summaryrefslogtreecommitdiff
path: root/block/io.c
diff options
context:
space:
mode:
Diffstat (limited to 'block/io.c')
-rw-r--r--block/io.c180
1 files changed, 116 insertions, 64 deletions
diff --git a/block/io.c b/block/io.c
index e00fb5d690..a7dbf85b19 100644
--- a/block/io.c
+++ b/block/io.c
@@ -22,11 +22,14 @@
* THE SOFTWARE.
*/
+#include "qemu/osdep.h"
#include "trace.h"
#include "sysemu/block-backend.h"
#include "block/blockjob.h"
#include "block/block_int.h"
#include "block/throttle-groups.h"
+#include "qemu/cutils.h"
+#include "qapi/error.h"
#include "qemu/error-report.h"
#define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */
@@ -43,12 +46,6 @@ static int coroutine_fn bdrv_co_readv_em(BlockDriverState *bs,
static int coroutine_fn bdrv_co_writev_em(BlockDriverState *bs,
int64_t sector_num, int nb_sectors,
QEMUIOVector *iov);
-static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
- int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
- BdrvRequestFlags flags);
-static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
- int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
- BdrvRequestFlags flags);
static BlockAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
int64_t sector_num,
QEMUIOVector *qiov,
@@ -166,9 +163,13 @@ void bdrv_refresh_limits(BlockDriverState *bs, Error **errp)
bs->bl.max_transfer_length = bs->file->bs->bl.max_transfer_length;
bs->bl.min_mem_alignment = bs->file->bs->bl.min_mem_alignment;
bs->bl.opt_mem_alignment = bs->file->bs->bl.opt_mem_alignment;
+ bs->bl.max_iov = bs->file->bs->bl.max_iov;
} else {
bs->bl.min_mem_alignment = 512;
bs->bl.opt_mem_alignment = getpagesize();
+
+ /* Safe default since most protocols use readv()/writev()/etc */
+ bs->bl.max_iov = IOV_MAX;
}
if (bs->backing) {
@@ -189,6 +190,9 @@ void bdrv_refresh_limits(BlockDriverState *bs, Error **errp)
bs->bl.min_mem_alignment =
MAX(bs->bl.min_mem_alignment,
bs->backing->bs->bl.min_mem_alignment);
+ bs->bl.max_iov =
+ MIN(bs->bl.max_iov,
+ bs->backing->bs->bl.max_iov);
}
/* Then let the driver override it */
@@ -249,6 +253,47 @@ static void bdrv_drain_recurse(BlockDriverState *bs)
}
}
+typedef struct {
+ Coroutine *co;
+ BlockDriverState *bs;
+ QEMUBH *bh;
+ bool done;
+} BdrvCoDrainData;
+
+static void bdrv_co_drain_bh_cb(void *opaque)
+{
+ BdrvCoDrainData *data = opaque;
+ Coroutine *co = data->co;
+
+ qemu_bh_delete(data->bh);
+ bdrv_drain(data->bs);
+ data->done = true;
+ qemu_coroutine_enter(co, NULL);
+}
+
+void coroutine_fn bdrv_co_drain(BlockDriverState *bs)
+{
+ BdrvCoDrainData data;
+
+ /* Calling bdrv_drain() from a BH ensures the current coroutine yields and
+ * other coroutines run if they were queued from
+ * qemu_co_queue_run_restart(). */
+
+ assert(qemu_in_coroutine());
+ data = (BdrvCoDrainData) {
+ .co = qemu_coroutine_self(),
+ .bs = bs,
+ .done = false,
+ .bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_co_drain_bh_cb, &data),
+ };
+ qemu_bh_schedule(data.bh);
+
+ qemu_coroutine_yield();
+ /* If we are resumed from some other event (such as an aio completion or a
+ * timer callback), it is a bug in the caller that should be fixed. */
+ assert(data.done);
+}
+
/*
* Wait for pending requests to complete on a single BlockDriverState subtree,
* and suspend block driver's internal I/O until next request arrives.
