From: Kevin Wolf <kwolf@redhat.com>
To: qemu-block@nongnu.org
Cc: kwolf@redhat.com, peter.maydell@linaro.org, qemu-devel@nongnu.org
Subject: [Qemu-devel] [PULL 11/29] file-posix: Make .bdrv_co_truncate asynchronous
Date: Fri, 29 Jun 2018 16:09:41 +0200 [thread overview]
Message-ID: <20180629140959.6690-12-kwolf@redhat.com> (raw)
In-Reply-To: <20180629140959.6690-1-kwolf@redhat.com>
This moves the code to resize an image file to the thread pool to avoid
blocking.
Creating large images with preallocation with blockdev-create is now
actually a background job instead of blocking the monitor (and most
other things) until the preallocation has completed.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
include/block/raw-aio.h | 4 +-
block/file-posix.c | 266 +++++++++++++++++++++++++++---------------------
2 files changed, 154 insertions(+), 116 deletions(-)
diff --git a/include/block/raw-aio.h b/include/block/raw-aio.h
index 8d698ccd31..6799614e56 100644
--- a/include/block/raw-aio.h
+++ b/include/block/raw-aio.h
@@ -26,6 +26,7 @@
#define QEMU_AIO_DISCARD 0x0010
#define QEMU_AIO_WRITE_ZEROES 0x0020
#define QEMU_AIO_COPY_RANGE 0x0040
+#define QEMU_AIO_TRUNCATE 0x0080
#define QEMU_AIO_TYPE_MASK \
(QEMU_AIO_READ | \
QEMU_AIO_WRITE | \
@@ -33,7 +34,8 @@
QEMU_AIO_FLUSH | \
QEMU_AIO_DISCARD | \
QEMU_AIO_WRITE_ZEROES | \
- QEMU_AIO_COPY_RANGE)
+ QEMU_AIO_COPY_RANGE | \
+ QEMU_AIO_TRUNCATE)
/* AIO flags */
#define QEMU_AIO_MISALIGNED 0x1000
diff --git a/block/file-posix.c b/block/file-posix.c
index 53c5833659..639265d53b 100644
--- a/block/file-posix.c
+++ b/block/file-posix.c
@@ -188,8 +188,16 @@ typedef struct RawPosixAIOData {
#define aio_ioctl_cmd aio_nbytes /* for QEMU_AIO_IOCTL */
off_t aio_offset;
int aio_type;
- int aio_fd2;
- off_t aio_offset2;
+ union {
+ struct {
+ int aio_fd2;
+ off_t aio_offset2;
+ };
+ struct {
+ PreallocMode prealloc;
+ Error **errp;
+ };
+ };
} RawPosixAIOData;
#if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
@@ -1539,6 +1547,122 @@ static ssize_t handle_aiocb_discard(RawPosixAIOData *aiocb)
return ret;
}
+static int handle_aiocb_truncate(RawPosixAIOData *aiocb)
+{
+ int result = 0;
+ int64_t current_length = 0;
+ char *buf = NULL;
+ struct stat st;
+ int fd = aiocb->aio_fildes;
+ int64_t offset = aiocb->aio_offset;
+ Error **errp = aiocb->errp;
+
+ if (fstat(fd, &st) < 0) {
+ result = -errno;
+ error_setg_errno(errp, -result, "Could not stat file");
+ return result;
+ }
+
+ current_length = st.st_size;
+ if (current_length > offset && aiocb->prealloc != PREALLOC_MODE_OFF) {
+ error_setg(errp, "Cannot use preallocation for shrinking files");
+ return -ENOTSUP;
+ }
+
+ switch (aiocb->prealloc) {
+#ifdef CONFIG_POSIX_FALLOCATE
+ case PREALLOC_MODE_FALLOC:
+ /*
+ * Truncating before posix_fallocate() makes it about twice slower on
+ * file systems that do not support fallocate(), trying to check if a
+ * block is allocated before allocating it, so don't do that here.
+ */
+ if (offset != current_length) {
+ result = -posix_fallocate(fd, current_length,
+ offset - current_length);
+ if (result != 0) {
+ /* posix_fallocate() doesn't set errno. */
+ error_setg_errno(errp, -result,
+ "Could not preallocate new data");
+ }
+ } else {
+ result = 0;
+ }
+ goto out;
+#endif
+ case PREALLOC_MODE_FULL:
+ {
+ int64_t num = 0, left = offset - current_length;
+ off_t seek_result;
+
+ /*
+ * Knowing the final size from the beginning could allow the file
+ * system driver to do less allocations and possibly avoid
+ * fragmentation of the file.
