* [Qemu-devel] [PATCH v7 1/5] block: Support Archipelago as a QEMU block backend
2014-07-23 14:07 [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Chrysostomos Nanakos
@ 2014-07-23 14:07 ` Chrysostomos Nanakos
2014-08-04 14:35 ` [Qemu-devel] [PATCH v8 " Chrysostomos Nanakos
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 2/5] block/archipelago: Implement bdrv_parse_filename() Chrysostomos Nanakos
` (5 subsequent siblings)
6 siblings, 1 reply; 13+ messages in thread
From: Chrysostomos Nanakos @ 2014-07-23 14:07 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, Chrysostomos Nanakos, stefanha
VM Image on Archipelago volume is specified like this:
file.driver=archipelago,file.volume=<volumename>[,file.mport=<mapperd_port>[,
file.vport=<vlmcd_port>][,file.segment=<segment_name>]]
'archipelago' is the protocol.
'mport' is the port number on which mapperd is listening. This is optional
and if not specified, QEMU will make Archipelago to use the default port.
'vport' is the port number on which vlmcd is listening. This is optional
and if not specified, QEMU will make Archipelago to use the default port.
'segment' is the name of the shared memory segment Archipelago stack is using.
This is optional and if not specified, QEMU will make Archipelago to use the
default value, 'archipelago'.
Examples:
file.driver=archipelago,file.volume=my_vm_volume
file.driver=archipelago,file.volume=my_vm_volume,file.mport=123
file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
file.vport=1234
file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
file.vport=1234,file.segment=my_segment
Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
---
MAINTAINERS | 6 +
block/Makefile.objs | 2 +
block/archipelago.c | 785 +++++++++++++++++++++++++++++++++++++++++++++++++++
configure | 40 +++
4 files changed, 833 insertions(+)
create mode 100644 block/archipelago.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 906f252..59940f9 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1000,3 +1000,9 @@ SSH
M: Richard W.M. Jones <rjones@redhat.com>
S: Supported
F: block/ssh.c
+
+ARCHIPELAGO
+M: Chrysostomos Nanakos <cnanakos@grnet.gr>
+M: Chrysostomos Nanakos <chris@include.gr>
+S: Maintained
+F: block/archipelago.c
diff --git a/block/Makefile.objs b/block/Makefile.objs
index fd88c03..858d2b3 100644
--- a/block/Makefile.objs
+++ b/block/Makefile.objs
@@ -17,6 +17,7 @@ block-obj-$(CONFIG_LIBNFS) += nfs.o
block-obj-$(CONFIG_CURL) += curl.o
block-obj-$(CONFIG_RBD) += rbd.o
block-obj-$(CONFIG_GLUSTERFS) += gluster.o
+block-obj-$(CONFIG_ARCHIPELAGO) += archipelago.o
block-obj-$(CONFIG_LIBSSH2) += ssh.o
endif
@@ -35,5 +36,6 @@ gluster.o-cflags := $(GLUSTERFS_CFLAGS)
gluster.o-libs := $(GLUSTERFS_LIBS)
ssh.o-cflags := $(LIBSSH2_CFLAGS)
ssh.o-libs := $(LIBSSH2_LIBS)
+archipelago.o-libs := $(ARCHIPELAGO_LIBS)
qcow.o-libs := -lz
linux-aio.o-libs := -laio
diff --git a/block/archipelago.c b/block/archipelago.c
new file mode 100644
index 0000000..1c21d36
--- /dev/null
+++ b/block/archipelago.c
@@ -0,0 +1,785 @@
+/*
+ * QEMU Block driver for Archipelago
+ *
+ * Copyright (C) 2014 Chrysostomos Nanakos <cnanakos@grnet.gr>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+/*
+ * VM Image on Archipelago volume is specified like this:
+ *
+ * file.driver=archipelago,file.volume=<volumename>
+ * [,file.mport=<mapperd_port>[,file.vport=<vlmcd_port>]
+ * [,file.segment=<segment_name>]]
+ *
+ * 'archipelago' is the protocol.
+ *
+ * 'mport' is the port number on which mapperd is listening. This is optional
+ * and if not specified, QEMU will make Archipelago to use the default port.
+ *
+ * 'vport' is the port number on which vlmcd is listening. This is optional
+ * and if not specified, QEMU will make Archipelago to use the default port.
+ *
+ * 'segment' is the name of the shared memory segment Archipelago stack
+ * is using. This is optional and if not specified, QEMU will make Archipelago
+ * to use the default value, 'archipelago'.
+ *
+ * Examples:
+ *
+ * file.driver=archipelago,file.volume=my_vm_volume
+ * file.driver=archipelago,file.volume=my_vm_volume,file.mport=123
+ * file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
+ * file.vport=1234
+ * file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
+ * file.vport=1234,file.segment=my_segment
+ */
+
+#include "block/block_int.h"
+#include "qemu/error-report.h"
+#include "qemu/thread.h"
+#include "qapi/qmp/qint.h"
+#include "qapi/qmp/qstring.h"
+#include "qapi/qmp/qjson.h"
+
+#include <inttypes.h>
+#include <xseg/xseg.h>
+#include <xseg/protocol.h>
+
+#define ARCHIP_FD_READ 0
+#define ARCHIP_FD_WRITE 1
+#define MAX_REQUEST_SIZE 524288
+
+#define ARCHIPELAGO_OPT_VOLUME "volume"
+#define ARCHIPELAGO_OPT_SEGMENT "segment"
+#define ARCHIPELAGO_OPT_MPORT "mport"
+#define ARCHIPELAGO_OPT_VPORT "vport"
+#define ARCHIPELAGO_DFL_MPORT 1001
+#define ARCHIPELAGO_DFL_VPORT 501
+
+#define archipelagolog(fmt, ...) \
+ do { \
+ fprintf(stderr, "archipelago\t%-24s: " fmt, __func__, ##__VA_ARGS__); \
+ } while (0)
+
+typedef enum {
+ ARCHIP_OP_READ,
+ ARCHIP_OP_WRITE,
+ ARCHIP_OP_FLUSH,
+ ARCHIP_OP_VOLINFO,
+} ARCHIPCmd;
+
+typedef struct ArchipelagoAIOCB {
+ BlockDriverAIOCB common;
+ QEMUBH *bh;
+ struct BDRVArchipelagoState *s;
+ QEMUIOVector *qiov;
+ ARCHIPCmd cmd;
+ bool cancelled;
+ int status;
+ int64_t size;
+ int64_t ret;
+} ArchipelagoAIOCB;
+
+typedef struct BDRVArchipelagoState {
+ ArchipelagoAIOCB *event_acb;
+ char *volname;
+ char *segment_name;
+ uint64_t size;
+ /* Archipelago specific */
+ struct xseg *xseg;
+ struct xseg_port *port;
+ xport srcport;
+ xport sport;
+ xport mportno;
+ xport vportno;
+ QemuMutex archip_mutex;
+ QemuCond archip_cond;
+ bool is_signaled;
+ /* Request handler specific */
+ QemuThread request_th;
+ QemuCond request_cond;
+ QemuMutex request_mutex;
+ bool th_is_signaled;
+ bool stopping;
+} BDRVArchipelagoState;
+
+typedef struct ArchipelagoSegmentedRequest {
+ size_t count;
+ size_t total;
+ int ref;
+ int failed;
+} ArchipelagoSegmentedRequest;
+
+typedef struct AIORequestData {
+ const char *volname;
+ off_t offset;
+ size_t size;
+ uint64_t bufidx;
+ int ret;
+ int op;
+ ArchipelagoAIOCB *aio_cb;
+ ArchipelagoSegmentedRequest *segreq;
+} AIORequestData;
+
+static void qemu_archipelago_complete_aio(void *opaque);
+
+static void init_local_signal(struct xseg *xseg, xport sport, xport srcport)
+{
+ if (xseg && (sport != srcport)) {
+ xseg_init_local_signal(xseg, srcport);
+ sport = srcport;
+ }
+}
+
+static void archipelago_finish_aiocb(AIORequestData *reqdata)
+{
+ if (reqdata->aio_cb->ret != reqdata->segreq->total) {
+ reqdata->aio_cb->ret = -EIO;
+ } else if (reqdata->aio_cb->ret == reqdata->segreq->total) {
+ reqdata->aio_cb->ret = 0;
+ }
+ reqdata->aio_cb->bh = aio_bh_new(
+ bdrv_get_aio_context(reqdata->aio_cb->common.bs),
+ qemu_archipelago_complete_aio, reqdata
+ );
+ qemu_bh_schedule(reqdata->aio_cb->bh);
+}
+
+static int wait_reply(struct xseg *xseg, xport srcport, struct xseg_port *port,
+ struct xseg_request *expected_req)
+{
+ struct xseg_request *req;
+ xseg_prepare_wait(xseg, srcport);
+ void *psd = xseg_get_signal_desc(xseg, port);
+ while (1) {
+ req = xseg_receive(xseg, srcport, X_NONBLOCK);
+ if (req) {
+ if (req != expected_req) {
+ archipelagolog("Unknown received request\n");
+ xseg_put_request(xseg, req, srcport);
+ } else if (!(req->state & XS_SERVED)) {
+ return -1;
+ } else {
+ break;
+ }
+ }
+ xseg_wait_signal(xseg, psd, 100000UL);
+ }
+ xseg_cancel_wait(xseg, srcport);
+ return 0;
+}
+
+static void xseg_request_handler(void *state)
+{
+ BDRVArchipelagoState *s = (BDRVArchipelagoState *) state;
+ void *psd = xseg_get_signal_desc(s->xseg, s->port);
+ qemu_mutex_lock(&s->request_mutex);
+
+ while (!s->stopping) {
+ struct xseg_request *req;
+ void *data;
+ xseg_prepare_wait(s->xseg, s->srcport);
+ req = xseg_receive(s->xseg, s->srcport, X_NONBLOCK);
+ if (req) {
+ AIORequestData *reqdata;
+ ArchipelagoSegmentedRequest *segreq;
+ xseg_get_req_data(s->xseg, req, (void **)&reqdata);
+
+ switch (reqdata->op) {
+ case ARCHIP_OP_READ:
+ data = xseg_get_data(s->xseg, req);
+ segreq = reqdata->segreq;
+ segreq->count += req->serviced;
+
+ qemu_iovec_from_buf(reqdata->aio_cb->qiov, reqdata->bufidx,
+ data,
+ req->serviced);
+
+ xseg_put_request(s->xseg, req, s->srcport);
+
+ if ((__sync_add_and_fetch(&segreq->ref, -1)) == 0) {
+ if (!segreq->failed) {
+ reqdata->aio_cb->ret = segreq->count;
+ archipelago_finish_aiocb(reqdata);
+ g_free(segreq);
+ } else {
+ g_free(segreq);
+ g_free(reqdata);
+ }
+ } else {
+ g_free(reqdata);
+ }
+ break;
+ case ARCHIP_OP_WRITE:
+ case ARCHIP_OP_FLUSH:
+ segreq = reqdata->segreq;
+ segreq->count += req->serviced;
+ xseg_put_request(s->xseg, req, s->srcport);
+
+ if ((__sync_add_and_fetch(&segreq->ref, -1)) == 0) {
+ if (!segreq->failed) {
+ reqdata->aio_cb->ret = segreq->count;
+ archipelago_finish_aiocb(reqdata);
+ g_free(segreq);
+ } else {
+ g_free(segreq);
+ g_free(reqdata);
+ }
+ } else {
+ g_free(reqdata);
+ }
+ break;
+ case ARCHIP_OP_VOLINFO:
+ s->is_signaled = true;
+ qemu_cond_signal(&s->archip_cond);
+ break;
+ }
+ } else {
+ xseg_wait_signal(s->xseg, psd, 100000UL);
+ }
+ xseg_cancel_wait(s->xseg, s->srcport);
+ }
+
+ s->th_is_signaled = true;
+ qemu_cond_signal(&s->request_cond);
+ qemu_mutex_unlock(&s->request_mutex);
+ qemu_thread_exit(NULL);
+}
+
+static int qemu_archipelago_xseg_init(BDRVArchipelagoState *s)
+{
+ if (xseg_initialize()) {
+ archipelagolog("Cannot initialize XSEG\n");
+ goto err_exit;
+ }
+
+ s->xseg = xseg_join("posix", s->segment_name,
+ "posixfd", NULL);
+ if (!s->xseg) {
+ archipelagolog("Cannot join XSEG shared memory segment\n");
