[PATCH RFC 1/3] nvme: add support for extended LBAs

[Klaus Jensen]

From: Gollu Appalanaidu <anaidu.gollu@samsung.com>

This allows logical blocks to be extended with a number of metadata
bytes specified by the new namespace parameter 'ms'. The additional
bytes are stored immediately after each logical block.

The Deallocated or Unwritten Logical Block Error recovery feature is not
supported for namespaces with extended LBAs since the extended logical
blocks are not aligned with the blocks of the underlying device and the
allocation status of blocks can thus not be detemined by the
BDRV_BLOCK_ZERO bdrv_block_status flag. Similary, the DLFEAT field will
not report any read behavior for deallocated logical blocks reported.

Signed-off-by: Gollu Appalanaidu <anaidu.gollu@samsung.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
---
 hw/block/nvme-ns.h | 19 ++++++++++++++++---
 hw/block/nvme-ns.c | 21 +++++++++++++++++----
 hw/block/nvme.c    |  6 ++++--
 3 files changed, 37 insertions(+), 9 deletions(-)

diff --git a/hw/block/nvme-ns.h b/hw/block/nvme-ns.h
index 44bf6271b744..1e621fb130a3 100644
--- a/hw/block/nvme-ns.h
+++ b/hw/block/nvme-ns.h
@@ -21,6 +21,7 @@
 
 typedef struct NvmeNamespaceParams {
     uint32_t nsid;
+    uint16_t ms;
 } NvmeNamespaceParams;
 
 typedef struct NvmeNamespace {
@@ -57,18 +58,30 @@ static inline uint8_t nvme_ns_lbads(NvmeNamespace *ns)
     return nvme_ns_lbaf(ns)->ds;
 }
 
-/* calculate the number of LBAs that the namespace can accomodate */
-static inline uint64_t nvme_ns_nlbas(NvmeNamespace *ns)
+static inline uint16_t nvme_ns_ms(NvmeNamespace *ns)
 {
-    return ns->size >> nvme_ns_lbads(ns);
+    return nvme_ns_lbaf(ns)->ms;
 }
 
 /* convert an LBA to the equivalent in bytes */
 static inline size_t nvme_l2b(NvmeNamespace *ns, uint64_t lba)
 {
+    if (NVME_ID_NS_FLBAS_EXTENDED(ns->id_ns.flbas)) {
+        return (lba << nvme_ns_lbads(ns)) + (lba * nvme_ns_ms(ns));
+    }
+
     return lba << nvme_ns_lbads(ns);
 }
 
+/* calculate the number of LBAs that the namespace can accomodate */
+static inline uint64_t nvme_ns_nlbas(NvmeNamespace *ns)
+{
+    if (NVME_ID_NS_FLBAS_EXTENDED(ns->id_ns.flbas)) {
+        return ns->size / nvme_l2b(ns, 1);
+    }
+    return ns->size >> nvme_ns_lbads(ns);
+}
+
 typedef struct NvmeCtrl NvmeCtrl;
 
 int nvme_ns_setup(NvmeCtrl *n, NvmeNamespace *ns, Error **errp);
diff --git a/hw/block/nvme-ns.c b/hw/block/nvme-ns.c
index 2d69b5177b51..a9785a12eb13 100644
--- a/hw/block/nvme-ns.c
+++ b/hw/block/nvme-ns.c
@@ -37,9 +37,24 @@ static int nvme_ns_init(NvmeNamespace *ns, Error **errp)
     int lba_index = NVME_ID_NS_FLBAS_INDEX(ns->id_ns.flbas);
     int npdg;
 
-    ns->id_ns.dlfeat = 0x9;
+    id_ns->dlfeat = 0x10;
 
     id_ns->lbaf[lba_index].ds = 31 - clz32(ns->blkconf.logical_block_size);
+    id_ns->lbaf[lba_index].ms = ns->params.ms;
+
+    /* support DULBE and I/O optimization fields */
+    id_ns->nsfeat |= 0x10;
+
+    if (!ns->params.ms) {
+        /* zeroes are guaranteed to be read from deallocated blocks */
+        id_ns->dlfeat |= 0x1 | 0x8;
+
+        /* support DULBE */
+        id_ns->nsfeat |= 0x4;
+    } else {
+        id_ns->mc = 0x1;
+        id_ns->flbas |= 0x10;
+    }
 
     id_ns->nsze = cpu_to_le64(nvme_ns_nlbas(ns));
 
@@ -47,9 +62,6 @@ static int nvme_ns_init(NvmeNamespace *ns, Error **errp)
     id_ns->ncap = id_ns->nsze;
     id_ns->nuse = id_ns->ncap;
 
-    /* support DULBE and I/O optimization fields */
-    id_ns->nsfeat |= (0x4 | 0x10);
-
     npdg = ns->blkconf.discard_granularity / ns->blkconf.logical_block_size;
 
     if (bdrv_get_info(blk_bs(ns->blkconf.blk), &bdi) >= 0 &&
@@ -150,6 +162,7 @@ static void nvme_ns_realize(DeviceState *dev, Error **errp)
 static Property nvme_ns_props[] = {
     DEFINE_BLOCK_PROPERTIES(NvmeNamespace, blkconf),
     DEFINE_PROP_UINT32("nsid", NvmeNamespace, params.nsid, 0),
+    DEFINE_PROP_UINT16("ms", NvmeNamespace, params.ms, 0),
     DEFINE_PROP_END_OF_LIST(),
 };
 
diff --git a/hw/block/nvme.c b/hw/block/nvme.c
index 28416b18a5c0..e4922c37c94d 100644
--- a/hw/block/nvme.c
+++ b/hw/block/nvme.c
@@ -1214,6 +1214,7 @@ static uint16_t nvme_rw(NvmeCtrl *n, NvmeRequest *req)
         BLOCK_ACCT_WRITE : BLOCK_ACCT_READ;
     BlockBackend *blk = ns->blkconf.blk;
     uint16_t status;
+    uint32_t sector_size;
 
     trace_pci_nvme_rw(nvme_cid(req), nvme_io_opc_str(rw->opcode),
                       nvme_nsid(ns), nlb, data_size, slba);
@@ -1246,12 +1247,13 @@ static uint16_t nvme_rw(NvmeCtrl *n, NvmeRequest *req)
 
     block_acct_start(blk_get_stats(blk), &req->acct, data_size, acct);
     if (req->qsg.sg) {
+        sector_size = nvme_l2b(ns, 1);
         if (acct == BLOCK_ACCT_WRITE) {
             req->aiocb = dma_blk_write(blk, &req->qsg, data_offset,
-                                       BDRV_SECTOR_SIZE, nvme_rw_cb, req);
+                                       sector_size, nvme_rw_cb, req);
         } else {
             req->aiocb = dma_blk_read(blk, &req->qsg, data_offset,
-                                      BDRV_SECTOR_SIZE, nvme_rw_cb, req);
+                                      sector_size, nvme_rw_cb, req);
         }
     } else {
         if (acct == BLOCK_ACCT_WRITE) {
-- 
2.29.2