[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v5 22/26] nvme: support multiple namespaces
|
From: |
Klaus Birkelund Jensen |
|
Subject: |
Re: [PATCH v5 22/26] nvme: support multiple namespaces |
|
Date: |
Mon, 16 Mar 2020 00:55:41 -0700 |
On Feb 12 14:34, Maxim Levitsky wrote:
> On Tue, 2020-02-04 at 10:52 +0100, Klaus Jensen wrote:
> > This adds support for multiple namespaces by introducing a new 'nvme-ns'
> > device model. The nvme device creates a bus named from the device name
> > ('id'). The nvme-ns devices then connect to this and registers
> > themselves with the nvme device.
> >
> > This changes how an nvme device is created. Example with two namespaces:
> >
> > -drive file=nvme0n1.img,if=none,id=disk1
> > -drive file=nvme0n2.img,if=none,id=disk2
> > -device nvme,serial=deadbeef,id=nvme0
> > -device nvme-ns,drive=disk1,bus=nvme0,nsid=1
> > -device nvme-ns,drive=disk2,bus=nvme0,nsid=2
> >
> > The drive property is kept on the nvme device to keep the change
> > backward compatible, but the property is now optional. Specifying a
> > drive for the nvme device will always create the namespace with nsid 1.
> Very reasonable way to do it.
> >
> > Signed-off-by: Klaus Jensen <address@hidden>
> > Signed-off-by: Klaus Jensen <address@hidden>
> > ---
> > hw/block/Makefile.objs | 2 +-
> > hw/block/nvme-ns.c | 158 +++++++++++++++++++++++++++
> > hw/block/nvme-ns.h | 60 +++++++++++
> > hw/block/nvme.c | 235 +++++++++++++++++++++++++----------------
> > hw/block/nvme.h | 47 ++++-----
> > hw/block/trace-events | 6 +-
> > 6 files changed, 389 insertions(+), 119 deletions(-)
> > create mode 100644 hw/block/nvme-ns.c
> > create mode 100644 hw/block/nvme-ns.h
> >
> > diff --git a/hw/block/Makefile.objs b/hw/block/Makefile.objs
> > index 28c2495a00dc..45f463462f1e 100644
> > --- a/hw/block/Makefile.objs
> > +++ b/hw/block/Makefile.objs
> > @@ -7,7 +7,7 @@ common-obj-$(CONFIG_PFLASH_CFI02) += pflash_cfi02.o
> > common-obj-$(CONFIG_XEN) += xen-block.o
> > common-obj-$(CONFIG_ECC) += ecc.o
> > common-obj-$(CONFIG_ONENAND) += onenand.o
> > -common-obj-$(CONFIG_NVME_PCI) += nvme.o
> > +common-obj-$(CONFIG_NVME_PCI) += nvme.o nvme-ns.o
> > common-obj-$(CONFIG_SWIM) += swim.o
> >
> > obj-$(CONFIG_SH4) += tc58128.o
> > diff --git a/hw/block/nvme-ns.c b/hw/block/nvme-ns.c
> > new file mode 100644
> > index 000000000000..0e5be44486f4
> > --- /dev/null
> > +++ b/hw/block/nvme-ns.c
> > @@ -0,0 +1,158 @@
> > +#include "qemu/osdep.h"
> > +#include "qemu/units.h"
> > +#include "qemu/cutils.h"
> > +#include "qemu/log.h"
> > +#include "hw/block/block.h"
> > +#include "hw/pci/msix.h"
> Do you need this include?
No, I needed hw/pci/pci.h instead :)
> > +#include "sysemu/sysemu.h"
> > +#include "sysemu/block-backend.h"
> > +#include "qapi/error.h"
> > +
> > +#include "hw/qdev-properties.h"
> > +#include "hw/qdev-core.h"
> > +
> > +#include "nvme.h"
> > +#include "nvme-ns.h"
> > +
> > +static int nvme_ns_init(NvmeNamespace *ns)
> > +{
> > + NvmeIdNs *id_ns = &ns->id_ns;
> > +
> > + id_ns->lbaf[0].ds = BDRV_SECTOR_BITS;
> > + id_ns->nuse = id_ns->ncap = id_ns->nsze =
> > + cpu_to_le64(nvme_ns_nlbas(ns));
> Nitpick: To be honest I don't really like that chain assignment,
> especially since it forces to wrap the line, but that is just my
> personal taste.
Fixed, and also added a comment as to why they are the same.
> > +
> > + return 0;
> > +}
> > +
> > +static int nvme_ns_init_blk(NvmeCtrl *n, NvmeNamespace *ns, NvmeIdCtrl *id,
> > + Error **errp)
> > +{
> > + uint64_t perm, shared_perm;
> > +
> > + Error *local_err = NULL;
> > + int ret;
> > +
> > + perm = BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE;
> > + shared_perm = BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED |
> > + BLK_PERM_GRAPH_MOD;
> > +
> > + ret = blk_set_perm(ns->blk, perm, shared_perm, &local_err);
> > + if (ret) {
> > + error_propagate_prepend(errp, local_err, "blk_set_perm: ");
> > + return ret;
> > + }
>
> You should consider using blkconf_apply_backend_options.
> Take a look at for example virtio_blk_device_realize.
> That will give you support for read only block devices as well.
So, yeah. There is a reason for this. And I will add that as a comment,
but I will write it here for posterity.
The problem is when the nvme-ns device starts getting more than just a
single drive attached (I have patches ready that will add a "metadata"
and a "state" drive). The blkconf_ functions work on a BlockConf that
embeds a BlockBackend, so you can't have one BlockConf with multiple
BlockBackend's. That is why I'm kinda copying the "good parts" of
the blkconf_apply_backend_options code here.
>
> I personally only once grazed the area of block permissions,
> so I prefer someone from the block layer to review this as well.
