[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-block] [PATCH 1/5] block: Add blklogwrites
From: |
Ari Sundholm |
Subject: |
[Qemu-block] [PATCH 1/5] block: Add blklogwrites |
Date: |
Fri, 1 Jun 2018 00:17:19 +0300 |
From: Aapo Vienamo <address@hidden>
Implements a block device write logging system, similar to Linux kernel
device mapper dm-log-writes. The write operations that are performed
on a block device are logged to a file or another block device. The
write log format is identical to the dm-log-writes format. Currently,
log markers are not supported.
This functionality can be used for fail-safe and fs consistency
testing. By implementing it in qemu, tests utilizing write logs can be
be used to test non-Linux drivers and older kernels.
The implementation is based on the blkverify and blkdebug block drivers.
Signed-off-by: Aapo Vienamo <address@hidden>
Signed-off-by: Ari Sundholm <address@hidden>
---
block.c | 6 -
block/Makefile.objs | 1 +
block/blklogwrites.c | 441 ++++++++++++++++++++++++++++++++++++++++++++++++++
include/block/block.h | 7 +
4 files changed, 449 insertions(+), 6 deletions(-)
create mode 100644 block/blklogwrites.c
diff --git a/block.c b/block.c
index 501b64c..c8cffe1 100644
--- a/block.c
+++ b/block.c
@@ -1914,12 +1914,6 @@ int bdrv_child_try_set_perm(BdrvChild *c, uint64_t perm,
uint64_t shared,
return 0;
}
-#define DEFAULT_PERM_PASSTHROUGH (BLK_PERM_CONSISTENT_READ \
- | BLK_PERM_WRITE \
- | BLK_PERM_WRITE_UNCHANGED \
- | BLK_PERM_RESIZE)
-#define DEFAULT_PERM_UNCHANGED (BLK_PERM_ALL & ~DEFAULT_PERM_PASSTHROUGH)
-
void bdrv_filter_default_perms(BlockDriverState *bs, BdrvChild *c,
const BdrvChildRole *role,
BlockReopenQueue *reopen_queue,
diff --git a/block/Makefile.objs b/block/Makefile.objs
index 899bfb5..c8337bf 100644
--- a/block/Makefile.objs
+++ b/block/Makefile.objs
@@ -5,6 +5,7 @@ block-obj-y += qed-check.o
block-obj-y += vhdx.o vhdx-endian.o vhdx-log.o
block-obj-y += quorum.o
block-obj-y += parallels.o blkdebug.o blkverify.o blkreplay.o
+block-obj-y += blklogwrites.o
block-obj-y += block-backend.o snapshot.o qapi.o
block-obj-$(CONFIG_WIN32) += file-win32.o win32-aio.o
block-obj-$(CONFIG_POSIX) += file-posix.o
diff --git a/block/blklogwrites.c b/block/blklogwrites.c
new file mode 100644
index 0000000..78835bf
--- /dev/null
+++ b/block/blklogwrites.c
@@ -0,0 +1,441 @@
+/*
+ * Write logging blk driver based on blkverify and blkdebug.