+ goto err_exit;
+ }
+ s->port = xseg_bind_dynport(s->xseg);
+ s->srcport = s->port->portno;
+ init_local_signal(s->xseg, s->sport, s->srcport);
+ return 0;
+
+err_exit:
+ return -1;
+}
+
+static int qemu_archipelago_init(BDRVArchipelagoState *s)
+{
+ int ret;
+
+ ret = qemu_archipelago_xseg_init(s);
+ if (ret < 0) {
+ error_report("Cannot initialize XSEG. Aborting...\n");
+ goto err_exit;
+ }
+
+ qemu_cond_init(&s->archip_cond);
+ qemu_mutex_init(&s->archip_mutex);
+ qemu_cond_init(&s->request_cond);
+ qemu_mutex_init(&s->request_mutex);
+ s->th_is_signaled = false;
+ qemu_thread_create(&s->request_th, "xseg_io_th",
+ (void *) xseg_request_handler,
+ (void *) s, QEMU_THREAD_JOINABLE);
+
+err_exit:
+ return ret;
+}
+
+static void qemu_archipelago_complete_aio(void *opaque)
+{
+ AIORequestData *reqdata = (AIORequestData *) opaque;
+ ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) reqdata->aio_cb;
+
+ qemu_bh_delete(aio_cb->bh);
+ aio_cb->common.cb(aio_cb->common.opaque, aio_cb->ret);
+ aio_cb->status = 0;
+
+ if (!aio_cb->cancelled) {
+ qemu_aio_release(aio_cb);
+ }
+ g_free(reqdata);
+}
+
+static QemuOptsList archipelago_runtime_opts = {
+ .name = "archipelago",
+ .head = QTAILQ_HEAD_INITIALIZER(archipelago_runtime_opts.head),
+ .desc = {
+ {
+ .name = ARCHIPELAGO_OPT_VOLUME,
+ .type = QEMU_OPT_STRING,
+ .help = "Name of the volume image",
+ },
+ {
+ .name = ARCHIPELAGO_OPT_SEGMENT,
+ .type = QEMU_OPT_STRING,
+ .help = "Name of the Archipelago shared memory segment",
+ },
+ {
+ .name = ARCHIPELAGO_OPT_MPORT,
+ .type = QEMU_OPT_NUMBER,
+ .help = "Archipelago mapperd port number"
+ },
+ {
+ .name = ARCHIPELAGO_OPT_VPORT,
+ .type = QEMU_OPT_NUMBER,
+ .help = "Archipelago vlmcd port number"
+
+ },
+ { /* end of list */ }
+ },
+};
+
+static int qemu_archipelago_open(BlockDriverState *bs,
+ QDict *options,
+ int bdrv_flags,
+ Error **errp)
+{
+ int ret = 0;
+ const char *volume, *segment_name;
+ QemuOpts *opts;
+ Error *local_err = NULL;
+ BDRVArchipelagoState *s = bs->opaque;
+
+ opts = qemu_opts_create(&archipelago_runtime_opts, NULL, 0, &error_abort);
+ qemu_opts_absorb_qdict(opts, options, &local_err);
+ if (local_err) {
+ error_propagate(errp, local_err);
+ ret = -EINVAL;
+ goto err_exit;
+ }
+
+ s->mportno = qemu_opt_get_number(opts, ARCHIPELAGO_OPT_MPORT,
+ ARCHIPELAGO_DFL_MPORT);
+ s->vportno = qemu_opt_get_number(opts, ARCHIPELAGO_OPT_VPORT,
+ ARCHIPELAGO_DFL_VPORT);
+
+ segment_name = qemu_opt_get(opts, ARCHIPELAGO_OPT_SEGMENT);
+ if (segment_name == NULL) {
+ s->segment_name = g_strdup("archipelago");
+ } else {
+ s->segment_name = g_strdup(segment_name);
+ }
+
+ volume = qemu_opt_get(opts, ARCHIPELAGO_OPT_VOLUME);
+ if (volume == NULL) {
+ error_setg(errp, "archipelago block driver requires the 'volume'"
+ " option");
+ ret = -EINVAL;
+ goto err_exit;
+ }
+ s->volname = g_strdup(volume);
+
+ /* Initialize XSEG, join shared memory segment */
+ ret = qemu_archipelago_init(s);
+ if (ret < 0) {
+ error_setg(errp, "cannot initialize XSEG and join shared "
+ "memory segment");
+ goto err_exit;
+ }
+
+ qemu_opts_del(opts);
+ return 0;
+
+err_exit:
+ g_free(s->volname);
+ g_free(s->segment_name);
+ qemu_opts_del(opts);
+ return ret;
+}
+
+static void qemu_archipelago_close(BlockDriverState *bs)
+{
+ int r, targetlen;
+ char *target;
+ struct xseg_request *req;
+ BDRVArchipelagoState *s = bs->opaque;
+
+ s->stopping = true;
+
+ qemu_mutex_lock(&s->request_mutex);
+ while (!s->th_is_signaled) {
+ qemu_cond_wait(&s->request_cond,
+ &s->request_mutex);
+ }
+ qemu_mutex_unlock(&s->request_mutex);
+ qemu_thread_join(&s->request_th);
+ qemu_cond_destroy(&s->request_cond);
+ qemu_mutex_destroy(&s->request_mutex);
+
+ qemu_cond_destroy(&s->archip_cond);
+ qemu_mutex_destroy(&s->archip_mutex);
+
+ targetlen = strlen(s->volname);
+ req = xseg_get_request(s->xseg, s->srcport, s->vportno, X_ALLOC);
+ if (!req) {
+ archipelagolog("Cannot get XSEG request\n");
+ goto err_exit;
+ }
+ r = xseg_prep_request(s->xseg, req, targetlen, 0);
+ if (r < 0) {
+ xseg_put_request(s->xseg, req, s->srcport);
+ archipelagolog("Cannot prepare XSEG close request\n");
+ goto err_exit;
+ }
+
+ target = xseg_get_target(s->xseg, req);
+ memcpy(target, s->volname, targetlen);
+ req->size = req->datalen;
+ req->offset = 0;
+ req->op = X_CLOSE;
+
+ xport p = xseg_submit(s->xseg, req, s->srcport, X_ALLOC);
+ if (p == NoPort) {
+ xseg_put_request(s->xseg, req, s->srcport);
+ archipelagolog("Cannot submit XSEG close request\n");
+ goto err_exit;
+ }
+
+ xseg_signal(s->xseg, p);
+ wait_reply(s->xseg, s->srcport, s->port, req);
+
+ xseg_put_request(s->xseg, req, s->srcport);
+
+err_exit:
+ g_free(s->volname);
+ g_free(s->segment_name);
+ xseg_quit_local_signal(s->xseg, s->srcport);
+ xseg_leave_dynport(s->xseg, s->port);
+ xseg_leave(s->xseg);
+}
+
+static void qemu_archipelago_aio_cancel(BlockDriverAIOCB *blockacb)
+{
+ ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) blockacb;
+ aio_cb->cancelled = true;
+ while (aio_cb->status == -EINPROGRESS) {
+ aio_poll(bdrv_get_aio_context(aio_cb->common.bs), true);
+ }
+ qemu_aio_release(aio_cb);
+}
+
+static const AIOCBInfo archipelago_aiocb_info = {
+ .aiocb_size = sizeof(ArchipelagoAIOCB),
+ .cancel = qemu_archipelago_aio_cancel,
+};
+
+static int archipelago_submit_request(BDRVArchipelagoState *s,
+ uint64_t bufidx,
+ size_t count,
+ off_t offset,
+ ArchipelagoAIOCB *aio_cb,
+ ArchipelagoSegmentedRequest *segreq,
+ int op)
+{
+ int ret, targetlen;
+ char *target;
+ void *data = NULL;
+ struct xseg_request *req;
+ AIORequestData *reqdata = g_malloc(sizeof(AIORequestData));
+
+ targetlen = strlen(s->volname);
+ req = xseg_get_request(s->xseg, s->srcport, s->vportno, X_ALLOC);
+ if (!req) {
+ archipelagolog("Cannot get XSEG request\n");
+ goto err_exit2;
+ }
+ ret = xseg_prep_request(s->xseg, req, targetlen, count);
+ if (ret < 0) {
+ archipelagolog("Cannot prepare XSEG request\n");
+ goto err_exit;
+ }
+ target = xseg_get_target(s->xseg, req);
+ if (!target) {
+ archipelagolog("Cannot get XSEG target\n");
+ goto err_exit;
+ }
+ memcpy(target, s->volname, targetlen);
+ req->size = count;
+ req->offset = offset;
+
+ switch (op) {
+ case ARCHIP_OP_READ:
+ req->op = X_READ;
+ break;
+ case ARCHIP_OP_WRITE:
+ case ARCHIP_OP_FLUSH:
+ req->op = X_WRITE;
+ break;
+ }
+ reqdata->volname = s->volname;
+ reqdata->offset = offset;
+ reqdata->size = count;
+ reqdata->bufidx = bufidx;
+ reqdata->aio_cb = aio_cb;
+ reqdata->segreq = segreq;
+ reqdata->op = op;
+
+ xseg_set_req_data(s->xseg, req, reqdata);
+ if (op == ARCHIP_OP_WRITE) {
+ data = xseg_get_data(s->xseg, req);
+ if (!data) {
+ archipelagolog("Cannot get XSEG data\n");
+ goto err_exit;
+ }
+ qemu_iovec_to_buf(aio_cb->qiov, bufidx, data, count);
+ }
+
+ xport p = xseg_submit(s->xseg, req, s->srcport, X_ALLOC);
+ if (p == NoPort) {
+ archipelagolog("Could not submit XSEG request\n");
+ goto err_exit;
+ }
+ xseg_signal(s->xseg, p);
+ return 0;
+
+err_exit:
+ g_free(reqdata);
+ xseg_put_request(s->xseg, req, s->srcport);
+ return -EIO;
+err_exit2:
+ g_free(reqdata);
+ return -EIO;
+}
+
+static int archipelago_aio_segmented_rw(BDRVArchipelagoState *s,
+ size_t count,
+ off_t offset,
+ ArchipelagoAIOCB *aio_cb,
+ int op)
+{
+ int i, ret, segments_nr, last_segment_size;
+ ArchipelagoSegmentedRequest *segreq;
+
+ segreq = g_malloc(sizeof(ArchipelagoSegmentedRequest));
+
+ if (op == ARCHIP_OP_FLUSH) {
+ segments_nr = 1;
+ segreq->ref = segments_nr;
+ segreq->total = count;
+ segreq->count = 0;
+ segreq->failed = 0;
+ ret = archipelago_submit_request(s, 0, count, offset, aio_cb,
+ segreq, ARCHIP_OP_FLUSH);
+ if (ret < 0) {
+ goto err_exit;
+ }
+ return 0;
+ }
+
+ segments_nr = (int)(count / MAX_REQUEST_SIZE) + \
+ ((count % MAX_REQUEST_SIZE) ? 1 : 0);
+ last_segment_size = (int)(count % MAX_REQUEST_SIZE);
+
+ segreq->ref = segments_nr;
+ segreq->total = count;
+ segreq->count = 0;
+ segreq->failed = 0;
+
+ for (i = 0; i < segments_nr - 1; i++) {
+ ret = archipelago_submit_request(s, i * MAX_REQUEST_SIZE,
+ MAX_REQUEST_SIZE,
+ offset + i * MAX_REQUEST_SIZE,
+ aio_cb, segreq, op);
+
+ if (ret < 0) {
+ goto err_exit;
+ }
+ }
+
+ if ((segments_nr > 1) && last_segment_size) {
+ ret = archipelago_submit_request(s, i * MAX_REQUEST_SIZE,
+ last_segment_size,
+ offset + i * MAX_REQUEST_SIZE,
+ aio_cb, segreq, op);
+ } else if ((segments_nr > 1) && !last_segment_size) {
+ ret = archipelago_submit_request(s, i * MAX_REQUEST_SIZE,
+ MAX_REQUEST_SIZE,
+ offset + i * MAX_REQUEST_SIZE,
+ aio_cb, segreq, op);
+ } else if (segments_nr == 1) {
+ ret = archipelago_submit_request(s, 0, count, offset, aio_cb,
+ segreq, op);
+ }
+
+ if (ret < 0) {
+ goto err_exit;
+ }
+
+ return 0;
+
+err_exit:
+ __sync_add_and_fetch(&segreq->failed, 1);
+ if (segments_nr == 1) {
+ if (__sync_add_and_fetch(&segreq->ref, -1) == 0) {
+ g_free(segreq);
+ }
+ } else {
+ if ((__sync_add_and_fetch(&segreq->ref, -segments_nr + i)) == 0) {
+ g_free(segreq);
+ }
+ }
+
+ return ret;
+}
+
+static BlockDriverAIOCB *qemu_archipelago_aio_rw(BlockDriverState *bs,
+ int64_t sector_num,
+ QEMUIOVector *qiov,
+ int nb_sectors,
+ BlockDriverCompletionFunc *cb,
+ void *opaque,
+ int op)
+{
+ ArchipelagoAIOCB *aio_cb;
+ BDRVArchipelagoState *s = bs->opaque;
+ int64_t size, off;
+ int ret;
+
+ aio_cb = qemu_aio_get(&archipelago_aiocb_info, bs, cb, opaque);
+ aio_cb->cmd = op;
+ aio_cb->qiov = qiov;
+
+ aio_cb->ret = 0;
+ aio_cb->s = s;
+ aio_cb->cancelled = false;
+ aio_cb->status = -EINPROGRESS;
+
+ off = sector_num * BDRV_SECTOR_SIZE;
+ size = nb_sectors * BDRV_SECTOR_SIZE;
+ aio_cb->size = size;
+
+ ret = archipelago_aio_segmented_rw(s, size, off,
+ aio_cb, op);
+ if (ret < 0) {