>
> > +
> > + ns->size = blk_getlength(ns->blk);
> > + if (ns->size < 0) {
> > + error_setg_errno(errp, -ns->size, "blk_getlength");
> > + return 1;
> > + }
> > +
> > + switch (n->conf.wce) {
> > + case ON_OFF_AUTO_ON:
> > + n->features.volatile_wc = 1;
> > + break;
> > + case ON_OFF_AUTO_OFF:
> > + n->features.volatile_wc = 0;
> > + case ON_OFF_AUTO_AUTO:
> > + n->features.volatile_wc = blk_enable_write_cache(ns->blk);
> > + break;
> > + default:
> > + abort();
> > + }
> > +
> > + blk_set_enable_write_cache(ns->blk, n->features.volatile_wc);
> > +
> > + return 0;
>
> Nitpick: also I just noticed that you call the controller 'n' I didn't paid
> attention to this
> before. I think something like 'ctrl' or ctl would be more readable.
>
Yeah, but using 'n' is done in all the existing code, so I think we
should stick with it.
> > +}
> > +
> > +static int nvme_ns_check_constraints(NvmeNamespace *ns, Error **errp)
> > +{
> > + if (!ns->blk) {
> > + error_setg(errp, "block backend not configured");
> > + return 1;
> > + }
> > +
> > + return 0;
> > +}
> > +
> > +int nvme_ns_setup(NvmeCtrl *n, NvmeNamespace *ns, Error **errp)
> > +{
> > + Error *local_err = NULL;
> > +
> > + if (nvme_ns_check_constraints(ns, &local_err)) {
> > + error_propagate_prepend(errp, local_err,
> > + "nvme_ns_check_constraints: ");
> > + return 1;
> > + }
> > +
> > + if (nvme_ns_init_blk(n, ns, &n->id_ctrl, &local_err)) {
> > + error_propagate_prepend(errp, local_err, "nvme_ns_init_blk: ");
> > + return 1;
> > + }
> > +
> > + nvme_ns_init(ns);
> > + if (nvme_register_namespace(n, ns, &local_err)) {
> > + error_propagate_prepend(errp, local_err, "nvme_register_namespace:
> > ");
> > + return 1;
> > + }
> > +
> > + return 0;
>
> Nitipick: to be honest I am not sure we want to expose internal function
> names like that in
> error hints are supposed to be readable to user that doesn't look at the
> source.
>
Fixed.
> > +}
> > +
> > +static void nvme_ns_realize(DeviceState *dev, Error **errp)
> > +{
> > + NvmeNamespace *ns = NVME_NS(dev);
> > + BusState *s = qdev_get_parent_bus(dev);
> > + NvmeCtrl *n = NVME(s->parent);
>
> Nitpick: Don't know if you defined this or it was like that always,
> but I would prefer something like NVME_CTL instead.
>
This is also grandfathered from the nvme device.
> > + Error *local_err = NULL;
> > +
> > + if (nvme_ns_setup(n, ns, &local_err)) {
> > + error_propagate_prepend(errp, local_err, "nvme_ns_setup: ");
> > + return;
> > + }
> > +}
> > +
> > +static Property nvme_ns_props[] = {
> > + DEFINE_NVME_NS_PROPERTIES(NvmeNamespace, params),
>
> If you go with my suggestion to use blkconf you will use here the
> DEFINE_BLOCK_PROPERTIES_BASE
>
See my comment above about that.
> > + DEFINE_PROP_END_OF_LIST(),
> > +};
> > +
> > +static void nvme_ns_class_init(ObjectClass *oc, void *data)
> > +{
> > + DeviceClass *dc = DEVICE_CLASS(oc);
> > +
> > + set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);
> > +
> > + dc->bus_type = TYPE_NVME_BUS;
> > + dc->realize = nvme_ns_realize;
> > + device_class_set_props(dc, nvme_ns_props);
> > + dc->desc = "virtual nvme namespace";
> > +}
>
> Looks reasonable.
> I don't know the device/bus model in depth to be honest
> (I learned it for few days some time ago though)
> so a review from someone that knows this area better that I do
> is very welcome.
>
> > +
> > +static void nvme_ns_instance_init(Object *obj)
> > +{
> > + NvmeNamespace *ns = NVME_NS(obj);
> > + char *bootindex = g_strdup_printf("/namespace@%d,0", ns->params.nsid);
> > +
> > + device_add_bootindex_property(obj, &ns->bootindex, "bootindex",
> > + bootindex, DEVICE(obj), &error_abort);
> > +
> > + g_free(bootindex);
> > +}
> > +
> > +static const TypeInfo nvme_ns_info = {
> > + .name = TYPE_NVME_NS,
> > + .parent = TYPE_DEVICE,
> > + .class_init = nvme_ns_class_init,
> > + .instance_size = sizeof(NvmeNamespace),
> > + .instance_init = nvme_ns_instance_init,
> > +};
> > +
> > +static void nvme_ns_register_types(void)
> > +{
> > + type_register_static(&nvme_ns_info);
> > +}
> > +
> > +type_init(nvme_ns_register_types)
> > diff --git a/hw/block/nvme-ns.h b/hw/block/nvme-ns.h
> > new file mode 100644
> > index 000000000000..b564bac25f6d
> > --- /dev/null
> > +++ b/hw/block/nvme-ns.h
> > @@ -0,0 +1,60 @@
> > +#ifndef NVME_NS_H
> > +#define NVME_NS_H
> > +
> > +#define TYPE_NVME_NS "nvme-ns"
> > +#define NVME_NS(obj) \
> > + OBJECT_CHECK(NvmeNamespace, (obj), TYPE_NVME_NS)
> > +
> > +#define DEFINE_NVME_NS_PROPERTIES(_state, _props) \
> > + DEFINE_PROP_DRIVE("drive", _state, blk), \
> > + DEFINE_PROP_UINT32("nsid", _state, _props.nsid, 0)
> > +
> > +typedef struct NvmeNamespaceParams {
> > + uint32_t nsid;
> > +} NvmeNamespaceParams;
> > +
> > +typedef struct NvmeNamespace {
> > + DeviceState parent_obj;
> > + BlockBackend *blk;
> > + int32_t bootindex;
> > + int64_t size;
> > +
> > + NvmeIdNs id_ns;
> > + NvmeNamespaceParams params;
> > +} NvmeNamespace;
> > +
> > +static inline uint32_t nvme_nsid(NvmeNamespace *ns)
> > +{
> > + if (ns) {
> > + return ns->params.nsid;
> > + }
> > +
> > + return -1;
> > +}
>
> To be honest I would allow user to omit nsid,
> and in this case pick a free slot out of valid namespaces.
