state_name(i
 	f(TRUNCATE,	__CEPH_OSD_OP(WR, DATA, 3),	"truncate")	    \
 	f(ZERO,		__CEPH_OSD_OP(WR, DATA, 4),	"zero")		    \
 	f(DELETE,	__CEPH_OSD_OP(WR, DATA, 5),	"delete")	    \
+	f(WRITESAME,	__CEPH_OSD_OP(WR, DATA, 36),	"write-same")	    \
 									    \
 	/* fancy write */						    \
 	f(APPEND,	__CEPH_OSD_OP(WR, DATA, 6),	"append")	    \
@@ -499,6 +500,11 @@ struct ceph_osd_op {
 			__le64 expected_object_size;
 			__le64 expected_write_size;
 		} __attribute__ ((packed)) alloc_hint;
+		struct {
+			__le64 offset;
+			__le64 length;
+			__le64 data_length;
+		} __attribute__ ((packed)) writesame;
 	};
 	__le32 payload_len;
 } __attribute__ ((packed));
--- a/net/ceph/osd_client.c
+++ b/net/ceph/osd_client.c
@@ -358,6 +358,17 @@ void osd_req_op_cls_response_data_pages(
 }
 EXPORT_SYMBOL(osd_req_op_cls_response_data_pages);
 
+void osd_req_op_writesame_osd_data_sg(struct ceph_osd_request *osd_req,
+			unsigned int which, struct scatterlist *sgl,
+			unsigned int init_sg_offset, u64 length)
+{
+	struct ceph_osd_data *osd_data;
+
+	osd_data = osd_req_op_data(osd_req, which, writesame, request_data);
+	ceph_osd_data_sg_init(osd_data, sgl, init_sg_offset, length);
+}
+EXPORT_SYMBOL(osd_req_op_writesame_osd_data_sg);
+
 static u64 ceph_osd_data_length(struct ceph_osd_data *osd_data)
 {
 	switch (osd_data->type) {
@@ -410,6 +421,9 @@ static void osd_req_op_data_release(stru
 	case CEPH_OSD_OP_CMPEXT:
 		ceph_osd_data_release(&op->extent.osd_data);
 		break;
+	case CEPH_OSD_OP_WRITESAME:
+		ceph_osd_data_release(&op->writesame.request_data);
+		break;
 	case CEPH_OSD_OP_CALL:
 		ceph_osd_data_release(&op->cls.request_info);
 		ceph_osd_data_release(&op->cls.request_data);
@@ -726,6 +740,7 @@ static void get_num_data_items(struct ce
 		case CEPH_OSD_OP_CMPXATTR:
 		case CEPH_OSD_OP_NOTIFY_ACK:
 		case CEPH_OSD_OP_CMPEXT:
+		case CEPH_OSD_OP_WRITESAME:
 			*num_request_data_items += 1;
 			break;
 
@@ -796,6 +811,22 @@ void osd_req_op_init(struct ceph_osd_req
 }
 EXPORT_SYMBOL(osd_req_op_init);
 
+void osd_req_op_writesame_init(struct ceph_osd_request *osd_req,
+			       unsigned int which, u16 opcode,
+			       u64 offset, u64 length, u64 data_length)
+{
+	struct ceph_osd_req_op *op = _osd_req_op_init(osd_req, which, opcode,
+						      0);
+
+	BUG_ON(opcode != CEPH_OSD_OP_WRITESAME);
+
+	op->writesame.offset = offset;
+	op->writesame.length = length;
+	op->writesame.data_length = data_length;
+	op->indata_len = data_length;
+}
+EXPORT_SYMBOL(osd_req_op_writesame_init);
+
 void osd_req_op_extent_init(struct ceph_osd_request *osd_req,
 				unsigned int which, u16 opcode,
 				u64 offset, u64 length,
@@ -1054,6 +1085,12 @@ static u32 osd_req_encode_op(struct ceph
 	case CEPH_OSD_OP_CREATE:
 	case CEPH_OSD_OP_DELETE:
 		break;
+	case CEPH_OSD_OP_WRITESAME:
+		dst->writesame.offset = cpu_to_le64(src->writesame.offset);
+		dst->writesame.length = cpu_to_le64(src->writesame.length);
+		dst->writesame.data_length =
+				cpu_to_le64(src->writesame.data_length);
+		break;
 	default:
 		pr_err("unsupported osd opcode %s\n",
 			ceph_osd_op_name(src->op));
@@ -1976,6 +2013,10 @@ static void setup_request_data(struct ce
 			ceph_osdc_msg_data_add(request_msg,
 					       &op->extent.osd_data);
 			break;
+		case CEPH_OSD_OP_WRITESAME:
+			ceph_osdc_msg_data_add(request_msg,
+					       &op->writesame.request_data);
+			break;
 		case CEPH_OSD_OP_SETXATTR:
 		case CEPH_OSD_OP_CMPXATTR:
 			WARN_ON(op->indata_len != op->xattr.name_len +