2 * Copyright (c) 2011 NetApp, Inc.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
14 * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED. IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
32 #include <sys/param.h>
33 #include <sys/linker_set.h>
36 #include <sys/ioctl.h>
55 #define VTBLK_RINGSZ 64
57 #define VTBLK_MAXSEGS 32
60 #define VTBLK_S_IOERR 1
61 #define VTBLK_S_UNSUPP 2
63 #define VTBLK_BLK_ID_BYTES 20
66 #define VTBLK_F_SEG_MAX (1 << 2) /* Maximum request segments */
67 #define VTBLK_F_BLK_SIZE (1 << 6) /* cfg block size valid */
68 #define VTBLK_F_TOPOLOGY (1 << 10) /* Optimal I/O alignment */
73 #define VTBLK_S_HOSTCAPS \
77 VIRTIO_RING_F_INDIRECT_DESC ) /* indirect descriptors */
80 * Config space "registers"
83 uint64_t vbc_capacity;
84 uint32_t vbc_size_max;
91 uint32_t vbc_blk_size;
93 uint8_t physical_block_exp;
94 uint8_t alignment_offset;
98 uint8_t vbc_writeback;
102 * Fixed-size block header
104 struct virtio_blk_hdr {
105 #define VBH_OP_READ 0
106 #define VBH_OP_WRITE 1
107 #define VBH_OP_FLUSH 4
108 #define VBH_OP_FLUSH_OUT 5
109 #define VBH_OP_IDENT 8
110 #define VBH_FLAG_BARRIER 0x80000000 /* OR'ed into vbh_type */
119 static int pci_vtblk_debug;
120 #define DPRINTF(params) if (pci_vtblk_debug) printf params
121 #define WPRINTF(params) printf params
126 struct pci_vtblk_softc {
127 struct virtio_softc vbsc_vs;
128 pthread_mutex_t vsc_mtx;
129 struct vqueue_info vbsc_vq;
132 struct vtblk_config vbsc_cfg;
133 char vbsc_ident[VTBLK_BLK_ID_BYTES];
136 static void pci_vtblk_reset(void *);
137 static void pci_vtblk_notify(void *, struct vqueue_info *);
138 static int pci_vtblk_cfgread(void *, int, int, uint32_t *);
139 static int pci_vtblk_cfgwrite(void *, int, int, uint32_t);
141 static struct virtio_consts vtblk_vi_consts = {
142 "vtblk", /* our name */
143 1, /* we support 1 virtqueue */
144 sizeof(struct vtblk_config), /* config reg size */
145 pci_vtblk_reset, /* reset */
146 pci_vtblk_notify, /* device-wide qnotify */
147 pci_vtblk_cfgread, /* read PCI config */
148 pci_vtblk_cfgwrite, /* write PCI config */
149 NULL, /* apply negotiated features */
150 VTBLK_S_HOSTCAPS, /* our capabilities */
154 pci_vtblk_reset(void *vsc)
156 struct pci_vtblk_softc *sc = vsc;
158 DPRINTF(("vtblk: device reset requested !\n"));
159 vi_reset_dev(&sc->vbsc_vs);
163 pci_vtblk_proc(struct pci_vtblk_softc *sc, struct vqueue_info *vq)
165 struct virtio_blk_hdr *vbh;
172 struct iovec iov[VTBLK_MAXSEGS + 2];
173 uint16_t flags[VTBLK_MAXSEGS + 2];
175 n = vq_getchain(vq, iov, VTBLK_MAXSEGS + 2, flags);
178 * The first descriptor will be the read-only fixed header,
179 * and the last is for status (hence +2 above and below).
180 * The remaining iov's are the actual data I/O vectors.
182 * XXX - note - this fails on crash dump, which does a
183 * VIRTIO_BLK_T_FLUSH with a zero transfer length
185 assert(n >= 2 && n <= VTBLK_MAXSEGS + 2);
187 assert((flags[0] & VRING_DESC_F_WRITE) == 0);
188 assert(iov[0].iov_len == sizeof(struct virtio_blk_hdr));
189 vbh = iov[0].iov_base;
191 status = iov[--n].iov_base;
192 assert(iov[n].iov_len == 1);
193 assert(flags[n] & VRING_DESC_F_WRITE);
197 * The guest should not be setting the BARRIER flag because
198 * we don't advertise the capability.
200 type = vbh->vbh_type & ~VBH_FLAG_BARRIER;
201 writeop = (type == VBH_OP_WRITE);
203 offset = vbh->vbh_sector * DEV_BSIZE;
206 for (i = 1; i < n; i++) {
208 * - write op implies read-only descriptor,
209 * - read/ident op implies write-only descriptor,
210 * therefore test the inverse of the descriptor bit
213 assert(((flags[i] & VRING_DESC_F_WRITE) == 0) == writeop);
214 iolen += iov[i].iov_len;
217 DPRINTF(("virtio-block: %s op, %d bytes, %d segs, offset %ld\n\r",
218 writeop ? "write" : "read/ident", iolen, i - 1, offset));
223 if (pwritev(sc->vbsc_fd, iov + 1, i - 1, offset) < 0)
227 if (preadv(sc->vbsc_fd, iov + 1, i - 1, offset) < 0)
231 /* Assume a single buffer */
232 strlcpy(iov[1].iov_base, sc->vbsc_ident,
233 MIN(iov[1].iov_len, sizeof(sc->vbsc_ident)));
237 case VBH_OP_FLUSH_OUT:
238 if (sc->vbsc_ischr) {
239 if (ioctl(sc->vbsc_fd, DIOCGFLUSH))
241 } else if (fsync(sc->vbsc_fd))
249 /* convert errno into a virtio block error return */
251 *status = VTBLK_S_UNSUPP;
253 *status = VTBLK_S_IOERR;
255 *status = VTBLK_S_OK;
258 * Return the descriptor back to the host.