@@ -265,6 +310,10 @@ void bdrv_drain(BlockDriverState *bs)
bool busy = true;
bdrv_drain_recurse(bs);
+ if (qemu_in_coroutine()) {
+ bdrv_co_drain(bs);
+ return;
+ }
while (busy) {
/* Keep iterating */
bdrv_flush_io_queue(bs);
@@ -293,6 +342,7 @@ void bdrv_drain_all(void)
if (bs->job) {
block_job_pause(bs->job);
}
+ bdrv_drain_recurse(bs);
aio_context_release(aio_context);
if (!g_slist_find(aio_ctxs, aio_context)) {
@@ -612,20 +662,6 @@ int bdrv_read(BlockDriverState *bs, int64_t sector_num,
return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false, 0);
}
-/* Just like bdrv_read(), but with I/O throttling temporarily disabled */
-int bdrv_read_unthrottled(BlockDriverState *bs, int64_t sector_num,
- uint8_t *buf, int nb_sectors)
-{
- bool enabled;
- int ret;
-
- enabled = bs->io_limits_enabled;
- bs->io_limits_enabled = false;
- ret = bdrv_read(bs, sector_num, buf, nb_sectors);
- bs->io_limits_enabled = enabled;
- return ret;
-}
-
/* Return < 0 if error. Important errors are:
-EIO generic I/O error (may happen for all errors)
-ENOMEDIUM No media inserted.
@@ -656,6 +692,7 @@ int bdrv_write_zeroes(BlockDriverState *bs, int64_t sector_num,
int bdrv_make_zero(BlockDriverState *bs, BdrvRequestFlags flags)
{
int64_t target_sectors, ret, nb_sectors, sector_num = 0;
+ BlockDriverState *file;
int n;
target_sectors = bdrv_nb_sectors(bs);
@@ -668,7 +705,7 @@ int bdrv_make_zero(BlockDriverState *bs, BdrvRequestFlags flags)
if (nb_sectors <= 0) {
return 0;
}
- ret = bdrv_get_block_status(bs, sector_num, nb_sectors, &n);
+ ret = bdrv_get_block_status(bs, sector_num, nb_sectors, &n, &file);
if (ret < 0) {
error_report("error getting block status at sector %" PRId64 ": %s",
sector_num, strerror(-ret));
@@ -755,9 +792,9 @@ int bdrv_pwrite_sync(BlockDriverState *bs, int64_t offset,
return ret;
}
- /* No flush needed for cache modes that already do it */
- if (bs->enable_write_cache) {
- bdrv_flush(bs);
+ ret = bdrv_flush(bs);
+ if (ret < 0) {
+ return ret;
}
return 0;
@@ -852,6 +889,7 @@ static int coroutine_fn bdrv_aligned_preadv(BlockDriverState *bs,
assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
assert(!qiov || bytes == qiov->size);
+ assert((bs->open_flags & BDRV_O_NO_IO) == 0);
/* Handle Copy on Read and associated serialisation */
if (flags & BDRV_REQ_COPY_ON_READ) {
@@ -929,7 +967,7 @@ out:
/*
* Handle a read request in coroutine context
*/
-static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
+int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
BdrvRequestFlags flags)
{
@@ -1138,6 +1176,7 @@ static int coroutine_fn bdrv_aligned_pwritev(BlockDriverState *bs,
assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
assert(!qiov || bytes == qiov->size);
+ assert((bs->open_flags & BDRV_O_NO_IO) == 0);
waited = wait_serialising_requests(req);
assert(!waited || !req->serialising);
@@ -1160,13 +1199,20 @@ static int coroutine_fn bdrv_aligned_pwritev(BlockDriverState *bs,
} else if (flags & BDRV_REQ_ZERO_WRITE) {
bdrv_debug_event(bs, BLKDBG_PWRITEV_ZERO);
ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors, flags);
+ } else if (drv->bdrv_co_writev_flags) {
+ bdrv_debug_event(bs, BLKDBG_PWRITEV);
+ ret = drv->bdrv_co_writev_flags(bs, sector_num, nb_sectors, qiov,
