From e076f3383b08a563d76c8beb9a716788a3987df9 Mon Sep 17 00:00:00 2001 From: Kevin Wolf Date: Tue, 29 Jun 2010 11:43:13 +0200 Subject: qemu-img check: Distinguish different kinds of errors People think that their images are corrupted when in fact there are just some leaked clusters. Differentiating several error cases should make the messages more comprehensible. Signed-off-by: Kevin Wolf --- block.c | 10 +++++++--- block.h | 10 +++++++++- qemu-img.c | 63 +++++++++++++++++++++++++++++++++++++++++++++++--------------- 3 files changed, 64 insertions(+), 19 deletions(-) diff --git a/block.c b/block.c index dd6dd76c6d..b0ceef0438 100644 --- a/block.c +++ b/block.c @@ -710,15 +710,19 @@ DeviceState *bdrv_get_attached(BlockDriverState *bs) /* * Run consistency checks on an image * - * Returns the number of errors or -errno when an internal error occurs + * Returns 0 if the check could be completed (it doesn't mean that the image is + * free of errors) or -errno when an internal error occured. The results of the + * check are stored in res. */ -int bdrv_check(BlockDriverState *bs) +int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res) { if (bs->drv->bdrv_check == NULL) { return -ENOTSUP; } - return bs->drv->bdrv_check(bs); + memset(res, 0, sizeof(*res)); + res->corruptions = bs->drv->bdrv_check(bs); + return res->corruptions < 0 ? res->corruptions : 0; } /* commit COW file into the raw image */ diff --git a/block.h b/block.h index 3d03b3e041..c2a7e4c762 100644 --- a/block.h +++ b/block.h @@ -74,7 +74,6 @@ void bdrv_close(BlockDriverState *bs); int bdrv_attach(BlockDriverState *bs, DeviceState *qdev); void bdrv_detach(BlockDriverState *bs, DeviceState *qdev); DeviceState *bdrv_get_attached(BlockDriverState *bs); -int bdrv_check(BlockDriverState *bs); int bdrv_read(BlockDriverState *bs, int64_t sector_num, uint8_t *buf, int nb_sectors); int bdrv_write(BlockDriverState *bs, int64_t sector_num, @@ -97,6 +96,15 @@ int bdrv_change_backing_file(BlockDriverState *bs, const char *backing_file, const char *backing_fmt); void bdrv_register(BlockDriver *bdrv); + +typedef struct BdrvCheckResult { + int corruptions; + int leaks; + int check_errors; +} BdrvCheckResult; + +int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res); + /* async block I/O */ typedef struct BlockDriverAIOCB BlockDriverAIOCB; typedef void BlockDriverCompletionFunc(void *opaque, int ret); diff --git a/qemu-img.c b/qemu-img.c index 700af21841..e300f911cb 100644 --- a/qemu-img.c +++ b/qemu-img.c @@ -425,11 +425,20 @@ out: return 0; } +/* + * Checks an image for consistency. Exit codes: + * + * 0 - Check completed, image is good + * 1 - Check not completed because of internal errors + * 2 - Check completed, image is corrupted + * 3 - Check completed, image has leaked clusters, but is good otherwise + */ static int img_check(int argc, char **argv) { int c, ret; const char *filename, *fmt; BlockDriverState *bs; + BdrvCheckResult result; fmt = NULL; for(;;) { @@ -453,28 +462,52 @@ static int img_check(int argc, char **argv) if (!bs) { return 1; } - ret = bdrv_check(bs); - switch(ret) { - case 0: - printf("No errors were found on the image.\n"); - break; - case -ENOTSUP: + ret = bdrv_check(bs, &result); + + if (ret == -ENOTSUP) { error("This image format does not support checks"); - break; - default: - if (ret < 0) { - error("An error occurred during the check"); - } else { - printf("%d errors were found on the image.\n", ret); + bdrv_delete(bs); + return 1; + } + + if (!(result.corruptions || result.leaks || result.check_errors)) { + printf("No errors were found on the image.\n"); + } else { + if (result.corruptions) { + printf("\n%d errors were found on the image.\n" + "Data may be corrupted, or further writes to the image " + "may corrupt it.\n", + result.corruptions); + } + + if (result.leaks) { + printf("\n%d leaked clusters were found on the image.\n" + "This means waste of disk space, but no harm to data.\n", + result.leaks); + } + + if (result.check_errors) { + printf("\n%d internal errors have occurred during the check.\n", + result.check_errors); } - break; } bdrv_delete(bs); - if (ret) { + + if (ret < 0 || result.check_errors) { + printf("\nAn error has occurred during the check: %s\n" + "The check is not complete and may have missed error.\n", + strerror(-ret)); return 1; } - return 0; + + if (result.corruptions) { + return 2; + } else if (result.leaks) { + return 3; + } else { + return 0; + } } static int img_commit(int argc, char **argv) -- cgit v1.2.1 From 9ac228e02cf16202547e7025ef300369e0db7781 Mon Sep 17 00:00:00 2001 From: Kevin Wolf Date: Tue, 29 Jun 2010 12:37:54 +0200 Subject: qcow2/vdi: Change check to distinguish error cases This distinguishes between harmless leaks and real corruption. Hopefully users better understand what qemu-img check wants to tell them. Signed-off-by: Kevin Wolf --- block.c | 3 +- block/qcow2-refcount.c | 120 ++++++++++++++++++++++++++----------------------- block/qcow2.c | 4 +- block/qcow2.h | 2 +- block/vdi.c | 10 ++--- block_int.h | 7 ++- 6 files changed, 79 insertions(+), 67 deletions(-) diff --git a/block.c b/block.c index b0ceef0438..65cf4dc9a4 100644 --- a/block.c +++ b/block.c @@ -721,8 +721,7 @@ int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res) } memset(res, 0, sizeof(*res)); - res->corruptions = bs->drv->bdrv_check(bs); - return res->corruptions < 0 ? res->corruptions : 0; + return bs->drv->bdrv_check(bs, res); } /* commit COW file into the raw image */ diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c index 4a96d986c9..4c19e7ebd8 100644 --- a/block/qcow2-refcount.c +++ b/block/qcow2-refcount.c @@ -884,9 +884,10 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, * This is used to construct a temporary refcount table out of L1 and L2 tables * which can be compared the the refcount table saved in the image. * - * Returns the number of errors in the image that were found + * Modifies the number of errors in res. */ -static int inc_refcounts(BlockDriverState *bs, +static void inc_refcounts(BlockDriverState *bs, + BdrvCheckResult *res, uint16_t *refcount_table, int refcount_table_size, int64_t offset, int64_t size) @@ -894,30 +895,32 @@ static int inc_refcounts(BlockDriverState *bs, BDRVQcowState *s = bs->opaque; int64_t start, last, cluster_offset; int k; - int errors = 0; if (size <= 0) - return 0; + return; start = offset & ~(s->cluster_size - 1); last = (offset + size - 1) & ~(s->cluster_size - 1); for(cluster_offset = start; cluster_offset <= last; cluster_offset += s->cluster_size) { k = cluster_offset >> s->cluster_bits; - if (k < 0 || k >= refcount_table_size) { + if (k < 0) { fprintf(stderr, "ERROR: invalid cluster offset=0x%" PRIx64 "\n", cluster_offset); - errors++; + res->corruptions++; + } else if (k >= refcount_table_size) { + fprintf(stderr, "Warning: cluster offset=0x%" PRIx64 " is after " + "the end of the image file, can't properly check refcounts.\n", + cluster_offset); + res->check_errors++; } else { if (++refcount_table[k] == 0) { fprintf(stderr, "ERROR: overflow cluster offset=0x%" PRIx64 "\n", cluster_offset); - errors++; + res->corruptions++; } } } - - return errors; } /* @@ -928,14 +931,13 @@ static int inc_refcounts(BlockDriverState *bs, * Returns the number of errors found by the checks or -errno if an internal * error occurred. */ -static int check_refcounts_l2(BlockDriverState *bs, +static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res, uint16_t *refcount_table, int refcount_table_size, int64_t l2_offset, int check_copied) { BDRVQcowState *s = bs->opaque; uint64_t *l2_table, offset; int i, l2_size, nb_csectors, refcount; - int errors = 0; /* Read L2 table from disk */ l2_size = s->l2_size * sizeof(uint64_t); @@ -955,16 +957,15 @@ static int check_refcounts_l2(BlockDriverState *bs, "copied flag must never be set for compressed " "clusters\n", offset >> s->cluster_bits); offset &= ~QCOW_OFLAG_COPIED; - errors++; + res->corruptions++; } /* Mark cluster as used */ nb_csectors = ((offset >> s->csize_shift) & s->csize_mask) + 1; offset &= s->cluster_offset_mask; - errors += inc_refcounts(bs, refcount_table, - refcount_table_size, - offset & ~511, nb_csectors * 512); + inc_refcounts(bs, res, refcount_table, refcount_table_size, + offset & ~511, nb_csectors * 512); } else { /* QCOW_OFLAG_COPIED must be set iff refcount == 1 */ if (check_copied) { @@ -974,35 +975,35 @@ static int check_refcounts_l2(BlockDriverState *bs, if (refcount < 0) { fprintf(stderr, "Can't get refcount for offset %" PRIx64 ": %s\n", entry, strerror(-refcount)); + goto fail; } if ((refcount == 1) != ((entry & QCOW_OFLAG_COPIED) != 0)) { fprintf(stderr, "ERROR OFLAG_COPIED: offset=%" PRIx64 " refcount=%d\n", entry, refcount); - errors++; + res->corruptions++; } } /* Mark cluster as used */ offset &= ~QCOW_OFLAG_COPIED; - errors += inc_refcounts(bs, refcount_table, - refcount_table_size, - offset, s->cluster_size); + inc_refcounts(bs, res, refcount_table,refcount_table_size, + offset, s->cluster_size); /* Correct offsets are cluster aligned */ if (offset & (s->cluster_size - 1)) { fprintf(stderr, "ERROR offset=%" PRIx64 ": Cluster is not " "properly aligned; L2 entry corrupted.\n", offset); - errors++; + res->corruptions++; } } } } qemu_free(l2_table); - return errors; + return 0; fail: - fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n"); + fprintf(stderr, "ERROR: I/O error in check_refcounts_l2\n"); qemu_free(l2_table); return -EIO; } @@ -1016,6 +1017,7 @@ fail: * error occurred. */ static int check_refcounts_l1(BlockDriverState *bs, + BdrvCheckResult *res, uint16_t *refcount_table, int refcount_table_size, int64_t l1_table_offset, int l1_size, @@ -1024,13 +1026,12 @@ static int check_refcounts_l1(BlockDriverState *bs, BDRVQcowState *s = bs->opaque; uint64_t *l1_table, l2_offset, l1_size2; int i, refcount, ret; - int errors = 0; l1_size2 = l1_size * sizeof(uint64_t); /* Mark L1 table as used */ - errors += inc_refcounts(bs, refcount_table, refcount_table_size, - l1_table_offset, l1_size2); + inc_refcounts(bs, res, refcount_table, refcount_table_size, + l1_table_offset, l1_size2); /* Read L1 table entries from disk */ if (l1_size2 == 0) { @@ -1055,42 +1056,41 @@ static int check_refcounts_l1(BlockDriverState *bs, if (refcount < 0) { fprintf(stderr, "Can't get refcount for l2_offset %" PRIx64 ": %s\n", l2_offset, strerror(-refcount)); + goto fail; } if ((refcount == 1) != ((l2_offset & QCOW_OFLAG_COPIED) != 0)) { fprintf(stderr, "ERROR OFLAG_COPIED: l2_offset=%" PRIx64 " refcount=%d\n", l2_offset, refcount); - errors++; + res->corruptions++; } } /* Mark L2 table as used */ l2_offset &= ~QCOW_OFLAG_COPIED; - errors += inc_refcounts(bs, refcount_table, - refcount_table_size, - l2_offset, - s->cluster_size); + inc_refcounts(bs, res, refcount_table, refcount_table_size, + l2_offset, s->cluster_size); /* L2 tables are cluster aligned */ if (l2_offset & (s->cluster_size - 1)) { fprintf(stderr, "ERROR l2_offset=%" PRIx64 ": Table is not " "cluster aligned; L1 entry corrupted\n", l2_offset); - errors++; + res->corruptions++; } /* Process and check L2 entries */ - ret = check_refcounts_l2(bs, refcount_table, refcount_table_size, - l2_offset, check_copied); + ret = check_refcounts_l2(bs, res, refcount_table, + refcount_table_size, l2_offset, check_copied); if (ret < 0) { goto fail; } - errors += ret; } } qemu_free(l1_table); - return errors; + return 0; fail: fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n"); + res->check_errors++; qemu_free(l1_table); return -EIO; } @@ -1101,44 +1101,47 @@ fail: * Returns 0 if no errors are found, the number of errors in case the image is * detected as corrupted, and -errno when an internal error occured. */ -int qcow2_check_refcounts(BlockDriverState *bs) +int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res) { BDRVQcowState *s = bs->opaque; int64_t size; int