>
> Let me explain the concept of valid/allocated/active namespaces
> from the spec as written in my summary:
>
> Valid namespaces are 1..N range of namespaces as reported in IDCTRL.NN
> That value is static, and it should be either set to some arbitrary large
> value (say 256)
> or set using qemu device parameter, and not changed dynamically as you
> currently do.
> As I understand it, IDCTRL output should not change during lifetime of the
> controller,
> although I didn't find exact confirmation on this in spec.
>
> Allocated namespaces are not relevant to us (this is only used for namespace
> management)
> (these are namespaces that exist but are not attached to the controller)
>
> And then you have Active namespaces which are the namespaces the user can
> actually address.
>
> However If I understand this correctly, currently the NVME 'bus' doesn't
> support hotplug, thus all namespaces will be already plugged in on
> VM startup, thus the issue doesn't really exist yet.
>
>
>
I added support for this. It's a nice addition and it makes the code
much nicer.
> > +
> > +static inline NvmeLBAF nvme_ns_lbaf(NvmeNamespace *ns)
> > +{
> > + NvmeIdNs *id_ns = &ns->id_ns;
> > + return id_ns->lbaf[NVME_ID_NS_FLBAS_INDEX(id_ns->flbas)];
> > +}
> > +
> > +static inline uint8_t nvme_ns_lbads(NvmeNamespace *ns)
> > +{
> > + return nvme_ns_lbaf(ns).ds;
> > +}
> > +
> > +static inline size_t nvme_ns_lbads_bytes(NvmeNamespace *ns)
> > +{
> > + return 1 << nvme_ns_lbads(ns);
> > +}
> > +
> > +static inline uint64_t nvme_ns_nlbas(NvmeNamespace *ns)
> > +{
> > + return ns->size >> nvme_ns_lbads(ns);
> > +}
> > +
> > +typedef struct NvmeCtrl NvmeCtrl;
> > +
> > +int nvme_ns_setup(NvmeCtrl *n, NvmeNamespace *ns, Error **errp);
> > +
> > +#endif /* NVME_NS_H */
> > diff --git a/hw/block/nvme.c b/hw/block/nvme.c
> > index a91c60fdc111..3a377bc56734 100644
> > --- a/hw/block/nvme.c
> > +++ b/hw/block/nvme.c
> > @@ -17,10 +17,11 @@
> > /**
> > * Usage: add options:
> > * -drive file=<file>,if=none,id=<drive_id>
> > - * -device nvme,drive=<drive_id>,serial=<serial>,id=<id[optional]>, \
> > + * -device nvme,serial=<serial>,id=<bus_name>, \
> > * cmb_size_mb=<cmb_size_mb[optional]>, \
> > * num_queues=<N[optional]>, \
> > * mdts=<mdts[optional]>
> > + * -device nvme-ns,drive=<drive_id>,bus=bus_name,nsid=1
> > *
> > * Note cmb_size_mb denotes size of CMB in MB. CMB is assumed to be at
> > * offset 0 in BAR2 and supports only WDS, RDS and SQS for now.
> > @@ -28,6 +29,7 @@
> >
> > #include "qemu/osdep.h"
> > #include "qemu/units.h"
> > +#include "qemu/error-report.h"
> > #include "hw/block/block.h"
> > #include "hw/pci/msix.h"
> > #include "hw/pci/pci.h"
> > @@ -43,6 +45,7 @@
> > #include "qemu/cutils.h"
> > #include "trace.h"
> > #include "nvme.h"
> > +#include "nvme-ns.h"
> >
> > #define NVME_SPEC_VER 0x00010300
> > #define NVME_MAX_QS PCI_MSIX_FLAGS_QSIZE
> > @@ -85,6 +88,17 @@ static int nvme_addr_read(NvmeCtrl *n, hwaddr addr, void
> > *buf, int size)
> > return pci_dma_read(&n->parent_obj, addr, buf, size);
> > }
> >
> > +static uint16_t nvme_nsid_err(NvmeCtrl *n, uint32_t nsid)
> > +{
> > + if (nsid && nsid < n->num_namespaces) {
> > + trace_nvme_dev_err_inactive_ns(nsid, n->num_namespaces);
> > + return NVME_INVALID_FIELD | NVME_DNR;
> > + }
> > +
> > + trace_nvme_dev_err_invalid_ns(nsid, n->num_namespaces);
> > + return NVME_INVALID_NSID | NVME_DNR;
> > +}
>
> I don't like that function to be honest.
> This function is called when nvme_ns returns NULL.
> IMHO it would be better to make nvme_ns return both namespace pointer and
> error code instead.
> In kernel we encode error values into the returned pointer.
>
I'm not sure how you want me to do this? I'm not familiar with the way
the kernel does it.