+ flags);
} else {
+ assert(drv->supported_write_flags == 0);
bdrv_debug_event(bs, BLKDBG_PWRITEV);
ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
}
bdrv_debug_event(bs, BLKDBG_PWRITEV_DONE);
- if (ret == 0 && !bs->enable_write_cache) {
+ if (ret == 0 && (flags & BDRV_REQ_FUA) &&
+ !(drv->supported_write_flags & BDRV_REQ_FUA))
+ {
ret = bdrv_co_flush(bs);
}
@@ -1274,7 +1320,7 @@ fail:
/*
* Handle a write request in coroutine context
*/
-static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
+int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
BdrvRequestFlags flags)
{
@@ -1293,6 +1339,7 @@ static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
if (bs->read_only) {
return -EPERM;
}
+ assert(!(bs->open_flags & BDRV_O_INACTIVE));
ret = bdrv_check_byte_request(bs, offset, bytes);
if (ret < 0) {
@@ -1434,29 +1481,10 @@ int coroutine_fn bdrv_co_write_zeroes(BlockDriverState *bs,
BDRV_REQ_ZERO_WRITE | flags);
}
-int bdrv_flush_all(void)
-{
- BlockDriverState *bs = NULL;
- int result = 0;
-
- while ((bs = bdrv_next(bs))) {
- AioContext *aio_context = bdrv_get_aio_context(bs);
- int ret;
-
- aio_context_acquire(aio_context);
- ret = bdrv_flush(bs);
- if (ret < 0 && !result) {
- result = ret;
- }
- aio_context_release(aio_context);
- }
-
- return result;
-}
-
typedef struct BdrvCoGetBlockStatusData {
BlockDriverState *bs;
BlockDriverState *base;
+ BlockDriverState **file;
int64_t sector_num;
int nb_sectors;
int *pnum;
@@ -1478,10 +1506,14 @@ typedef struct BdrvCoGetBlockStatusData {
*
* 'nb_sectors' is the max value 'pnum' should be set to. If nb_sectors goes
* beyond the end of the disk image it will be clamped.
+ *
+ * If returned value is positive and BDRV_BLOCK_OFFSET_VALID bit is set, 'file'
+ * points to the BDS which the sector range is allocated in.
*/
static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
int64_t sector_num,
- int nb_sectors, int *pnum)
+ int nb_sectors, int *pnum,
+ BlockDriverState **file)
{
int64_t total_sectors;
int64_t n;
@@ -1511,7 +1543,9 @@ static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
return ret;
}
- ret = bs->drv->bdrv_co_get_block_status(bs, sector_num, nb_sectors, pnum);
+ *file = NULL;
+ ret = bs->drv->bdrv_co_get_block_status(bs, sector_num, nb_sectors, pnum,
+ file);
if (ret < 0) {
*pnum = 0;
return ret;
@@ -1520,7 +1554,7 @@ static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
if (ret & BDRV_BLOCK_RAW) {
assert(ret & BDRV_BLOCK_OFFSET_VALID);
return bdrv_get_block_status(bs->file->bs, ret >> BDRV_SECTOR_BITS,
- *pnum, pnum);
+ *pnum, pnum, file);
}
if (ret & (BDRV_BLOCK_DATA | BDRV_BLOCK_ZERO)) {
@@ -1537,13 +1571,14 @@ static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
}
}
- if (bs->file &&
+ if (*file && *file != bs &&
(ret & BDRV_BLOCK_DATA) && !(ret & BDRV_BLOCK_ZERO) &&
(ret & BDRV_BLOCK_OFFSET_VALID)) {
+ BlockDriverState *file2;
int file_pnum;
- ret2 = bdrv_co_get_block_status(bs->file->bs, ret >> BDRV_SECTOR_BITS,
- *pnum, &file_pnum);
+ ret2 = bdrv_co_get_block_status(*file, ret >> BDRV_SECTOR_BITS,
+ *pnum, &file_pnum, &file2);
if (ret2 >= 0) {
/* Ignore errors. This is just providing extra information, it
* is useful but not necessary.