259 * We wrote 1 byte (our status) to host.
265 pci_vtblk_notify(void *vsc, struct vqueue_info *vq)
267 struct pci_vtblk_softc *sc = vsc;
270 while (vq_has_descs(vq))
271 pci_vtblk_proc(sc, vq);
272 vq_endchains(vq, 1); /* Generate interrupt if appropriate. */
276 pci_vtblk_init(struct vmctx *ctx, struct pci_devinst *pi, char *opts)
281 struct pci_vtblk_softc *sc;
282 off_t size, sts, sto;
287 printf("virtio-block: backing device required\n");
292 * The supplied backing file has to exist
294 fd = open(opts, O_RDWR);
296 perror("Could not open backing file");
300 if (fstat(fd, &sbuf) < 0) {
301 perror("Could not stat backing file");
307 * Deal with raw devices
312 if (S_ISCHR(sbuf.st_mode)) {
313 if (ioctl(fd, DIOCGMEDIASIZE, &size) < 0 ||
314 ioctl(fd, DIOCGSECTORSIZE, §sz)) {
315 perror("Could not fetch dev blk/sector size");
321 if (ioctl(fd, DIOCGSTRIPESIZE, &sts) == 0 && sts > 0)
322 ioctl(fd, DIOCGSTRIPEOFFSET, &sto);
324 sts = sbuf.st_blksize;
326 sc = calloc(1, sizeof(struct pci_vtblk_softc));
328 /* record fd of storage device/file */
330 sc->vbsc_ischr = S_ISCHR(sbuf.st_mode);
332 pthread_mutex_init(&sc->vsc_mtx, NULL);
334 /* init virtio softc and virtqueues */
335 vi_softc_linkup(&sc->vbsc_vs, &vtblk_vi_consts, sc, pi, &sc->vbsc_vq);
336 sc->vbsc_vs.vs_mtx = &sc->vsc_mtx;
338 sc->vbsc_vq.vq_qsize = VTBLK_RINGSZ;
339 /* sc->vbsc_vq.vq_notify = we have no per-queue notify */
342 * Create an identifier for the backing file. Use parts of the
343 * md5 sum of the filename
346 MD5Update(&mdctx, opts, strlen(opts));
347 MD5Final(digest, &mdctx);
348 sprintf(sc->vbsc_ident, "BHYVE-%02X%02X-%02X%02X-%02X%02X",
349 digest[0], digest[1], digest[2], digest[3], digest[4], digest[5]);
351 /* setup virtio block config space */
352 sc->vbsc_cfg.vbc_capacity = size / DEV_BSIZE; /* 512-byte units */
353 sc->vbsc_cfg.vbc_size_max = 0; /* not negotiated */
354 sc->vbsc_cfg.vbc_seg_max = VTBLK_MAXSEGS;
355 sc->vbsc_cfg.vbc_geometry.cylinders = 0; /* no geometry */
356 sc->vbsc_cfg.vbc_geometry.heads = 0;
357 sc->vbsc_cfg.vbc_geometry.sectors = 0;
358 sc->vbsc_cfg.vbc_blk_size = sectsz;
359 sc->vbsc_cfg.vbc_topology.physical_block_exp =
360 (sts > sectsz) ? (ffsll(sts / sectsz) - 1) : 0;
361 sc->vbsc_cfg.vbc_topology.alignment_offset =
362 (sto != 0) ? ((sts - sto) / sectsz) : 0;
363 sc->vbsc_cfg.vbc_topology.min_io_size = 0;
364 sc->vbsc_cfg.vbc_topology.opt_io_size = 0;
365 sc->vbsc_cfg.vbc_writeback = 0;
368 * Should we move some of this into virtio.c? Could
369 * have the device, class, and subdev_0 as fields in
370 * the virtio constants structure.
372 pci_set_cfgdata16(pi, PCIR_DEVICE, VIRTIO_DEV_BLOCK);
373 pci_set_cfgdata16(pi, PCIR_VENDOR, VIRTIO_VENDOR);
374 pci_set_cfgdata8(pi, PCIR_CLASS, PCIC_STORAGE);
375 pci_set_cfgdata16(pi, PCIR_SUBDEV_0, VIRTIO_TYPE_BLOCK);
377 pci_lintr_request(pi);
379 if (vi_intr_init(&sc->vbsc_vs, 1, fbsdrun_virtio_msix()))
381 vi_set_io_bar(&sc->vbsc_vs, 0);
386 pci_vtblk_cfgwrite(void *vsc, int offset, int size, uint32_t value)
389 DPRINTF(("vtblk: write to readonly reg %d\n\r", offset));
394 pci_vtblk_cfgread(void *vsc, int offset, int size, uint32_t *retval)
396 struct pci_vtblk_softc *sc = vsc;
399 /* our caller has already verified offset and size */
400 ptr = (uint8_t *)&sc->vbsc_cfg + offset;
401 memcpy(retval, ptr, size);
405 struct pci_devemu pci_de_vblk = {
406 .pe_emu = "virtio-blk",
407 .pe_init = pci_vtblk_init,
408 .pe_barwrite = vi_pci_write,
409 .pe_barread = vi_pci_read
411 PCI_EMUL_SET(pci_de_vblk);