+ */
+ if (ftruncate(fd, offset) != 0) {
+ result = -errno;
+ error_setg_errno(errp, -result, "Could not resize file");
+ goto out;
+ }
+
+ buf = g_malloc0(65536);
+
+ seek_result = lseek(fd, current_length, SEEK_SET);
+ if (seek_result < 0) {
+ result = -errno;
+ error_setg_errno(errp, -result,
+ "Failed to seek to the old end of file");
+ goto out;
+ }
+
+ while (left > 0) {
+ num = MIN(left, 65536);
+ result = write(fd, buf, num);
+ if (result < 0) {
+ result = -errno;
+ error_setg_errno(errp, -result,
+ "Could not write zeros for preallocation");
+ goto out;
+ }
+ left -= result;
+ }
+ if (result >= 0) {
+ result = fsync(fd);
+ if (result < 0) {
+ result = -errno;
+ error_setg_errno(errp, -result,
+ "Could not flush file to disk");
+ goto out;
+ }
+ }
+ goto out;
+ }
+ case PREALLOC_MODE_OFF:
+ if (ftruncate(fd, offset) != 0) {
+ result = -errno;
+ error_setg_errno(errp, -result, "Could not resize file");
+ }
+ return result;
+ default:
+ result = -ENOTSUP;
+ error_setg(errp, "Unsupported preallocation mode: %s",
+ PreallocMode_str(aiocb->prealloc));
+ return result;
+ }
+
+out:
+ if (result < 0) {
+ if (ftruncate(fd, current_length) < 0) {
+ error_report("Failed to restore old file length: %s",
+ strerror(errno));
+ }
+ }
+
+ g_free(buf);
+ return result;
+}
+
static int aio_worker(void *arg)
{
RawPosixAIOData *aiocb = arg;
@@ -1582,6 +1706,9 @@ static int aio_worker(void *arg)
case QEMU_AIO_COPY_RANGE:
ret = handle_aiocb_copy_range(aiocb);
break;
+ case QEMU_AIO_TRUNCATE:
+ ret = handle_aiocb_truncate(aiocb);
+ break;
default:
fprintf(stderr, "invalid aio request (0x%x)\n", aiocb->aio_type);
ret = -EINVAL;
@@ -1765,117 +1892,25 @@ static void raw_close(BlockDriverState *bs)
*
* Returns: 0 on success, -errno on failure.
*/
-static int raw_regular_truncate(int fd, int64_t offset, PreallocMode prealloc,
- Error **errp)
+static int coroutine_fn
+raw_regular_truncate(BlockDriverState *bs, int fd, int64_t offset,
+ PreallocMode prealloc, Error **errp)
{
- int result = 0;
- int64_t current_length = 0;
- char *buf = NULL;
- struct stat st;
-
- if (fstat(fd, &st) < 0) {
- result = -errno;
- error_setg_errno(errp, -result, "Could not stat file");
- return result;
- }
-
- current_length = st.st_size;
- if (current_length > offset && prealloc != PREALLOC_MODE_OFF) {
- error_setg(errp, "Cannot use preallocation for shrinking files");
- return -ENOTSUP;
- }
-
- switch (prealloc) {
-#ifdef CONFIG_POSIX_FALLOCATE
- case PREALLOC_MODE_FALLOC:
- /*
- * Truncating before posix_fallocate() makes it about twice slower on
- * file systems that do not support fallocate(), trying to check if a
- * block is allocated before allocating it, so don't do that here.
- */
- if (offset != current_length) {
- result = -posix_fallocate(fd, current_length, offset - current_length);
- if (result != 0) {
- /* posix_fallocate() doesn't set errno. */
- error_setg_errno(errp, -result,
- "Could not preallocate new data");
- }
- } else {
- result = 0;
- }
- goto out;
-#endif
- case PREALLOC_MODE_FULL:
- {
- int64_t num = 0, left = offset - current_length;
- off_t seek_result;
-
- /*
- * Knowing the final size from the beginning could allow the file
- * system driver to do less allocations and possibly avoid
- * fragmentation of the file.
