From: Gollu Appalanaidu <anaidu.gollu@samsung.com>
See NVM Express 1.4, section 6.14 ("Verify Command").
Signed-off-by: Gollu Appalanaidu <anaidu.gollu@samsung.com>
[k.jensen: rebased, refactored for e2e]
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
---
hw/block/nvme.h | 1 +
include/block/nvme.h | 2 +
hw/block/nvme.c | 148 +++++++++++++++++++++++++++++++++++++++++-
hw/block/trace-events | 3 +
4 files changed, 153 insertions(+), 1 deletion(-)
diff --git a/hw/block/nvme.h b/hw/block/nvme.h
index 86f000235cf0..3b8ed73d2a04 100644
--- a/hw/block/nvme.h
+++ b/hw/block/nvme.h
@@ -82,6 +82,7 @@ static inline const char *nvme_io_opc_str(uint8_t opc)
case NVME_CMD_COMPARE: return "NVME_NVM_CMD_COMPARE";
case NVME_CMD_WRITE_ZEROES: return "NVME_NVM_CMD_WRITE_ZEROES";
case NVME_CMD_DSM: return "NVME_NVM_CMD_DSM";
+ case NVME_CMD_VERIFY: return "NVME_NVM_CMD_VERIFY";
case NVME_CMD_COPY: return "NVME_NVM_CMD_COPY";
case NVME_CMD_ZONE_MGMT_SEND: return "NVME_ZONED_CMD_MGMT_SEND";
case NVME_CMD_ZONE_MGMT_RECV: return "NVME_ZONED_CMD_MGMT_RECV";
diff --git a/include/block/nvme.h b/include/block/nvme.h
index a7debf29c644..c2fd7e817e5d 100644
--- a/include/block/nvme.h
+++ b/include/block/nvme.h
@@ -579,6 +579,7 @@ enum NvmeIoCommands {
NVME_CMD_COMPARE = 0x05,
NVME_CMD_WRITE_ZEROES = 0x08,
NVME_CMD_DSM = 0x09,
+ NVME_CMD_VERIFY = 0x0c,
NVME_CMD_COPY = 0x19,
NVME_CMD_ZONE_MGMT_SEND = 0x79,
NVME_CMD_ZONE_MGMT_RECV = 0x7a,
@@ -1060,6 +1061,7 @@ enum NvmeIdCtrlOncs {
NVME_ONCS_FEATURES = 1 << 4,
NVME_ONCS_RESRVATIONS = 1 << 5,
NVME_ONCS_TIMESTAMP = 1 << 6,
+ NVME_ONCS_VERIFY = 1 << 7,
NVME_ONCS_COPY = 1 << 8,
};
diff --git a/hw/block/nvme.c b/hw/block/nvme.c
index ab45e58b2a3b..7b366f979a1c 100644
--- a/hw/block/nvme.c
+++ b/hw/block/nvme.c
@@ -186,6 +186,7 @@ static const uint32_t nvme_cse_iocs_nvm[256] = {
[NVME_CMD_WRITE] = NVME_CMD_EFF_CSUPP | NVME_CMD_EFF_LBCC,
[NVME_CMD_READ] = NVME_CMD_EFF_CSUPP,
[NVME_CMD_DSM] = NVME_CMD_EFF_CSUPP | NVME_CMD_EFF_LBCC,
+ [NVME_CMD_VERIFY] = NVME_CMD_EFF_CSUPP,
[NVME_CMD_COPY] = NVME_CMD_EFF_CSUPP | NVME_CMD_EFF_LBCC,
[NVME_CMD_COMPARE] = NVME_CMD_EFF_CSUPP,
};
@@ -196,6 +197,7 @@ static const uint32_t nvme_cse_iocs_zoned[256] = {
[NVME_CMD_WRITE] = NVME_CMD_EFF_CSUPP | NVME_CMD_EFF_LBCC,
[NVME_CMD_READ] = NVME_CMD_EFF_CSUPP,
[NVME_CMD_DSM] = NVME_CMD_EFF_CSUPP | NVME_CMD_EFF_LBCC,
+ [NVME_CMD_VERIFY] = NVME_CMD_EFF_CSUPP,
[NVME_CMD_COPY] = NVME_CMD_EFF_CSUPP | NVME_CMD_EFF_LBCC,
[NVME_CMD_COMPARE] = NVME_CMD_EFF_CSUPP,
[NVME_CMD_ZONE_APPEND] = NVME_CMD_EFF_CSUPP | NVME_CMD_EFF_LBCC,