+ *
+ * Copyright (c) 2017 Tuomas Tynkkynen <address@hidden>
+ * Copyright (c) 2018 Aapo Vienamo <address@hidden>
+ * Copyright (c) 2018 Ari Sundholm <address@hidden>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#include "qemu/osdep.h"
+#include "qapi/error.h"
+#include "qemu/sockets.h" /* for EINPROGRESS on Windows */
+#include "block/block_int.h"
+#include "qapi/qmp/qdict.h"
+#include "qapi/qmp/qstring.h"
+#include "qemu/cutils.h"
+#include "qemu/option.h"
+
+/* Disk format stuff - taken from Linux drivers/md/dm-log-writes.c */
+
+#define LOG_FLUSH_FLAG (1 << 0)
+#define LOG_FUA_FLAG (1 << 1)
+#define LOG_DISCARD_FLAG (1 << 2)
+#define LOG_MARK_FLAG (1 << 3)
+
+#define WRITE_LOG_VERSION 1ULL
+#define WRITE_LOG_MAGIC 0x6a736677736872ULL
+
+/* All fields are little-endian. */
+struct log_write_super {
+ uint64_t magic;
+ uint64_t version;
+ uint64_t nr_entries;
+ uint32_t sectorsize;
+};
+
+struct log_write_entry {
+ uint64_t sector;
+ uint64_t nr_sectors;
+ uint64_t flags;
+ uint64_t data_len;
+};
+
+/* End of disk format structures. */
+
+typedef struct {
+ BdrvChild *log_file;
+ uint64_t cur_log_sector;
+ uint64_t nr_entries;
+} BDRVBlkLogWritesState;
+
+/* Valid blk_log_writes filenames look like:
+ * blk_log_writes:path/to/raw_image:path/to/logfile */
+static void blk_log_writes_parse_filename(const char *filename, QDict *options,
+ Error **errp)
+{
+ const char *c;
+ QString *raw_path;
+
+ /* Parse the blk_log_writes: prefix */
+ if (!strstart(filename, "blk_log_writes:", &filename)) {
+ /* There was no prefix; therefore, all options have to be already
+ * present in the QDict (except for the filename) */
+ qdict_put(options, "x-log", qstring_from_str(filename));
+ return;
+ }
+
+ /* Parse the raw image filename */
+ c = strchr(filename, ':');
+ if (c == NULL) {
+ error_setg(errp,
+ "blk_log_writes requires paths to both image and log");
+ return;
+ }
+
+ raw_path = qstring_from_substr(filename, 0, c - filename - 1);
+ qdict_put(options, "x-raw", raw_path);
+
+ filename = c + 1;
+ qdict_put(options, "x-log", qstring_from_str(filename));
+}
+
+static QemuOptsList runtime_opts = {
+ .name = "blk_log_writes",
+ .head = QTAILQ_HEAD_INITIALIZER(runtime_opts.head),
+ .desc = {
+ {
+ .name = "x-raw",
+ .type = QEMU_OPT_STRING,
+ .help = "[internal use only, will be removed]",
+ },
+ {
+ .name = "x-log",
+ .type = QEMU_OPT_STRING,
+ .help = "[internal use only, will be removed]",
+ },
+ { /* end of list */ }
+ },
+};
+
+static int blk_log_writes_open(BlockDriverState *bs, QDict *options, int flags,
+ Error **errp)
+{
+ BDRVBlkLogWritesState *s = bs->opaque;
+ QemuOpts *opts;
+ Error *local_err = NULL;
+ int ret;
+
+ opts = qemu_opts_create(&runtime_opts, NULL, 0, &error_abort);
+ qemu_opts_absorb_qdict(opts, options, &local_err);
+ if (local_err) {
+ ret = -EINVAL;
+ error_propagate(errp, local_err);
+ goto fail;
+ }
+
+ /* Open the raw file */
+ bs->file = bdrv_open_child(qemu_opt_get(opts, "x-raw"), options, "raw",
+ bs, &child_file, false, &local_err);
+ if (local_err) {
+ ret = -EINVAL;