>
> > +
> > static int nvme_check_sqid(NvmeCtrl *n, uint16_t sqid)
> > {
> > return sqid < n->params.num_queues && n->sq[sqid] != NULL ? 0 : -1;
> > @@ -889,7 +903,7 @@ static inline uint16_t nvme_check_bounds(NvmeCtrl *n,
> > uint64_t slba,
> > uint64_t nsze = le64_to_cpu(ns->id_ns.nsze);
> >
> > if (unlikely((slba + nlb) > nsze)) {
> > - block_acct_invalid(blk_get_stats(n->conf.blk),
> > + block_acct_invalid(blk_get_stats(ns->blk),
> > nvme_req_is_write(req) ? BLOCK_ACCT_WRITE : BLOCK_ACCT_READ);
> > trace_nvme_dev_err_invalid_lba_range(slba, nlb, nsze);
> > return NVME_LBA_RANGE | NVME_DNR;
> > @@ -924,11 +938,12 @@ static uint16_t nvme_check_rw(NvmeCtrl *n,
> > NvmeRequest *req)
> >
> > static void nvme_rw_cb(NvmeRequest *req, void *opaque)
> > {
> > + NvmeNamespace *ns = req->ns;
> > NvmeSQueue *sq = req->sq;
> > NvmeCtrl *n = sq->ctrl;
> > NvmeCQueue *cq = n->cq[sq->cqid];
> >
> > - trace_nvme_dev_rw_cb(nvme_cid(req), req->cmd.nsid);
> > + trace_nvme_dev_rw_cb(nvme_cid(req), nvme_nsid(ns));
> >
> > nvme_enqueue_req_completion(cq, req);
> > }
> > @@ -1011,10 +1026,11 @@ static void nvme_aio_cb(void *opaque, int ret)
> >
> > static uint16_t nvme_flush(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest *req)
> > {
> > + NvmeNamespace *ns = req->ns;
> > NvmeAIO *aio = g_new0(NvmeAIO, 1);
> >
> > *aio = (NvmeAIO) {
> > - .blk = n->conf.blk,
> > + .blk = ns->blk,
> > .req = req,
> > };
> >
> > @@ -1038,12 +1054,12 @@ static uint16_t nvme_write_zeros(NvmeCtrl *n,
> > NvmeCmd *cmd, NvmeRequest *req)
> > req->slba = le64_to_cpu(rw->slba);
> > req->nlb = le16_to_cpu(rw->nlb) + 1;
> >
> > - trace_nvme_dev_write_zeros(nvme_cid(req), le32_to_cpu(cmd->nsid),
> > - req->slba, req->nlb);
> > + trace_nvme_dev_write_zeros(nvme_cid(req), nvme_nsid(ns), req->slba,
> > + req->nlb);
> >
> > status = nvme_check_bounds(n, req->slba, req->nlb, req);
> > if (unlikely(status)) {
> > - block_acct_invalid(blk_get_stats(n->conf.blk), BLOCK_ACCT_WRITE);
> > + block_acct_invalid(blk_get_stats(ns->blk), BLOCK_ACCT_WRITE);
> > return status;
> > }
> >
> > @@ -1053,7 +1069,7 @@ static uint16_t nvme_write_zeros(NvmeCtrl *n, NvmeCmd
> > *cmd, NvmeRequest *req)
> > aio = g_new0(NvmeAIO, 1);
> >
> > *aio = (NvmeAIO) {
> > - .blk = n->conf.blk,
> > + .blk = ns->blk,
> > .offset = offset,
> > .len = count,
> > .req = req,
> > @@ -1077,22 +1093,23 @@ static uint16_t nvme_rw(NvmeCtrl *n, NvmeCmd *cmd,
> > NvmeRequest *req)
> > req->nlb = le16_to_cpu(rw->nlb) + 1;
> > req->slba = le64_to_cpu(rw->slba);
> >
> > - trace_nvme_dev_rw(nvme_req_is_write(req) ? "write" : "read", req->nlb,
> > - req->nlb << nvme_ns_lbads(req->ns), req->slba);
> > + trace_nvme_dev_rw(nvme_cid(req), nvme_req_is_write(req) ? "write" :
> > "read",
> > + nvme_nsid(ns), req->nlb, req->nlb << nvme_ns_lbads(ns),
> > + req->slba);
> >
> > status = nvme_check_rw(n, req);
> > if (status) {
> > - block_acct_invalid(blk_get_stats(n->conf.blk), acct);
> > + block_acct_invalid(blk_get_stats(ns->blk), acct);
> > return status;
> > }
> >
> > status = nvme_map(n, cmd, req);
> > if (status) {
> > - block_acct_invalid(blk_get_stats(n->conf.blk), acct);
> > + block_acct_invalid(blk_get_stats(ns->blk), acct);
> > return status;
> > }
> >
> > - nvme_rw_aio(n->conf.blk, req->slba << nvme_ns_lbads(ns), req);
> > + nvme_rw_aio(ns->blk, req->slba << nvme_ns_lbads(ns), req);
> > nvme_req_set_cb(req, nvme_rw_cb, NULL);
> >
> > return NVME_NO_COMPLETE;
> > @@ -1105,12 +1122,11 @@ static uint16_t nvme_io_cmd(NvmeCtrl *n, NvmeCmd
> > *cmd, NvmeRequest *req)
> > trace_nvme_dev_io_cmd(nvme_cid(req), nsid, le16_to_cpu(req->sq->sqid),
> > cmd->opcode);
> >
> > - if (unlikely(nsid == 0 || nsid > n->num_namespaces)) {
> > - trace_nvme_dev_err_invalid_ns(nsid, n->num_namespaces);
> > - return NVME_INVALID_NSID | NVME_DNR;
> > - }
> > + req->ns = nvme_ns(n, nsid);
> >
> > - req->ns = &n->namespaces[nsid - 1];
> > + if (unlikely(!req->ns)) {
> > + return nvme_nsid_err(n, nsid);
> > + }
> >
> > switch (cmd->opcode) {
> > case NVME_CMD_FLUSH:
> > @@ -1256,18 +1272,24 @@ static uint16_t nvme_smart_info(NvmeCtrl *n,
> > NvmeCmd *cmd, uint8_t rae,
> > uint64_t units_read = 0, units_written = 0, read_commands = 0,
> > write_commands = 0;
> > NvmeSmartLog smart;
> > - BlockAcctStats *s;
> >
> > if (nsid && nsid != 0xffffffff) {
> > return NVME_INVALID_FIELD | NVME_DNR;
> > }
> >
> > - s = blk_get_stats(n->conf.blk);
> > + for (int i = 1; i <= n->num_namespaces; i++) {
> > + NvmeNamespace *ns = nvme_ns(n, i);
> > + if (!ns) {
> > + continue;
> > + }
> >
> > - units_read = s->nr_bytes[BLOCK_ACCT_READ] >> BDRV_SECTOR_BITS;
> > - units_written = s->nr_bytes[BLOCK_ACCT_WRITE] >> BDRV_SECTOR_BITS;
> > - read_commands = s->nr_ops[BLOCK_ACCT_READ];
> > - write_commands = s->nr_ops[BLOCK_ACCT_WRITE];
> > + BlockAcctStats *s = blk_get_stats(ns->blk);
> > +
> > + units_read += s->nr_bytes[BLOCK_ACCT_READ] >> BDRV_SECTOR_BITS;
> > + units_written += s->nr_bytes[BLOCK_ACCT_WRITE] >> BDRV_SECTOR_BITS;
> > + read_commands += s->nr_ops[BLOCK_ACCT_READ];
> > + write_commands += s->nr_ops[BLOCK_ACCT_WRITE];
> > + }
> Very minor nitpick: As something minor to do in the future, is to
> report the statistics per namespace.