nb_clusters, refcount1, refcount2, i; QCowSnapshot *sn; uint16_t *refcount_table; - int ret, errors = 0; + int ret; size = bdrv_getlength(bs->file); nb_clusters = size_to_clusters(s, size); refcount_table = qemu_mallocz(nb_clusters * sizeof(uint16_t)); /* header */ - errors += inc_refcounts(bs, refcount_table, nb_clusters, - 0, s->cluster_size); + inc_refcounts(bs, res, refcount_table, nb_clusters, + 0, s->cluster_size); /* current L1 table */ - ret = check_refcounts_l1(bs, refcount_table, nb_clusters, + ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters, s->l1_table_offset, s->l1_size, 1); if (ret < 0) { return ret; } - errors += ret; /* snapshots */ for(i = 0; i < s->nb_snapshots; i++) { sn = s->snapshots + i; - check_refcounts_l1(bs, refcount_table, nb_clusters, - sn->l1_table_offset, sn->l1_size, 0); + ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters, + sn->l1_table_offset, sn->l1_size, 0); + if (ret < 0) { + return ret; + } } - errors += inc_refcounts(bs, refcount_table, nb_clusters, - s->snapshots_offset, s->snapshots_size); + inc_refcounts(bs, res, refcount_table, nb_clusters, + s->snapshots_offset, s->snapshots_size); /* refcount data */ - errors += inc_refcounts(bs, refcount_table, nb_clusters, - s->refcount_table_offset, - s->refcount_table_size * sizeof(uint64_t)); + inc_refcounts(bs, res, refcount_table, nb_clusters, + s->refcount_table_offset, + s->refcount_table_size * sizeof(uint64_t)); + for(i = 0; i < s->refcount_table_size; i++) { uint64_t offset, cluster; offset = s->refcount_table[i]; @@ -1148,22 +1151,23 @@ int qcow2_check_refcounts(BlockDriverState *bs) if (offset & (s->cluster_size - 1)) { fprintf(stderr, "ERROR refcount block %d is not " "cluster aligned; refcount table entry corrupted\n", i); - errors++; + res->corruptions++; continue; } if (cluster >= nb_clusters) { fprintf(stderr, "ERROR refcount block %d is outside image\n", i); - errors++; + res->corruptions++; continue; } if (offset != 0) { - errors += inc_refcounts(bs, refcount_table, nb_clusters, - offset, s->cluster_size); + inc_refcounts(bs, res, refcount_table, nb_clusters, + offset, s->cluster_size); if (refcount_table[cluster] != 1) { fprintf(stderr, "ERROR refcount block %d refcount=%d\n", i, refcount_table[cluster]); + res->corruptions++; } } } @@ -1174,19 +1178,25 @@ int qcow2_check_refcounts(BlockDriverState *bs) if (refcount1 < 0) { fprintf(stderr, "Can't get refcount for cluster %d: %s\n", i, strerror(-refcount1)); + res->check_errors++; continue; } refcount2 = refcount_table[i]; if (refcount1 != refcount2) { - fprintf(stderr, "ERROR cluster %d refcount=%d reference=%d\n", + fprintf(stderr, "%s cluster %d refcount=%d reference=%d\n", + refcount1 < refcount2 ? "ERROR" : "Leaked", i, refcount1, refcount2); - errors++; + if (refcount1 < refcount2) { + res->corruptions++; + } else { + res->leaks++; + } } } qemu_free(refcount_table); - return errors; + return 0; } diff --git a/block/qcow2.c b/block/qcow2.c index 9ee34b6dd0..a53014dbda 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -1239,9 +1239,9 @@ static int qcow_get_info(BlockDriverState *bs, BlockDriverInfo *bdi) } -static int qcow_check(BlockDriverState *bs) +static int qcow_check(BlockDriverState *bs, BdrvCheckResult *result) { - return qcow2_check_refcounts(bs); + return qcow2_check_refcounts(bs, result); } #if 0 diff --git a/block/qcow2.h b/block/qcow2.h index c59b827da8..3ff162efcd 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -185,7 +185,7 @@ void qcow2_create_refcount_update(QCowCreateState *s, int64_t offset, int qcow2_update_snapshot_refcount(BlockDriverState *bs, int64_t l1_table_offset, int l1_size, int addend); -int qcow2_check_refcounts(BlockDriverState *bs); +int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res); /* qcow2-cluster.c functions */ int qcow2_grow_l1_table(BlockDriverState *bs, int min_size); diff --git a/block/vdi.c b/block/vdi.c index ee8cc7b1aa..f72633cf19 100644 --- a/block/vdi.c +++ b/block/vdi.c @@ -291,11 +291,10 @@ static void vdi_header_print(VdiHeader *header) } #endif -static int vdi_check(BlockDriverState *bs) +static int vdi_check(BlockDriverState *bs, BdrvCheckResult *res) { /* TODO: additional checks possible. */ BDRVVdiState *s = (BDRVVdiState *)bs->opaque; - int n_errors = 0; uint32_t blocks_allocated = 0; uint32_t block; uint32_t *bmap; @@ -315,11 +314,12 @@ static int vdi_check(BlockDriverState *bs) } else { fprintf(stderr, "ERROR: block index %" PRIu32 " also used by %" PRIu32 "\n", bmap[bmap_entry], bmap_entry); + res->corruptions++; } } else { fprintf(stderr, "ERROR: block index %" PRIu32 " too large, is %" PRIu32 "\n", block, bmap_entry); - n_errors++; + res->corruptions++; } } } @@ -327,12 +327,12 @@ static int vdi_check(BlockDriverState *bs) fprintf(stderr, "ERROR: allocated blocks mismatch, is %" PRIu32 ", should be %" PRIu32 "\n", blocks_allocated, s->header.blocks_allocated); - n_errors++; + res->corruptions++; } qemu_free(bmap); - return n_errors; + return 0; } static int vdi_get_info(BlockDriverState *bs, BlockDriverInfo *bdi) diff --git a/block_int.h b/block_int.h index a94b80152f..877e1e5943 100644 --- a/block_int.h +++ b/block_int.h @@ -119,8 +119,11 @@ struct BlockDriver { QEMUOptionParameter *create_options; - /* Returns number of errors in image, -errno for internal errors */ - int (*bdrv_check)(BlockDriverState* bs); + /* + * Returns 0 for completed check, -errno for internal errors. + * The check results are stored in result. + */ + int (*bdrv_check)(BlockDriverState* bs, BdrvCheckResult *result); void (*bdrv_debug_event)(BlockDriverState *bs, BlkDebugEvent event); -- cgit v1.2.1 From 2930b313dd602d67a568815b0b031b824916cec9 Mon Sep 17 00:00:00 2001 From: john cooper Date: Fri, 2 Jul 2010 13:44:25 -0400 Subject: Add virtio disk identification support This patch adds the final missing bits for support of passing a serial/id string to a virtio-blk guest driver. The guest-side component already exists in the virtio driver, and has recently been reworked by Ryan to export a /sys interface for retrieval of the id from guest userland. Signed-off-by: john cooper Signed-off-by: Kevin Wolf --- hw/virtio-blk.c | 14 ++++++++++++++ hw/virtio-blk.h | 3 +++ 2 files changed, 17 insertions(+) diff --git a/hw/virtio-blk.c b/hw/virtio-blk.c index f0b3ba5cb1..0bd57b5055 100644 --- a/hw/virtio-blk.c +++ b/hw/virtio-blk.c @@ -26,6 +26,7 @@ typedef struct VirtIOBlock QEMUBH *bh; BlockConf *conf; unsigned short sector_mask; + char sn[BLOCK_SERIAL_STRLEN]; } VirtIOBlock; static VirtIOBlock *to_virtio_blk(VirtIODevice *vdev) @@ -324,6 +325,12 @@ static void virtio_blk_handle_request(VirtIOBlockReq *req, virtio_blk_handle_flush(req, mrb); } else if (req->out->type & VIRTIO_BLK_T_SCSI_CMD) { virtio_blk_handle_scsi(req); + } else if (req->out->type & VIRTIO_BLK_T_GET_ID) { + VirtIOBlock *s = req->dev; + + memcpy(req->elem.in_sg[0].iov_base, s->sn, + MIN(req->elem.in_sg[0].iov_len, sizeof(s->sn))); + virtio_blk_req_complete(req, VIRTIO_BLK_S_OK); } else if (req->out->type & VIRTIO_BLK_T_OUT) { qemu_iovec_init_external(&req->qiov, &req->elem.out_sg[1], req->elem.out_num - 1); @@ -481,6 +488,7 @@ VirtIODevice *virtio_blk_init(DeviceState *dev, BlockConf *conf) VirtIOBlock *s; int cylinders, heads, secs; static int virtio_blk_id; + DriveInfo *dinfo; s = (VirtIOBlock *)virtio_common_init("virtio-blk", VIRTIO_ID_BLOCK, sizeof(struct virtio_blk_config), @@ -495,6 +503,12 @@ VirtIODevice *virtio_blk_init(DeviceState *dev, BlockConf *conf) s->sector_mask = (s->conf->logical_block_size / BDRV_SECTOR_SIZE) - 1; bdrv_guess_geometry(s->bs, &cylinders, &heads, &secs); + /* NB: per existing s/n string convention the string is terminated + * by '\0' only when less than sizeof (s->sn) + */ + dinfo = drive_get_by_blockdev(s->bs); + strncpy(s->sn, dinfo->serial, sizeof (s->sn)); + s->vq = virtio_add_queue(&s->vdev, 128, virtio_blk_handle_output); qemu_add_vm_change_state_handler(virtio_blk_dma_restart_cb, s); diff --git a/hw/virtio-blk.h b/hw/virtio-blk.h index 7a7ece3d93..fff46da7db 100644 --- a/hw/virtio-blk.h +++ b/hw/virtio-blk.h @@ -59,6 +59,9 @@ struct virtio_blk_config /* Flush the volatile write cache */ #define VIRTIO_BLK_T_FLUSH 4 +/* return the device ID string */ +#define VIRTIO_BLK_T_GET_ID 8 + /* Barrier before this op. */ #define VIRTIO_BLK_T_BARRIER 0x80000000 -- cgit v1.2.1 From 528f7663473e146fd1b435d73ca21b219451877c Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Fri, 25 Jun 2010 15:26:48 +0200 Subject: blockdev: Clean up how readonly persists across virtual media change Since commit cb4e5f8e, monitor command change makes the new media readonly iff the type hint is BDRV_TYPE_CDROM, i.e. the drive was created with media=cdrom. The intention is to avoid changing a block device's read-only-ness. However, BDRV_TYPE_CDROM is only a hint. It is currently sufficent for read-only. But it's not necessary, and it may not remain sufficient. Use bdrv_is_read_only() instead. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- blockdev.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/blockdev.c b/blockdev.c index be88098d53..0a9dec364e 100644 --- a/blockdev.c +++ b/blockdev.c @@ -589,7 +589,7 @@ int do_change_block(Monitor *mon, const char *device, if (eject_device(mon, bs, 0) < 0) { return -1; } - bdrv_flags = bdrv_get_type_hint(bs) == BDRV_TYPE_CDROM ? 0 : BDRV_O_RDWR; + bdrv_flags = bdrv_is_read_only(bs) ? 0 : BDRV_O_RDWR; if (bdrv_open(bs, filename, bdrv_flags, drv) < 0) { qerror_report(QERR_OPEN_FILE_FAILED, filename); return -1; -- cgit v1.2.1 From d246673dcb9911218ff555bcdf28b250e38fa46c Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Mon, 28 Jun 2010 10:45:02 +0200 Subject: block migration: Fix test for read-only drive init_blk_migration_it() skips drives with type hint BDRV_TYPE_CDROM. The intention is to skip read-only drives. However, BDRV_TYPE_CDROM is only a hint. It is currently sufficent for read-only. But it's not necessary, and it may not remain sufficient. Use bdrv_is_read_only() instead. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- block-migration.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/block-migration.c b/block-migration.c index 7d04d6db0a..7337349e83 100644 --- a/block-migration.c +++ b/block-migration.c @@ -236,7 +236,7 @@ static void init_blk_migration_it(void *opaque, BlockDriverState *bs) BlkMigDevState *bmds; int64_t sectors; - if (bs->type == BDRV_TYPE_HD) { + if (!bdrv_is_read_only(bs)) { sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS; if (sectors == 0) { return; -- cgit v1.2.1 From 65d21bc73bda6515fd9b4ff5b2e90454f7a0b419 Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Mon, 28 Jun 2010 10:25:20 +0200 Subject: raw-posix: Fix test for host CD-ROM raw_pread_aligned() retries up to two times if the block device backs a virtual CD-ROM (a drive with media=cdrom and if=ide, scsi, xen or none). This makes no sense. Whether retrying reads can correct read errors can only depend on what we're reading, not on how the result gets used. We need to check what whether we're reading from a physical CD-ROM or floppy here. I doubt retrying is useful even then. Left for another day. Impact: * Virtual CD-ROM backed by host_cdrom behaves the same. * Virtual CD-ROM backed by file or host_device no longer retries. * A drive backed by host_cdrom now retries even if it's not a virtual CD-ROM. * Any drive backed by host_floppy now retries. While there, clean up gratuitous use of goto. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- block/raw-posix.c | 17 ++++++----------- 1 file changed, 6 insertions(+), 11 deletions(-) diff --git a/block/raw-posix.c b/block/raw-posix.c index 3f0701b8a4..291699fbc3 100644 --- a/block/raw-posix.c +++ b/block/raw-posix.c @@ -242,15 +242,14 @@ static int raw_pread_aligned(BlockDriverState *bs, int64_t offset, ret = pread(s->fd, buf, count, offset); if (ret == count) - goto label__raw_read__success; + return ret; /* Allow reads beyond the end (needed for pwrite) */ if ((ret == 0) && bs->growable) { int64_t size = raw_getlength(bs); if (offset >= size) { memset(buf, 0, count); - ret = count; - goto label__raw_read__success; + return count; } } @@ -260,13 +259,13 @@ static int raw_pread_aligned(BlockDriverState *bs, int64_t offset, bs->total_sectors, ret, errno, strerror(errno)); /* Try harder for CDrom. */ - if (bs->type == BDRV_TYPE_CDROM) { + if (s->type != FTYPE_FILE) { ret = pread(s->fd, buf, count, offset); if (ret == count) - goto label__raw_read__success; + return ret; ret = pread(s->fd, buf, count, offset); if (ret == count) - goto label__raw_read__success; + return ret; DEBUG_BLOCK_PRINT("raw_pread(%d:%s, %" PRId64 ", %p, %d) [%" PRId64 "] retry read failed %d : %d = %s\n", @@ -274,8 +273,6 @@ static int raw_pread_aligned(BlockDriverState *bs, int64_t offset, bs->total_sectors, ret, errno, strerror(errno)); } -label__raw_read__success: - return (ret < 0) ? -errno : ret; } @@ -298,15 +295,13 @@ static int raw_pwrite_aligned(BlockDriverState *bs, int64_t offset, ret = pwrite(s->fd, buf, count, offset); if (ret == count) - goto label__raw_write__success; + return ret; DEBUG_BLOCK_PRINT("raw_pwrite(%d:%s, %" PRId64 ", %p, %d) [%" PRId64 "] write failed %d : %d = %s\n", s->fd, bs->filename, offset, buf, count, bs->total_sectors, ret, errno, strerror(errno)); -label__raw_write__success: - return (ret < 0) ? -errno : ret; } -- cgit v1.2.1 From b47b35250fbfa062aedf6ab6e5faab84c4a76f4f Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Thu, 27 May 2010 20:06:12 +0200 Subject: fdc: Reject unimplemented error actions drive_init() doesn't permit them for if=floppy, but that's worthless: we get them via if=none and -global. This can make device initialization fail. Since all callers of fdctrl_init_isa() ignore its value, change it to die instead of returning failure. Without this, some callers would ignore the failure, and others would crash. Wart: unlike drive_init(), we don't reject the default action when it's explicitly specified. That's because we can't distinguish "no rerror option" from "rerror=report", or "no werror" from "rerror=enospc". Left for another day. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/fdc.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/hw/fdc.c b/hw/fdc.c index 6c748782e3..2d50bd6a39 100644 --- a/hw/fdc.c +++ b/hw/fdc.c @@ -29,6 +29,7 @@ #include "hw.h" #include "fdc.h" +#include "qemu-error.h" #include "qemu-timer.h" #include "isa.h" #include "sysbus.h" @@ -1844,7 +1845,7 @@ static void fdctrl_result_timer(void *opaque) } /* Init functions */ -static void fdctrl_connect_drives(FDCtrl *fdctrl) +static int fdctrl_connect_drives(FDCtrl *fdctrl) { unsigned int i; FDrive *drive; @@ -1852,12 +1853,24 @@ static void fdctrl_connect_drives(FDCtrl *fdctrl) for (i = 0; i < MAX_FD; i++) { drive = &fdctrl->drives[i]; + if (drive->bs) { + if (bdrv_get_on_error(drive->bs, 0) != BLOCK_ERR_STOP_ENOSPC) { + error_report("fdc doesn't support drive option werror"); + return -1; + } + if (bdrv_get_on_error(drive->bs, 1) != BLOCK_ERR_REPORT) { + error_report("fdc doesn't support drive option rerror"); + return -1; + } + } + fd_init(drive); fd_revalidate(drive); if (drive->bs) { bdrv_set_removable(drive->bs, 1); } } + return 0; } FDCtrl *fdctrl_init_isa(DriveInfo **fds) @@ -1871,8 +1884,7 @@ FDCtrl *fdctrl_init_isa(DriveInfo **fds) if (fds[1]) { qdev_prop_set_drive_nofail(&dev->qdev, "driveB", fds[1]->bdrv); } - if (qdev_init(&dev->qdev) < 0) - return NULL; + qdev_init_nofail(&dev->qdev); return &(DO_UPCAST(FDCtrlISABus, busdev, dev)->state); } @@ -1950,9 +1962,7 @@ static int fdctrl_init_common(FDCtrl *fdctrl) if (fdctrl->dma_chann != -1) DMA_register_channel(fdctrl->dma_chann, &fdctrl_transfer_handler, fdctrl); - fdctrl_connect_drives(fdctrl); - - return 0; + return fdctrl_connect_drives(fdctrl); } static int isabus_fdc_init1(ISADevice *dev) -- cgit v1.2.1 From bd6c9a617d969752c9d3663f6ad29ae6d6d6c402 Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Thu, 27 May 2010 21:23:08 +0200 Subject: qdev: Don't hw_error() in qdev_init_nofail() Some of the failures are internal errors, and hw_error() is okay then. But the common way to fail is bad user input, e.g. -global isa-fdc.driveA=foo where drive foo has an unsupported rerror value. exit(1) instead. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/qdev.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/hw/qdev.c b/hw/qdev.c index 61f999c023..00ceada615 100644 --- a/hw/qdev.c +++ b/hw/qdev.c @@ -326,8 +326,10 @@ void qdev_init_nofail(DeviceState *dev) { DeviceInfo *info = dev->info; - if (qdev_init(dev) < 0) - hw_error("Initialization of device %s failed\n", info->name); + if (qdev_init(dev) < 0) { + error_report("Initialization of device %s failed\n", info->name); + exit(1); + } } /* Unlink device from bus and free the structure. */ -- cgit v1.2.1 From 620f862e1c3d0862923997c8d6632c5916da997c Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Thu, 27 May 2010 20:02:28 +0200 Subject: scsi: Reject unimplemented error actions drive_init() doesn't permit rerror for if=scsi, but that's worthless: we get it via if=none and -device. Moreover, scsi-generic doesn't support werror. Since drive_init() doesn't catch that, option werror was silently ignored even with if=scsi. Wart: unlike drive_init(), we don't reject the default action when it's explicitly specified. That's because we can't distinguish "no rerror option" from "rerror=report", or "no werror" from "rerror=enospc". Left for another day. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/scsi-disk.c | 5 +++++ hw/scsi-generic.c | 9 +++++++++ 2 files changed, 14 insertions(+) diff --git a/hw/scsi-disk.c b/hw/scsi-disk.c index 3e41011ccb..c30709c550 100644 --- a/hw/scsi-disk.c +++ b/hw/scsi-disk.c @@ -1059,6 +1059,11 @@ static int scsi_disk_initfn(SCSIDevice *dev) s->bs = s->qdev.conf.bs; is_cd = bdrv_get_type_hint(s->bs) == BDRV_TYPE_CDROM; + if (bdrv_get_on_error(s->bs, 1) != BLOCK_ERR_REPORT) { + error_report("Device doesn't support drive option rerror"); + return -1; + } + if (!s->serial) { /* try to fall back to value set with legacy -drive serial=... */ dinfo = drive_get_by_blockdev(s->bs); diff --git a/hw/scsi-generic.c b/hw/scsi-generic.c index 3915e7844e..a8b4176d80 100644 --- a/hw/scsi-generic.c +++ b/hw/scsi-generic.c @@ -474,6 +474,15 @@ static int scsi_generic_initfn(SCSIDevice *dev) return -1; } + if (bdrv_get_on_error(s->bs, 0) != BLOCK_ERR_STOP_ENOSPC) { + error_report("Device doesn't support drive option werror"); + return -1; + } + if (bdrv_get_on_error(s->bs, 1) != BLOCK_ERR_REPORT) { + error_report("Device doesn't support drive option rerror"); + return -1; + } + /* check we are using a driver managing SG_IO (version 3 and after */ if (bdrv_ioctl(s->bs, SG_GET_VERSION_NUM, &sg_version) < 0 || sg_version < 30000) { -- cgit v1.2.1 From 94ac726842d37b50c7742ecb643bebe2e5d72cd2 Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Thu, 27 May 2010 21:06:04 +0200 Subject: error: New qemu_opts_loc_restore() Needed for decent error locations when complaining about options outside of qemu_opts_foreach(). That one sets the location already. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- qemu-option.c | 5 +++++ qemu-option.h | 1 + 2 files changed, 6 insertions(+) diff --git a/qemu-option.c b/qemu-option.c index 30327d4804..1f8f41ae7d 100644 --- a/qemu-option.c +++ b/qemu-option.c @@ -728,6 +728,11 @@ void qemu_opts_reset(QemuOptsList *list) } } +void qemu_opts_loc_restore(QemuOpts *opts) +{ + loc_restore(&opts->loc); +} + int qemu_opts_set(QemuOptsList *list, const char *id, const char *name, const char *value) { diff --git a/qemu-option.h b/qemu-option.h index 9e2406c562..b515813891 100644 --- a/qemu-option.h +++ b/qemu-option.h @@ -116,6 +116,7 @@ int qemu_opt_foreach(QemuOpts *opts, qemu_opt_loopfunc func, void *opaque, QemuOpts *qemu_opts_find(QemuOptsList *list, const char *id); QemuOpts *qemu_opts_create(QemuOptsList *list, const char *id, int fail_if_exists); void qemu_opts_reset(QemuOptsList *list); +void qemu_opts_loc_restore(QemuOpts *opts); int qemu_opts_set(QemuOptsList *list, const char *id, const char *name, const char *value); const char *qemu_opts_id(QemuOpts *opts); -- cgit v1.2.1 From 42e766a24b53ff5042576449dd526f734ca8f659 Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Thu, 27 May 2010 21:08:18 +0200 Subject: scsi: Error locations for -drive if=scsi device initialization Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/scsi-bus.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/hw/scsi-bus.c b/hw/scsi-bus.c index b84b9b98b5..d69c74c4ef 100644 --- a/hw/scsi-bus.c +++ b/hw/scsi-bus.c @@ -102,19 +102,23 @@ SCSIDevice *scsi_bus_legacy_add_drive(SCSIBus *bus, BlockDriverState *bdrv, int int scsi_bus_legacy_handle_cmdline(SCSIBus *bus) { + Location loc; DriveInfo *dinfo; int res = 0, unit; + loc_push_none(&loc); for (unit = 0; unit < MAX_SCSI_DEVS; unit++) { dinfo = drive_get(IF_SCSI, bus->busnr, unit); if (dinfo == NULL) { continue; } + qemu_opts_loc_restore(dinfo->opts); if (!scsi_bus_legacy_add_drive(bus, dinfo->bdrv, unit)) { res = -1; break; } } + loc_pop(&loc); return res; } -- cgit v1.2.1 From f597627ff5eb683501d65cf169f467bb4e894626 Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Mon, 28 Jun 2010 08:36:53 +0200 Subject: ide: Improve error messages Use error_report(), because it points to the error location. Reword "tried to assign twice" messages to make it clear that we're complaining about the unit property. Report invalid unit property instead of failing silently. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/ide/qdev.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/hw/ide/qdev.c b/hw/ide/qdev.c index 2977a168e5..221f387af1 100644 --- a/hw/ide/qdev.c +++ b/hw/ide/qdev.c @@ -18,7 +18,7 @@ */ #include #include "dma.h" - +#include "qemu-error.h" #include /* --------------------------------- */ @@ -40,7 +40,7 @@ static int ide_qdev_init(DeviceState *qdev, DeviceInfo *base) IDEBus *bus = DO_UPCAST(IDEBus, qbus, qdev->parent_bus); if (!dev->conf.bs) { - fprintf(stderr, "%s: no drive specified\n", qdev->info->name); + error_report("No drive specified"); goto err; } if (dev->unit == -1) { @@ -49,19 +49,20 @@ static int ide_qdev_init(DeviceState *qdev, DeviceInfo *base) switch (dev->unit) { case 0: if (bus->master) { - fprintf(stderr, "ide: tried to assign master twice\n"); + error_report("IDE unit %d is in use", dev->unit); goto err; } bus->master = dev; break; case 1: if (bus->slave) { - fprintf(stderr, "ide: tried to assign slave twice\n"); + error_report("IDE unit %d is in use", dev->unit); goto err; } bus->slave = dev; break; default: + error_report("Invalid IDE unit %d", dev->unit); goto err; } return info->init(dev); -- cgit v1.2.1 From cd8722bb22ed41586ef489b3521173f1a80c9357 Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Fri, 28 May 2010 13:32:45 +0200 Subject: ide: Replace IDEState members is_cdrom, is_cf by drive_kind The two aren't independent variables. Make that obvious. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/ide/core.c | 40 ++++++++++++++++++++-------------------- hw/ide/internal.h | 5 +++-- hw/ide/macio.c | 2 +- hw/ide/microdrive.c | 2 +- 4 files changed, 25 insertions(+), 24 deletions(-) diff --git a/hw/ide/core.c b/hw/ide/core.c index ebdceb5fec..58b88ee227 100644 --- a/hw/ide/core.c +++ b/hw/ide/core.c @@ -292,7 +292,7 @@ static void ide_set_signature(IDEState *s) /* put signature */ s->nsector = 1; s->sector = 1; - if (s->is_cdrom) { + if (s->drive_kind == IDE_CD) { s->lcyl = 0x14; s->hcyl = 0xeb; } else if (s->bs) { @@ -1827,15 +1827,15 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) switch(val) { case WIN_IDENTIFY: - if (s->bs && !s->is_cdrom) { - if (!s->is_cf) + if (s->bs && s->drive_kind != IDE_CD) { + if (s->drive_kind != IDE_CFATA) ide_identify(s); else ide_cfata_identify(s); s->status = READY_STAT | SEEK_STAT; ide_transfer_start(s, s->io_buffer, 512, ide_transfer_stop); } else { - if (s->is_cdrom) { + if (s->drive_kind == IDE_CD) { ide_set_signature(s); } ide_abort_command(s); @@ -1849,7 +1849,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) ide_set_irq(s->bus); break; case WIN_SETMULT: - if (s->is_cf && s->nsector == 0) { + if (s->drive_kind == IDE_CFATA && s->nsector == 0) { /* Disable Read and Write Multiple */ s->mult_sectors = 0; s->status = READY_STAT | SEEK_STAT; @@ -2033,7 +2033,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) ide_set_irq(s->bus); break; case WIN_SEEK: - if(s->is_cdrom) + if(s->drive_kind == IDE_CD) goto abort_cmd; /* XXX: Check that seek is within bounds */ s->status = READY_STAT | SEEK_STAT; @@ -2041,7 +2041,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) break; /* ATAPI commands */ case WIN_PIDENTIFY: - if (s->is_cdrom) { + if (s->drive_kind == IDE_CD) { ide_atapi_identify(s); s->status = READY_STAT | SEEK_STAT; ide_transfer_start(s, s->io_buffer, 512, ide_transfer_stop); @@ -2052,7 +2052,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) break; case WIN_DIAGNOSE: ide_set_signature(s); - if (s->is_cdrom) + if (s->drive_kind == IDE_CD) s->status = 0; /* ATAPI spec (v6) section 9.10 defines packet * devices to return a clear status register * with READY_STAT *not* set. */ @@ -2064,14 +2064,14 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) ide_set_irq(s->bus); break; case WIN_SRST: - if (!s->is_cdrom) + if (s->drive_kind != IDE_CD) goto abort_cmd; ide_set_signature(s); s->status = 0x00; /* NOTE: READY is _not_ set */ s->error = 0x01; break; case WIN_PACKETCMD: - if (!s->is_cdrom) + if (s->drive_kind != IDE_CD) goto abort_cmd; /* overlapping commands not supported */ if (s->feature & 0x02) @@ -2084,7 +2084,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) break; /* CF-ATA commands */ case CFA_REQ_EXT_ERROR_CODE: - if (!s->is_cf) + if (s->drive_kind != IDE_CFATA) goto abort_cmd; s->error = 0x09; /* miscellaneous error */ s->status = READY_STAT | SEEK_STAT; @@ -2092,7 +2092,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) break; case CFA_ERASE_SECTORS: case CFA_WEAR_LEVEL: - if (!s->is_cf) + if (s->drive_kind != IDE_CFATA) goto abort_cmd; if (val == CFA_WEAR_LEVEL) s->nsector = 0; @@ -2103,7 +2103,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) ide_set_irq(s->bus); break; case CFA_TRANSLATE_SECTOR: - if (!s->is_cf) + if (s->drive_kind != IDE_CFATA) goto abort_cmd; s->error = 0x00; s->status = READY_STAT | SEEK_STAT; @@ -2123,7 +2123,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) ide_set_irq(s->bus); break; case CFA_ACCESS_METADATA_STORAGE: - if (!s->is_cf) + if (s->drive_kind != IDE_CFATA) goto abort_cmd; switch (s->feature) { case 0x02: /* Inquiry Metadata Storage */ @@ -2143,7 +2143,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) ide_set_irq(s->bus); break; case IBM_SENSE_CONDITION: - if (!s->is_cf) + if (s->drive_kind != IDE_CFATA) goto abort_cmd; switch (s->feature) { case 0x01: /* sense temperature in device */ @@ -2157,7 +2157,7 @@ void ide_ioport_write(void *opaque, uint32_t addr, uint32_t val) break; case WIN_SMART: - if (s->is_cdrom) + if (s->drive_kind == IDE_CD) goto abort_cmd; if (s->hcyl != 0xc2 || s->lcyl != 0x4f) goto abort_cmd; @@ -2438,7 +2438,7 @@ void ide_cmd_write(void *opaque, uint32_t addr, uint32_t val) /* high to low */ for(i = 0;i < 2; i++) { s = &bus->ifs[i]; - if (s->is_cdrom) + if (s->drive_kind == IDE_CD) s->status = 0x00; /* NOTE: READY is _not_ set */ else s->status = READY_STAT | SEEK_STAT; @@ -2540,7 +2540,7 @@ static void ide_reset(IDEState *s) #ifdef DEBUG_IDE printf("ide: reset\n"); #endif - if (s->is_cf) + if (s->drive_kind == IDE_CFATA) s->mult_sectors = 0; else s->mult_sectors = MAX_MULT_SECTORS; @@ -2614,7 +2614,7 @@ void ide_init_drive(IDEState *s, BlockDriverState *bs, s->smart_errors = 0; s->smart_selftest_count = 0; if (bdrv_get_type_hint(bs) == BDRV_TYPE_CDROM) { - s->is_cdrom = 1; + s->drive_kind = IDE_CD; bdrv_set_change_cb(bs, cdrom_change_cb, s); } if (serial) { @@ -2629,7 +2629,7 @@ void ide_init_drive(IDEState *s, BlockDriverState *bs, pstrcpy(s->version, sizeof(s->version), QEMU_VERSION); } ide_reset(s); - bdrv_set_removable(bs, s->is_cdrom); + bdrv_set_removable(bs, s->drive_kind == IDE_CD); } static void ide_init1(IDEBus *bus, int unit) diff --git a/hw/ide/internal.h b/hw/ide/internal.h index 0125a9f0b9..bd5366066c 100644 --- a/hw/ide/internal.h +++ b/hw/ide/internal.h @@ -362,6 +362,8 @@ typedef struct BMDMAState BMDMAState; #define SMART_DISABLE 0xd9 #define SMART_STATUS 0xda +typedef enum { IDE_HD, IDE_CD, IDE_CFATA } IDEDriveKind; + typedef void EndTransferFunc(IDEState *); /* NOTE: IDEState represents in fact one drive */ @@ -369,8 +371,7 @@ struct IDEState { IDEBus *bus; uint8_t unit; /* ide config */ - int is_cdrom; - int is_cf; + IDEDriveKind drive_kind; int cylinders, heads, sectors; int64_t nb_sectors; int mult_sectors; diff --git a/hw/ide/macio.c b/hw/ide/macio.c index f76c0fa31c..539c067b00 100644 --- a/hw/ide/macio.c +++ b/hw/ide/macio.c @@ -162,7 +162,7 @@ static void pmac_ide_transfer(DBDMA_io *io) IDEState *s = idebus_active_if(&m->bus); s->io_buffer_size = 0; - if (s->is_cdrom) { + if (s->drive_kind == IDE_CD) { pmac_ide_atapi_transfer_cb(io, 0); return; } diff --git a/hw/ide/microdrive.c b/hw/ide/microdrive.c index a7beac552a..47c3c8cfa5 100644 --- a/hw/ide/microdrive.c +++ b/hw/ide/microdrive.c @@ -541,7 +541,7 @@ PCMCIACardState *dscm1xxxx_init(DriveInfo *bdrv) ide_init2_with_non_qdev_drives(&md->bus, bdrv, NULL, qemu_allocate_irqs(md_set_irq, md, 1)[0]); - md->bus.ifs[0].is_cf = 1; + md->bus.ifs[0].drive_kind = IDE_CFATA; md->bus.ifs[0].mdata_size = METADATA_SIZE; md->bus.ifs[0].mdata_storage = (uint8_t *) qemu_mallocz(METADATA_SIZE); -- cgit v1.2.1 From c4d74df726cb3791d9f1661b58067df5608b627e Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Mon, 28 Jun 2010 19:07:51 +0200 Subject: ide: Make ide_init_drive() return success It still always succeeds. The next commits will add failures. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/ide/core.c | 13 +++++++++---- hw/ide/internal.h | 4 ++-- hw/ide/qdev.c | 4 +++- 3 files changed, 14 insertions(+), 7 deletions(-) diff --git a/hw/ide/core.c b/hw/ide/core.c index 58b88ee227..dbb7accca0 100644 --- a/hw/ide/core.c +++ b/hw/ide/core.c @@ -26,6 +26,7 @@ #include #include #include +#include "qemu-error.h" #include "qemu-timer.h" #include "sysemu.h" #include "dma.h" @@ -2594,8 +2595,8 @@ void ide_bus_reset(IDEBus *bus) ide_clear_hob(bus); } -void ide_init_drive(IDEState *s, BlockDriverState *bs, - const char *version, const char *serial) +int ide_init_drive(IDEState *s, BlockDriverState *bs, + const char *version, const char *serial) { int cylinders, heads, secs; uint64_t nb_sectors; @@ -2630,6 +2631,7 @@ void ide_init_drive(IDEState *s, BlockDriverState *bs, } ide_reset(s); bdrv_set_removable(bs, s->drive_kind == IDE_CD); + return 0; } static void ide_init1(IDEBus *bus, int unit) @@ -2669,8 +2671,11 @@ void ide_init2_with_non_qdev_drives(IDEBus *bus, DriveInfo *hd0, dinfo = i == 0 ? hd0 : hd1; ide_init1(bus, i); if (dinfo) { - ide_init_drive(&bus->ifs[i], dinfo->bdrv, NULL, - *dinfo->serial ? dinfo->serial : NULL); + if (ide_init_drive(&bus->ifs[i], dinfo->bdrv, NULL, + *dinfo->serial ? dinfo->serial : NULL) < 0) { + error_report("Can't set up IDE drive %s", dinfo->id); + exit(1); + } } else { ide_reset(&bus->ifs[i]); } diff --git a/hw/ide/internal.h b/hw/ide/internal.h index bd5366066c..416554324c 100644 --- a/hw/ide/internal.h +++ b/hw/ide/internal.h @@ -556,8 +556,8 @@ uint32_t ide_data_readw(void *opaque, uint32_t addr); void ide_data_writel(void *opaque, uint32_t addr, uint32_t val); uint32_t ide_data_readl(void *opaque, uint32_t addr); -void ide_init_drive(IDEState *s, BlockDriverState *bs, - const char *version, const char *serial); +int ide_init_drive(IDEState *s, BlockDriverState *bs, + const char *version, const char *serial); void ide_init2(IDEBus *bus, qemu_irq irq); void ide_init2_with_non_qdev_drives(IDEBus *bus, DriveInfo *hd0, DriveInfo *hd1, qemu_irq irq); diff --git a/hw/ide/qdev.c b/hw/ide/qdev.c index 221f387af1..53468edcbc 100644 --- a/hw/ide/qdev.c +++ b/hw/ide/qdev.c @@ -118,7 +118,9 @@ static int ide_drive_initfn(IDEDevice *dev) } } - ide_init_drive(s, dev->conf.bs, dev->version, serial); + if (ide_init_drive(s, dev->conf.bs, dev->version, serial) < 0) { + return -1; + } if (!dev->version) { dev->version = qemu_strdup(s->version); -- cgit v1.2.1 From 7aa9c811ca0761918a0252d0f923a80224953fa6 Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Mon, 28 Jun 2010 19:10:36 +0200 Subject: ide: Reject readonly drives unless CD-ROM drive_init() doesn't permit option readonly for if=ide, but that's worthless: we get it via if=none and -device. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/ide/core.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/hw/ide/core.c b/hw/ide/core.c index dbb7accca0..3b84bea5cf 100644 --- a/hw/ide/core.c +++ b/hw/ide/core.c @@ -2617,6 +2617,11 @@ int ide_init_drive(IDEState *s, BlockDriverState *bs, if (bdrv_get_type_hint(bs) == BDRV_TYPE_CDROM) { s->drive_kind = IDE_CD; bdrv_set_change_cb(bs, cdrom_change_cb, s); + } else { + if (bdrv_is_read_only(bs)) { + error_report("Can't use a read-only drive"); + return -1; + } } if (serial) { strncpy(s->drive_serial_str, serial, sizeof(s->drive_serial_str)); -- cgit v1.2.1 From dce9e92834cc4f962e547cae46b73ca559d05b0c Mon Sep 17 00:00:00 2001 From: Markus Armbruster Date: Mon, 28 Jun 2010 21:25:17 +0200 Subject: ide: Reject invalid CHS geometry drive_init() doesn't permit invalid CHS for if=ide, but that's worthless: we get it via if=none and -device. Signed-off-by: Markus Armbruster Signed-off-by: Kevin Wolf --- hw/ide/core.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/hw/ide/core.c b/hw/ide/core.c index 3b84bea5cf..af52c2cb2d 100644 --- a/hw/ide/core.c +++ b/hw/ide/core.c @@ -2604,6 +2604,18 @@ int ide_init_drive(IDEState *s, BlockDriverState *bs, s->bs = bs; bdrv_get_geometry(bs, &nb_sectors); bdrv_guess_geometry(bs, &cylinders, &heads, &secs); + if (cylinders < 1 || cylinders > 16383) { + error_report("cyls must be between 1 and 16383"); + return -1; + } + if (heads < 1 || heads > 16) { + error_report("heads must be between 1 and 16"); + return -1; + } + if (secs < 1 || secs > 63) { + error_report("secs must be between 1 and 63"); + return -1; + } s->cylinders = cylinders; s->heads = heads; s->sectors = secs; -- cgit v1.2.1 From 33b1db1c8888b77e06c720ebef0482ed598eb384 Mon Sep 17 00:00:00 2001 From: MORITA Kazutaka Date: Mon, 21 Jun 2010 05:01:00 +0900 Subject: block: add sheepdog driver for distributed storage support Sheepdog is a distributed storage system for QEMU. It provides highly available block level storage volumes to VMs like Amazon EBS. This patch adds a qemu block driver for Sheepdog. Sheepdog features are: - No node in the cluster is special (no metadata node, no control node, etc) - Linear scalability in performance and capacity - No single point of failure - Autonomous management (zero configuration) - Useful volume management support such as snapshot and cloning - Thin provisioning - Autonomous load balancing The more details are available at the project site: http://www.osrg.net/sheepdog/ Signed-off-by: MORITA Kazutaka Signed-off-by: Kevin Wolf --- Makefile.objs | 2 +- block/sheepdog.c | 2036 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 2037 insertions(+), 1 deletion(-) create mode 100644 block/sheepdog.c diff --git a/Makefile.objs b/Makefile.objs index 55417c9941..67f1b215b1 100644 --- a/Makefile.objs +++ b/Makefile.objs @@ -14,7 +14,7 @@ block-obj-$(CONFIG_LINUX_AIO) += linux-aio.o block-nested-y += raw.o cow.o qcow.o vdi.o vmdk.o cloop.o dmg.o bochs.o vpc.o vvfat.o block-nested-y += qcow2.o qcow2-refcount.o qcow2-cluster.o qcow2-snapshot.o -block-nested-y += parallels.o nbd.o blkdebug.o +block-nested-y += parallels.o nbd.o blkdebug.o sheepdog.o block-nested-$(CONFIG_WIN32) += raw-win32.o block-nested-$(CONFIG_POSIX) += raw-posix.o block-nested-$(CONFIG_CURL) += curl.o diff --git a/block/sheepdog.c b/block/sheepdog.c new file mode 100644 index 0000000000..69a24940aa --- /dev/null +++ b/block/sheepdog.c @@ -0,0 +1,2036 @@ +/* + * Copyright (C) 2009-2010 Nippon Telegraph and Telephone Corporation. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ +#ifdef _WIN32 +#include +#include +#include +#else +#include +#include + +#define closesocket(s) close(s) +#endif + +#include "qemu-common.h" +#include "qemu-error.h" +#include "qemu_socket.h" +#include "block_int.h" + +#define SD_PROTO_VER 0x01 + +#define SD_DEFAULT_ADDR "localhost" +#define SD_DEFAULT_PORT "7000" + +#define SD_OP_CREATE_AND_WRITE_OBJ 0x01 +#define SD_OP_READ_OBJ 0x02 +#define SD_OP_WRITE_OBJ 0x03 + +#define SD_OP_NEW_VDI 0x11 +#define SD_OP_LOCK_VDI 0x12 +#define SD_OP_RELEASE_VDI 0x13 +#define SD_OP_GET_VDI_INFO 0x14 +#define SD_OP_READ_VDIS 0x15 + +#define SD_FLAG_CMD_WRITE 0x01 +#define SD_FLAG_CMD_COW 0x02 + +#define SD_RES_SUCCESS 0x00 /* Success */ +#define SD_RES_UNKNOWN 0x01 /* Unknown error */ +#define SD_RES_NO_OBJ 0x02 /* No object found */ +#define SD_RES_EIO 0x03 /* I/O error */ +#define SD_RES_VDI_EXIST 0x04 /* Vdi exists already */ +#define SD_RES_INVALID_PARMS 0x05 /* Invalid parameters */ +#define SD_RES_SYSTEM_ERROR 0x06 /* System error */ +#define SD_RES_VDI_LOCKED 0x07 /* Vdi is locked */ +#define SD_RES_NO_VDI 0x08 /* No vdi found */ +#define SD_RES_NO_BASE_VDI 0x09 /* No base vdi found */ +#define SD_RES_VDI_READ 0x0A /* Cannot read requested vdi */ +#define SD_RES_VDI_WRITE 0x0B /* Cannot write requested vdi */ +#define SD_RES_BASE_VDI_READ 0x0C /* Cannot read base vdi */ +#define SD_RES_BASE_VDI_WRITE 0x0D /* Cannot write base vdi */ +#define SD_RES_NO_TAG 0x0E /* Requested tag is not found */ +#define SD_RES_STARTUP 0x0F /* Sheepdog is on starting up */ +#define SD_RES_VDI_NOT_LOCKED 0x10 /* Vdi is not locked */ +#define SD_RES_SHUTDOWN 0x11 /* Sheepdog is shutting down */ +#define SD_RES_NO_MEM 0x12 /* Cannot allocate memory */ +#define SD_RES_FULL_VDI 0x13 /* we already have the maximum vdis */ +#define SD_RES_VER_MISMATCH 0x14 /* Protocol version mismatch */ +#define SD_RES_NO_SPACE 0x15 /* Server has no room for new objects */ +#define SD_RES_WAIT_FOR_FORMAT 0x16 /* Waiting for a format operation */ +#define SD_RES_WAIT_FOR_JOIN 0x17 /* Waiting for other nodes joining */ +#define SD_RES_JOIN_FAILED 0x18 /* Target node had failed to join sheepdog */ + +/* + * Object ID rules + * + * 0 - 19 (20 bits): data object space + * 20 - 31 (12 bits): reserved data object space + * 32 - 55 (24 bits): vdi object space + * 56 - 59 ( 4 bits): reserved vdi object space + * 60 - 63 ( 4 bits): object type indentifier space + */ + +#define VDI_SPACE_SHIFT 32 +#define VDI_BIT (UINT64_C(1) << 63) +#define VMSTATE_BIT (UINT64_C(1) << 62) +#define MAX_DATA_OBJS (UINT64_C(1) << 20) +#define MAX_CHILDREN 1024 +#define SD_MAX_VDI_LEN 256 +#define SD_MAX_VDI_TAG_LEN 256 +#define SD_NR_VDIS (1U << 24) +#define SD_DATA_OBJ_SIZE (UINT64_C(1) << 22) +#define SD_MAX_VDI_SIZE (SD_DATA_OBJ_SIZE * MAX_DATA_OBJS) +#define SECTOR_SIZE 512 + +#define SD_INODE_SIZE (sizeof(SheepdogInode)) +#define CURRENT_VDI_ID 0 + +typedef struct SheepdogReq { + uint8_t proto_ver; + uint8_t opcode; + uint16_t flags; + uint32_t epoch; + uint32_t id; + uint32_t data_length; + uint32_t opcode_specific[8]; +} SheepdogReq; + +typedef struct SheepdogRsp { + uint8_t proto_ver; + uint8_t opcode; + uint16_t flags; + uint32_t epoch; + uint32_t id; + uint32_t data_length; + uint32_t result; + uint32_t opcode_specific[7]; +} SheepdogRsp; + +typedef struct SheepdogObjReq { + uint8_t proto_ver; + uint8_t opcode; + uint16_t flags; + uint32_t epoch; + uint32_t id; + uint32_t data_length; + uint64_t oid; + uint64_t cow_oid; + uint32_t copies; + uint32_t rsvd; + uint64_t offset; +} SheepdogObjReq; + +typedef struct SheepdogObjRsp { + uint8_t proto_ver; + uint8_t opcode; + uint16_t flags; + uint32_t epoch; + uint32_t id; + uint32_t data_length; + uint32_t result; + uint32_t copies; + uint32_t pad[6]; +} SheepdogObjRsp; + +typedef struct SheepdogVdiReq { + uint8_t proto_ver; + uint8_t opcode; + uint16_t flags; + uint32_t epoch; + uint32_t id; + uint32_t data_length; + uint64_t vdi_size; + uint32_t base_vdi_id; + uint32_t copies; + uint32_t snapid; + uint32_t pad[3]; +} SheepdogVdiReq; + +typedef struct SheepdogVdiRsp { + uint8_t proto_ver; + uint8_t opcode; + uint16_t flags; + uint32_t epoch; + uint32_t id; + uint32_t data_length; + uint32_t result; + uint32_t rsvd; + uint32_t vdi_id; + uint32_t pad[5]; +} SheepdogVdiRsp; + +typedef struct SheepdogInode { + char name[SD_MAX_VDI_LEN]; + char tag[SD_MAX_VDI_TAG_LEN]; + uint64_t ctime; + uint64_t snap_ctime; + uint64_t vm_clock_nsec; + uint64_t vdi_size; + uint64_t vm_state_size; + uint16_t copy_policy; + uint8_t nr_copies; + uint8_t block_size_shift; + uint32_t snap_id; + uint32_t vdi_id; + uint32_t parent_vdi_id; + uint32_t child_vdi_id[MAX_CHILDREN]; + uint32_t data_vdi_id[MAX_DATA_OBJS]; +} SheepdogInode; + +/* + * 64 bit FNV-1a non-zero initial basis + */ +#define FNV1A_64_INIT ((uint64_t)0xcbf29ce484222325ULL) + +/* + * 64 bit Fowler/Noll/Vo FNV-1a hash code + */ +static inline uint64_t fnv_64a_buf(void *buf, size_t len, uint64_t hval) +{ + unsigned char *bp = buf; + unsigned char *be = bp + len; + while (bp < be) { + hval ^= (uint64_t) *bp++; + hval += (hval << 1) + (hval << 4) + (hval << 5) + + (hval << 7) + (hval << 8) + (hval << 40); + } + return hval; +} + +static inline int is_data_obj_writeable(SheepdogInode *inode, unsigned int idx) +{ + return inode->vdi_id == inode->data_vdi_id[idx]; +} + +static inline int is_data_obj(uint64_t oid) +{ + return !(VDI_BIT & oid); +} + +static inline uint64_t data_oid_to_idx(uint64_t oid) +{ + return oid & (MAX_DATA_OBJS - 1); +} + +static inline uint64_t vid_to_vdi_oid(uint32_t vid) +{ + return VDI_BIT | ((uint64_t)vid << VDI_SPACE_SHIFT); +} + +static inline uint64_t vid_to_vmstate_oid(uint32_t vid, uint32_t idx) +{ + return VMSTATE_BIT | ((uint64_t)vid << VDI_SPACE_SHIFT) | idx; +} + +static inline uint64_t vid_to_data_oid(uint32_t vid, uint32_t idx) +{ + return ((uint64_t)vid << VDI_SPACE_SHIFT) | idx; +} + +static inline int is_snapshot(struct SheepdogInode *inode) +{ + return !!inode->snap_ctime; +} + +#undef dprintf +#ifdef DEBUG_SDOG +#define dprintf(fmt, args...) \ + do { \ + fprintf(stdout, "%s %d: " fmt, __func__, __LINE__, ##args); \ + } while (0) +#else +#define dprintf(fmt, args...) +#endif + +typedef struct SheepdogAIOCB SheepdogAIOCB; + +typedef struct AIOReq { + SheepdogAIOCB *aiocb; + unsigned int iov_offset; + + uint64_t oid; + uint64_t base_oid; + uint64_t offset; + unsigned int data_len; + uint8_t flags; + uint32_t id; + + QLIST_ENTRY(AIOReq) outstanding_aio_siblings; + QLIST_ENTRY(AIOReq) aioreq_siblings; +} AIOReq; + +enum AIOCBState { + AIOCB_WRITE_UDATA, + AIOCB_READ_UDATA, +}; + +struct SheepdogAIOCB { + BlockDriverAIOCB common; + + QEMUIOVector *qiov; + + int64_t sector_num; + int nb_sectors; + + int ret; + enum AIOCBState aiocb_type; + + QEMUBH *bh; + void (*aio_done_func)(SheepdogAIOCB *); + + int canceled; + + QLIST_HEAD(aioreq_head, AIOReq) aioreq_head; +}; + +typedef struct BDRVSheepdogState { + SheepdogInode inode; + + uint32_t min_dirty_data_idx; + uint32_t max_dirty_data_idx; + + char name[SD_MAX_VDI_LEN]; + int is_snapshot; + + char *addr; + char *port; + int fd; + + uint32_t aioreq_seq_num; + QLIST_HEAD(outstanding_aio_head, AIOReq) outstanding_aio_head; +} BDRVSheepdogState; + +static const char * sd_strerror(int err) +{ + int i; + + static const struct { + int err; + const char *desc; + } errors[] = { + {SD_RES_SUCCESS, "Success"}, + {SD_RES_UNKNOWN, "Unknown error"}, + {SD_RES_NO_OBJ, "No object found"}, + {SD_RES_EIO, "I/O error"}, + {SD_RES_VDI_EXIST, "VDI exists already"}, + {SD_RES_INVALID_PARMS, "Invalid parameters"}, + {SD_RES_SYSTEM_ERROR, "System error"}, + {SD_RES_VDI_LOCKED, "VDI is already locked"}, + {SD_RES_NO_VDI, "No vdi found"}, + {SD_RES_NO_BASE_VDI, "No base VDI found"}, + {SD_RES_VDI_READ, "Failed read the requested VDI"}, + {SD_RES_VDI_WRITE, "Failed to write the requested VDI"}, + {SD_RES_BASE_VDI_READ, "Failed to read the base VDI"}, + {SD_RES_BASE_VDI_WRITE, "Failed to write the base VDI"}, + {SD_RES_NO_TAG, "Failed to find the requested tag"}, + {SD_RES_STARTUP, "The system is still booting"}, + {SD_RES_VDI_NOT_LOCKED, "VDI isn't locked"}, + {SD_RES_SHUTDOWN, "The system is shutting down"}, + {SD_RES_NO_MEM, "Out of memory on the server"}, + {SD_RES_FULL_VDI, "We already have the maximum vdis"}, + {SD_RES_VER_MISMATCH, "Protocol version mismatch"}, + {SD_RES_NO_SPACE, "Server has no space for new objects"}, + {SD_RES_WAIT_FOR_FORMAT, "Sheepdog is waiting for a format operation"}, + {SD_RES_WAIT_FOR_JOIN, "Sheepdog is waiting for other nodes joining"}, + {SD_RES_JOIN_FAILED, "Target node had failed to join sheepdog"}, + }; + + for (i = 0; i < ARRAY_SIZE(errors); ++i) { + if (errors[i].err == err) { + return errors[i].desc; + } + } + + return "Invalid error code"; +} + +/* + * Sheepdog I/O handling: + * + * 1. In the sd_aio_readv/writev, read/write requests are added to the + * QEMU Bottom Halves. + * + * 2. In sd_readv_writev_bh_cb, the callbacks of BHs, we send the I/O + * requests to the server and link the requests to the + * outstanding_list in the BDRVSheepdogState. we exits the + * function without waiting for receiving the response. + * + * 3. We receive the response in aio_read_response, the fd handler to + * the sheepdog connection. If metadata update is needed, we send + * the write request to the vdi object in sd_write_done, the write + * completion function. The AIOCB callback is not called until all + * the requests belonging to the AIOCB are finished. + */ + +static inline AIOReq *alloc_aio_req(BDRVSheepdogState *s, SheepdogAIOCB *acb, + uint64_t oid, unsigned int data_len, + uint64_t offset, uint8_t flags, + uint64_t base_oid, unsigned int iov_offset) +{ + AIOReq *aio_req; + + aio_req = qemu_malloc(sizeof(*aio_req)); + aio_req->aiocb = acb; + aio_req->iov_offset = iov_offset; + aio_req->oid = oid; + aio_req->base_oid = base_oid; + aio_req->offset = offset; + aio_req->data_len = data_len; + aio_req->flags = flags; + aio_req->id = s->aioreq_seq_num++; + + QLIST_INSERT_HEAD(&s->outstanding_aio_head, aio_req, + outstanding_aio_siblings); + QLIST_INSERT_HEAD(&acb->aioreq_head, aio_req, aioreq_siblings); + + return aio_req; +} + +static inline int free_aio_req(BDRVSheepdogState *s, AIOReq *aio_req) +{ + SheepdogAIOCB *acb = aio_req->aiocb; + QLIST_REMOVE(aio_req, outstanding_aio_siblings); + QLIST_REMOVE(aio_req, aioreq_siblings); + qemu_free(aio_req); + + return !QLIST_EMPTY(&acb->aioreq_head); +} + +static void sd_finish_aiocb(SheepdogAIOCB *acb) +{ + if (!acb->canceled) { + acb->common.cb(acb->common.opaque, acb->ret); + } + qemu_aio_release(acb); +} + +static void sd_aio_cancel(BlockDriverAIOCB *blockacb) +{ + SheepdogAIOCB *acb = (SheepdogAIOCB *)blockacb; + + /* + * Sheepdog cannot cancel the requests which are already sent to + * the servers, so we just complete the request with -EIO here. + */ + acb->common.cb(acb->common.opaque, -EIO); + acb->canceled = 1; +} + +static AIOPool sd_aio_pool = { + .aiocb_size = sizeof(SheepdogAIOCB), + .cancel = sd_aio_cancel, +}; + +static SheepdogAIOCB *sd_aio_setup(BlockDriverState *bs, QEMUIOVector *qiov, + int64_t sector_num, int