+ goto err_exit;
+ }
+ return &aio_cb->common;
+
+err_exit:
+ error_report("qemu_archipelago_aio_rw(): I/O Error\n");
+ qemu_aio_release(aio_cb);
+ return NULL;
+}
+
+static BlockDriverAIOCB *qemu_archipelago_aio_readv(BlockDriverState *bs,
+ int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+ BlockDriverCompletionFunc *cb, void *opaque)
+{
+ return qemu_archipelago_aio_rw(bs, sector_num, qiov, nb_sectors, cb,
+ opaque, ARCHIP_OP_READ);
+}
+
+static BlockDriverAIOCB *qemu_archipelago_aio_writev(BlockDriverState *bs,
+ int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+ BlockDriverCompletionFunc *cb, void *opaque)
+{
+ return qemu_archipelago_aio_rw(bs, sector_num, qiov, nb_sectors, cb,
+ opaque, ARCHIP_OP_WRITE);
+}
+
+static int64_t archipelago_volume_info(BDRVArchipelagoState *s)
+{
+ uint64_t size;
+ int ret, targetlen;
+ struct xseg_request *req;
+ struct xseg_reply_info *xinfo;
+ AIORequestData *reqdata = g_malloc(sizeof(AIORequestData));
+
+ const char *volname = s->volname;
+ targetlen = strlen(volname);
+ req = xseg_get_request(s->xseg, s->srcport, s->mportno, X_ALLOC);
+ if (!req) {
+ archipelagolog("Cannot get XSEG request\n");
+ goto err_exit2;
+ }
+ ret = xseg_prep_request(s->xseg, req, targetlen,
+ sizeof(struct xseg_reply_info));
+ if (ret < 0) {
+ archipelagolog("Cannot prepare XSEG request\n");
+ goto err_exit;
+ }
+ char *target = xseg_get_target(s->xseg, req);
+ if (!target) {
+ archipelagolog("Cannot get XSEG target\n");
+ goto err_exit;
+ }
+ memcpy(target, volname, targetlen);
+ req->size = req->datalen;
+ req->offset = 0;
+ req->op = X_INFO;
+
+ reqdata->op = ARCHIP_OP_VOLINFO;
+ reqdata->volname = volname;
+ xseg_set_req_data(s->xseg, req, reqdata);
+
+ xport p = xseg_submit(s->xseg, req, s->srcport, X_ALLOC);
+ if (p == NoPort) {
+ archipelagolog("Cannot submit XSEG request\n");
+ goto err_exit;
+ }
+ xseg_signal(s->xseg, p);
+ qemu_mutex_lock(&s->archip_mutex);
+ while (!s->is_signaled) {
+ qemu_cond_wait(&s->archip_cond, &s->archip_mutex);
+ }
+ s->is_signaled = false;
+ qemu_mutex_unlock(&s->archip_mutex);
+
+ xinfo = (struct xseg_reply_info *) xseg_get_data(s->xseg, req);
+ size = xinfo->size;
+ xseg_put_request(s->xseg, req, s->srcport);
+ g_free(reqdata);
+ s->size = size;
+ return size;
+
+err_exit:
+ xseg_put_request(s->xseg, req, s->srcport);
+err_exit2:
+ g_free(reqdata);
+ return -EIO;
+}
+
+static int64_t qemu_archipelago_getlength(BlockDriverState *bs)
+{
+ int64_t ret;
+ BDRVArchipelagoState *s = bs->opaque;
+
+ ret = archipelago_volume_info(s);
+ return ret;
+}
+
+static BlockDriverAIOCB *qemu_archipelago_aio_flush(BlockDriverState *bs,
+ BlockDriverCompletionFunc *cb, void *opaque)
+{
+ return qemu_archipelago_aio_rw(bs, 0, NULL, 0, cb, opaque,
+ ARCHIP_OP_FLUSH);
+}
+
+static BlockDriver bdrv_archipelago = {
+ .format_name = "archipelago",
+ .protocol_name = "archipelago",
+ .instance_size = sizeof(BDRVArchipelagoState),
+ .bdrv_file_open = qemu_archipelago_open,
+ .bdrv_close = qemu_archipelago_close,
+ .bdrv_getlength = qemu_archipelago_getlength,
+ .bdrv_aio_readv = qemu_archipelago_aio_readv,
+ .bdrv_aio_writev = qemu_archipelago_aio_writev,
+ .bdrv_aio_flush = qemu_archipelago_aio_flush,
+ .bdrv_has_zero_init = bdrv_has_zero_init_1,
+};
+
+static void bdrv_archipelago_init(void)
+{
+ bdrv_register(&bdrv_archipelago);
+}
+
+block_init(bdrv_archipelago_init);
diff --git a/configure b/configure
index f7685b5..8a5cc5e 100755
--- a/configure
+++ b/configure
@@ -326,6 +326,7 @@ seccomp=""
glusterfs=""
glusterfs_discard="no"
glusterfs_zerofill="no"
+archipelago=""
virtio_blk_data_plane=""
gtk=""
gtkabi=""
@@ -1087,6 +1088,10 @@ for opt do
;;
--enable-glusterfs) glusterfs="yes"
;;
+ --disable-archipelago) archipelago="no"
+ ;;
+ --enable-archipelago) archipelago="yes"
+ ;;
--disable-virtio-blk-data-plane) virtio_blk_data_plane="no"
;;
--enable-virtio-blk-data-plane) virtio_blk_data_plane="yes"
@@ -1382,6 +1387,8 @@ Advanced options (experts only):
--enable-coroutine-pool enable coroutine freelist (better performance)
--enable-glusterfs enable GlusterFS backend
--disable-glusterfs disable GlusterFS backend
+ --enable-archipelago enable Archipelago backend
+ --disable-archipelago disable Archipelago backend
--enable-gcov enable test coverage analysis with gcov
--gcov=GCOV use specified gcov [$gcov_tool]
--disable-tpm disable TPM support
@@ -3071,6 +3078,33 @@ EOF
fi
fi
+
+##########################################
+# archipelago probe
+if test "$archipelago" != "no" ; then
+ cat > $TMPC <<EOF
+#include <stdio.h>
+#include <xseg/xseg.h>
+#include <xseg/protocol.h>
+int main(void) {
+ xseg_initialize();
+ return 0;
+}
+EOF
+ archipelago_libs=-lxseg
+ if compile_prog "" "$archipelago_libs"; then
+ archipelago="yes"
+ libs_tools="$archipelago_libs $libs_tools"
+ libs_softmmu="$archipelago_libs $libs_softmmu"
+ else
+ if test "$archipelago" = "yes" ; then
+ feature_not_found "Archipelago backend support" "Install libxseg devel"
+ fi
+ archipelago="no"
+ fi
+fi
+
+
##########################################
# glusterfs probe
if test "$glusterfs" != "no" ; then
@@ -4250,6 +4284,7 @@ echo "seccomp support $seccomp"
echo "coroutine backend $coroutine"
echo "coroutine pool $coroutine_pool"
echo "GlusterFS support $glusterfs"
+echo "Archipelago support $archipelago"
echo "virtio-blk-data-plane $virtio_blk_data_plane"
echo "gcov $gcov_tool"
echo "gcov enabled $gcov"
@@ -4688,6 +4723,11 @@ if test "$glusterfs_zerofill" = "yes" ; then
echo "CONFIG_GLUSTERFS_ZEROFILL=y" >> $config_host_mak
fi
+if test "$archipelago" = "yes" ; then
+ echo "CONFIG_ARCHIPELAGO=m" >> $config_host_mak
+ echo "ARCHIPELAGO_LIBS=$archipelago_libs" >> $config_host_mak
+fi
+
if test "$libssh2" = "yes" ; then
echo "CONFIG_LIBSSH2=m" >> $config_host_mak
echo "LIBSSH2_CFLAGS=$libssh2_cflags" >> $config_host_mak
--
1.7.10.4
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PATCH v8 1/5] block: Support Archipelago as a QEMU block backend
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 1/5] block: " Chrysostomos Nanakos
@ 2014-08-04 14:35 ` Chrysostomos Nanakos
2014-08-05 8:20 ` Stefan Hajnoczi
0 siblings, 1 reply; 13+ messages in thread
From: Chrysostomos Nanakos @ 2014-08-04 14:35 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, Chrysostomos Nanakos, stefanha
VM Image on Archipelago volume is specified like this:
file.driver=archipelago,file.volume=<volumename>[,file.mport=<mapperd_port>[,
file.vport=<vlmcd_port>][,file.segment=<segment_name>]]
'archipelago' is the protocol.
'mport' is the port number on which mapperd is listening. This is optional
and if not specified, QEMU will make Archipelago to use the default port.
'vport' is the port number on which vlmcd is listening. This is optional
and if not specified, QEMU will make Archipelago to use the default port.
'segment' is the name of the shared memory segment Archipelago stack is using.
This is optional and if not specified, QEMU will make Archipelago to use the
default value, 'archipelago'.
Examples:
file.driver=archipelago,file.volume=my_vm_volume
file.driver=archipelago,file.volume=my_vm_volume,file.mport=123
file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
file.vport=1234
file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
file.vport=1234,file.segment=my_segment
Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
---
MAINTAINERS | 6 +
block/Makefile.objs | 2 +
block/archipelago.c | 787 +++++++++++++++++++++++++++++++++++++++++++++++++++
configure | 40 +++
4 files changed, 835 insertions(+)
create mode 100644 block/archipelago.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 906f252..59940f9 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1000,3 +1000,9 @@ SSH
M: Richard W.M. Jones <rjones@redhat.com>
S: Supported
F: block/ssh.c
+
+ARCHIPELAGO
+M: Chrysostomos Nanakos <cnanakos@grnet.gr>
+M: Chrysostomos Nanakos <chris@include.gr>
+S: Maintained
+F: block/archipelago.c
diff --git a/block/Makefile.objs b/block/Makefile.objs
index fd88c03..858d2b3 100644
--- a/block/Makefile.objs
+++ b/block/Makefile.objs
@@ -17,6 +17,7 @@ block-obj-$(CONFIG_LIBNFS) += nfs.o
block-obj-$(CONFIG_CURL) += curl.o
block-obj-$(CONFIG_RBD) += rbd.o
block-obj-$(CONFIG_GLUSTERFS) += gluster.o
+block-obj-$(CONFIG_ARCHIPELAGO) += archipelago.o
block-obj-$(CONFIG_LIBSSH2) += ssh.o
endif
@@ -35,5 +36,6 @@ gluster.o-cflags := $(GLUSTERFS_CFLAGS)
gluster.o-libs := $(GLUSTERFS_LIBS)
ssh.o-cflags := $(LIBSSH2_CFLAGS)
ssh.o-libs := $(LIBSSH2_LIBS)
+archipelago.o-libs := $(ARCHIPELAGO_LIBS)
qcow.o-libs := -lz
linux-aio.o-libs := -laio
diff --git a/block/archipelago.c b/block/archipelago.c
new file mode 100644
index 0000000..d91872d
--- /dev/null
+++ b/block/archipelago.c
@@ -0,0 +1,787 @@
+/*
+ * QEMU Block driver for Archipelago
+ *
+ * Copyright (C) 2014 Chrysostomos Nanakos <cnanakos@grnet.gr>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+/*
+ * VM Image on Archipelago volume is specified like this:
+ *
+ * file.driver=archipelago,file.volume=<volumename>
+ * [,file.mport=<mapperd_port>[,file.vport=<vlmcd_port>]
+ * [,file.segment=<segment_name>]]
+ *
+ * 'archipelago' is the protocol.
+ *
+ * 'mport' is the port number on which mapperd is listening. This is optional
+ * and if not specified, QEMU will make Archipelago to use the default port.
+ *
+ * 'vport' is the port number on which vlmcd is listening. This is optional
+ * and if not specified, QEMU will make Archipelago to use the default port.
+ *
+ * 'segment' is the name of the shared memory segment Archipelago stack
+ * is using. This is optional and if not specified, QEMU will make Archipelago
+ * to use the default value, 'archipelago'.