In NVMe v1.4 there is no namespace specific information in the
SMART/Health log page, so this is valid for both v1.3 and v1.4.
> >
> > if (off > sizeof(smart)) {
> > return NVME_INVALID_FIELD | NVME_DNR;
> > @@ -1477,19 +1499,25 @@ static uint16_t nvme_identify_ctrl(NvmeCtrl *n,
> > NvmeCmd *cmd, NvmeRequest *req)
> >
> > static uint16_t nvme_identify_ns(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest
> > *req)
> > {
> > - NvmeNamespace *ns;
> > + NvmeIdNs *id_ns, inactive = { 0 };
> > uint32_t nsid = le32_to_cpu(cmd->nsid);
> > + NvmeNamespace *ns = nvme_ns(n, nsid);
> >
> > trace_nvme_dev_identify_ns(nsid);
> >
> > - if (unlikely(nsid == 0 || nsid > n->num_namespaces)) {
> > - trace_nvme_dev_err_invalid_ns(nsid, n->num_namespaces);
> > - return NVME_INVALID_NSID | NVME_DNR;
> > + if (unlikely(!ns)) {
> > + uint16_t status = nvme_nsid_err(n, nsid);
> > +
> > + if (!nvme_status_is_error(status, NVME_INVALID_FIELD)) {
> > + return status;
> > + }
> I really don't like checking the error value like that.
> It would be better IMHO to have something like
> nvme_is_valid_ns, nvme_is_active_ns or something like that.
>
Fixed.
> > +
> > + id_ns = &inactive;
> > + } else {
> > + id_ns = &ns->id_ns;
> > }
> >
> > - ns = &n->namespaces[nsid - 1];
> > -
> > - return nvme_dma(n, (uint8_t *) &ns->id_ns, sizeof(ns->id_ns), cmd,
> > + return nvme_dma(n, (uint8_t *) id_ns, sizeof(NvmeIdNs), cmd,
> > DMA_DIRECTION_FROM_DEVICE, req);
> > }
> >
> > @@ -1505,11 +1533,11 @@ static uint16_t nvme_identify_ns_list(NvmeCtrl *n,
> > NvmeCmd *cmd,
> > trace_nvme_dev_identify_ns_list(min_nsid);
> >
> > list = g_malloc0(data_len);
> > - for (i = 0; i < n->num_namespaces; i++) {
> > - if (i < min_nsid) {
> > + for (i = 1; i <= n->num_namespaces; i++) {
> > + if (i <= min_nsid || !nvme_ns(n, i)) {
> > continue;
> > }
> > - list[j++] = cpu_to_le32(i + 1);
> > + list[j++] = cpu_to_le32(i);
> > if (j == data_len / sizeof(uint32_t)) {
> > break;
> > }
> The refactoring part (removing that +1) which is very nice IMHO should move
> to one of earlier refactoring patches.
>
Done.
> > @@ -1539,9 +1567,8 @@ static uint16_t nvme_identify_ns_descr_list(NvmeCtrl
> > *n, NvmeCmd *cmd,
> >
> > trace_nvme_dev_identify_ns_descr_list(nsid);
> >
> > - if (unlikely(nsid == 0 || nsid > n->num_namespaces)) {
> > - trace_nvme_dev_err_invalid_ns(nsid, n->num_namespaces);
> > - return NVME_INVALID_NSID | NVME_DNR;
> > + if (unlikely(!nvme_ns(n, nsid))) {
> > + return nvme_nsid_err(n, nsid);
> > }
> >
> > /*
> > @@ -1681,7 +1708,7 @@ static uint16_t nvme_get_feature(NvmeCtrl *n, NvmeCmd
> > *cmd, NvmeRequest *req)
> > result = cpu_to_le32(n->features.err_rec);
> > break;
> > case NVME_VOLATILE_WRITE_CACHE:
> > - result = blk_enable_write_cache(n->conf.blk);
> > + result = cpu_to_le32(n->features.volatile_wc);
> OK, this fixes lack of endianess conversion I pointed out in patch 12.
> > trace_nvme_dev_getfeat_vwcache(result ? "enabled" : "disabled");
> > break;
> > case NVME_NUMBER_OF_QUEUES:
> > @@ -1735,6 +1762,8 @@ static uint16_t nvme_set_feature_timestamp(NvmeCtrl
> > *n, NvmeCmd *cmd,
> >
> > static uint16_t nvme_set_feature(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest
> > *req)
> > {
> > + NvmeNamespace *ns;
> > +
> > uint32_t dw10 = le32_to_cpu(cmd->cdw10);
> > uint32_t dw11 = le32_to_cpu(cmd->cdw11);
> >
> > @@ -1766,8 +1795,19 @@ static uint16_t nvme_set_feature(NvmeCtrl *n,
> > NvmeCmd *cmd, NvmeRequest *req)
> >
> > break;
> > case NVME_VOLATILE_WRITE_CACHE:
> > - blk_set_enable_write_cache(n->conf.blk, dw11 & 1);
> > + n->features.volatile_wc = dw11;
> > +
> > + for (int i = 1; i <= n->num_namespaces; i++) {
> > + ns = nvme_ns(n, i);
> > + if (!ns) {
> > + continue;
> > + }
> > +
> > + blk_set_enable_write_cache(ns->blk, dw11 & 1);
> > + }
> > +
> Features are per namespace (page 79 in the spec), so this
> is a good candidate of per namespace feature
>
Some features are, but the Volatile Write Cache feature is actually not.