@@ -2154,6 +2156,91 @@ out:
nvme_e2e_rw_cb(ctx, ret);
}
+
+static void nvme_verify_cb(void *opaque, int ret)
+{
+ struct nvme_bounce_ctx *ctx = opaque;
+ NvmeRequest *req = ctx->req;
+ NvmeNamespace *ns = req->ns;
+ BlockBackend *blk = ns->blkconf.blk;
+ BlockAcctCookie *acct = &req->acct;
+ BlockAcctStats *stats = blk_get_stats(blk);
+ NvmeRwCmd *rw = (NvmeRwCmd *)&req->cmd;
+ uint64_t slba = le64_to_cpu(rw->slba);
+ uint16_t ctrl = le16_to_cpu(rw->control);
+ uint16_t apptag = le16_to_cpu(rw->apptag);
+ uint16_t appmask = le16_to_cpu(rw->appmask);
+ uint32_t reftag = le32_to_cpu(rw->reftag);
+ uint16_t status;
+
+ trace_pci_nvme_verify_cb(nvme_cid(req), NVME_RW_PRINFO(ctrl), apptag,
+ appmask, reftag);
+
+ if (ret) {
+ block_acct_failed(stats, acct);
+ nvme_aio_err(req, ret);
+ goto out;
+ }
+
+ block_acct_done(stats, acct);
+
+ if (NVME_ID_NS_DPS_TYPE(ns->id_ns.dps)) {
+ status = nvme_e2e_mangle_mdata(ns, ctx->mdata.bounce,
+ ctx->mdata.iov.size, slba);
+ if (status) {
+ req->status = status;
+ goto out;
+ }
+
+ req->status = nvme_e2e_check(ns, ctx->data.bounce, ctx->data.iov.size,
+ ctx->mdata.bounce, ctx->mdata.iov.size,
+ ctrl, slba, apptag, appmask, reftag);
+ }
+
+out:
+ qemu_iovec_destroy(&ctx->data.iov);
+ g_free(ctx->data.bounce);
+
+ qemu_iovec_destroy(&ctx->mdata.iov);
+ g_free(ctx->mdata.bounce);
+
+ g_free(ctx);
+
+ nvme_enqueue_req_completion(nvme_cq(req), req);
+}
+
+
+static void nvme_verify_mdata_in_cb(void *opaque, int ret)
+{
+ struct nvme_bounce_ctx *ctx = opaque;
+ NvmeRequest *req = ctx->req;
+ NvmeNamespace *ns = req->ns;
+ NvmeRwCmd *rw = (NvmeRwCmd *)&req->cmd;
+ uint64_t slba = le64_to_cpu(rw->slba);
+ uint32_t nlb = le16_to_cpu(rw->nlb) + 1;
+ size_t mlen = nvme_m2b(ns, nlb);
+ uint64_t offset = ns->mdata_offset + nvme_m2b(ns, slba);
+ BlockBackend *blk = ns->blkconf.blk;
+
+ trace_pci_nvme_verify_mdata_in_cb(nvme_cid(req), blk_name(blk));
+
+ if (ret) {
+ goto out;
+ }
+
+ ctx->mdata.bounce = g_malloc(mlen);
+
+ qemu_iovec_reset(&ctx->mdata.iov);
+ qemu_iovec_add(&ctx->mdata.iov, ctx->mdata.bounce, mlen);
+
+ req->aiocb = blk_aio_preadv(blk, offset, &ctx->mdata.iov, 0,
+ nvme_verify_cb, ctx);
+ return;
+
+out:
+ nvme_verify_cb(ctx, ret);
+}
+
static void nvme_e2e_rw_mdata_in_cb(void *opaque, int ret)
{
struct nvme_bounce_ctx *ctx = opaque;
@@ -2752,6 +2839,62 @@ static uint16_t nvme_dsm(NvmeCtrl *n, NvmeRequest *req)
return status;
}
+static uint16_t nvme_verify(NvmeCtrl *n, NvmeRequest *req)
+{
+ NvmeRwCmd *rw = (NvmeRwCmd *)&req->cmd;
+ NvmeNamespace *ns = req->ns;
+ BlockBackend *blk = ns->blkconf.blk;