+ *
+ * Examples:
+ *
+ * file.driver=archipelago,file.volume=my_vm_volume
+ * file.driver=archipelago,file.volume=my_vm_volume,file.mport=123
+ * file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
+ * file.vport=1234
+ * file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
+ * file.vport=1234,file.segment=my_segment
+ */
+
+#include "block/block_int.h"
+#include "qemu/error-report.h"
+#include "qemu/thread.h"
+#include "qapi/qmp/qint.h"
+#include "qapi/qmp/qstring.h"
+#include "qapi/qmp/qjson.h"
+
+#include <inttypes.h>
+#include <xseg/xseg.h>
+#include <xseg/protocol.h>
+
+#define ARCHIP_FD_READ 0
+#define ARCHIP_FD_WRITE 1
+#define MAX_REQUEST_SIZE 524288
+
+#define ARCHIPELAGO_OPT_VOLUME "volume"
+#define ARCHIPELAGO_OPT_SEGMENT "segment"
+#define ARCHIPELAGO_OPT_MPORT "mport"
+#define ARCHIPELAGO_OPT_VPORT "vport"
+#define ARCHIPELAGO_DFL_MPORT 1001
+#define ARCHIPELAGO_DFL_VPORT 501
+
+#define archipelagolog(fmt, ...) \
+ do { \
+ fprintf(stderr, "archipelago\t%-24s: " fmt, __func__, ##__VA_ARGS__); \
+ } while (0)
+
+typedef enum {
+ ARCHIP_OP_READ,
+ ARCHIP_OP_WRITE,
+ ARCHIP_OP_FLUSH,
+ ARCHIP_OP_VOLINFO,
+} ARCHIPCmd;
+
+typedef struct ArchipelagoAIOCB {
+ BlockDriverAIOCB common;
+ QEMUBH *bh;
+ struct BDRVArchipelagoState *s;
+ QEMUIOVector *qiov;
+ ARCHIPCmd cmd;
+ bool cancelled;
+ int status;
+ int64_t size;
+ int64_t ret;
+} ArchipelagoAIOCB;
+
+typedef struct BDRVArchipelagoState {
+ ArchipelagoAIOCB *event_acb;
+ char *volname;
+ char *segment_name;
+ uint64_t size;
+ /* Archipelago specific */
+ struct xseg *xseg;
+ struct xseg_port *port;
+ xport srcport;
+ xport sport;
+ xport mportno;
+ xport vportno;
+ QemuMutex archip_mutex;
+ QemuCond archip_cond;
+ bool is_signaled;
+ /* Request handler specific */
+ QemuThread request_th;
+ QemuCond request_cond;
+ QemuMutex request_mutex;
+ bool th_is_signaled;
+ bool stopping;
+} BDRVArchipelagoState;
+
+typedef struct ArchipelagoSegmentedRequest {
+ size_t count;
+ size_t total;
+ int ref;
+ int failed;
+} ArchipelagoSegmentedRequest;
+
+typedef struct AIORequestData {
+ const char *volname;
+ off_t offset;
+ size_t size;
+ uint64_t bufidx;
+ int ret;
+ int op;
+ ArchipelagoAIOCB *aio_cb;
+ ArchipelagoSegmentedRequest *segreq;
+} AIORequestData;
+
+static void qemu_archipelago_complete_aio(void *opaque);
+
+static void init_local_signal(struct xseg *xseg, xport sport, xport srcport)
+{
+ if (xseg && (sport != srcport)) {
+ xseg_init_local_signal(xseg, srcport);
+ sport = srcport;
+ }
+}
+
+static void archipelago_finish_aiocb(AIORequestData *reqdata)
+{
+ if (reqdata->aio_cb->ret != reqdata->segreq->total) {
+ reqdata->aio_cb->ret = -EIO;
+ } else if (reqdata->aio_cb->ret == reqdata->segreq->total) {
+ reqdata->aio_cb->ret = 0;
+ }
+ reqdata->aio_cb->bh = aio_bh_new(
+ bdrv_get_aio_context(reqdata->aio_cb->common.bs),
+ qemu_archipelago_complete_aio, reqdata
+ );
+ qemu_bh_schedule(reqdata->aio_cb->bh);
+}
+
+static int wait_reply(struct xseg *xseg, xport srcport, struct xseg_port *port,
+ struct xseg_request *expected_req)
+{
+ struct xseg_request *req;
+ xseg_prepare_wait(xseg, srcport);
+ void *psd = xseg_get_signal_desc(xseg, port);
+ while (1) {
+ req = xseg_receive(xseg, srcport, X_NONBLOCK);
+ if (req) {
+ if (req != expected_req) {
+ archipelagolog("Unknown received request\n");
+ xseg_put_request(xseg, req, srcport);
+ } else if (!(req->state & XS_SERVED)) {
+ return -1;
+ } else {
+ break;
+ }
+ }
+ xseg_wait_signal(xseg, psd, 100000UL);
+ }
+ xseg_cancel_wait(xseg, srcport);
+ return 0;
+}
+
+static void xseg_request_handler(void *state)
+{
+ BDRVArchipelagoState *s = (BDRVArchipelagoState *) state;
+ void *psd = xseg_get_signal_desc(s->xseg, s->port);
+ qemu_mutex_lock(&s->request_mutex);
+
+ while (!s->stopping) {
+ struct xseg_request *req;
+ void *data;
+ xseg_prepare_wait(s->xseg, s->srcport);
+ req = xseg_receive(s->xseg, s->srcport, X_NONBLOCK);
+ if (req) {
+ AIORequestData *reqdata;
+ ArchipelagoSegmentedRequest *segreq;
+ xseg_get_req_data(s->xseg, req, (void **)&reqdata);
+
+ switch (reqdata->op) {
+ case ARCHIP_OP_READ:
+ data = xseg_get_data(s->xseg, req);
+ segreq = reqdata->segreq;
+ segreq->count += req->serviced;
+
+ qemu_iovec_from_buf(reqdata->aio_cb->qiov, reqdata->bufidx,
+ data,
+ req->serviced);
+
+ xseg_put_request(s->xseg, req, s->srcport);
+
+ if ((__sync_add_and_fetch(&segreq->ref, -1)) == 0) {
+ if (!segreq->failed) {
+ reqdata->aio_cb->ret = segreq->count;
+ archipelago_finish_aiocb(reqdata);
+ g_free(segreq);
+ } else {
+ g_free(segreq);
+ g_free(reqdata);
+ }
+ } else {
+ g_free(reqdata);
+ }
+ break;
+ case ARCHIP_OP_WRITE:
+ case ARCHIP_OP_FLUSH:
+ segreq = reqdata->segreq;
+ segreq->count += req->serviced;
+ xseg_put_request(s->xseg, req, s->srcport);
+
+ if ((__sync_add_and_fetch(&segreq->ref, -1)) == 0) {
+ if (!segreq->failed) {
+ reqdata->aio_cb->ret = segreq->count;
+ archipelago_finish_aiocb(reqdata);
+ g_free(segreq);
+ } else {
+ g_free(segreq);
+ g_free(reqdata);
+ }
+ } else {
+ g_free(reqdata);
+ }
+ break;
+ case ARCHIP_OP_VOLINFO:
+ s->is_signaled = true;
+ qemu_cond_signal(&s->archip_cond);
+ break;
+ }
+ } else {
+ xseg_wait_signal(s->xseg, psd, 100000UL);
+ }
+ xseg_cancel_wait(s->xseg, s->srcport);
+ }
+
+ s->th_is_signaled = true;
+ qemu_cond_signal(&s->request_cond);
+ qemu_mutex_unlock(&s->request_mutex);
+ qemu_thread_exit(NULL);
+}
+
+static int qemu_archipelago_xseg_init(BDRVArchipelagoState *s)
+{
+ if (xseg_initialize()) {
+ archipelagolog("Cannot initialize XSEG\n");
+ goto err_exit;
+ }
+
+ s->xseg = xseg_join("posix", s->segment_name,
+ "posixfd", NULL);
+ if (!s->xseg) {
+ archipelagolog("Cannot join XSEG shared memory segment\n");
+ goto err_exit;
+ }
+ s->port = xseg_bind_dynport(s->xseg);
+ s->srcport = s->port->portno;
+ init_local_signal(s->xseg, s->sport, s->srcport);
+ return 0;
+
+err_exit:
+ return -1;
+}
+
+static int qemu_archipelago_init(BDRVArchipelagoState *s)
+{
+ int ret;
+
+ ret = qemu_archipelago_xseg_init(s);
+ if (ret < 0) {
+ error_report("Cannot initialize XSEG. Aborting...\n");
+ goto err_exit;
+ }
+
+ qemu_cond_init(&s->archip_cond);
+ qemu_mutex_init(&s->archip_mutex);
+ qemu_cond_init(&s->request_cond);
+ qemu_mutex_init(&s->request_mutex);
+ s->th_is_signaled = false;
+ qemu_thread_create(&s->request_th, "xseg_io_th",
+ (void *) xseg_request_handler,
+ (void *) s, QEMU_THREAD_JOINABLE);
+
+err_exit:
+ return ret;
+}
+
+static void qemu_archipelago_complete_aio(void *opaque)
+{
+ AIORequestData *reqdata = (AIORequestData *) opaque;
+ ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) reqdata->aio_cb;
+
+ qemu_bh_delete(aio_cb->bh);
+ aio_cb->common.cb(aio_cb->common.opaque, aio_cb->ret);
+ aio_cb->status = 0;
+
+ if (!aio_cb->cancelled) {
+ qemu_aio_release(aio_cb);
+ }
+ g_free(reqdata);
+}
+
+static QemuOptsList archipelago_runtime_opts = {
+ .name = "archipelago",
+ .head = QTAILQ_HEAD_INITIALIZER(archipelago_runtime_opts.head),
+ .desc = {
+ {
+ .name = ARCHIPELAGO_OPT_VOLUME,
+ .type = QEMU_OPT_STRING,
+ .help = "Name of the volume image",
+ },
+ {
+ .name = ARCHIPELAGO_OPT_SEGMENT,
+ .type = QEMU_OPT_STRING,
+ .help = "Name of the Archipelago shared memory segment",
+ },
+ {
+ .name = ARCHIPELAGO_OPT_MPORT,
+ .type = QEMU_OPT_NUMBER,
+ .help = "Archipelago mapperd port number"
+ },
+ {
+ .name = ARCHIPELAGO_OPT_VPORT,
+ .type = QEMU_OPT_NUMBER,
+ .help = "Archipelago vlmcd port number"
+
+ },
+ { /* end of list */ }
+ },
+};
+
+static int qemu_archipelago_open(BlockDriverState *bs,
+ QDict *options,
+ int bdrv_flags,
+ Error **errp)
+{
+ int ret = 0;
+ const char *volume, *segment_name;
+ QemuOpts *opts;
+ Error *local_err = NULL;
+ BDRVArchipelagoState *s = bs->opaque;
+
+ opts = qemu_opts_create(&archipelago_runtime_opts, NULL, 0, &error_abort);
+ qemu_opts_absorb_qdict(opts, options, &local_err);
+ if (local_err) {
+ error_propagate(errp, local_err);
+ ret = -EINVAL;
+ goto err_exit;
+ }
+
+ s->mportno = qemu_opt_get_number(opts, ARCHIPELAGO_OPT_MPORT,
+ ARCHIPELAGO_DFL_MPORT);
+ s->vportno = qemu_opt_get_number(opts, ARCHIPELAGO_OPT_VPORT,
+ ARCHIPELAGO_DFL_VPORT);
+
+ segment_name = qemu_opt_get(opts, ARCHIPELAGO_OPT_SEGMENT);
+ if (segment_name == NULL) {
+ s->segment_name = g_strdup("archipelago");
+ } else {
+ s->segment_name = g_strdup(segment_name);
+ }
+
+ volume = qemu_opt_get(opts, ARCHIPELAGO_OPT_VOLUME);
+ if (volume == NULL) {
+ error_setg(errp, "archipelago block driver requires the 'volume'"
+ " option");
+ ret = -EINVAL;
+ goto err_exit;
+ }
+ s->volname = g_strdup(volume);
+
+ /* Initialize XSEG, join shared memory segment */
+ ret = qemu_archipelago_init(s);
+ if (ret < 0) {
+ error_setg(errp, "cannot initialize XSEG and join shared "
+ "memory segment");
+ goto err_exit;
+ }
+
+ qemu_opts_del(opts);
+ return 0;
+
+err_exit:
+ g_free(s->volname);
+ g_free(s->segment_name);
+ qemu_opts_del(opts);
+ return ret;
+}
+
+static void qemu_archipelago_close(BlockDriverState *bs)
+{
+ int r, targetlen;
+ char *target;
+ struct xseg_request *req;
+ BDRVArchipelagoState *s = bs->opaque;
+
+ s->stopping = true;
+
+ qemu_mutex_lock(&s->request_mutex);
+ while (!s->th_is_signaled) {
+ qemu_cond_wait(&s->request_cond,
+ &s->request_mutex);
+ }
+ qemu_mutex_unlock(&s->request_mutex);
+ qemu_thread_join(&s->request_th);
+ qemu_cond_destroy(&s->request_cond);
+ qemu_mutex_destroy(&s->request_mutex);
+
+ qemu_cond_destroy(&s->archip_cond);
+ qemu_mutex_destroy(&s->archip_mutex);
+
+ targetlen = strlen(s->volname);
+ req = xseg_get_request(s->xseg, s->srcport, s->vportno, X_ALLOC);
+ if (!req) {
+ archipelagolog("Cannot get XSEG request\n");
+ goto err_exit;
+ }
+ r = xseg_prep_request(s->xseg, req, targetlen, 0);
+ if (r < 0) {
+ xseg_put_request(s->xseg, req, s->srcport);
+ archipelagolog("Cannot prepare XSEG close request\n");
+ goto err_exit;
+ }
+
+ target = xseg_get_target(s->xseg, req);
+ memcpy(target, s->volname, targetlen);
+ req->size = req->datalen;
+ req->offset = 0;
+ req->op = X_CLOSE;
+
+ xport p = xseg_submit(s->xseg, req, s->srcport, X_ALLOC);
+ if (p == NoPort) {
+ xseg_put_request(s->xseg, req, s->srcport);
+ archipelagolog("Cannot submit XSEG close request\n");
+ goto err_exit;
+ }
+
+ xseg_signal(s->xseg, p);
+ wait_reply(s->xseg, s->srcport, s->port, req);
+
+ xseg_put_request(s->xseg, req, s->srcport);
+
+err_exit:
+ g_free(s->volname);
+ g_free(s->segment_name);
+ xseg_quit_local_signal(s->xseg, s->srcport);
+ xseg_leave_dynport(s->xseg, s->port);
+ xseg_leave(s->xseg);
+}
+
+static void qemu_archipelago_aio_cancel(BlockDriverAIOCB *blockacb)
+{
+ ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) blockacb;
+ aio_cb->cancelled = true;
+ while (aio_cb->status == -EINPROGRESS) {
+ aio_poll(bdrv_get_aio_context(aio_cb->common.bs), true);
+ }
+ qemu_aio_release(aio_cb);
+}
+
+static const AIOCBInfo archipelago_aiocb_info = {
+ .aiocb_size = sizeof(ArchipelagoAIOCB),
+ .cancel = qemu_archipelago_aio_cancel,
+};
+
+static int archipelago_submit_request(BDRVArchipelagoState *s,