+ error_propagate(errp, local_err);
+ goto fail;
+ }
+
+ s->cur_log_sector = 1;
+ s->nr_entries = 0;
+ qdict_put_str(options, "log.driver", "raw");
+ qdict_put_str(options, "log.read-only", "off");
+
+ /* Open the log file */
+ s->log_file = bdrv_open_child(qemu_opt_get(opts, "x-log"), options,
+ "log", bs, &child_file, false,
+ &local_err);
+ if (local_err) {
+ ret = -EINVAL;
+ error_propagate(errp, local_err);
+ goto fail;
+ }
+
+ ret = 0;
+fail:
+ if (ret < 0) {
+ bdrv_unref_child(bs, bs->file);
+ bs->file = NULL;
+ }
+ qemu_opts_del(opts);
+ return ret;
+}
+
+static void blk_log_writes_close(BlockDriverState *bs)
+{
+ BDRVBlkLogWritesState *s = bs->opaque;
+
+ bdrv_unref_child(bs, s->log_file);
+ s->log_file = NULL;
+}
+
+static int64_t blk_log_writes_getlength(BlockDriverState *bs)
+{
+ return bdrv_getlength(bs->file->bs);
+}
+
+static void blk_log_writes_refresh_filename(BlockDriverState *bs,
+ QDict *options)
+{
+ BDRVBlkLogWritesState *s = bs->opaque;
+
+ /* bs->file->bs has already been refreshed */
+ bdrv_refresh_filename(s->log_file->bs);
+
+ if (bs->file->bs->full_open_options
+ && s->log_file->bs->full_open_options)
+ {
+ QDict *opts = qdict_new();
+ qdict_put_obj(opts, "driver",
+ QOBJECT(qstring_from_str("blk_log_writes")));
+
+ qobject_ref(bs->file->bs->full_open_options);
+ qdict_put_obj(opts, "raw", QOBJECT(bs->file->bs->full_open_options));
+ qobject_ref(s->log_file->bs->full_open_options);
+ qdict_put_obj(opts, "log",
+ QOBJECT(s->log_file->bs->full_open_options));
+
+ bs->full_open_options = opts;
+ }
+
+ if (bs->file->bs->exact_filename[0]
+ && s->log_file->bs->exact_filename[0])
+ {
+ snprintf(bs->exact_filename, sizeof(bs->exact_filename),
+ "blk_log_writes:%s:%s",
+ bs->file->bs->exact_filename,
+ s->log_file->bs->exact_filename);
+ }
+}
+
+static void blk_log_writes_child_perm(BlockDriverState *bs, BdrvChild *c,
+ const BdrvChildRole *role,
+ BlockReopenQueue *ro_q,
+ uint64_t perm, uint64_t shrd,
+ uint64_t *nperm, uint64_t *nshrd)
+{
+ if (!c) {
+ *nperm = perm & DEFAULT_PERM_PASSTHROUGH;
+ *nshrd = (shrd & DEFAULT_PERM_PASSTHROUGH) | DEFAULT_PERM_UNCHANGED;
+ return;
+ }
+
+ if (!strcmp(c->name, "log"))
+ bdrv_format_default_perms(bs, c, role, ro_q, perm, shrd, nperm, nshrd);
+ else
+ bdrv_filter_default_perms(bs, c, role, ro_q, perm, shrd, nperm, nshrd);
+}
+
+static void blk_log_writes_refresh_limits(BlockDriverState *bs, Error **errp)
+{
+ if (bs->bl.request_alignment < BDRV_SECTOR_SIZE) {
+ bs->bl.request_alignment = BDRV_SECTOR_SIZE;
+
+ if (bs->bl.pdiscard_alignment &&
+ bs->bl.pdiscard_alignment < bs->bl.request_alignment)
+ bs->bl.pdiscard_alignment = bs->bl.request_alignment;
+ if (bs->bl.pwrite_zeroes_alignment &&
+ bs->bl.pwrite_zeroes_alignment < bs->bl.request_alignment)
+ bs->bl.pwrite_zeroes_alignment = bs->bl.request_alignment;
+ }
+}
+
+static int coroutine_fn
+blk_log_writes_co_preadv(BlockDriverState *bs, uint64_t offset, uint64_t bytes,
+ QEMUIOVector *qiov, int flags)
+{
+ return bdrv_co_preadv(bs->file, offset, bytes, qiov, flags);