nb_sectors, + BlockDriverCompletionFunc *cb, void *opaque) +{ + SheepdogAIOCB *acb; + + acb = qemu_aio_get(&sd_aio_pool, bs, cb, opaque); + + acb->qiov = qiov; + + acb->sector_num = sector_num; + acb->nb_sectors = nb_sectors; + + acb->aio_done_func = NULL; + acb->canceled = 0; + acb->bh = NULL; + acb->ret = 0; + QLIST_INIT(&acb->aioreq_head); + return acb; +} + +static int sd_schedule_bh(QEMUBHFunc *cb, SheepdogAIOCB *acb) +{ + if (acb->bh) { + error_report("bug: %d %d\n", acb->aiocb_type, acb->aiocb_type); + return -EIO; + } + + acb->bh = qemu_bh_new(cb, acb); + if (!acb->bh) { + error_report("oom: %d %d\n", acb->aiocb_type, acb->aiocb_type); + return -EIO; + } + + qemu_bh_schedule(acb->bh); + + return 0; +} + +#ifdef _WIN32 + +struct msghdr { + struct iovec *msg_iov; + size_t msg_iovlen; +}; + +static ssize_t sendmsg(int s, const struct msghdr *msg, int flags) +{ + size_t size = 0; + char *buf, *p; + int i, ret; + + /* count the msg size */ + for (i = 0; i < msg->msg_iovlen; i++) { + size += msg->msg_iov[i].iov_len; + } + buf = qemu_malloc(size); + + p = buf; + for (i = 0; i < msg->msg_iovlen; i++) { + memcpy(p, msg->msg_iov[i].iov_base, msg->msg_iov[i].iov_len); + p += msg->msg_iov[i].iov_len; + } + + ret = send(s, buf, size, flags); + + qemu_free(buf); + return ret; +} + +static ssize_t recvmsg(int s, struct msghdr *msg, int flags) +{ + size_t size = 0; + char *buf, *p; + int i, ret; + + /* count the msg size */ + for (i = 0; i < msg->msg_iovlen; i++) { + size += msg->msg_iov[i].iov_len; + } + buf = qemu_malloc(size); + + ret = recv(s, buf, size, flags); + if (ret < 0) { + goto out; + } + + p = buf; + for (i = 0; i < msg->msg_iovlen; i++) { + memcpy(msg->msg_iov[i].iov_base, p, msg->msg_iov[i].iov_len); + p += msg->msg_iov[i].iov_len; + } +out: + qemu_free(buf); + return ret; +} + +#endif + +/* + * Send/recv data with iovec buffers + * + * This function send/recv data from/to the iovec buffer directly. + * The first `offset' bytes in the iovec buffer are skipped and next + * `len' bytes are used. + * + * For example, + * + * do_send_recv(sockfd, iov, len, offset, 1); + * + * is equals to + * + * char *buf = malloc(size); + * iov_to_buf(iov, iovcnt, buf, offset, size); + * send(sockfd, buf, size, 0); + * free(buf); + */ +static int do_send_recv(int sockfd, struct iovec *iov, int len, int offset, + int write) +{ + struct msghdr msg; + int ret, diff; + + memset(&msg, 0, sizeof(msg)); + msg.msg_iov = iov; + msg.msg_iovlen = 1; + + len += offset; + + while (iov->iov_len < len) { + len -= iov->iov_len; + + iov++; + msg.msg_iovlen++; + } + + diff = iov->iov_len - len; + iov->iov_len -= diff; + + while (msg.msg_iov->iov_len <= offset) { + offset -= msg.msg_iov->iov_len; + + msg.msg_iov++; + msg.msg_iovlen--; + } + + msg.msg_iov->iov_base = (char *) msg.msg_iov->iov_base + offset; + msg.msg_iov->iov_len -= offset; + + if (write) { + ret = sendmsg(sockfd, &msg, 0); + } else { + ret = recvmsg(sockfd, &msg, 0); + } + + msg.msg_iov->iov_base = (char *) msg.msg_iov->iov_base - offset; + msg.msg_iov->iov_len += offset; + + iov->iov_len += diff; + return ret; +} + +static int connect_to_sdog(const char *addr, const char *port) +{ + char hbuf[NI_MAXHOST], sbuf[NI_MAXSERV]; + int fd, ret; + struct addrinfo hints, *res, *res0; + + if (!addr) { + addr = SD_DEFAULT_ADDR; + port = SD_DEFAULT_PORT; + } + + memset(&hints, 0, sizeof(hints)); + hints.ai_socktype = SOCK_STREAM; + + ret = getaddrinfo(addr, port, &hints, &res0); + if (ret) { + error_report("unable to get address info %s, %s\n", + addr, strerror(errno)); + return -1; + } + + for (res = res0; res; res = res->ai_next) { + ret = getnameinfo(res->ai_addr, res->ai_addrlen, hbuf, sizeof(hbuf), + sbuf, sizeof(sbuf), NI_NUMERICHOST | NI_NUMERICSERV); + if (ret) { + continue; + } + + fd = socket(res->ai_family, res->ai_socktype, res->ai_protocol); + if (fd < 0) { + continue; + } + + reconnect: + ret = connect(fd, res->ai_addr, res->ai_addrlen); + if (ret < 0) { + if (errno == EINTR) { + goto reconnect; + } + break; + } + + dprintf("connected to %s:%s\n", addr, port); + goto success; + } + fd = -1; + error_report("failed connect to %s:%s\n", addr, port); +success: + freeaddrinfo(res0); + return fd; +} + +static int do_readv_writev(int sockfd, struct iovec *iov, int len, + int iov_offset, int write) +{ + int ret; +again: + ret = do_send_recv(sockfd, iov, len, iov_offset, write); + if (ret < 0) { + if (errno == EINTR || errno == EAGAIN) { + goto again; + } + error_report("failed to recv a rsp, %s\n", strerror(errno)); + return 1; + } + + iov_offset += ret; + len -= ret; + if (len) { + goto again; + } + + return 0; +} + +static int do_readv(int sockfd, struct iovec *iov, int len, int iov_offset) +{ + return do_readv_writev(sockfd, iov, len, iov_offset, 0); +} + +static int do_writev(int sockfd, struct iovec *iov, int len, int iov_offset) +{ + return do_readv_writev(sockfd, iov, len, iov_offset, 1); +} + +static int do_read_write(int sockfd, void *buf, int len, int write) +{ + struct iovec iov; + + iov.iov_base = buf; + iov.iov_len = len; + + return do_readv_writev(sockfd, &iov, len, 0, write); +} + +static int do_read(int sockfd, void *buf, int len) +{ + return do_read_write(sockfd, buf, len, 0); +} + +static int do_write(int sockfd, void *buf, int len) +{ + return do_read_write(sockfd, buf, len, 1); +} + +static int send_req(int sockfd, SheepdogReq *hdr, void *data, + unsigned int *wlen) +{ + int ret; + struct iovec iov[2]; + + iov[0].iov_base = hdr; + iov[0].iov_len = sizeof(*hdr); + + if (*wlen) { + iov[1].iov_base = data; + iov[1].iov_len = *wlen; + } + + ret = do_writev(sockfd, iov, sizeof(*hdr) + *wlen, 0); + if (ret) { + error_report("failed to send a req, %s\n", strerror(errno)); + ret = -1; + } + + return ret; +} + +static int do_req(int sockfd, SheepdogReq *hdr, void *data, + unsigned int *wlen, unsigned int *rlen) +{ + int ret; + + ret = send_req(sockfd, hdr, data, wlen); + if (ret) { + ret = -1; + goto out; + } + + ret = do_read(sockfd, hdr, sizeof(*hdr)); + if (ret) { + error_report("failed to get a rsp, %s\n", strerror(errno)); + ret = -1; + goto out; + } + + if (*rlen > hdr->data_length) { + *rlen = hdr->data_length; + } + + if (*rlen) { + ret = do_read(sockfd, data, *rlen); + if (ret) { + error_report("failed to get the data, %s\n", strerror(errno)); + ret = -1; + goto out; + } + } + ret = 0; +out: + return ret; +} + +static int add_aio_request(BDRVSheepdogState *s, AIOReq *aio_req, + struct iovec *iov, int niov, int create, + enum AIOCBState aiocb_type); + +/* + * This function searchs pending requests to the object `oid', and + * sends them. + */ +static void send_pending_req(BDRVSheepdogState *s, uint64_t oid, uint32_t id) +{ + AIOReq *aio_req, *next; + SheepdogAIOCB *acb; + int ret; + + QLIST_FOREACH_SAFE(aio_req, &s->outstanding_aio_head, + outstanding_aio_siblings, next) { + if (id == aio_req->id) { + continue; + } + if (aio_req->oid != oid) { + continue; + } + + acb = aio_req->aiocb; + ret = add_aio_request(s, aio_req, acb->qiov->iov, + acb->qiov->niov, 0, acb->aiocb_type); + if (ret < 0) { + error_report("add_aio_request is failed\n"); + free_aio_req(s, aio_req); + if (QLIST_EMPTY(&acb->aioreq_head)) { + sd_finish_aiocb(acb); + } + } + } +} + +/* + * Receive responses of the I/O requests. + * + * This function is registered as a fd handler, and called from the + * main loop when s->fd is ready for reading responses. + */ +static void aio_read_response(void *opaque) +{ + SheepdogObjRsp rsp; + BDRVSheepdogState *s = opaque; + int fd = s->fd; + int ret; + AIOReq *aio_req = NULL; + SheepdogAIOCB *acb; + int rest; + unsigned long idx; + + if (QLIST_EMPTY(&s->outstanding_aio_head)) { + return; + } + + /* read a header */ + ret = do_read(fd, &rsp, sizeof(rsp)); + if (ret) { + error_report("failed to get the header, %s\n", strerror(errno)); + return; + } + + /* find the right aio_req from the outstanding_aio list */ + QLIST_FOREACH(aio_req, &s->outstanding_aio_head, outstanding_aio_siblings) { + if (aio_req->id == rsp.id) { + break; + } + } + if (!aio_req) { + error_report("cannot find aio_req %x\n", rsp.id); + return; + } + + acb = aio_req->aiocb; + + switch (acb->aiocb_type) { + case AIOCB_WRITE_UDATA: + if (!is_data_obj(aio_req->oid)) { + break; + } + idx = data_oid_to_idx(aio_req->oid); + + if (s->inode.data_vdi_id[idx] != s->inode.vdi_id) { + /* + * If the object is newly created one, we need to update + * the vdi object (metadata object). min_dirty_data_idx + * and max_dirty_data_idx are changed to include updated + * index between them. + */ + s->inode.data_vdi_id[idx] = s->inode.vdi_id; + s->max_dirty_data_idx = MAX(idx, s->max_dirty_data_idx); + s->min_dirty_data_idx = MIN(idx, s->min_dirty_data_idx); + + /* + * Some requests may be blocked because simultaneous + * create requests are not allowed, so we search the + * pending requests here. + */ + send_pending_req(s, vid_to_data_oid(s->inode.vdi_id, idx), rsp.id); + } + break; + case AIOCB_READ_UDATA: + ret = do_readv(fd, acb->qiov->iov, rsp.data_length, + aio_req->iov_offset); + if (ret) { + error_report("failed to get the data, %s\n", strerror(errno)); + return; + } + break; + } + + if (rsp.result != SD_RES_SUCCESS) { + acb->ret = -EIO; + error_report("%s\n", sd_strerror(rsp.result)); + } + + rest = free_aio_req(s, aio_req); + if (!rest) { + /* + * We've finished all requests which belong to the AIOCB, so + * we can call the callback now. + */ + acb->aio_done_func(acb); + } +} + +static int aio_flush_request(void *opaque) +{ + BDRVSheepdogState *s = opaque; + + return !QLIST_EMPTY(&s->outstanding_aio_head); +} + +#ifdef _WIN32 + +static int set_cork(int fd, int v) +{ + return 0; +} + +#else + +static int set_cork(int fd, int v) +{ + return setsockopt(fd, SOL_TCP, TCP_CORK, &v, sizeof(v)); +} + +#endif + +static int set_nodelay(int fd) +{ + int ret, opt; + + opt = 1; + ret = setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, (char *)&opt, sizeof(opt)); + return ret; +} + +/* + * Return a socket discriptor to read/write objects. + * + * We cannot use this discriptor for other operations because + * the block driver may be on waiting response from the server. + */ +static int get_sheep_fd(BDRVSheepdogState *s) +{ + int ret, fd; + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + error_report("%s\n", strerror(errno)); + return -1; + } + + socket_set_nonblock(fd); + + ret = set_nodelay(fd); + if (ret) { + error_report("%s\n", strerror(errno)); + closesocket(fd); + return -1; + } + + qemu_aio_set_fd_handler(fd, aio_read_response, NULL, aio_flush_request, + NULL, s); + return fd; +} + +/* + * Parse a filename + * + * filename must be one of the following formats: + * 1. [vdiname] + * 2. [vdiname]:[snapid] + * 3. [vdiname]:[tag] + * 4. [hostname]:[port]:[vdiname] + * 5. [hostname]:[port]:[vdiname]:[snapid] + * 6. [hostname]:[port]:[vdiname]:[tag] + * + * You can boot from the snapshot images by specifying `snapid` or + * `tag'. + * + * You can run VMs outside the Sheepdog cluster by specifying + * `hostname' and `port' (experimental). + */ +static int parse_vdiname(BDRVSheepdogState *s, const char *filename, + char *vdi, uint32_t *snapid, char *tag) +{ + char *p, *q; + int nr_sep; + + p = q = qemu_strdup(filename); + + /* count the number of separators */ + nr_sep = 0; + while (*p) { + if (*p == ':') { + nr_sep++; + } + p++; + } + p = q; + + /* use the first two tokens as hostname and port number. */ + if (nr_sep >= 2) { + s->addr = p; + p = strchr(p, ':'); + *p++ = '\0'; + + s->port = p; + p = strchr(p, ':'); + *p++ = '\0'; + } else { + s->addr = NULL; + s->port = 0; + } + + strncpy(vdi, p, SD_MAX_VDI_LEN); + + p = strchr(vdi, ':'); + if (p) { + *p++ = '\0'; + *snapid = strtoul(p, NULL, 10); + if (*snapid == 0) { + strncpy(tag, p, SD_MAX_VDI_TAG_LEN); + } + } else { + *snapid = CURRENT_VDI_ID; /* search current vdi */ + } + + if (s->addr == NULL) { + qemu_free(q); + } + + return 0; +} + +static int find_vdi_name(BDRVSheepdogState *s, char *filename, uint32_t snapid, + char *tag, uint32_t *vid, int for_snapshot) +{ + int ret, fd; + SheepdogVdiReq hdr; + SheepdogVdiRsp *rsp = (SheepdogVdiRsp *)&hdr; + unsigned int wlen, rlen = 0; + char buf[SD_MAX_VDI_LEN + SD_MAX_VDI_TAG_LEN]; + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + return -1; + } + + memset(buf, 0, sizeof(buf)); + strncpy(buf, filename, SD_MAX_VDI_LEN); + strncpy(buf + SD_MAX_VDI_LEN, tag, SD_MAX_VDI_TAG_LEN); + + memset(&hdr, 0, sizeof(hdr)); + if (for_snapshot) { + hdr.opcode = SD_OP_GET_VDI_INFO; + } else { + hdr.opcode = SD_OP_LOCK_VDI; + } + wlen = SD_MAX_VDI_LEN + SD_MAX_VDI_TAG_LEN; + hdr.proto_ver = SD_PROTO_VER; + hdr.data_length = wlen; + hdr.snapid = snapid; + hdr.flags = SD_FLAG_CMD_WRITE; + + ret = do_req(fd, (SheepdogReq *)&hdr, buf, &wlen, &rlen); + if (ret) { + ret = -1; + goto out; + } + + if (rsp->result != SD_RES_SUCCESS) { + error_report("cannot get vdi info, %s, %s %d %s\n", + sd_strerror(rsp->result), filename, snapid, tag); + ret = -1; + goto out; + } + *vid = rsp->vdi_id; + + ret = 0; +out: + closesocket(fd); + return ret; +} + +static int add_aio_request(BDRVSheepdogState *s, AIOReq *aio_req, + struct iovec *iov, int niov, int create, + enum AIOCBState aiocb_type) +{ + int nr_copies = s->inode.nr_copies; + SheepdogObjReq hdr; + unsigned int wlen; + int ret; + uint64_t oid = aio_req->oid; + unsigned int datalen = aio_req->data_len; + uint64_t offset = aio_req->offset; + uint8_t flags = aio_req->flags; + uint64_t old_oid = aio_req->base_oid; + + if (!nr_copies) { + error_report("bug\n"); + } + + memset(&hdr, 0, sizeof(hdr)); + + if (aiocb_type == AIOCB_READ_UDATA) { + wlen = 0; + hdr.opcode = SD_OP_READ_OBJ; + hdr.flags = flags; + } else if (create) { + wlen = datalen; + hdr.opcode = SD_OP_CREATE_AND_WRITE_OBJ; + hdr.flags = SD_FLAG_CMD_WRITE | flags; + } else { + wlen = datalen; + hdr.opcode = SD_OP_WRITE_OBJ; + hdr.flags = SD_FLAG_CMD_WRITE | flags; + } + + hdr.oid = oid; + hdr.cow_oid = old_oid; + hdr.copies = s->inode.nr_copies; + + hdr.data_length = datalen; + hdr.offset = offset; + + hdr.id = aio_req->id; + + set_cork(s->fd, 1); + + /* send a header */ + ret = do_write(s->fd, &hdr, sizeof(hdr)); + if (ret) { + error_report("failed to send a req, %s\n", strerror(errno)); + return -EIO; + } + + if (wlen) { + ret = do_writev(s->fd, iov, wlen, aio_req->iov_offset); + if (ret) { + error_report("failed to send a data, %s\n", strerror(errno)); + return -EIO; + } + } + + set_cork(s->fd, 0); + + return 0; +} + +static int read_write_object(int fd, char *buf, uint64_t oid, int copies, + unsigned int datalen, uint64_t offset, + int write, int create) +{ + SheepdogObjReq hdr; + SheepdogObjRsp *rsp = (SheepdogObjRsp *)&hdr; + unsigned int wlen, rlen; + int ret; + + memset(&hdr, 0, sizeof(hdr)); + + if (write) { + wlen = datalen; + rlen = 0; + hdr.flags = SD_FLAG_CMD_WRITE; + if (create) { + hdr.opcode = SD_OP_CREATE_AND_WRITE_OBJ; + } else { + hdr.opcode = SD_OP_WRITE_OBJ; + } + } else { + wlen = 0; + rlen = datalen; + hdr.opcode = SD_OP_READ_OBJ; + } + hdr.oid = oid; + hdr.data_length = datalen; + hdr.offset = offset; + hdr.copies = copies; + + ret = do_req(fd, (SheepdogReq *)&hdr, buf, &wlen, &rlen); + if (ret) { + error_report("failed to send a request to the sheep\n"); + return -1; + } + + switch (rsp->result) { + case SD_RES_SUCCESS: + return 0; + default: + error_report("%s\n", sd_strerror(rsp->result)); + return -1; + } +} + +static int read_object(int fd, char *buf, uint64_t oid, int copies, + unsigned int datalen, uint64_t offset) +{ + return read_write_object(fd, buf, oid, copies, datalen, offset, 0, 0); +} + +static int write_object(int fd, char *buf, uint64_t oid, int copies, + unsigned int datalen, uint64_t offset, int create) +{ + return read_write_object(fd, buf, oid, copies, datalen, offset, 1, create); +} + +static int sd_open(BlockDriverState *bs, const char *filename, int flags) +{ + int ret, fd; + uint32_t vid = 0; + BDRVSheepdogState *s = bs->opaque; + char vdi[SD_MAX_VDI_LEN], tag[SD_MAX_VDI_TAG_LEN]; + uint32_t snapid; + char *buf = NULL; + + strstart(filename, "sheepdog:", (const char **)&filename); + + QLIST_INIT(&s->outstanding_aio_head); + s->fd = -1; + + memset(vdi, 0, sizeof(vdi)); + memset(tag, 0, sizeof(tag)); + if (parse_vdiname(s, filename, vdi, &snapid, tag) < 0) { + goto out; + } + s->fd = get_sheep_fd(s); + if (s->fd < 0) { + goto out; + } + + ret = find_vdi_name(s, vdi, snapid, tag, &vid, 0); + if (ret) { + goto out; + } + + if (snapid) { + dprintf("%" PRIx32 " snapshot inode was open.\n", vid); + s->is_snapshot = 1; + } + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + error_report("failed to connect\n"); + goto out; + } + + buf = qemu_malloc(SD_INODE_SIZE); + ret = read_object(fd, buf, vid_to_vdi_oid(vid), 0, SD_INODE_SIZE, 0); + + closesocket(fd); + + if (ret) { + goto out; + } + + memcpy(&s->inode, buf, sizeof(s->inode)); + s->min_dirty_data_idx = UINT32_MAX; + s->max_dirty_data_idx = 0; + + bs->total_sectors = s->inode.vdi_size / SECTOR_SIZE; + strncpy(s->name, vdi, sizeof(s->name)); + qemu_free(buf); + return 0; +out: + qemu_aio_set_fd_handler(s->fd, NULL, NULL, NULL, NULL, NULL); + if (s->fd >= 0) { + closesocket(s->fd); + } + qemu_free(buf); + return -1; +} + +static int do_sd_create(char *filename, int64_t vdi_size, + uint32_t base_vid, uint32_t *vdi_id, int snapshot, + const char *addr, const char *port) +{ + SheepdogVdiReq hdr; + SheepdogVdiRsp *rsp = (SheepdogVdiRsp *)&hdr; + int fd, ret; + unsigned int wlen, rlen = 0; + char buf[SD_MAX_VDI_LEN]; + + fd = connect_to_sdog(addr, port); + if (fd < 0) { + return -EIO; + } + + memset(buf, 0, sizeof(buf)); + strncpy(buf, filename, SD_MAX_VDI_LEN); + + memset(&hdr, 0, sizeof(hdr)); + hdr.opcode = SD_OP_NEW_VDI; + hdr.base_vdi_id = base_vid; + + wlen = SD_MAX_VDI_LEN; + + hdr.flags = SD_FLAG_CMD_WRITE; + hdr.snapid = snapshot; + + hdr.data_length = wlen; + hdr.vdi_size = vdi_size; + + ret = do_req(fd, (SheepdogReq *)&hdr, buf, &wlen, &rlen); + + closesocket(fd); + + if (ret) { + return -EIO; + } + + if (rsp->result != SD_RES_SUCCESS) { + error_report("%s, %s\n", sd_strerror(rsp->result), filename); + return -EIO; + } + + if (vdi_id) { + *vdi_id = rsp->vdi_id; + } + + return 0; +} + +static int sd_create(const char *filename, QEMUOptionParameter *options) +{ + int ret; + uint32_t vid = 0; + int64_t vdi_size = 0; + char *backing_file = NULL; + + strstart(filename, "sheepdog:", (const char **)&filename); + + while (options && options->name) { + if (!strcmp(options->name, BLOCK_OPT_SIZE)) { + vdi_size = options->value.n; + } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FILE)) { + backing_file = options->value.s; + } + options++; + } + + if (vdi_size > SD_MAX_VDI_SIZE) { + error_report("too big image size\n"); + return -EINVAL; + } + + if (backing_file) { + BlockDriverState *bs; + BDRVSheepdogState *s; + BlockDriver *drv; + + /* Currently, only Sheepdog backing image is supported. */ + drv = bdrv_find_protocol(backing_file); + if (!drv || strcmp(drv->protocol_name, "sheepdog") != 0) { + error_report("backing_file must be a sheepdog image\n"); + return -EINVAL; + } + + ret = bdrv_file_open(&bs, backing_file, 0); + if (ret < 0) + return -EIO; + + s = bs->opaque; + + if (!is_snapshot(&s->inode)) { + error_report("cannot clone from a non snapshot vdi\n"); + bdrv_delete(bs); + return -EINVAL; + } + + vid = s->inode.vdi_id; + bdrv_delete(bs); + } + + return do_sd_create((char *)filename, vdi_size, vid, NULL, 0, NULL, NULL); +} + +static void sd_close(BlockDriverState *bs) +{ + BDRVSheepdogState *s = bs->opaque; + SheepdogVdiReq hdr; + SheepdogVdiRsp *rsp = (SheepdogVdiRsp *)&hdr; + unsigned int wlen, rlen = 0; + int fd, ret; + + dprintf("%s\n", s->name); + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + return; + } + + memset(&hdr, 0, sizeof(hdr)); + + hdr.opcode = SD_OP_RELEASE_VDI; + wlen = strlen(s->name) + 1; + hdr.data_length = wlen; + hdr.flags = SD_FLAG_CMD_WRITE; + + ret = do_req(fd, (SheepdogReq *)&hdr, s->name, &wlen, &rlen); + + closesocket(fd); + + if (!ret && rsp->result != SD_RES_SUCCESS && + rsp->result != SD_RES_VDI_NOT_LOCKED) { + error_report("%s, %s\n", sd_strerror(rsp->result), s->name); + } + + qemu_aio_set_fd_handler(s->fd, NULL, NULL, NULL, NULL, NULL); + closesocket(s->fd); + qemu_free(s->addr); +} + +static int64_t sd_getlength(BlockDriverState *bs) +{ + BDRVSheepdogState *s = bs->opaque; + + return s->inode.vdi_size; +} + +static int sd_truncate(BlockDriverState *bs, int64_t offset) +{ + BDRVSheepdogState *s = bs->opaque; + int ret, fd; + unsigned int datalen; + + if (offset < s->inode.vdi_size) { + error_report("shrinking is not supported\n"); + return -EINVAL; + } else if (offset > SD_MAX_VDI_SIZE) { + error_report("too big image size\n"); + return -EINVAL; + } + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + return -EIO; + } + + /* we don't need to update entire object */ + datalen = SD_INODE_SIZE - sizeof(s->inode.data_vdi_id); + s->inode.vdi_size = offset; + ret = write_object(fd, (char *)&s->inode, vid_to_vdi_oid(s->inode.vdi_id), + s->inode.nr_copies, datalen, 0, 0); + close(fd); + + if (ret < 0) { + error_report("failed to update an inode.\n"); + return -EIO; + } + + return 0; +} + +/* + * This function is called after writing data objects. If we need to + * update metadata, this sends a write request to the vdi object. + * Otherwise, this calls the AIOCB callback. + */ +static void sd_write_done(SheepdogAIOCB *acb) +{ + int ret; + BDRVSheepdogState *s = acb->common.bs->opaque; + struct iovec iov; + AIOReq *aio_req; + uint32_t offset, data_len, mn, mx; + + mn = s->min_dirty_data_idx; + mx = s->max_dirty_data_idx; + if (mn <= mx) { + /* we need to update the vdi object. */ + offset = sizeof(s->inode) - sizeof(s->inode.data_vdi_id) + + mn * sizeof(s->inode.data_vdi_id[0]); + data_len = (mx - mn + 1) * sizeof(s->inode.data_vdi_id[0]); + + s->min_dirty_data_idx = UINT32_MAX; + s->max_dirty_data_idx = 0; + + iov.iov_base = &s->inode; + iov.iov_len = sizeof(s->inode); + aio_req = alloc_aio_req(s, acb, vid_to_vdi_oid(s->inode.vdi_id), + data_len, offset, 0, 0, offset); + ret = add_aio_request(s, aio_req, &iov, 1, 0, AIOCB_WRITE_UDATA); + if (ret) { + free_aio_req(s, aio_req); + acb->ret = -EIO; + goto out; + } + + acb->aio_done_func = sd_finish_aiocb; + acb->aiocb_type = AIOCB_WRITE_UDATA; + return; + } +out: + sd_finish_aiocb(acb); +} + +/* + * Create a writable VDI from a snapshot + */ +static int sd_create_branch(BDRVSheepdogState *s) +{ + int ret, fd; + uint32_t vid; + char *buf; + + dprintf("%" PRIx32 " is snapshot.\n", s->inode.vdi_id); + + buf = qemu_malloc(SD_INODE_SIZE); + + ret = do_sd_create(s->name, s->inode.vdi_size, s->inode.vdi_id, &vid, 1, + s->addr, s->port); + if (ret) { + goto out; + } + + dprintf("%" PRIx32 " is created.\n", vid); + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + error_report("failed to connect\n"); + goto out; + } + + ret = read_object(fd, buf, vid_to_vdi_oid(vid), s->inode.nr_copies, + SD_INODE_SIZE, 0); + + closesocket(fd); + + if (ret < 0) { + goto out; + } + + memcpy(&s->inode, buf, sizeof(s->inode)); + + s->is_snapshot = 0; + ret = 0; + dprintf("%" PRIx32 " was newly created.