> > break;
> > +
> > case NVME_NUMBER_OF_QUEUES:
> > if (n->qs_created) {
> > return NVME_CMD_SEQ_ERROR | NVME_DNR;
> > @@ -1890,9 +1930,17 @@ static void nvme_process_sq(void *opaque)
> >
> > static void nvme_clear_ctrl(NvmeCtrl *n)
> > {
> > + NvmeNamespace *ns;
> > int i;
> >
> > - blk_drain(n->conf.blk);
> > + for (i = 1; i <= n->num_namespaces; i++) {
> > + ns = nvme_ns(n, i);
> > + if (!ns) {
> > + continue;
> > + }
> > +
> > + blk_drain(ns->blk);
> > + }
> >
> > for (i = 0; i < n->params.num_queues; i++) {
> > if (n->sq[i] != NULL) {
> > @@ -1915,7 +1963,15 @@ static void nvme_clear_ctrl(NvmeCtrl *n)
> > n->outstanding_aers = 0;
> > n->qs_created = false;
> >
> > - blk_flush(n->conf.blk);
> > + for (i = 1; i <= n->num_namespaces; i++) {
> > + ns = nvme_ns(n, i);
> > + if (!ns) {
> > + continue;
> > + }
> > +
> > + blk_flush(ns->blk);
> > + }
> > +
> > n->bar.cc = 0;
> > }
> >
> > @@ -2335,8 +2391,8 @@ static int nvme_check_constraints(NvmeCtrl *n, Error
> > **errp)
> > {
> > NvmeParams *params = &n->params;
> >
> > - if (!n->conf.blk) {
> > - error_setg(errp, "nvme: block backend not configured");
> > + if (!n->namespace.blk && !n->parent_obj.qdev.id) {
> > + error_setg(errp, "nvme: invalid 'id' parameter");
>
> Nitpick: I think that usually qemu allows user to shoot him in the foot and
> allow to specify a device without ID,
> to which you can't attach devices, so I think that this check is not needed.
> You also probably mean 'missing ID'
>
Right. I added a deprecation warning when the drive parameter is used
instead.
> > return 1;
> > }
> >
> > @@ -2353,22 +2409,10 @@ static int nvme_check_constraints(NvmeCtrl *n,
> > Error **errp)
> > return 0;
> > }
> >
> > -static int nvme_init_blk(NvmeCtrl *n, Error **errp)
> > -{
> > - blkconf_blocksizes(&n->conf);
> > - if (!blkconf_apply_backend_options(&n->conf,
> > blk_is_read_only(n->conf.blk),
> > - false, errp)) {
> > - return 1;
> > - }
> > -
> > - return 0;
> > -}
> > -
> > static void nvme_init_state(NvmeCtrl *n)
> > {
> > - n->num_namespaces = 1;
> > + n->num_namespaces = 0;
>
> And to say that again since number of valid namespaces should remain static,
> here you should just initialize this NVME_MAX_NAMESPACES, and remove the code
> that changes IDCTRL.NN dynamically.
>
Done.
>
> > n->reg_size = pow2ceil(0x1004 + 2 * (n->params.num_queues + 1) * 4);
> > - n->namespaces = g_new0(NvmeNamespace, n->num_namespaces);
> > n->sq = g_new0(NvmeSQueue *, n->params.num_queues);
> > n->cq = g_new0(NvmeCQueue *, n->params.num_queues);
> >
> > @@ -2483,12 +2527,7 @@ static void nvme_init_ctrl(NvmeCtrl *n)
> > id->cqes = (0x4 << 4) | 0x4;
> > id->nn = cpu_to_le32(n->num_namespaces);
> > id->oncs = cpu_to_le16(NVME_ONCS_WRITE_ZEROS | NVME_ONCS_TIMESTAMP);
> > -
> > -
> > - if (blk_enable_write_cache(n->conf.blk)) {
> > - id->vwc = 1;
> > - }
> > -
> > + id->vwc = 1;
> > id->sgls = cpu_to_le32(0x1);
> >
> > strcpy((char *) id->subnqn, "nqn.2019-08.org.qemu:");
> > @@ -2509,22 +2548,25 @@ static void nvme_init_ctrl(NvmeCtrl *n)
> > n->bar.intmc = n->bar.intms = 0;
> > }
> >
> > -static int nvme_init_namespace(NvmeCtrl *n, NvmeNamespace *ns, Error
> > **errp)
> > +int nvme_register_namespace(NvmeCtrl *n, NvmeNamespace *ns, Error **errp)
> > {
> > - int64_t bs_size;
> > - NvmeIdNs *id_ns = &ns->id_ns;
> > + uint32_t nsid = nvme_nsid(ns);
> >
> > - bs_size = blk_getlength(n->conf.blk);
> > - if (bs_size < 0) {
> > - error_setg_errno(errp, -bs_size, "blk_getlength");
> > + if (nsid == 0 || nsid > NVME_MAX_NAMESPACES) {
> > + error_setg(errp, "invalid nsid");
> > return 1;
> > }
>
> As I said above it would be nice to find a valid namespace slot instead
> of erroring out when nsid == 0.
> Also the error message can be a bit improved IMHO.
>
Done.
> >
> > - id_ns->lbaf[0].ds = BDRV_SECTOR_BITS;
> > - n->ns_size = bs_size;
> > + if (n->namespaces[nsid - 1]) {
> > + error_setg(errp, "nsid must be unique");
> > + return 1;
> > + }
> > +
> > + trace_nvme_dev_register_namespace(nsid);
> >
> > - id_ns->ncap = id_ns->nuse = id_ns->nsze =
> > - cpu_to_le64(nvme_ns_nlbas(n, ns));
> > + n->namespaces[nsid - 1] = ns;
>
> > + n->num_namespaces = MAX(n->num_namespaces, nsid);
> > + n->id_ctrl.nn = cpu_to_le32(n->num_namespaces);
>
> These should be removed once you set num_namespaces to be fixed number.
>
Done.
