bool page_cache_inconsistent:1;
bool has_fallocate;
bool needs_alignment;
+ bool drop_cache;
bool check_cache_dropped;
PRManager *pr_mgr;
typedef struct BDRVRawReopenState {
int fd;
int open_flags;
+ bool drop_cache;
bool check_cache_dropped;
} BDRVRawReopenState;
.type = QEMU_OPT_STRING,
.help = "id of persistent reservation manager object (default: none)",
},
+#if defined(__linux__)
+ {
+ .name = "drop-cache",
+ .type = QEMU_OPT_BOOL,
+ .help = "invalidate page cache during live migration (default: on)",
+ },
+#endif
{
.name = "x-check-cache-dropped",
.type = QEMU_OPT_BOOL,
}
}
+ s->drop_cache = qemu_opt_get_bool(opts, "drop-cache", true);
s->check_cache_dropped = qemu_opt_get_bool(opts, "x-check-cache-dropped",
false);
}
#endif
- bs->supported_zero_flags = BDRV_REQ_MAY_UNMAP;
+ bs->supported_zero_flags = BDRV_REQ_MAY_UNMAP | BDRV_REQ_NO_FALLBACK;
ret = 0;
fail:
if (filename && (bdrv_flags & BDRV_O_TEMPORARY)) {
switch (op) {
case RAW_PL_PREPARE:
+ if ((s->perm | new_perm) == s->perm &&
+ (s->shared_perm & new_shared) == s->shared_perm)
+ {
+ /*
+ * We are going to unlock bytes, it should not fail. If it fail due
+ * to some fs-dependent permission-unrelated reasons (which occurs
+ * sometimes on NFS and leads to abort in bdrv_replace_child) we
+ * can't prevent such errors by any check here. And we ignore them
+ * anyway in ABORT and COMMIT.
+ */
+ return 0;
+ }
ret = raw_apply_lock_bytes(s, s->fd, s->perm | new_perm,
~s->shared_perm | ~new_shared,
false, errp);
goto out;
}
+ rs->drop_cache = qemu_opt_get_bool_del(opts, "drop-cache", true);
rs->check_cache_dropped =
qemu_opt_get_bool_del(opts, "x-check-cache-dropped", false);
BDRVRawReopenState *rs = state->opaque;
BDRVRawState *s = state->bs->opaque;
+ s->drop_cache = rs->drop_cache;
s->check_cache_dropped = rs->check_cache_dropped;
s->open_flags = rs->open_flags;
#ifdef CONFIG_XFS
static int xfs_write_zeroes(BDRVRawState *s, int64_t offset, uint64_t bytes)
{
+ int64_t len;
struct xfs_flock64 fl;
int err;
+ len = lseek(s->fd, 0, SEEK_END);
+ if (len < 0) {
+ return -errno;
+ }
+
+ if (offset + bytes > len) {
+ /* XFS_IOC_ZERO_RANGE does not increase the file length */
+ if (ftruncate(s->fd, offset + bytes) < 0) {
+ return -errno;
+ }
+ }
+
memset(&fl, 0, sizeof(fl));
fl.l_whence = SEEK_SET;
fl.l_start = offset;
}
#ifdef BLKZEROOUT
- do {
- uint64_t range[2] = { aiocb->aio_offset, aiocb->aio_nbytes };
- if (ioctl(aiocb->aio_fildes, BLKZEROOUT, range) == 0) {
- return 0;
- }
- } while (errno == EINTR);
+ /* The BLKZEROOUT implementation in the kernel doesn't set
+ * BLKDEV_ZERO_NOFALLBACK, so we can't call this if we have to avoid slow
+ * fallbacks. */
+ if (!(aiocb->aio_type & QEMU_AIO_NO_FALLBACK)) {
+ do {
+ uint64_t range[2] = { aiocb->aio_offset, aiocb->aio_nbytes };
+ if (ioctl(aiocb->aio_fildes, BLKZEROOUT, range) == 0) {
+ return 0;
+ }
+ } while (errno == EINTR);
- ret = translate_err(-errno);
+ ret = translate_err(-errno);
+ }
#endif
if (ret == -ENOTSUP) {
off_t data = 0, hole = 0;
int ret;
+ assert(QEMU_IS_ALIGNED(offset | bytes, bs->bl.request_alignment));
+
ret = fd_open(bs);
if (ret < 0) {
return ret;
/* On a data extent, compute bytes to the end of the extent,
* possibly including a partial sector at EOF. */
*pnum = MIN(bytes, hole - offset);
+
+ /*
+ * We are not allowed to return partial sectors, though, so
+ * round up if necessary.
+ */
+ if (!QEMU_IS_ALIGNED(*pnum, bs->bl.request_alignment)) {
+ int64_t file_length = raw_getlength(bs);
+ if (file_length > 0) {
+ /* Ignore errors, this is just a safeguard */
+ assert(hole == file_length);
+ }
+ *pnum = ROUND_UP(*pnum, bs->bl.request_alignment);
+ }
+
ret = BDRV_BLOCK_DATA;
} else {
/* On a hole, compute bytes to the beginning of the next extent. */
return;
}
+ if (!s->drop_cache) {
+ return;
+ }
+
if (s->open_flags & O_DIRECT) {
return; /* No host kernel page cache */
}
if (blkdev) {
acb.aio_type |= QEMU_AIO_BLKDEV;
}
+ if (flags & BDRV_REQ_NO_FALLBACK) {
+ acb.aio_type |= QEMU_AIO_NO_FALLBACK;
+ }
if (flags & BDRV_REQ_MAY_UNMAP) {
acb.aio_type |= QEMU_AIO_DISCARD;