+ uint64_t bufidx,
+ size_t count,
+ off_t offset,
+ ArchipelagoAIOCB *aio_cb,
+ ArchipelagoSegmentedRequest *segreq,
+ int op)
+{
+ int ret, targetlen;
+ char *target;
+ void *data = NULL;
+ struct xseg_request *req;
+ AIORequestData *reqdata = g_malloc(sizeof(AIORequestData));
+
+ targetlen = strlen(s->volname);
+ req = xseg_get_request(s->xseg, s->srcport, s->vportno, X_ALLOC);
+ if (!req) {
+ archipelagolog("Cannot get XSEG request\n");
+ goto err_exit2;
+ }
+ ret = xseg_prep_request(s->xseg, req, targetlen, count);
+ if (ret < 0) {
+ archipelagolog("Cannot prepare XSEG request\n");
+ goto err_exit;
+ }
+ target = xseg_get_target(s->xseg, req);
+ if (!target) {
+ archipelagolog("Cannot get XSEG target\n");
+ goto err_exit;
+ }
+ memcpy(target, s->volname, targetlen);
+ req->size = count;
+ req->offset = offset;
+
+ switch (op) {
+ case ARCHIP_OP_READ:
+ req->op = X_READ;
+ break;
+ case ARCHIP_OP_WRITE:
+ req->op = X_WRITE;
+ break;
+ case ARCHIP_OP_FLUSH:
+ req->op = X_FLUSH;
+ break;
+ }
+ reqdata->volname = s->volname;
+ reqdata->offset = offset;
+ reqdata->size = count;
+ reqdata->bufidx = bufidx;
+ reqdata->aio_cb = aio_cb;
+ reqdata->segreq = segreq;
+ reqdata->op = op;
+
+ xseg_set_req_data(s->xseg, req, reqdata);
+ if (op == ARCHIP_OP_WRITE) {
+ data = xseg_get_data(s->xseg, req);
+ if (!data) {
+ archipelagolog("Cannot get XSEG data\n");
+ goto err_exit;
+ }
+ qemu_iovec_to_buf(aio_cb->qiov, bufidx, data, count);
+ }
+
+ xport p = xseg_submit(s->xseg, req, s->srcport, X_ALLOC);
+ if (p == NoPort) {
+ archipelagolog("Could not submit XSEG request\n");
+ goto err_exit;
+ }
+ xseg_signal(s->xseg, p);
+ return 0;
+
+err_exit:
+ g_free(reqdata);
+ xseg_put_request(s->xseg, req, s->srcport);
+ return -EIO;
+err_exit2:
+ g_free(reqdata);
+ return -EIO;
+}
+
+static int archipelago_aio_segmented_rw(BDRVArchipelagoState *s,
+ size_t count,
+ off_t offset,
+ ArchipelagoAIOCB *aio_cb,
+ int op)
+{
+ int i, ret, segments_nr, last_segment_size;
+ ArchipelagoSegmentedRequest *segreq;
+
+ segreq = g_malloc(sizeof(ArchipelagoSegmentedRequest));
+
+ if (op == ARCHIP_OP_FLUSH) {
+ segments_nr = 1;
+ segreq->ref = segments_nr;
+ segreq->total = count;
+ segreq->count = 0;
+ segreq->failed = 0;
+ ret = archipelago_submit_request(s, 0, count, offset, aio_cb,
+ segreq, ARCHIP_OP_FLUSH);
+ if (ret < 0) {
+ goto err_exit;
+ }
+ return 0;
+ }
+
+ segments_nr = (int)(count / MAX_REQUEST_SIZE) + \
+ ((count % MAX_REQUEST_SIZE) ? 1 : 0);
+ last_segment_size = (int)(count % MAX_REQUEST_SIZE);
+
+ segreq->ref = segments_nr;
+ segreq->total = count;
+ segreq->count = 0;
+ segreq->failed = 0;
+
+ for (i = 0; i < segments_nr - 1; i++) {
+ ret = archipelago_submit_request(s, i * MAX_REQUEST_SIZE,
+ MAX_REQUEST_SIZE,
+ offset + i * MAX_REQUEST_SIZE,
+ aio_cb, segreq, op);
+
+ if (ret < 0) {
+ goto err_exit;
+ }
+ }
+
+ if ((segments_nr > 1) && last_segment_size) {
+ ret = archipelago_submit_request(s, i * MAX_REQUEST_SIZE,
+ last_segment_size,
+ offset + i * MAX_REQUEST_SIZE,
+ aio_cb, segreq, op);
+ } else if ((segments_nr > 1) && !last_segment_size) {
+ ret = archipelago_submit_request(s, i * MAX_REQUEST_SIZE,
+ MAX_REQUEST_SIZE,
+ offset + i * MAX_REQUEST_SIZE,
+ aio_cb, segreq, op);
+ } else if (segments_nr == 1) {
+ ret = archipelago_submit_request(s, 0, count, offset, aio_cb,
+ segreq, op);
+ }
+
+ if (ret < 0) {
+ goto err_exit;
+ }
+
+ return 0;
+
+err_exit:
+ __sync_add_and_fetch(&segreq->failed, 1);
+ if (segments_nr == 1) {
+ if (__sync_add_and_fetch(&segreq->ref, -1) == 0) {
+ g_free(segreq);
+ }
+ } else {
+ if ((__sync_add_and_fetch(&segreq->ref, -segments_nr + i)) == 0) {
+ g_free(segreq);
+ }
+ }
+
+ return ret;
+}
+
+static BlockDriverAIOCB *qemu_archipelago_aio_rw(BlockDriverState *bs,
+ int64_t sector_num,
+ QEMUIOVector *qiov,
+ int nb_sectors,
+ BlockDriverCompletionFunc *cb,
+ void *opaque,
+ int op)
+{
+ ArchipelagoAIOCB *aio_cb;
+ BDRVArchipelagoState *s = bs->opaque;
+ int64_t size, off;
+ int ret;
+
+ aio_cb = qemu_aio_get(&archipelago_aiocb_info, bs, cb, opaque);
+ aio_cb->cmd = op;
+ aio_cb->qiov = qiov;
+
+ aio_cb->ret = 0;
+ aio_cb->s = s;
+ aio_cb->cancelled = false;
+ aio_cb->status = -EINPROGRESS;
+
+ off = sector_num * BDRV_SECTOR_SIZE;
+ size = nb_sectors * BDRV_SECTOR_SIZE;
+ aio_cb->size = size;
+
+ ret = archipelago_aio_segmented_rw(s, size, off,
+ aio_cb, op);
+ if (ret < 0) {
+ goto err_exit;
+ }
+ return &aio_cb->common;
+
+err_exit:
+ error_report("qemu_archipelago_aio_rw(): I/O Error\n");
+ qemu_aio_release(aio_cb);
+ return NULL;
+}
+
+static BlockDriverAIOCB *qemu_archipelago_aio_readv(BlockDriverState *bs,
+ int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+ BlockDriverCompletionFunc *cb, void *opaque)
+{
+ return qemu_archipelago_aio_rw(bs, sector_num, qiov, nb_sectors, cb,
+ opaque, ARCHIP_OP_READ);
+}
+
+static BlockDriverAIOCB *qemu_archipelago_aio_writev(BlockDriverState *bs,
+ int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+ BlockDriverCompletionFunc *cb, void *opaque)
+{
+ return qemu_archipelago_aio_rw(bs, sector_num, qiov, nb_sectors, cb,
+ opaque, ARCHIP_OP_WRITE);
+}
+
+static int64_t archipelago_volume_info(BDRVArchipelagoState *s)
+{
+ uint64_t size;
+ int ret, targetlen;
+ struct xseg_request *req;
+ struct xseg_reply_info *xinfo;
+ AIORequestData *reqdata = g_malloc(sizeof(AIORequestData));
+
+ const char *volname = s->volname;
+ targetlen = strlen(volname);
+ req = xseg_get_request(s->xseg, s->srcport, s->mportno, X_ALLOC);
+ if (!req) {
+ archipelagolog("Cannot get XSEG request\n");
+ goto err_exit2;
+ }
+ ret = xseg_prep_request(s->xseg, req, targetlen,
+ sizeof(struct xseg_reply_info));
+ if (ret < 0) {
+ archipelagolog("Cannot prepare XSEG request\n");
+ goto err_exit;
+ }
+ char *target = xseg_get_target(s->xseg, req);
+ if (!target) {
+ archipelagolog("Cannot get XSEG target\n");
+ goto err_exit;
+ }
+ memcpy(target, volname, targetlen);
+ req->size = req->datalen;
+ req->offset = 0;
+ req->op = X_INFO;
+
+ reqdata->op = ARCHIP_OP_VOLINFO;
+ reqdata->volname = volname;
+ xseg_set_req_data(s->xseg, req, reqdata);
+
+ xport p = xseg_submit(s->xseg, req, s->srcport, X_ALLOC);
+ if (p == NoPort) {
+ archipelagolog("Cannot submit XSEG request\n");
+ goto err_exit;
+ }
+ xseg_signal(s->xseg, p);
+ qemu_mutex_lock(&s->archip_mutex);
+ while (!s->is_signaled) {
+ qemu_cond_wait(&s->archip_cond, &s->archip_mutex);
+ }
+ s->is_signaled = false;
+ qemu_mutex_unlock(&s->archip_mutex);
+
+ xinfo = (struct xseg_reply_info *) xseg_get_data(s->xseg, req);
+ size = xinfo->size;
+ xseg_put_request(s->xseg, req, s->srcport);
+ g_free(reqdata);
+ s->size = size;
+ return size;
+
+err_exit:
+ xseg_put_request(s->xseg, req, s->srcport);
+err_exit2:
+ g_free(reqdata);
+ return -EIO;
+}
+
+static int64_t qemu_archipelago_getlength(BlockDriverState *bs)
+{
+ int64_t ret;
+ BDRVArchipelagoState *s = bs->opaque;
+
+ ret = archipelago_volume_info(s);
+ return ret;
+}
+
+static BlockDriverAIOCB *qemu_archipelago_aio_flush(BlockDriverState *bs,
+ BlockDriverCompletionFunc *cb, void *opaque)
+{
+ return qemu_archipelago_aio_rw(bs, 0, NULL, 0, cb, opaque,
+ ARCHIP_OP_FLUSH);
+}
+
+static BlockDriver bdrv_archipelago = {
+ .format_name = "archipelago",
+ .protocol_name = "archipelago",
+ .instance_size = sizeof(BDRVArchipelagoState),
+ .bdrv_file_open = qemu_archipelago_open,
+ .bdrv_close = qemu_archipelago_close,
+ .bdrv_getlength = qemu_archipelago_getlength,
+ .bdrv_aio_readv = qemu_archipelago_aio_readv,
+ .bdrv_aio_writev = qemu_archipelago_aio_writev,
+ .bdrv_aio_flush = qemu_archipelago_aio_flush,
+ .bdrv_has_zero_init = bdrv_has_zero_init_1,
+};
+
+static void bdrv_archipelago_init(void)
+{
+ bdrv_register(&bdrv_archipelago);
+}
+
+block_init(bdrv_archipelago_init);
diff --git a/configure b/configure
index f7685b5..8a5cc5e 100755
--- a/configure
+++ b/configure
@@ -326,6 +326,7 @@ seccomp=""
glusterfs=""
glusterfs_discard="no"
glusterfs_zerofill="no"
+archipelago=""
virtio_blk_data_plane=""
gtk=""
gtkabi=""
@@ -1087,6 +1088,10 @@ for opt do
;;
--enable-glusterfs) glusterfs="yes"
;;
+ --disable-archipelago) archipelago="no"
+ ;;
+ --enable-archipelago) archipelago="yes"
+ ;;
--disable-virtio-blk-data-plane) virtio_blk_data_plane="no"
;;
--enable-virtio-blk-data-plane) virtio_blk_data_plane="yes"
@@ -1382,6 +1387,8 @@ Advanced options (experts only):
--enable-coroutine-pool enable coroutine freelist (better performance)
--enable-glusterfs enable GlusterFS backend
--disable-glusterfs disable GlusterFS backend
+ --enable-archipelago enable Archipelago backend
+ --disable-archipelago disable Archipelago backend
--enable-gcov enable test coverage analysis with gcov
--gcov=GCOV use specified gcov [$gcov_tool]
--disable-tpm disable TPM support
@@ -3071,6 +3078,33 @@ EOF
fi
fi
+
+##########################################
+# archipelago probe
+if test "$archipelago" != "no" ; then
+ cat > $TMPC <<EOF
+#include <stdio.h>
+#include <xseg/xseg.h>
+#include <xseg/protocol.h>
+int main(void) {
+ xseg_initialize();
+ return 0;
+}
+EOF
+ archipelago_libs=-lxseg
+ if compile_prog "" "$archipelago_libs"; then
+ archipelago="yes"
+ libs_tools="$archipelago_libs $libs_tools"
+ libs_softmmu="$archipelago_libs $libs_softmmu"
+ else
+ if test "$archipelago" = "yes" ; then
+ feature_not_found "Archipelago backend support" "Install libxseg devel"
+ fi
+ archipelago="no"
+ fi
+fi
+
+
##########################################
# glusterfs probe
if test "$glusterfs" != "no" ; then
@@ -4250,6 +4284,7 @@ echo "seccomp support $seccomp"
echo "coroutine backend $coroutine"
echo "coroutine pool $coroutine_pool"
echo "GlusterFS support $glusterfs"
+echo "Archipelago support $archipelago"
echo "virtio-blk-data-plane $virtio_blk_data_plane"
echo "gcov $gcov_tool"
echo "gcov enabled $gcov"
@@ -4688,6 +4723,11 @@ if test "$glusterfs_zerofill" = "yes" ; then
echo "CONFIG_GLUSTERFS_ZEROFILL=y" >> $config_host_mak
fi
+if test "$archipelago" = "yes" ; then
+ echo "CONFIG_ARCHIPELAGO=m" >> $config_host_mak
+ echo "ARCHIPELAGO_LIBS=$archipelago_libs" >> $config_host_mak
+fi
+
if test "$libssh2" = "yes" ; then
echo "CONFIG_LIBSSH2=m" >> $config_host_mak
echo "LIBSSH2_CFLAGS=$libssh2_cflags" >> $config_host_mak
--
1.7.10.4
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [Qemu-devel] [PATCH v8 1/5] block: Support Archipelago as a QEMU block backend
2014-08-04 14:35 ` [Qemu-devel] [PATCH v8 " Chrysostomos Nanakos
@ 2014-08-05 8:20 ` Stefan Hajnoczi
2014-08-05 8:50 ` Kevin Wolf
0 siblings, 1 reply; 13+ messages in thread
From: Stefan Hajnoczi @ 2014-08-05 8:20 UTC (permalink / raw)
To: Chrysostomos Nanakos; +Cc: kwolf, qemu-devel
[-- Attachment #1: Type: text/plain, Size: 1753 bytes --]
On Mon, Aug 04, 2014 at 05:35:32PM +0300, Chrysostomos Nanakos wrote:
> VM Image on Archipelago volume is specified like this:
>
> file.driver=archipelago,file.volume=<volumename>[,file.mport=<mapperd_port>[,
> file.vport=<vlmcd_port>][,file.segment=<segment_name>]]
>
> 'archipelago' is the protocol.