> >
> > return 0;
> > }
> > @@ -2532,30 +2574,31 @@ static int nvme_init_namespace(NvmeCtrl *n,
> > NvmeNamespace *ns, Error **errp)
> > static void nvme_realize(PCIDevice *pci_dev, Error **errp)
> > {
> > NvmeCtrl *n = NVME(pci_dev);
> > + NvmeNamespace *ns;
> > Error *local_err = NULL;
> > - int i;
> >
> > if (nvme_check_constraints(n, &local_err)) {
> > error_propagate_prepend(errp, local_err, "nvme_check_constraints:
> > ");
> > return;
> > }
> >
> > + qbus_create_inplace(&n->bus, sizeof(NvmeBus), TYPE_NVME_BUS,
> > + &pci_dev->qdev, n->parent_obj.qdev.id);
> > +
> > nvme_init_state(n);
> > -
> > - if (nvme_init_blk(n, &local_err)) {
> > - error_propagate_prepend(errp, local_err, "nvme_init_blk: ");
> > - return;
> > - }
> > -
> > - for (i = 0; i < n->num_namespaces; i++) {
> > - if (nvme_init_namespace(n, &n->namespaces[i], &local_err)) {
> > - error_propagate_prepend(errp, local_err, "nvme_init_namespace:
> > ");
> > - return;
> > - }
> > - }
> > -
> > nvme_init_pci(n, pci_dev);
> > nvme_init_ctrl(n);
> > +
> > + /* setup a namespace if the controller drive property was given */
> > + if (n->namespace.blk) {
> > + ns = &n->namespace;
> > + ns->params.nsid = 1;
> > +
> > + if (nvme_ns_setup(n, ns, &local_err)) {
> > + error_propagate_prepend(errp, local_err, "nvme_ns_setup: ");
> > + return;
> > + }
> > + }
> > }
> >
> > static void nvme_exit(PCIDevice *pci_dev)
> > @@ -2576,7 +2619,8 @@ static void nvme_exit(PCIDevice *pci_dev)
> > }
> >
> > static Property nvme_props[] = {
> > - DEFINE_BLOCK_PROPERTIES(NvmeCtrl, conf),
> > + DEFINE_BLOCK_PROPERTIES_BASE(NvmeCtrl, conf), \
> > + DEFINE_PROP_DRIVE("drive", NvmeCtrl, namespace.blk), \
> > DEFINE_NVME_PROPERTIES(NvmeCtrl, params),
> > DEFINE_PROP_END_OF_LIST(),
> > };
> > @@ -2608,26 +2652,35 @@ static void nvme_instance_init(Object *obj)
> > {
> > NvmeCtrl *s = NVME(obj);
> >
> > - device_add_bootindex_property(obj, &s->conf.bootindex,
> > - "bootindex", "/namespace@1,0",
> > - DEVICE(obj), &error_abort);
> > + if (s->namespace.blk) {
> > + device_add_bootindex_property(obj, &s->conf.bootindex,
> > + "bootindex", "/namespace@1,0",
> > + DEVICE(obj), &error_abort);
> > + }
> > }
> >
> > static const TypeInfo nvme_info = {
> > .name = TYPE_NVME,
> > .parent = TYPE_PCI_DEVICE,
> > .instance_size = sizeof(NvmeCtrl),
> > - .class_init = nvme_class_init,
> > .instance_init = nvme_instance_init,
> > + .class_init = nvme_class_init,
> > .interfaces = (InterfaceInfo[]) {
> > { INTERFACE_PCIE_DEVICE },
> > { }
> > },
> > };
> >
> > +static const TypeInfo nvme_bus_info = {
> > + .name = TYPE_NVME_BUS,
> > + .parent = TYPE_BUS,
> > + .instance_size = sizeof(NvmeBus),
> > +};
> > +
> > static void nvme_register_types(void)
> > {
> > type_register_static(&nvme_info);
> > + type_register_static(&nvme_bus_info);
> > }
> >
> > type_init(nvme_register_types)
> > diff --git a/hw/block/nvme.h b/hw/block/nvme.h
> > index 3319f8edd7e1..c3cef0f024da 100644
> > --- a/hw/block/nvme.h
> > +++ b/hw/block/nvme.h
> > @@ -2,6 +2,9 @@
> > #define HW_NVME_H
> >
> > #include "block/nvme.h"
> > +#include "nvme-ns.h"
> > +
> > +#define NVME_MAX_NAMESPACES 256
> >
> > #define DEFINE_NVME_PROPERTIES(_state, _props) \
> > DEFINE_PROP_STRING("serial", _state, _props.serial), \
> > @@ -108,26 +111,6 @@ typedef struct NvmeCQueue {
> > QTAILQ_HEAD(, NvmeRequest) req_list;
> > } NvmeCQueue;
> >
> > -typedef struct NvmeNamespace {
> > - NvmeIdNs id_ns;
> > -} NvmeNamespace;
> > -
> > -static inline NvmeLBAF nvme_ns_lbaf(NvmeNamespace *ns)
> > -{
> > - NvmeIdNs *id_ns = &ns->id_ns;
> > - return id_ns->lbaf[NVME_ID_NS_FLBAS_INDEX(id_ns->flbas)];
> > -}
> > -
> > -static inline uint8_t nvme_ns_lbads(NvmeNamespace *ns)
> > -{
> > - return nvme_ns_lbaf(ns).ds;
> > -}
> > -
> > -static inline size_t nvme_ns_lbads_bytes(NvmeNamespace *ns)
> > -{
> > - return 1 << nvme_ns_lbads(ns);
> > -}
> > -
> > typedef enum NvmeAIOOp {
> > NVME_AIO_OPC_NONE = 0x0,
> > NVME_AIO_OPC_FLUSH = 0x1,
> > @@ -182,6 +165,13 @@ static inline bool nvme_req_is_write(NvmeRequest *req)
> > }
> > }
> >
> > +#define TYPE_NVME_BUS "nvme-bus"
> > +#define NVME_BUS(obj) OBJECT_CHECK(NvmeBus, (obj), TYPE_NVME_BUS)
> > +
> > +typedef struct NvmeBus {
> > + BusState parent_bus;
> > +} NvmeBus;
> > +
> > #define TYPE_NVME "nvme"
> > #define NVME(obj) \
> > OBJECT_CHECK(NvmeCtrl, (obj), TYPE_NVME)
> > @@ -191,8 +181,9 @@ typedef struct NvmeCtrl {
> > MemoryRegion iomem;
> > MemoryRegion ctrl_mem;
> > NvmeBar bar;