+}
+
+typedef struct BlkLogWritesReq {
+ Coroutine *co;
+ BlockDriverState *bs;
+ unsigned done;
+} BlkLogWritesReq;
+
+typedef struct BlkLogWritesFileReq {
+ BlkLogWritesReq *r;
+ uint64_t offset;
+ uint64_t bytes;
+ int file_flags;
+ QEMUIOVector *qiov;
+ int (*func)(struct BlkLogWritesFileReq *r);
+ int file_ret;
+} BlkLogWritesFileReq;
+
+typedef struct {
+ BlkLogWritesReq *r;
+ QEMUIOVector *qiov;
+ struct log_write_entry entry;
+ int log_ret;
+} BlkLogWritesLogReq;
+
+static void coroutine_fn blk_log_writes_co_do_log(void *opaque)
+{
+ BlkLogWritesLogReq *lr = opaque;
+ BDRVBlkLogWritesState *s = lr->r->bs->opaque;
+ uint64_t cur_log_offset = s->cur_log_sector * BDRV_SECTOR_SIZE;
+
+ s->nr_entries++;
+ s->cur_log_sector +=
+ ROUND_UP(lr->qiov->size, BDRV_SECTOR_SIZE) >> BDRV_SECTOR_BITS;
+
+ lr->log_ret = bdrv_co_pwritev(s->log_file, cur_log_offset, lr->qiov->size,
+ lr->qiov, 0);
+
+ /* Update super block on flush */
+ if (lr->log_ret == 0 && lr->entry.flags & LOG_FLUSH_FLAG) {
+ struct log_write_super super = {
+ .magic = cpu_to_le64(WRITE_LOG_MAGIC),
+ .version = cpu_to_le64(WRITE_LOG_VERSION),
+ .nr_entries = cpu_to_le64(s->nr_entries),
+ .sectorsize = cpu_to_le32(1 << BDRV_SECTOR_BITS),
+ };
+ static const char zeroes[BDRV_SECTOR_SIZE - sizeof(super)] = { '\0' };
+ QEMUIOVector qiov;
+
+ qemu_iovec_init(&qiov, 2);
+ qemu_iovec_add(&qiov, &super, sizeof(super));
+ qemu_iovec_add(&qiov, (void *)zeroes, sizeof(zeroes));
+
+ lr->log_ret =
+ bdrv_co_pwritev(s->log_file, 0, BDRV_SECTOR_SIZE, &qiov, 0);
+ if (lr->log_ret == 0)
+ lr->log_ret = bdrv_co_flush(s->log_file->bs);
+ qemu_iovec_destroy(&qiov);
+ }
+
+ lr->r->done++;
+ qemu_coroutine_enter_if_inactive(lr->r->co);
+}
+
+static void blk_log_writes_co_do_file(void *opaque)
+{
+ BlkLogWritesFileReq *fr = opaque;
+
+ fr->file_ret = fr->func(fr);
+
+ fr->r->done++;
+ qemu_coroutine_enter_if_inactive(fr->r->co);
+}
+
+static int coroutine_fn
+blk_log_writes_co_log(BlockDriverState *bs, uint64_t offset, uint64_t bytes,
+ QEMUIOVector *qiov, int flags,
+ int (*file_func)(BlkLogWritesFileReq *r),
+ uint64_t entry_flags)
+{
+ QEMUIOVector log_qiov;
+ size_t niov = qiov ? qiov->niov : 0;
+ Coroutine *co_log, *co_file;
+ BlkLogWritesReq r = {
+ .co = qemu_coroutine_self(),
+ .bs = bs,
+ .done = 0,
+ };
+ BlkLogWritesFileReq fr = {
+ .r = &r,
+ .offset = offset,
+ .bytes = bytes,
+ .qiov = qiov,
+ .func = file_func,
+ };
+ BlkLogWritesLogReq lr = {
+ .r = &r,
+ .qiov = &log_qiov,
+ .entry = {
+ .sector = cpu_to_le64(offset >> BDRV_SECTOR_BITS),
+ .nr_sectors = cpu_to_le64(bytes >> BDRV_SECTOR_BITS),
+ .flags = cpu_to_le64(entry_flags),
+ .data_len = 0,
+ },
+ };
+ static const char zeroes[BDRV_SECTOR_SIZE - sizeof(struct log_write_entry)]
+ = { '\0' };
+
+ assert(QEMU_IS_ALIGNED(offset, bs->bl.request_alignment));
+ assert(QEMU_IS_ALIGNED(bytes, bs->bl.request_alignment));
+
+ qemu_iovec_init(&log_qiov, niov + 2);
+ qemu_iovec_add(&log_qiov, &lr.entry, sizeof(lr.entry));
+ qemu_iovec_add(&log_qiov, (void *)zeroes, sizeof(zeroes));