\n", s->inode.vdi_id); + +out: + qemu_free(buf); + + return ret; +} + +/* + * Send I/O requests to the server. + * + * This function sends requests to the server, links the requests to + * the outstanding_list in BDRVSheepdogState, and exits without + * waiting the response. The responses are received in the + * `aio_read_response' function which is called from the main loop as + * a fd handler. + */ +static void sd_readv_writev_bh_cb(void *p) +{ + SheepdogAIOCB *acb = p; + int ret = 0; + unsigned long len, done = 0, total = acb->nb_sectors * SECTOR_SIZE; + unsigned long idx = acb->sector_num * SECTOR_SIZE / SD_DATA_OBJ_SIZE; + uint64_t oid; + uint64_t offset = (acb->sector_num * SECTOR_SIZE) % SD_DATA_OBJ_SIZE; + BDRVSheepdogState *s = acb->common.bs->opaque; + SheepdogInode *inode = &s->inode; + AIOReq *aio_req; + + qemu_bh_delete(acb->bh); + acb->bh = NULL; + + if (acb->aiocb_type == AIOCB_WRITE_UDATA && s->is_snapshot) { + /* + * In the case we open the snapshot VDI, Sheepdog creates the + * writable VDI when we do a write operation first. + */ + ret = sd_create_branch(s); + if (ret) { + acb->ret = -EIO; + goto out; + } + } + + while (done != total) { + uint8_t flags = 0; + uint64_t old_oid = 0; + int create = 0; + + oid = vid_to_data_oid(inode->data_vdi_id[idx], idx); + + len = MIN(total - done, SD_DATA_OBJ_SIZE - offset); + + if (!inode->data_vdi_id[idx]) { + if (acb->aiocb_type == AIOCB_READ_UDATA) { + goto done; + } + + create = 1; + } else if (acb->aiocb_type == AIOCB_WRITE_UDATA + && !is_data_obj_writeable(inode, idx)) { + /* Copy-On-Write */ + create = 1; + old_oid = oid; + flags = SD_FLAG_CMD_COW; + } + + if (create) { + dprintf("update ino (%" PRIu32") %" PRIu64 " %" PRIu64 + " %" PRIu64 "\n", inode->vdi_id, oid, + vid_to_data_oid(inode->data_vdi_id[idx], idx), idx); + oid = vid_to_data_oid(inode->vdi_id, idx); + dprintf("new oid %lx\n", oid); + } + + aio_req = alloc_aio_req(s, acb, oid, len, offset, flags, old_oid, done); + + if (create) { + AIOReq *areq; + QLIST_FOREACH(areq, &s->outstanding_aio_head, + outstanding_aio_siblings) { + if (areq == aio_req) { + continue; + } + if (areq->oid == oid) { + /* + * Sheepdog cannot handle simultaneous create + * requests to the same object. So we cannot send + * the request until the previous request + * finishes. + */ + aio_req->flags = 0; + aio_req->base_oid = 0; + goto done; + } + } + } + + ret = add_aio_request(s, aio_req, acb->qiov->iov, acb->qiov->niov, + create, acb->aiocb_type); + if (ret < 0) { + error_report("add_aio_request is failed\n"); + free_aio_req(s, aio_req); + acb->ret = -EIO; + goto out; + } + done: + offset = 0; + idx++; + done += len; + } +out: + if (QLIST_EMPTY(&acb->aioreq_head)) { + sd_finish_aiocb(acb); + } +} + +static BlockDriverAIOCB *sd_aio_writev(BlockDriverState *bs, int64_t sector_num, + QEMUIOVector *qiov, int nb_sectors, + BlockDriverCompletionFunc *cb, + void *opaque) +{ + SheepdogAIOCB *acb; + + if (bs->growable && sector_num + nb_sectors > bs->total_sectors) { + /* TODO: shouldn't block here */ + if (sd_truncate(bs, (sector_num + nb_sectors) * SECTOR_SIZE) < 0) { + return NULL; + } + bs->total_sectors = sector_num + nb_sectors; + } + + acb = sd_aio_setup(bs, qiov, sector_num, nb_sectors, cb, opaque); + acb->aio_done_func = sd_write_done; + acb->aiocb_type = AIOCB_WRITE_UDATA; + + sd_schedule_bh(sd_readv_writev_bh_cb, acb); + return &acb->common; +} + +static BlockDriverAIOCB *sd_aio_readv(BlockDriverState *bs, int64_t sector_num, + QEMUIOVector *qiov, int nb_sectors, + BlockDriverCompletionFunc *cb, + void *opaque) +{ + SheepdogAIOCB *acb; + int i; + + acb = sd_aio_setup(bs, qiov, sector_num, nb_sectors, cb, opaque); + acb->aiocb_type = AIOCB_READ_UDATA; + acb->aio_done_func = sd_finish_aiocb; + + /* + * TODO: we can do better; we don't need to initialize + * blindly. + */ + for (i = 0; i < qiov->niov; i++) { + memset(qiov->iov[i].iov_base, 0, qiov->iov[i].iov_len); + } + + sd_schedule_bh(sd_readv_writev_bh_cb, acb); + return &acb->common; +} + +static int sd_snapshot_create(BlockDriverState *bs, QEMUSnapshotInfo *sn_info) +{ + BDRVSheepdogState *s = bs->opaque; + int ret, fd; + uint32_t new_vid; + SheepdogInode *inode; + unsigned int datalen; + + dprintf("sn_info: name %s id_str %s s: name %s vm_state_size %d " + "is_snapshot %d\n", sn_info->name, sn_info->id_str, + s->name, sn_info->vm_state_size, s->is_snapshot); + + if (s->is_snapshot) { + error_report("You can't create a snapshot of a snapshot VDI, " + "%s (%" PRIu32 ").\n", s->name, s->inode.vdi_id); + + return -EINVAL; + } + + dprintf("%s %s\n", sn_info->name, sn_info->id_str); + + s->inode.vm_state_size = sn_info->vm_state_size; + s->inode.vm_clock_nsec = sn_info->vm_clock_nsec; + strncpy(s->inode.tag, sn_info->name, sizeof(s->inode.tag)); + /* we don't need to update entire object */ + datalen = SD_INODE_SIZE - sizeof(s->inode.data_vdi_id); + + /* refresh inode. */ + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + ret = -EIO; + goto cleanup; + } + + ret = write_object(fd, (char *)&s->inode, vid_to_vdi_oid(s->inode.vdi_id), + s->inode.nr_copies, datalen, 0, 0); + if (ret < 0) { + error_report("failed to write snapshot's inode.\n"); + ret = -EIO; + goto cleanup; + } + + ret = do_sd_create(s->name, s->inode.vdi_size, s->inode.vdi_id, &new_vid, 1, + s->addr, s->port); + if (ret < 0) { + error_report("failed to create inode for snapshot. %s\n", + strerror(errno)); + ret = -EIO; + goto cleanup; + } + + inode = (SheepdogInode *)qemu_malloc(datalen); + + ret = read_object(fd, (char *)inode, vid_to_vdi_oid(new_vid), + s->inode.nr_copies, datalen, 0); + + if (ret < 0) { + error_report("failed to read new inode info. %s\n", strerror(errno)); + ret = -EIO; + goto cleanup; + } + + memcpy(&s->inode, inode, datalen); + dprintf("s->inode: name %s snap_id %x oid %x\n", + s->inode.name, s->inode.snap_id, s->inode.vdi_id); + +cleanup: + closesocket(fd); + return ret; +} + +static int sd_snapshot_goto(BlockDriverState *bs, const char *snapshot_id) +{ + BDRVSheepdogState *s = bs->opaque; + BDRVSheepdogState *old_s; + char vdi[SD_MAX_VDI_LEN], tag[SD_MAX_VDI_TAG_LEN]; + char *buf = NULL; + uint32_t vid; + uint32_t snapid = 0; + int ret = -ENOENT, fd; + + old_s = qemu_malloc(sizeof(BDRVSheepdogState)); + + memcpy(old_s, s, sizeof(BDRVSheepdogState)); + + memset(vdi, 0, sizeof(vdi)); + strncpy(vdi, s->name, sizeof(vdi)); + + memset(tag, 0, sizeof(tag)); + snapid = strtoul(snapshot_id, NULL, 10); + if (!snapid) { + strncpy(tag, s->name, sizeof(tag)); + } + + ret = find_vdi_name(s, vdi, snapid, tag, &vid, 1); + if (ret) { + error_report("Failed to find_vdi_name\n"); + ret = -ENOENT; + goto out; + } + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + error_report("failed to connect\n"); + goto out; + } + + buf = qemu_malloc(SD_INODE_SIZE); + ret = read_object(fd, buf, vid_to_vdi_oid(vid), s->inode.nr_copies, + SD_INODE_SIZE, 0); + + closesocket(fd); + + if (ret) { + ret = -ENOENT; + goto out; + } + + memcpy(&s->inode, buf, sizeof(s->inode)); + + if (!s->inode.vm_state_size) { + error_report("Invalid snapshot\n"); + ret = -ENOENT; + goto out; + } + + s->is_snapshot = 1; + + qemu_free(buf); + qemu_free(old_s); + + return 0; +out: + /* recover bdrv_sd_state */ + memcpy(s, old_s, sizeof(BDRVSheepdogState)); + qemu_free(buf); + qemu_free(old_s); + + error_report("failed to open. recover old bdrv_sd_state.\n"); + + return ret; +} + +static int sd_snapshot_delete(BlockDriverState *bs, const char *snapshot_id) +{ + /* FIXME: Delete specified snapshot id. */ + return 0; +} + +#define DIV_ROUND_UP(n,d) (((n) + (d) - 1) / (d)) +#define BITS_PER_BYTE 8 +#define BITS_TO_LONGS(nr) DIV_ROUND_UP(nr, BITS_PER_BYTE * sizeof(long)) +#define DECLARE_BITMAP(name,bits) \ + unsigned long name[BITS_TO_LONGS(bits)] + +#define BITS_PER_LONG (BITS_PER_BYTE * sizeof(long)) + +static inline int test_bit(unsigned int nr, const unsigned long *addr) +{ + return ((1UL << (nr % BITS_PER_LONG)) & + (((unsigned long *)addr)[nr / BITS_PER_LONG])) != 0; +} + +static int sd_snapshot_list(BlockDriverState *bs, QEMUSnapshotInfo **psn_tab) +{ + BDRVSheepdogState *s = bs->opaque; + SheepdogReq req; + int fd, nr = 1024, ret, max = BITS_TO_LONGS(SD_NR_VDIS) * sizeof(long); + QEMUSnapshotInfo *sn_tab = NULL; + unsigned wlen, rlen; + int found = 0; + static SheepdogInode inode; + unsigned long *vdi_inuse; + unsigned int start_nr; + uint64_t hval; + uint32_t vid; + + vdi_inuse = qemu_malloc(max); + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + goto out; + } + + rlen = max; + wlen = 0; + + memset(&req, 0, sizeof(req)); + + req.opcode = SD_OP_READ_VDIS; + req.data_length = max; + + ret = do_req(fd, (SheepdogReq *)&req, vdi_inuse, &wlen, &rlen); + + closesocket(fd); + if (ret) { + goto out; + } + + sn_tab = qemu_mallocz(nr * sizeof(*sn_tab)); + + /* calculate a vdi id with hash function */ + hval = fnv_64a_buf(s->name, strlen(s->name), FNV1A_64_INIT); + start_nr = hval & (SD_NR_VDIS - 1); + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + error_report("failed to connect\n"); + goto out; + } + + for (vid = start_nr; found < nr; vid = (vid + 1) % SD_NR_VDIS) { + if (!test_bit(vid, vdi_inuse)) { + break; + } + + /* we don't need to read entire object */ + ret = read_object(fd, (char *)&inode, vid_to_vdi_oid(vid), + 0, SD_INODE_SIZE - sizeof(inode.data_vdi_id), 0); + + if (ret) { + continue; + } + + if (!strcmp(inode.name, s->name) && is_snapshot(&inode)) { + sn_tab[found].date_sec = inode.snap_ctime >> 32; + sn_tab[found].date_nsec = inode.snap_ctime & 0xffffffff; + sn_tab[found].vm_state_size = inode.vm_state_size; + sn_tab[found].vm_clock_nsec = inode.vm_clock_nsec; + + snprintf(sn_tab[found].id_str, sizeof(sn_tab[found].id_str), "%u", + inode.snap_id); + strncpy(sn_tab[found].name, inode.tag, + MIN(sizeof(sn_tab[found].name), sizeof(inode.tag))); + found++; + } + } + + closesocket(fd); +out: + *psn_tab = sn_tab; + + qemu_free(vdi_inuse); + + return found; +} + +static int do_load_save_vmstate(BDRVSheepdogState *s, uint8_t *data, + int64_t pos, int size, int load) +{ + int fd, create; + int ret = 0; + unsigned int data_len; + uint64_t vmstate_oid; + uint32_t vdi_index; + uint64_t offset; + + fd = connect_to_sdog(s->addr, s->port); + if (fd < 0) { + ret = -EIO; + goto cleanup; + } + + while (size) { + vdi_index = pos / SD_DATA_OBJ_SIZE; + offset = pos % SD_DATA_OBJ_SIZE; + + data_len = MIN(size, SD_DATA_OBJ_SIZE); + + vmstate_oid = vid_to_vmstate_oid(s->inode.vdi_id, vdi_index); + + create = (offset == 0); + if (load) { + ret = read_object(fd, (char *)data, vmstate_oid, + s->inode.nr_copies, data_len, offset); + } else { + ret = write_object(fd, (char *)data, vmstate_oid, + s->inode.nr_copies, data_len, offset, create); + } + + if (ret < 0) { + error_report("failed to save vmstate %s\n", strerror(errno)); + ret = -EIO; + goto cleanup; + } + + pos += data_len; + size -= data_len; + ret += data_len; + } +cleanup: + closesocket(fd); + return ret; +} + +static int sd_save_vmstate(BlockDriverState *bs, const uint8_t *data, + int64_t pos, int size) +{ + BDRVSheepdogState *s = bs->opaque; + + return do_load_save_vmstate(s, (uint8_t *)data, pos, size, 0); +} + +static int sd_load_vmstate(BlockDriverState *bs, uint8_t *data, + int64_t pos, int size) +{ + BDRVSheepdogState *s = bs->opaque; + + return do_load_save_vmstate(s, data, pos, size, 1); +} + + +static QEMUOptionParameter sd_create_options[] = { + { + .name = BLOCK_OPT_SIZE, + .type = OPT_SIZE, + .help = "Virtual disk size" + }, + { + .name = BLOCK_OPT_BACKING_FILE, + .type = OPT_STRING, + .help = "File name of a base image" + }, + { NULL } +}; + +BlockDriver bdrv_sheepdog = { + .format_name = "sheepdog", + .protocol_name = "sheepdog", + .instance_size = sizeof(BDRVSheepdogState), + .bdrv_file_open = sd_open, + .bdrv_close = sd_close, + .bdrv_create = sd_create, + .bdrv_getlength = sd_getlength, + .bdrv_truncate = sd_truncate, + + .bdrv_aio_readv = sd_aio_readv, + .bdrv_aio_writev = sd_aio_writev, + + .bdrv_snapshot_create = sd_snapshot_create, + .bdrv_snapshot_goto = sd_snapshot_goto, + .bdrv_snapshot_delete = sd_snapshot_delete, + .bdrv_snapshot_list = sd_snapshot_list, + + .bdrv_save_vmstate = sd_save_vmstate, + .bdrv_load_vmstate = sd_load_vmstate, + + .create_options = sd_create_options, +}; + +static void bdrv_sheepdog_init(void) +{ + bdrv_register(&bdrv_sheepdog); +} +block_init(bdrv_sheepdog_init); -- cgit v1.2.1