> > - BlockConf conf;
> > NvmeParams params;
> > + NvmeBus bus;
> > + BlockConf conf;
> >
> > bool qs_created;
> > uint32_t page_size;
> > @@ -203,7 +194,6 @@ typedef struct NvmeCtrl {
> > uint32_t reg_size;
> > uint32_t num_namespaces;
> > uint32_t max_q_ents;
> > - uint64_t ns_size;
> > uint8_t outstanding_aers;
> > uint32_t cmbsz;
> > uint32_t cmbloc;
> > @@ -219,7 +209,8 @@ typedef struct NvmeCtrl {
> > QTAILQ_HEAD(, NvmeAsyncEvent) aer_queue;
> > int aer_queued;
> >
> > - NvmeNamespace *namespaces;
> > + NvmeNamespace namespace;
> > + NvmeNamespace *namespaces[NVME_MAX_NAMESPACES];
> > NvmeSQueue **sq;
> > NvmeCQueue **cq;
> > NvmeSQueue admin_sq;
> > @@ -228,9 +219,13 @@ typedef struct NvmeCtrl {
> > NvmeFeatureVal features;
> > } NvmeCtrl;
> >
> > -static inline uint64_t nvme_ns_nlbas(NvmeCtrl *n, NvmeNamespace *ns)
> > +static inline NvmeNamespace *nvme_ns(NvmeCtrl *n, uint32_t nsid)
> > {
> > - return n->ns_size >> nvme_ns_lbads(ns);
> > + if (!nsid || nsid > n->num_namespaces) {
> > + return NULL;
> > + }
> > +
> > + return n->namespaces[nsid - 1];
> > }
> >
> > static inline uint16_t nvme_cid(NvmeRequest *req)
> > @@ -253,4 +248,6 @@ static inline NvmeCtrl *nvme_ctrl(NvmeRequest *req)
> > return req->sq->ctrl;
> > }
> >
> > +int nvme_register_namespace(NvmeCtrl *n, NvmeNamespace *ns, Error **errp);
> > +
> > #endif /* HW_NVME_H */
> > diff --git a/hw/block/trace-events b/hw/block/trace-events
> > index 81d69e15fc32..aaf1fcda7923 100644
> > --- a/hw/block/trace-events
> > +++ b/hw/block/trace-events
> > @@ -29,6 +29,7 @@ hd_geometry_guess(void *blk, uint32_t cyls, uint32_t
> > heads, uint32_t secs, int t
> >
> > # nvme.c
> > # nvme traces for successful events
> > +nvme_dev_register_namespace(uint32_t nsid) "nsid %"PRIu32""
> > nvme_dev_irq_msix(uint32_t vector) "raising MSI-X IRQ vector %u"
> > nvme_dev_irq_pin(void) "pulsing IRQ pin"
> > nvme_dev_irq_masked(void) "IRQ is masked"
> > @@ -38,7 +39,7 @@ nvme_dev_map_sgl(uint16_t cid, uint8_t typ, uint32_t nlb,
> > uint64_t len) "cid %"P
> > nvme_dev_req_register_aio(uint16_t cid, void *aio, const char *blkname,
> > uint64_t offset, uint64_t count, const char *opc, void *req) "cid %"PRIu16"
> > aio %p blk \"%s\" offset %"PRIu64" count
> > %"PRIu64" opc \"%s\" req %p"
> > nvme_dev_aio_cb(uint16_t cid, void *aio, const char *blkname, uint64_t
> > offset, const char *opc, void *req) "cid %"PRIu16" aio %p blk \"%s\" offset
> > %"PRIu64" opc \"%s\" req %p"
> > nvme_dev_io_cmd(uint16_t cid, uint32_t nsid, uint16_t sqid, uint8_t
> > opcode) "cid %"PRIu16" nsid %"PRIu32" sqid %"PRIu16" opc 0x%"PRIx8""
> > -nvme_dev_rw(const char *verb, uint32_t blk_count, uint64_t byte_count,
> > uint64_t lba) "%s %"PRIu32" blocks (%"PRIu64" bytes) from LBA %"PRIu64""
> > +nvme_dev_rw(uint16_t cid, const char *verb, uint32_t nsid, uint32_t nlb,
> > uint64_t count, uint64_t lba) "cid %"PRIu16" %s nsid %"PRIu32" nlb
> > %"PRIu32" count %"PRIu64" lba 0x%"PRIx64""
> > nvme_dev_rw_cb(uint16_t cid, uint32_t nsid) "cid %"PRIu16" nsid %"PRIu32""
> > nvme_dev_write_zeros(uint16_t cid, uint32_t nsid, uint64_t slba, uint32_t
> > nlb) "cid %"PRIu16" nsid %"PRIu32" slba %"PRIu64" nlb %"PRIu32""
> > nvme_dev_create_sq(uint64_t addr, uint16_t sqid, uint16_t cqid, uint16_t
> > qsize, uint16_t qflags) "create submission queue, addr=0x%"PRIx64",
> > sqid=%"PRIu16", cqid=%"PRIu16", qsize=%"PRIu16",
> > qflags=%"PRIu16""
> > @@ -94,7 +95,8 @@ nvme_dev_err_invalid_prplist_ent(uint64_t prplist) "PRP
> > list entry is null or no
> > nvme_dev_err_invalid_prp2_align(uint64_t prp2) "PRP2 is not page aligned:
> > 0x%"PRIx64""
> > nvme_dev_err_invalid_prp2_missing(void) "PRP2 is null and more data to be
> > transferred"
> > nvme_dev_err_invalid_prp(void) "invalid PRP"
> > -nvme_dev_err_invalid_ns(uint32_t ns, uint32_t limit) "invalid namespace %u
> > not within 1-%u"
> > +nvme_dev_err_invalid_ns(uint32_t nsid, uint32_t nn) "nsid %"PRIu32" nn
> > %"PRIu32""
> > +nvme_dev_err_inactive_ns(uint32_t nsid, uint32_t nn) "nsid %"PRIu32" nn
> > %"PRIu32""
> > nvme_dev_err_invalid_opc(uint8_t opc) "invalid opcode 0x%"PRIx8""
> > nvme_dev_err_invalid_admin_opc(uint8_t opc) "invalid admin opcode
> > 0x%"PRIx8""
> > nvme_dev_err_invalid_lba_range(uint64_t start, uint64_t len, uint64_t
> > limit) "Invalid LBA start=%"PRIu64" len=%"PRIu64" limit=%"PRIu64""
>
>
> Best regards,
> Maxim Levitsky
>
>
>
- Re: [PATCH v5 22/26] nvme: support multiple namespaces,
Klaus Birkelund Jensen <=