>
> 'mport' is the port number on which mapperd is listening. This is optional
> and if not specified, QEMU will make Archipelago to use the default port.
>
> 'vport' is the port number on which vlmcd is listening. This is optional
> and if not specified, QEMU will make Archipelago to use the default port.
>
> 'segment' is the name of the shared memory segment Archipelago stack is using.
> This is optional and if not specified, QEMU will make Archipelago to use the
> default value, 'archipelago'.
>
> Examples:
>
> file.driver=archipelago,file.volume=my_vm_volume
> file.driver=archipelago,file.volume=my_vm_volume,file.mport=123
> file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
> file.vport=1234
> file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
> file.vport=1234,file.segment=my_segment
>
> Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
> ---
> MAINTAINERS | 6 +
> block/Makefile.objs | 2 +
> block/archipelago.c | 787 +++++++++++++++++++++++++++++++++++++++++++++++++++
> configure | 40 +++
> 4 files changed, 835 insertions(+)
> create mode 100644 block/archipelago.c
The diff is that a flush op is used now:
+ case ARCHIP_OP_FLUSH:
+ req->op = X_FLUSH;
+ break;
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Kevin: Please replace the v7 commit with this new patch on the block
branch.
[-- Attachment #2: Type: application/pgp-signature, Size: 473 bytes --]
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [Qemu-devel] [PATCH v8 1/5] block: Support Archipelago as a QEMU block backend
2014-08-05 8:20 ` Stefan Hajnoczi
@ 2014-08-05 8:50 ` Kevin Wolf
0 siblings, 0 replies; 13+ messages in thread
From: Kevin Wolf @ 2014-08-05 8:50 UTC (permalink / raw)
To: Stefan Hajnoczi; +Cc: Chrysostomos Nanakos, qemu-devel
[-- Attachment #1: Type: text/plain, Size: 1961 bytes --]
Am 05.08.2014 um 10:20 hat Stefan Hajnoczi geschrieben:
> On Mon, Aug 04, 2014 at 05:35:32PM +0300, Chrysostomos Nanakos wrote:
> > VM Image on Archipelago volume is specified like this:
> >
> > file.driver=archipelago,file.volume=<volumename>[,file.mport=<mapperd_port>[,
> > file.vport=<vlmcd_port>][,file.segment=<segment_name>]]
> >
> > 'archipelago' is the protocol.
> >
> > 'mport' is the port number on which mapperd is listening. This is optional
> > and if not specified, QEMU will make Archipelago to use the default port.
> >
> > 'vport' is the port number on which vlmcd is listening. This is optional
> > and if not specified, QEMU will make Archipelago to use the default port.
> >
> > 'segment' is the name of the shared memory segment Archipelago stack is using.
> > This is optional and if not specified, QEMU will make Archipelago to use the
> > default value, 'archipelago'.
> >
> > Examples:
> >
> > file.driver=archipelago,file.volume=my_vm_volume
> > file.driver=archipelago,file.volume=my_vm_volume,file.mport=123
> > file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
> > file.vport=1234
> > file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
> > file.vport=1234,file.segment=my_segment
> >
> > Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
> > ---
> > MAINTAINERS | 6 +
> > block/Makefile.objs | 2 +
> > block/archipelago.c | 787 +++++++++++++++++++++++++++++++++++++++++++++++++++
> > configure | 40 +++
> > 4 files changed, 835 insertions(+)
> > create mode 100644 block/archipelago.c
>
> The diff is that a flush op is used now:
> + case ARCHIP_OP_FLUSH:
> + req->op = X_FLUSH;
> + break;
>
> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
>
> Kevin: Please replace the v7 commit with this new patch on the block
> branch.
Already did that, will add your Reviewed-by now.
Kevin
[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Qemu-devel] [PATCH v7 2/5] block/archipelago: Implement bdrv_parse_filename()
2014-07-23 14:07 [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Chrysostomos Nanakos
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 1/5] block: " Chrysostomos Nanakos
@ 2014-07-23 14:07 ` Chrysostomos Nanakos
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 3/5] block/archipelago: Add support for creating images Chrysostomos Nanakos
` (4 subsequent siblings)
6 siblings, 0 replies; 13+ messages in thread
From: Chrysostomos Nanakos @ 2014-07-23 14:07 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, Chrysostomos Nanakos, stefanha
VM Image on Archipelago volume can also be specified like this:
file=archipelago:<volumename>[/mport=<mapperd_port>[:vport=<vlmcd_port>][:
segment=<segment_name>]]
Examples:
file=archipelago:my_vm_volume
file=archipelago:my_vm_volume/mport=123
file=archipelago:my_vm_volume/mport=123:vport=1234
file=archipelago:my_vm_volume/mport=123:vport=1234:segment=my_segment
Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
---
block/archipelago.c | 140 ++++++++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 138 insertions(+), 2 deletions(-)
diff --git a/block/archipelago.c b/block/archipelago.c
index 1c21d36..5a9fc68 100644
--- a/block/archipelago.c
+++ b/block/archipelago.c
@@ -15,6 +15,11 @@
* [,file.mport=<mapperd_port>[,file.vport=<vlmcd_port>]
* [,file.segment=<segment_name>]]
*
+ * or
+ *
+ * file=archipelago:<volumename>[/mport=<mapperd_port>[:vport=<vlmcd_port>][:
+ * segment=<segment_name>]]
+ *
* 'archipelago' is the protocol.
*
* 'mport' is the port number on which mapperd is listening. This is optional
@@ -32,11 +37,20 @@
* file.driver=archipelago,file.volume=my_vm_volume
* file.driver=archipelago,file.volume=my_vm_volume,file.mport=123
* file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
- * file.vport=1234
+ * file.vport=1234
* file.driver=archipelago,file.volume=my_vm_volume,file.mport=123,
- * file.vport=1234,file.segment=my_segment
+ * file.vport=1234,file.segment=my_segment
+ *
+ * or
+ *
+ * file=archipelago:my_vm_volume
+ * file=archipelago:my_vm_volume/mport=123
+ * file=archipelago:my_vm_volume/mport=123:vport=1234
+ * file=archipelago:my_vm_volume/mport=123:vport=1234:segment=my_segment
+ *
*/
+#include "qemu-common.h"
#include "block/block_int.h"
#include "qemu/error-report.h"
#include "qemu/thread.h"
@@ -309,6 +323,127 @@ static void qemu_archipelago_complete_aio(void *opaque)
g_free(reqdata);
}
+static void xseg_find_port(char *pstr, const char *needle, xport *aport)
+{
+ const char *a;
+ char *endptr = NULL;
+ unsigned long port;
+ if (strstart(pstr, needle, &a)) {
+ if (strlen(a) > 0) {
+ port = strtoul(a, &endptr, 10);
+ if (strlen(endptr)) {
+ *aport = -2;
+ return;
+ }
+ *aport = (xport) port;
+ }
+ }
+}
+
+static void xseg_find_segment(char *pstr, const char *needle,
+ char **segment_name)
+{
+ const char *a;
+ if (strstart(pstr, needle, &a)) {
+ if (strlen(a) > 0) {
+ *segment_name = g_strdup(a);
+ }
+ }
+}
+
+static void parse_filename_opts(const char *filename, Error **errp,
+ char **volume, char **segment_name,
+ xport *mport, xport *vport)
+{
+ const char *start;
+ char *tokens[4], *ds;
+ int idx;
+ xport lmport = NoPort, lvport = NoPort;
+
+ strstart(filename, "archipelago:", &start);
+
+ ds = g_strdup(start);
+ tokens[0] = strtok(ds, "/");
+ tokens[1] = strtok(NULL, ":");
+ tokens[2] = strtok(NULL, ":");
+ tokens[3] = strtok(NULL, "\0");
+
+ if (!strlen(tokens[0])) {
+ error_setg(errp, "volume name must be specified first");
+ g_free(ds);
+ return;
+ }
+
+ for (idx = 1; idx < 4; idx++) {
+ if (tokens[idx] != NULL) {
+ if (strstart(tokens[idx], "mport=", NULL)) {
+ xseg_find_port(tokens[idx], "mport=", &lmport);
+ }
+ if (strstart(tokens[idx], "vport=", NULL)) {
+ xseg_find_port(tokens[idx], "vport=", &lvport);
+ }
+ if (strstart(tokens[idx], "segment=", NULL)) {
+ xseg_find_segment(tokens[idx], "segment=", segment_name);
+ }
+ }
+ }
+
+ if ((lmport == -2) || (lvport == -2)) {
+ error_setg(errp, "mport and/or vport must be set");
+ g_free(ds);
+ return;
+ }
+ *volume = g_strdup(tokens[0]);
+ *mport = lmport;
+ *vport = lvport;
+ g_free(ds);
+}
+
+static void archipelago_parse_filename(const char *filename, QDict *options,
+ Error **errp)
+{
+ const char *start;
+ char *volume = NULL, *segment_name = NULL;
+ xport mport = NoPort, vport = NoPort;
+
+ if (qdict_haskey(options, ARCHIPELAGO_OPT_VOLUME)
+ || qdict_haskey(options, ARCHIPELAGO_OPT_SEGMENT)
+ || qdict_haskey(options, ARCHIPELAGO_OPT_MPORT)
+ || qdict_haskey(options, ARCHIPELAGO_OPT_VPORT)) {
+ error_setg(errp, "volume/mport/vport/segment and a file name may not"
+ " be specified at the same time");
+ return;
+ }
+
+ if (!strstart(filename, "archipelago:", &start)) {
+ error_setg(errp, "File name must start with 'archipelago:'");
+ return;
+ }
+
+ if (!strlen(start) || strstart(start, "/", NULL)) {
+ error_setg(errp, "volume name must be specified");
+ return;
+ }
+
+ parse_filename_opts(filename, errp, &volume, &segment_name, &mport, &vport);
+
+ if (volume) {
+ qdict_put(options, ARCHIPELAGO_OPT_VOLUME, qstring_from_str(volume));
+ g_free(volume);
+ }
+ if (segment_name) {
+ qdict_put(options, ARCHIPELAGO_OPT_SEGMENT,
+ qstring_from_str(segment_name));
+ g_free(segment_name);
+ }
+ if (mport != NoPort) {
+ qdict_put(options, ARCHIPELAGO_OPT_MPORT, qint_from_int(mport));
+ }
+ if (vport != NoPort) {
+ qdict_put(options, ARCHIPELAGO_OPT_VPORT, qint_from_int(vport));
+ }
+}
+
static QemuOptsList archipelago_runtime_opts = {
.name = "archipelago",
.head = QTAILQ_HEAD_INITIALIZER(archipelago_runtime_opts.head),
@@ -768,6 +903,7 @@ static BlockDriver bdrv_archipelago = {
.format_name = "archipelago",
.protocol_name = "archipelago",
.instance_size = sizeof(BDRVArchipelagoState),
+ .bdrv_parse_filename = archipelago_parse_filename,
.bdrv_file_open = qemu_archipelago_open,
.bdrv_close = qemu_archipelago_close,
.bdrv_getlength = qemu_archipelago_getlength,
--
1.7.10.4
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PATCH v7 3/5] block/archipelago: Add support for creating images
2014-07-23 14:07 [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Chrysostomos Nanakos
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 1/5] block: " Chrysostomos Nanakos
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 2/5] block/archipelago: Implement bdrv_parse_filename() Chrysostomos Nanakos
@ 2014-07-23 14:07 ` Chrysostomos Nanakos
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 4/5] QMP: Add support for Archipelago Chrysostomos Nanakos
` (3 subsequent siblings)
6 siblings, 0 replies; 13+ messages in thread
From: Chrysostomos Nanakos @ 2014-07-23 14:07 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, Chrysostomos Nanakos, stefanha
qemu-img archipelago:<volumename>[/mport=<mapperd_port>[:vport=<vlmcd_port>]
[:segment=<segment_name>]] [size]
Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
---
block/archipelago.c | 146 +++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 146 insertions(+)
diff --git a/block/archipelago.c b/block/archipelago.c
index 5a9fc68..b5c66fd 100644