@@ -1568,14 +1603,15 @@ static int64_t coroutine_fn bdrv_co_get_block_status_above(BlockDriverState *bs,
BlockDriverState *base,
int64_t sector_num,
int nb_sectors,
- int *pnum)
+ int *pnum,
+ BlockDriverState **file)
{
BlockDriverState *p;
int64_t ret = 0;
assert(bs != base);
for (p = bs; p != base; p = backing_bs(p)) {
- ret = bdrv_co_get_block_status(p, sector_num, nb_sectors, pnum);
+ ret = bdrv_co_get_block_status(p, sector_num, nb_sectors, pnum, file);
if (ret < 0 || ret & BDRV_BLOCK_ALLOCATED) {
break;
}
@@ -1594,7 +1630,8 @@ static void coroutine_fn bdrv_get_block_status_above_co_entry(void *opaque)
data->ret = bdrv_co_get_block_status_above(data->bs, data->base,
data->sector_num,
data->nb_sectors,
- data->pnum);
+ data->pnum,
+ data->file);
data->done = true;
}
@@ -1606,12 +1643,14 @@ static void coroutine_fn bdrv_get_block_status_above_co_entry(void *opaque)
int64_t bdrv_get_block_status_above(BlockDriverState *bs,
BlockDriverState *base,
int64_t sector_num,
- int nb_sectors, int *pnum)
+ int nb_sectors, int *pnum,
+ BlockDriverState **file)
{
Coroutine *co;
BdrvCoGetBlockStatusData data = {
.bs = bs,
.base = base,
+ .file = file,
.sector_num = sector_num,
.nb_sectors = nb_sectors,
.pnum = pnum,
@@ -1635,16 +1674,19 @@ int64_t bdrv_get_block_status_above(BlockDriverState *bs,
int64_t bdrv_get_block_status(BlockDriverState *bs,
int64_t sector_num,
- int nb_sectors, int *pnum)
+ int nb_sectors, int *pnum,
+ BlockDriverState **file)
{
return bdrv_get_block_status_above(bs, backing_bs(bs),
- sector_num, nb_sectors, pnum);
+ sector_num, nb_sectors, pnum, file);
}
int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num,
int nb_sectors, int *pnum)
{
- int64_t ret = bdrv_get_block_status(bs, sector_num, nb_sectors, pnum);
+ BlockDriverState *file;
+ int64_t ret = bdrv_get_block_status(bs, sector_num, nb_sectors, pnum,
+ &file);
if (ret < 0) {
return ret;
}
@@ -1882,7 +1924,8 @@ static int multiwrite_merge(BlockDriverState *bs, BlockRequest *reqs,
merge = 1;
}
- if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
+ if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 >
+ bs->bl.max_iov) {
merge = 0;
}
@@ -2342,6 +2385,13 @@ int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
}
tracked_request_begin(&req, bs, 0, 0, BDRV_TRACKED_FLUSH);
+
+ /* Write back all layers by calling one driver function */
+ if (bs->drv->bdrv_co_flush) {
+ ret = bs->drv->bdrv_co_flush(bs);
+ goto out;
+ }
+
/* Write back cached data to the OS even with cache=unsafe */
BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_OS);
if (bs->drv->bdrv_co_flush_to_os) {
@@ -2453,6 +2503,7 @@ int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
} else if (bs->read_only) {
return -EPERM;
}
+ assert(!(bs->open_flags & BDRV_O_INACTIVE));
/* Do nothing if disabled. */
if (!(bs->open_flags & BDRV_O_UNMAP)) {
@@ -2614,10 +2665,11 @@ int bdrv_ioctl(BlockDriverState *bs, unsigned long int req, void *buf)
bdrv_co_ioctl_entry(&data);
} else {
Coroutine *co = qemu_coroutine_create(bdrv_co_ioctl_entry);
+
qemu_coroutine_enter(co, &data);
- }
- while (data.ret == -EINPROGRESS) {
- aio_poll(bdrv_get_aio_context(bs), true);
+ while (data.ret == -EINPROGRESS) {
+ aio_poll(bdrv_get_aio_context(bs), true);
+ }
}
return data.ret;
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-23 15:01:09 +0000