+ uint64_t slba = le64_to_cpu(rw->slba);
+ uint32_t nlb = le16_to_cpu(rw->nlb) + 1;
+ size_t len = nvme_l2b(ns, nlb);
+ int64_t offset = nvme_l2b(ns, slba);
+ uint16_t ctrl = le16_to_cpu(rw->control);
+ uint32_t reftag = le32_to_cpu(rw->reftag);
+ struct nvme_bounce_ctx *ctx = NULL;
+ uint16_t status;
+
+ trace_pci_nvme_verify(nvme_cid(req), nvme_nsid(ns), slba, nlb);
+
+ if (NVME_ID_NS_DPS_TYPE(ns->id_ns.dps)) {
+ status = nvme_check_prinfo(ns, ctrl, slba, reftag);
+ if (status) {
+ return status;
+ }
+
+ if (ctrl & NVME_RW_PRINFO_PRACT) {
+ return NVME_INVALID_PROT_INFO | NVME_DNR;
+ }
+ }
+
+ status = nvme_check_bounds(ns, slba, nlb);
+ if (status) {
+ trace_pci_nvme_err_invalid_lba_range(slba, nlb, ns->id_ns.nsze);
+ return status;
+ }
+
+ if (NVME_ERR_REC_DULBE(ns->features.err_rec)) {
+ status = nvme_check_dulbe(ns, slba, nlb);
+ if (status) {
+ return status;
+ }
+ }
+
+ ctx = g_new0(struct nvme_bounce_ctx, 1);
+ ctx->req = req;
+
+ ctx->data.bounce = g_malloc(len);
+
+ qemu_iovec_init(&ctx->data.iov, 1);
+ qemu_iovec_add(&ctx->data.iov, ctx->data.bounce, len);
+
+ block_acct_start(blk_get_stats(blk), &req->acct, ctx->data.iov.size,
+ BLOCK_ACCT_READ);
+
+ req->aiocb = blk_aio_preadv(ns->blkconf.blk, offset, &ctx->data.iov, 0,
+ nvme_verify_mdata_in_cb, ctx);
+ return NVME_NO_COMPLETE;
+}
+
static uint16_t nvme_copy(NvmeCtrl *n, NvmeRequest *req)
{
NvmeNamespace *ns = req->ns;
@@ -3934,6 +4077,8 @@ static uint16_t nvme_io_cmd(NvmeCtrl *n, NvmeRequest *req)
return nvme_compare(n, req);
case NVME_CMD_DSM:
return nvme_dsm(n, req);
+ case NVME_CMD_VERIFY:
+ return nvme_verify(n, req);
case NVME_CMD_COPY:
return nvme_copy(n, req);
case NVME_CMD_ZONE_MGMT_SEND:
@@ -6045,7 +6190,8 @@ static void nvme_init_ctrl(NvmeCtrl *n, PCIDevice *pci_dev)
id->nn = cpu_to_le32(n->num_namespaces);
id->oncs = cpu_to_le16(NVME_ONCS_WRITE_ZEROES | NVME_ONCS_TIMESTAMP |
NVME_ONCS_FEATURES | NVME_ONCS_DSM |
- NVME_ONCS_COMPARE | NVME_ONCS_COPY);
+ NVME_ONCS_COMPARE | NVME_ONCS_COPY |
+ NVME_ONCS_VERIFY);
/*
* NOTE: If this device ever supports a command set that does NOT use 0x0
diff --git a/hw/block/trace-events b/hw/block/trace-events
index 243b0857b3d9..b5189aa88b85 100644
--- a/hw/block/trace-events
+++ b/hw/block/trace-events
@@ -59,6 +59,9 @@ pci_nvme_copy(uint16_t cid, uint32_t nsid, uint16_t nr, uint8_t format) "cid %"P
pci_nvme_copy_source_range(uint64_t slba, uint32_t nlb) "slba 0x%"PRIx64" nlb %"PRIu32""
pci_nvme_copy_in_complete(uint16_t cid) "cid %"PRIu16""
pci_nvme_copy_cb(uint16_t cid) "cid %"PRIu16""