--- a/block/archipelago.c
+++ b/block/archipelago.c
@@ -592,6 +592,137 @@ err_exit:
xseg_leave(s->xseg);
}
+static int qemu_archipelago_create_volume(Error **errp, const char *volname,
+ char *segment_name,
+ uint64_t size, xport mportno,
+ xport vportno)
+{
+ int ret, targetlen;
+ struct xseg *xseg = NULL;
+ struct xseg_request *req;
+ struct xseg_request_clone *xclone;
+ struct xseg_port *port;
+ xport srcport = NoPort, sport = NoPort;
+ char *target;
+
+ /* Try default values if none has been set */
+ if (mportno == (xport) -1) {
+ mportno = ARCHIPELAGO_DFL_MPORT;
+ }
+
+ if (vportno == (xport) -1) {
+ vportno = ARCHIPELAGO_DFL_VPORT;
+ }
+
+ if (xseg_initialize()) {
+ error_setg(errp, "Cannot initialize XSEG");
+ return -1;
+ }
+
+ xseg = xseg_join("posix", segment_name,
+ "posixfd", NULL);
+
+ if (!xseg) {
+ error_setg(errp, "Cannot join XSEG shared memory segment");
+ return -1;
+ }
+
+ port = xseg_bind_dynport(xseg);
+ srcport = port->portno;
+ init_local_signal(xseg, sport, srcport);
+
+ req = xseg_get_request(xseg, srcport, mportno, X_ALLOC);
+ if (!req) {
+ error_setg(errp, "Cannot get XSEG request");
+ return -1;
+ }
+
+ targetlen = strlen(volname);
+ ret = xseg_prep_request(xseg, req, targetlen,
+ sizeof(struct xseg_request_clone));
+ if (ret < 0) {
+ error_setg(errp, "Cannot prepare XSEG request");
+ goto err_exit;
+ }
+
+ target = xseg_get_target(xseg, req);
+ if (!target) {
+ error_setg(errp, "Cannot get XSEG target.\n");
+ goto err_exit;
+ }
+ memcpy(target, volname, targetlen);
+ xclone = (struct xseg_request_clone *) xseg_get_data(xseg, req);
+ memset(xclone->target, 0 , XSEG_MAX_TARGETLEN);
+ xclone->targetlen = 0;
+ xclone->size = size;
+ req->offset = 0;
+ req->size = req->datalen;
+ req->op = X_CLONE;
+
+ xport p = xseg_submit(xseg, req, srcport, X_ALLOC);
+ if (p == NoPort) {
+ error_setg(errp, "Could not submit XSEG request");
+ goto err_exit;
+ }
+ xseg_signal(xseg, p);
+
+ ret = wait_reply(xseg, srcport, port, req);
+ if (ret < 0) {
+ error_setg(errp, "wait_reply() error.");
+ }
+
+ xseg_put_request(xseg, req, srcport);
+ xseg_quit_local_signal(xseg, srcport);
+ xseg_leave_dynport(xseg, port);
+ xseg_leave(xseg);
+ return ret;
+
+err_exit:
+ xseg_put_request(xseg, req, srcport);
+ xseg_quit_local_signal(xseg, srcport);
+ xseg_leave_dynport(xseg, port);
+ xseg_leave(xseg);
+ return -1;
+}
+
+static int qemu_archipelago_create(const char *filename,
+ QemuOpts *options,
+ Error **errp)
+{
+ int ret = 0;
+ uint64_t total_size = 0;
+ char *volname = NULL, *segment_name = NULL;
+ const char *start;
+ xport mport = NoPort, vport = NoPort;
+
+ if (!strstart(filename, "archipelago:", &start)) {
+ error_setg(errp, "File name must start with 'archipelago:'");
+ return -1;
+ }
+
+ if (!strlen(start) || strstart(start, "/", NULL)) {
+ error_setg(errp, "volume name must be specified");
+ return -1;
+ }
+
+ parse_filename_opts(filename, errp, &volname, &segment_name, &mport,
+ &vport);
+ total_size = qemu_opt_get_size_del(options, BLOCK_OPT_SIZE, 0);
+
+ if (segment_name == NULL) {
+ segment_name = g_strdup("archipelago");
+ }
+
+ /* Create an Archipelago volume */
+ ret = qemu_archipelago_create_volume(errp, volname, segment_name,
+ total_size, mport,
+ vport);
+
+ g_free(volname);
+ g_free(segment_name);
+ return ret;
+}
+
static void qemu_archipelago_aio_cancel(BlockDriverAIOCB *blockacb)
{
ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) blockacb;
@@ -892,6 +1023,19 @@ static int64_t qemu_archipelago_getlength(BlockDriverState *bs)
return ret;
}
+static QemuOptsList qemu_archipelago_create_opts = {
+ .name = "archipelago-create-opts",
+ .head = QTAILQ_HEAD_INITIALIZER(qemu_archipelago_create_opts.head),
+ .desc = {
+ {
+ .name = BLOCK_OPT_SIZE,
+ .type = QEMU_OPT_SIZE,
+ .help = "Virtual disk size"
+ },
+ { /* end of list */ }
+ }
+};
+
static BlockDriverAIOCB *qemu_archipelago_aio_flush(BlockDriverState *bs,
BlockDriverCompletionFunc *cb, void *opaque)
{
@@ -906,11 +1050,13 @@ static BlockDriver bdrv_archipelago = {
.bdrv_parse_filename = archipelago_parse_filename,
.bdrv_file_open = qemu_archipelago_open,
.bdrv_close = qemu_archipelago_close,
+ .bdrv_create = qemu_archipelago_create,
.bdrv_getlength = qemu_archipelago_getlength,
.bdrv_aio_readv = qemu_archipelago_aio_readv,
.bdrv_aio_writev = qemu_archipelago_aio_writev,
.bdrv_aio_flush = qemu_archipelago_aio_flush,
.bdrv_has_zero_init = bdrv_has_zero_init_1,
+ .create_opts = &qemu_archipelago_create_opts,
};
static void bdrv_archipelago_init(void)
--
1.7.10.4
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PATCH v7 4/5] QMP: Add support for Archipelago
2014-07-23 14:07 [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Chrysostomos Nanakos
` (2 preceding siblings ...)
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 3/5] block/archipelago: Add support for creating images Chrysostomos Nanakos
@ 2014-07-23 14:07 ` Chrysostomos Nanakos
2014-07-30 16:33 ` Eric Blake
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 5/5] qemu-iotests: add support for Archipelago protocol Chrysostomos Nanakos
` (2 subsequent siblings)
6 siblings, 1 reply; 13+ messages in thread
From: Chrysostomos Nanakos @ 2014-07-23 14:07 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, Chrysostomos Nanakos, stefanha
Introduce new enum BlockdevOptionsArchipelago.
@volume: #Name of the Archipelago volume image
@mport: #'mport' is the port number on which mapperd is
listening. This is optional and if not specified,
QEMU will make Archipelago to use the default port.
@vport: #'vport' is the port number on which vlmcd is
listening. This is optional and if not specified,
QEMU will make Archipelago to use the default port.
@segment: #optional The name of the shared memory segment
Archipelago stack is using. This is optional
and if not specified, QEMU will make Archipelago
use the default value, 'archipelago'.
Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
---
qapi/block-core.json | 38 +++++++++++++++++++++++++++++++++++---
1 file changed, 35 insertions(+), 3 deletions(-)
diff --git a/qapi/block-core.json b/qapi/block-core.json
index e378653..0fa0c12 100644
--- a/qapi/block-core.json
+++ b/qapi/block-core.json
@@ -190,8 +190,8 @@
# @ro: true if the backing device was open read-only
#
# @drv: the name of the block format used to open the backing device. As of
-# 0.14.0 this can be: 'blkdebug', 'bochs', 'cloop', 'cow', 'dmg',
-# 'file', 'file', 'ftp', 'ftps', 'host_cdrom', 'host_device',
+# 0.14.0 this can be: 'archipelago', 'blkdebug', 'bochs', 'cloop', 'cow',
+# 'dmg', 'file', 'file', 'ftp', 'ftps', 'host_cdrom', 'host_device',
# 'host_floppy', 'http', 'https', 'nbd', 'parallels', 'qcow',
# 'qcow2', 'raw', 'tftp', 'vdi', 'vmdk', 'vpc', 'vvfat'
#
@@ -1143,7 +1143,7 @@
# Since: 2.0
##
{ 'enum': 'BlockdevDriver',
- 'data': [ 'file', 'host_device', 'host_cdrom', 'host_floppy',
+ 'data': [ 'archipelago', 'file', 'host_device', 'host_cdrom', 'host_floppy',
'http', 'https', 'ftp', 'ftps', 'tftp', 'vvfat', 'blkdebug',
'blkverify', 'bochs', 'cloop', 'cow', 'dmg', 'parallels', 'qcow',
'qcow2', 'qed', 'raw', 'vdi', 'vhdx', 'vmdk', 'vpc', 'quorum' ] }
@@ -1273,6 +1273,37 @@
'*pass-discard-snapshot': 'bool',
'*pass-discard-other': 'bool' } }
+
+##
+# @BlockdevOptionsArchipelago
+#
+# Driver specific block device options for Archipelago.
+#
+# @volume: Name of the Archipelago volume image
+#
+# @mport: #optional The port number on which mapperd is
+# listening. This is optional
+# and if not specified, QEMU will make Archipelago
+# use the default port (1001).
+#
+# @vport: #optional The port number on which vlmcd is
+# listening. This is optional
+# and if not specified, QEMU will make Archipelago
+# use the default port (501).
+#
+# @segment: #optional The name of the shared memory segment
+# Archipelago stack is using. This is optional
+# and if not specified, QEMU will make Archipelago
+# use the default value, 'archipelago'.
+# Since: 2.2
+##
+{ 'type': 'BlockdevOptionsArchipelago',
+ 'data': { 'volume': 'str',
+ '*mport': 'int',
+ '*vport': 'int',
+ '*segment': 'str' } }
+
+
##
# @BlkdebugEvent
#
@@ -1416,6 +1447,7 @@
'base': 'BlockdevOptionsBase',
'discriminator': 'driver',
'data': {
+ 'archipelago':'BlockdevOptionsArchipelago',
'file': 'BlockdevOptionsFile',
'host_device':'BlockdevOptionsFile',
'host_cdrom': 'BlockdevOptionsFile',
--
1.7.10.4
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [Qemu-devel] [PATCH v7 4/5] QMP: Add support for Archipelago
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 4/5] QMP: Add support for Archipelago Chrysostomos Nanakos
@ 2014-07-30 16:33 ` Eric Blake
0 siblings, 0 replies; 13+ messages in thread
From: Eric Blake @ 2014-07-30 16:33 UTC (permalink / raw)
To: Chrysostomos Nanakos, qemu-devel; +Cc: kwolf, stefanha
[-- Attachment #1: Type: text/plain, Size: 2124 bytes --]
On 07/23/2014 08:07 AM, Chrysostomos Nanakos wrote:
> Introduce new enum BlockdevOptionsArchipelago.
>
> @volume: #Name of the Archipelago volume image
>
> @mport: #'mport' is the port number on which mapperd is
> listening. This is optional and if not specified,
> QEMU will make Archipelago to use the default port.
>
> @vport: #'vport' is the port number on which vlmcd is
> listening. This is optional and if not specified,
> QEMU will make Archipelago to use the default port.
>
> @segment: #optional The name of the shared memory segment
> Archipelago stack is using. This is optional
> and if not specified, QEMU will make Archipelago
> use the default value, 'archipelago'.
>
> Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
> ---
> qapi/block-core.json | 38 +++++++++++++++++++++++++++++++++++---
> 1 file changed, 35 insertions(+), 3 deletions(-)
>
> diff --git a/qapi/block-core.json b/qapi/block-core.json
> index e378653..0fa0c12 100644
> --- a/qapi/block-core.json
> +++ b/qapi/block-core.json
> @@ -190,8 +190,8 @@
> # @ro: true if the backing device was open read-only
> #
> # @drv: the name of the block format used to open the backing device. As of
> -# 0.14.0 this can be: 'blkdebug', 'bochs', 'cloop', 'cow', 'dmg',
> -# 'file', 'file', 'ftp', 'ftps', 'host_cdrom', 'host_device',
> +# 0.14.0 this can be: 'archipelago', 'blkdebug', 'bochs', 'cloop', 'cow',
> +# 'dmg', 'file', 'file', 'ftp', 'ftps', 'host_cdrom', 'host_device',
> # 'host_floppy', 'http', 'https', 'nbd', 'parallels', 'qcow',
> # 'qcow2', 'raw', 'tftp', 'vdi', 'vmdk', 'vpc', 'vvfat'
This comment is wrong; 'archipelago' did not exist in 0.14.0. Please
split the list into:
0.14.0 ... existing
2.2 'archipelago'
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 539 bytes --]
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Qemu-devel] [PATCH v7 5/5] qemu-iotests: add support for Archipelago protocol
2014-07-23 14:07 [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Chrysostomos Nanakos
` (3 preceding siblings ...)