- */
- if (ftruncate(fd, offset) != 0) {
- result = -errno;
- error_setg_errno(errp, -result, "Could not resize file");
- goto out;
- }
-
- buf = g_malloc0(65536);
-
- seek_result = lseek(fd, current_length, SEEK_SET);
- if (seek_result < 0) {
- result = -errno;
- error_setg_errno(errp, -result,
- "Failed to seek to the old end of file");
- goto out;
- }
-
- while (left > 0) {
- num = MIN(left, 65536);
- result = write(fd, buf, num);
- if (result < 0) {
- result = -errno;
- error_setg_errno(errp, -result,
- "Could not write zeros for preallocation");
- goto out;
- }
- left -= result;
- }
- if (result >= 0) {
- result = fsync(fd);
- if (result < 0) {
- result = -errno;
- error_setg_errno(errp, -result,
- "Could not flush file to disk");
- goto out;
- }
- }
- goto out;
- }
- case PREALLOC_MODE_OFF:
- if (ftruncate(fd, offset) != 0) {
- result = -errno;
- error_setg_errno(errp, -result, "Could not resize file");
- }
- return result;
- default:
- result = -ENOTSUP;
- error_setg(errp, "Unsupported preallocation mode: %s",
- PreallocMode_str(prealloc));
- return result;
- }
+ RawPosixAIOData *acb = g_new(RawPosixAIOData, 1);
+ ThreadPool *pool;
-out:
- if (result < 0) {
- if (ftruncate(fd, current_length) < 0) {
- error_report("Failed to restore old file length: %s",
- strerror(errno));
- }
- }
+ *acb = (RawPosixAIOData) {
+ .bs = bs,
+ .aio_fildes = fd,
+ .aio_type = QEMU_AIO_TRUNCATE,
+ .aio_offset = offset,
+ .prealloc = prealloc,
+ .errp = errp,
+ };
- g_free(buf);
- return result;
+ /* @bs can be NULL, bdrv_get_aio_context() returns the main context then */
+ pool = aio_get_thread_pool(bdrv_get_aio_context(bs));
+ return thread_pool_submit_co(pool, aio_worker, acb);
}
static int coroutine_fn raw_co_truncate(BlockDriverState *bs, int64_t offset,
@@ -1892,7 +1927,7 @@ static int coroutine_fn raw_co_truncate(BlockDriverState *bs, int64_t offset,
}
if (S_ISREG(st.st_mode)) {
- return raw_regular_truncate(s->fd, offset, prealloc, errp);
+ return raw_regular_truncate(bs, s->fd, offset, prealloc, errp);
}
if (prealloc != PREALLOC_MODE_OFF) {
@@ -2094,7 +2129,8 @@ static int64_t raw_get_allocated_file_size(BlockDriverState *bs)
return (int64_t)st.st_blocks * 512;
}
-static int raw_co_create(BlockdevCreateOptions *options, Error **errp)
+static int coroutine_fn
+raw_co_create(BlockdevCreateOptions *options, Error **errp)
{
BlockdevCreateOptionsFile *file_opts;
int fd;
@@ -2146,7 +2182,7 @@ static int raw_co_create(BlockdevCreateOptions *options, Error **errp)
}
/* Clear the file by truncating it to 0 */
- result = raw_regular_truncate(fd, 0, PREALLOC_MODE_OFF, errp);
+ result = raw_regular_truncate(NULL, fd, 0, PREALLOC_MODE_OFF, errp);
if (result < 0) {
goto out_close;
}
@@ -2168,8 +2204,8 @@ static int raw_co_create(BlockdevCreateOptions *options, Error **errp)
/* Resize and potentially preallocate the file to the desired
* final size */
- result = raw_regular_truncate(fd, file_opts->size, file_opts->preallocation,
- errp);
+ result = raw_regular_truncate(NULL, fd, file_opts->size,
+ file_opts->preallocation, errp);
if (result < 0) {
goto out_close;
}
--
2.13.6
next prev parent reply other threads:[~2018-06-29 14:10 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-06-29 14:09 [Qemu-devel] [PULL 00/29] Block layer patches Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 01/29] block-qdict: Pacify Coverity after commit f1b34a248e9 Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 02/29] block/crypto: Pacify Coverity after commit f853465aacb Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 03/29] qapi/job: The next release will be 3.0 Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 04/29] usb-storage: Add rerror/werror properties Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 05/29] hw/block/nvme: add optional parameter num_queues for nvme device Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 06/29] qcow2: Fix qcow2_truncate() error return value Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 07/29] block: Convert .bdrv_truncate callback to coroutine_fn Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 08/29] qcow2: Remove coroutine trampoline for preallocate_co() Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 09/29] block: Move bdrv_truncate() implementation to io.c Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 10/29] block: Use tracked request for truncate Kevin Wolf
2018-06-29 14:09 ` Kevin Wolf [this message]
2018-06-29 14:09 ` [Qemu-devel] [PULL 12/29] qcow2: Remove dead check on !ret Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 13/29] block: Move request tracking to children in copy offloading Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 14/29] block/crypto: Simplify block_crypto_{open, create}_opts_init() Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 15/29] qemu-iotests: Update 026.out.nocache reference output Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 16/29] qcow2: Free allocated clusters on write error Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 17/29] qemu-iotests: Test qcow2 not leaking " Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 18/29] file-posix: Implement co versions of discard/flush Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 19/29] qcow2: Fix src_offset in copy offloading Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 20/29] iscsi: Don't blindly use designator length in response for memcpy Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 21/29] file-posix: Fix EINTR handling Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 22/29] parallels: Switch to byte-based calls Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 23/29] qcow: Switch get_cluster_offset to be byte-based Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 24/29] qcow: Switch qcow_co_readv to byte-based calls Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 25/29] qcow: Switch qcow_co_writev " Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 26/29] qcow: Switch to a byte-based driver Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 27/29] replication: Switch to byte-based calls Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 28/29] vhdx: " Kevin Wolf
2018-06-29 14:09 ` [Qemu-devel] [PULL 29/29] block: Remove unused sector-based vectored I/O Kevin Wolf
2018-06-30 10:54 ` [Qemu-devel] [PULL 00/29] Block layer patches Peter Maydell
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180629140959.6690-12-kwolf@redhat.com \
--to=kwolf@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=qemu-block@nongnu.org \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.