+ for (size_t i = 0; i < niov; ++i) {
+ qemu_iovec_add(&log_qiov, qiov->iov[i].iov_base, qiov->iov[i].iov_len);
+ }
+
+ co_log = qemu_coroutine_create(blk_log_writes_co_do_log, &lr);
+ co_file = qemu_coroutine_create(blk_log_writes_co_do_file, &fr);
+
+ qemu_coroutine_enter(co_file);
+ qemu_coroutine_enter(co_log);
+
+ while (r.done < 2) {
+ qemu_coroutine_yield();
+ }
+
+ qemu_iovec_destroy(&log_qiov);
+
+ if (lr.log_ret < 0) {
+ return lr.log_ret;
+ }
+
+ return fr.file_ret;
+}
+
+static int coroutine_fn
+blk_log_writes_co_do_file_pwritev(BlkLogWritesFileReq *fr)
+{
+ return bdrv_co_pwritev(fr->r->bs->file, fr->offset, fr->bytes,
+ fr->qiov, fr->file_flags);
+}
+
+static int coroutine_fn blk_log_writes_co_do_file_flush(BlkLogWritesFileReq
*fr)
+{
+ return bdrv_co_flush(fr->r->bs->file->bs);
+}
+
+static int coroutine_fn
+blk_log_writes_co_do_file_pdiscard(BlkLogWritesFileReq *fr)
+{
+ return bdrv_co_pdiscard(fr->r->bs->file->bs, fr->offset, fr->bytes);
+}
+
+static int coroutine_fn
+blk_log_writes_co_pwritev(BlockDriverState *bs, uint64_t offset, uint64_t
bytes,
+ QEMUIOVector *qiov, int flags)
+{
+ return blk_log_writes_co_log(bs, offset, bytes, qiov, flags,
+ blk_log_writes_co_do_file_pwritev, 0);
+}
+
+static int coroutine_fn blk_log_writes_co_flush_to_disk(BlockDriverState *bs)
+{
+ return blk_log_writes_co_log(bs, 0, 0, NULL, 0,
+ blk_log_writes_co_do_file_flush,
+ LOG_FLUSH_FLAG);
+}
+
+static int coroutine_fn
+blk_log_writes_co_pdiscard(BlockDriverState *bs, int64_t offset, int count)
+{
+ return blk_log_writes_co_log(bs, offset, count, NULL, 0,
+ blk_log_writes_co_do_file_pdiscard,
+ LOG_DISCARD_FLAG);
+}
+
+static BlockDriver bdrv_blk_log_writes = {
+ .format_name = "blk_log_writes",
+ .protocol_name = "blk_log_writes",
+ .instance_size = sizeof(BDRVBlkLogWritesState),
+
+ .bdrv_parse_filename = blk_log_writes_parse_filename,
+ .bdrv_file_open = blk_log_writes_open,
+ .bdrv_close = blk_log_writes_close,
+ .bdrv_getlength = blk_log_writes_getlength,
+ .bdrv_refresh_filename = blk_log_writes_refresh_filename,
+ .bdrv_child_perm = blk_log_writes_child_perm,
+ .bdrv_refresh_limits = blk_log_writes_refresh_limits,
+
+ .bdrv_co_preadv = blk_log_writes_co_preadv,
+ .bdrv_co_pwritev = blk_log_writes_co_pwritev,
+ .bdrv_co_flush_to_disk = blk_log_writes_co_flush_to_disk,
+ .bdrv_co_pdiscard = blk_log_writes_co_pdiscard,
+
+ .is_filter = true,
+};
+
+static void bdrv_blk_log_writes_init(void)
+{
+ bdrv_register(&bdrv_blk_log_writes);
+}
+
+block_init(bdrv_blk_log_writes_init);
diff --git a/include/block/block.h b/include/block/block.h
index 3894edd..fb7d379 100644
--- a/include/block/block.h
+++ b/include/block/block.h
@@ -225,6 +225,13 @@ enum {
BLK_PERM_GRAPH_MOD = 0x10,
BLK_PERM_ALL = 0x1f,
+
+ DEFAULT_PERM_PASSTHROUGH = BLK_PERM_CONSISTENT_READ
+ | BLK_PERM_WRITE
+ | BLK_PERM_WRITE_UNCHANGED
+ | BLK_PERM_RESIZE,
+
+ DEFAULT_PERM_UNCHANGED = BLK_PERM_ALL & ~DEFAULT_PERM_PASSTHROUGH,
};
char *bdrv_perm_names(uint64_t perm);
--
2.7.4
- [Qemu-block] [PATCH 1/5] block: Add blklogwrites,
Ari Sundholm <=