+pci_nvme_verify(uint16_t cid, uint32_t nsid, uint64_t slba, uint32_t nlb) "cid %"PRIu16" nsid %"PRIu32" slba 0x%"PRIx64" nlb %"PRIu32""
+pci_nvme_verify_mdata_in_cb(uint16_t cid, const char *blkname) "cid %"PRIu16" blk '%s'"
+pci_nvme_verify_cb(uint16_t cid, uint8_t prinfo, uint16_t apptag, uint16_t appmask, uint32_t reftag) "cid %"PRIu16" prinfo 0x%"PRIx8" apptag 0x%"PRIx16" appmask 0x%"PRIx16" reftag 0x%"PRIx32""
pci_nvme_rw_complete_cb(uint16_t cid, const char *blkname) "cid %"PRIu16" blk '%s'"
pci_nvme_block_status(int64_t offset, int64_t bytes, int64_t pnum, int ret, bool zeroed) "offset %"PRId64" bytes %"PRId64" pnum %"PRId64" ret 0x%x zeroed %d"
pci_nvme_dsm(uint16_t cid, uint32_t nsid, uint32_t nr, uint32_t attr) "cid %"PRIu16" nsid %"PRIu32" nr %"PRIu32" attr 0x%"PRIx32""
--
2.30.0
On Mon, Feb 15, 2021 at 12:02:37AM +0100, Klaus Jensen wrote:
> From: Gollu Appalanaidu <anaidu.gollu@samsung.com>
>
> See NVM Express 1.4, section 6.14 ("Verify Command").
>
> Signed-off-by: Gollu Appalanaidu <anaidu.gollu@samsung.com>
> [k.jensen: rebased, refactored for e2e]
> Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Verify is a generic block command supported in other protocols beyond
nvme. If we're going to support the command in nvme, I prefer the
implementation had generic backing out of the qemu block API rather than
emulate the entirety out of the nvme device.
On Feb 16 15:12, Keith Busch wrote:
> On Mon, Feb 15, 2021 at 12:02:37AM +0100, Klaus Jensen wrote:
> > From: Gollu Appalanaidu <anaidu.gollu@samsung.com>
> >
> > See NVM Express 1.4, section 6.14 ("Verify Command").
> >
> > Signed-off-by: Gollu Appalanaidu <anaidu.gollu@samsung.com>
> > [k.jensen: rebased, refactored for e2e]
> > Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
>
> Verify is a generic block command supported in other protocols beyond
> nvme. If we're going to support the command in nvme, I prefer the
> implementation had generic backing out of the qemu block API rather than
> emulate the entirety out of the nvme device.
You mean that the block API could provide a basic "check that we can
read this stuff without error"-call? Sounds reasonable enough, but since
the end-to-end data protection checks are performed in the device, we
need to pass the data buffers up anyway. If we had basic I/O (non-pi)
verify in the block API it would defeat the purpose if it provided those
buffers.
We've actually been asked directly on the availablity of Verify support
in QEMU, so I think this implementation as-is provides something useful
to users.
© 2016 - 2025 Red Hat, Inc.