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 4/5] QMP: Add support for Archipelago Chrysostomos Nanakos
@ 2014-07-23 14:07 ` Chrysostomos Nanakos
2014-07-30 15:58 ` [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Stefan Hajnoczi
2014-07-30 17:59 ` [Qemu-devel] [PATCH v8 4/5] QMP: Add support for Archipelago Chrysostomos Nanakos
6 siblings, 0 replies; 13+ messages in thread
From: Chrysostomos Nanakos @ 2014-07-23 14:07 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, Chrysostomos Nanakos, stefanha
Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
---
tests/qemu-iotests/common | 6 ++++++
tests/qemu-iotests/common.rc | 9 ++++++++-
2 files changed, 14 insertions(+), 1 deletion(-)
diff --git a/tests/qemu-iotests/common b/tests/qemu-iotests/common
index e4083f4..70df659 100644
--- a/tests/qemu-iotests/common
+++ b/tests/qemu-iotests/common
@@ -152,6 +152,7 @@ check options
-nbd test nbd
-ssh test ssh
-nfs test nfs
+ -archipelago test archipelago
-xdiff graphical mode diff
-nocache use O_DIRECT on backing file
-misalign misalign memory allocations
@@ -263,6 +264,11 @@ testlist options
xpand=false
;;
+ -archipelago)
+ IMGPROTO=archipelago
+ xpand=false
+ ;;
+
-nocache)
CACHEMODE="none"
CACHEMODE_IS_DEFAULT=false
diff --git a/tests/qemu-iotests/common.rc b/tests/qemu-iotests/common.rc
index e0ea7e3..3fd691e 100644
--- a/tests/qemu-iotests/common.rc
+++ b/tests/qemu-iotests/common.rc
@@ -64,6 +64,8 @@ elif [ "$IMGPROTO" = "ssh" ]; then
elif [ "$IMGPROTO" = "nfs" ]; then
TEST_DIR="nfs://127.0.0.1/$TEST_DIR"
TEST_IMG=$TEST_DIR/t.$IMGFMT
+elif [ "$IMGPROTO" = "archipelago" ]; then
+ TEST_IMG="archipelago:at.$IMGFMT"
else
TEST_IMG=$IMGPROTO:$TEST_DIR/t.$IMGFMT
fi
@@ -163,7 +165,8 @@ _make_test_img()
-e "s# lazy_refcounts=\\(on\\|off\\)##g" \
-e "s# block_size=[0-9]\\+##g" \
-e "s# block_state_zero=\\(on\\|off\\)##g" \
- -e "s# log_size=[0-9]\\+##g"
+ -e "s# log_size=[0-9]\\+##g" \
+ -e "s/archipelago:a/TEST_DIR\//g"
# Start an NBD server on the image file, which is what we'll be talking to
if [ $IMGPROTO = "nbd" ]; then
@@ -206,6 +209,10 @@ _cleanup_test_img()
rbd --no-progress rm "$TEST_DIR/t.$IMGFMT" > /dev/null
;;
+ archipelago)
+ vlmc remove "at.$IMGFMT" > /dev/null
+ ;;
+
sheepdog)
collie vdi delete "$TEST_DIR/t.$IMGFMT"
;;
--
1.7.10.4
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend
2014-07-23 14:07 [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Chrysostomos Nanakos
` (4 preceding siblings ...)
2014-07-23 14:07 ` [Qemu-devel] [PATCH v7 5/5] qemu-iotests: add support for Archipelago protocol Chrysostomos Nanakos
@ 2014-07-30 15:58 ` Stefan Hajnoczi
2014-07-30 17:59 ` [Qemu-devel] [PATCH v8 4/5] QMP: Add support for Archipelago Chrysostomos Nanakos
6 siblings, 0 replies; 13+ messages in thread
From: Stefan Hajnoczi @ 2014-07-30 15:58 UTC (permalink / raw)
To: Chrysostomos Nanakos; +Cc: kwolf, qemu-devel, stefanha
[-- Attachment #1: Type: text/plain, Size: 4221 bytes --]
On Wed, Jul 23, 2014 at 05:07:28PM +0300, Chrysostomos Nanakos wrote:
> v7:
> - Fix coding style issues.
> - Rename __archipelago_submit_request function to archipelago_submit_request.
> - Set X_NONBLOCK flag to xseg_receive().
> - Return -EIO to .bdrv_getlength() if archipelago_volume_info() fails.
> - Fix segment_name mem leak.
> - Bump version number from 2.1 to 2.2 in qapi/block-core.json file concerning
> QEMU Archipelago support.
> - Convert qemu_aio_wait() to aio_poll().
> - Remove qemu_blockalign() and memcpy() call and use qemu_iovec_to_buf()
> directly.
>
> v6:
> - Split v5 1/4 patch into two different patches. First one implements
> QMP structured options and the second one implements bdrv_parse_filename().
>
> v5:
> - Remove useless qemu_aio_count variable from BDRVArchipelagoState struct.
> - Cleanup xseg signal descriptor, call xseg_quit_local_signal() when closing
> block device.
> - Fix ds and volname leaks.
> - Make xseg request handler thread joinable and wait until exits before
> destroying condition variables and mutexes. Thanks to Stefan Hajnoczi for
> pointing this out.
> - Remove error_propagate() useless call.
> - Use memcpy instead of strncpy.
> - Remove check after trying to allocate memory with g_malloc().
> - Remove pipe code and complete AIO by introducing QEMU "bottom-half".
> - Add Archipelago shared memory segment name in options list and QMP.
> - Remove functions archipelago_aio_read()/_write() and introduce new
> and simpler function, __archipelago_submit_request().
> Refactor archipelago_aio_segmented_rw() function.
> - Enable Archipelago support in qemu-iotests
>
> v4:
> - Move Archipelago QMP support from qapi-schema.json file to
> qapi/block-core.json. Fixe various typographic errors, thanks to
> Kevin Wolf and Eric Blake.
> - Use new .create_opts format, define new QemuOptsList structure and refactor
> qemu_archipelago_create function.
>
> v3:
> - Break down initial patch from one to three. First patch implements
> Archipelago QEMU block backend with read/write functionality.
> Second patch implements .bdrv_create() and adds support for creating
> Archipelago images. Third patch adds QMP support.
> - Remove global variable g_xseg_init, make xseg_initialize(), xseg_join()
> and xseg_leave() reentrant and thread-safe.
> - Introduce new enum BlockdevOptionsArchipelago for the QMP support.
>
> v2:
> - Implement .bdrv_parse_filename() function to convert the shortuct version
> with a single string to the individual options.
> - Remove global variables and move relevant fields to ArchipelagoAIOCB struct.
> - Remove ArchipelagoConf struct and use the relevant fields as individual
> arguments.
> - Remove ArchipelagoCB struct and use ArchipelagoAIOCB instead.
> - Remove ArchipelagoThread struct and move relevant fields to
> ArchipelagoAIOCB instead. Now an I/O thread is spawned for per-device to
> handle all async I/O requests.
> - Remove double data copy, use qemu_iovec_from_buf() and copy data directly
> to the destination buffer.
> - Remove archipelago_aio_bh_cb() function, a full request is completed in
> qemu_archipelago_complete_aio() instead.
> - Resolve proposed changes from Kevin Wolf and miscellaneous style issues.
>
> Chrysostomos Nanakos (5):
> block: Support Archipelago as a QEMU block backend
> block/archipelago: Implement bdrv_parse_filename()
> block/archipelago: Add support for creating images
> QMP: Add support for Archipelago
> qemu-iotests: add support for Archipelago protocol
>
> MAINTAINERS | 6 +
> block/Makefile.objs | 2 +
> block/archipelago.c | 1064 ++++++++++++++++++++++++++++++++++++++++++
> configure | 40 ++
> qapi/block-core.json | 38 +-
> tests/qemu-iotests/common | 6 +
> tests/qemu-iotests/common.rc | 9 +-
> 7 files changed, 1161 insertions(+), 4 deletions(-)
> create mode 100644 block/archipelago.c
Thanks, applied to my block-next tree:
https://github.com/stefanha/qemu/commits/block-next
Stefan
[-- Attachment #2: Type: application/pgp-signature, Size: 473 bytes --]
^ permalink raw reply [flat|nested] 13+ messages in thread* [Qemu-devel] [PATCH v8 4/5] QMP: Add support for Archipelago
2014-07-23 14:07 [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Chrysostomos Nanakos
` (5 preceding siblings ...)
2014-07-30 15:58 ` [Qemu-devel] [PATCH v7 0/5] Support Archipelago as a QEMU block backend Stefan Hajnoczi
@ 2014-07-30 17:59 ` Chrysostomos Nanakos
2014-07-31 9:42 ` Stefan Hajnoczi
6 siblings, 1 reply; 13+ messages in thread
From: Chrysostomos Nanakos @ 2014-07-30 17:59 UTC (permalink / raw)
To: qemu-devel; +Cc: kwolf, Chrysostomos Nanakos, stefanha
Introduce new enum BlockdevOptionsArchipelago.
@volume: #Name of the Archipelago volume image
@mport: #'mport' is the port number on which mapperd is
listening. This is optional and if not specified,
QEMU will make Archipelago to use the default port.
@vport: #'vport' is the port number on which vlmcd is
listening. This is optional and if not specified,
QEMU will make Archipelago to use the default port.
@segment: #optional The name of the shared memory segment
Archipelago stack is using. This is optional
and if not specified, QEMU will make Archipelago
use the default value, 'archipelago'.
Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
---
qapi/block-core.json | 35 ++++++++++++++++++++++++++++++++++-
1 file changed, 34 insertions(+), 1 deletion(-)
diff --git a/qapi/block-core.json b/qapi/block-core.json
index e378653..d1c4fc2 100644
--- a/qapi/block-core.json
+++ b/qapi/block-core.json
@@ -194,6 +194,7 @@
# 'file', 'file', 'ftp', 'ftps', 'host_cdrom', 'host_device',
# 'host_floppy', 'http', 'https', 'nbd', 'parallels', 'qcow',
# 'qcow2', 'raw', 'tftp', 'vdi', 'vmdk', 'vpc', 'vvfat'
+# 2.2: 'archipelago'
#
# @backing_file: #optional the name of the backing file (for copy-on-write)
#
@@ -1143,7 +1144,7 @@
# Since: 2.0
##
{ 'enum': 'BlockdevDriver',
- 'data': [ 'file', 'host_device', 'host_cdrom', 'host_floppy',
+ 'data': [ 'archipelago', 'file', 'host_device', 'host_cdrom', 'host_floppy',
'http', 'https', 'ftp', 'ftps', 'tftp', 'vvfat', 'blkdebug',
'blkverify', 'bochs', 'cloop', 'cow', 'dmg', 'parallels', 'qcow',
'qcow2', 'qed', 'raw', 'vdi', 'vhdx', 'vmdk', 'vpc', 'quorum' ] }
@@ -1273,6 +1274,37 @@
'*pass-discard-snapshot': 'bool',
'*pass-discard-other': 'bool' } }
+
+##
+# @BlockdevOptionsArchipelago
+#
+# Driver specific block device options for Archipelago.
+#
+# @volume: Name of the Archipelago volume image
+#
+# @mport: #optional The port number on which mapperd is
+# listening. This is optional
+# and if not specified, QEMU will make Archipelago
+# use the default port (1001).
+#
+# @vport: #optional The port number on which vlmcd is
+# listening. This is optional
+# and if not specified, QEMU will make Archipelago
+# use the default port (501).
+#
+# @segment: #optional The name of the shared memory segment
+# Archipelago stack is using. This is optional
+# and if not specified, QEMU will make Archipelago
+# use the default value, 'archipelago'.
+# Since: 2.2
+##
+{ 'type': 'BlockdevOptionsArchipelago',
+ 'data': { 'volume': 'str',
+ '*mport': 'int',
+ '*vport': 'int',
+ '*segment': 'str' } }
+
+
##
# @BlkdebugEvent
#
@@ -1416,6 +1448,7 @@
'base': 'BlockdevOptionsBase',
'discriminator': 'driver',
'data': {
+ 'archipelago':'BlockdevOptionsArchipelago',
'file': 'BlockdevOptionsFile',
'host_device':'BlockdevOptionsFile',
'host_cdrom': 'BlockdevOptionsFile',
--
1.7.10.4
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [Qemu-devel] [PATCH v8 4/5] QMP: Add support for Archipelago
2014-07-30 17:59 ` [Qemu-devel] [PATCH v8 4/5] QMP: Add support for Archipelago Chrysostomos Nanakos
@ 2014-07-31 9:42 ` Stefan Hajnoczi
0 siblings, 0 replies; 13+ messages in thread
From: Stefan Hajnoczi @ 2014-07-31 9:42 UTC (permalink / raw)
To: Chrysostomos Nanakos; +Cc: kwolf, qemu-devel
[-- Attachment #1: Type: text/plain, Size: 1245 bytes --]
On Wed, Jul 30, 2014 at 08:59:09PM +0300, Chrysostomos Nanakos wrote:
> Introduce new enum BlockdevOptionsArchipelago.
>
> @volume: #Name of the Archipelago volume image
>
> @mport: #'mport' is the port number on which mapperd is
> listening. This is optional and if not specified,
> QEMU will make Archipelago to use the default port.
>
> @vport: #'vport' is the port number on which vlmcd is
> listening. This is optional and if not specified,
> QEMU will make Archipelago to use the default port.
>
> @segment: #optional The name of the shared memory segment
> Archipelago stack is using. This is optional
> and if not specified, QEMU will make Archipelago
> use the default value, 'archipelago'.
>
> Signed-off-by: Chrysostomos Nanakos <cnanakos@grnet.gr>
> ---
> qapi/block-core.json | 35 ++++++++++++++++++++++++++++++++++-
> 1 file changed, 34 insertions(+), 1 deletion(-)
Thanks, applied to my block-next tree and replaced v7 patch:
https://github.com/stefanha/qemu/commits/block-next
Stefan
[-- Attachment #2: Type: application/pgp-signature, Size: 473 bytes --]
^ permalink raw reply [flat|nested] 13+ messages in thread