On Fri, Aug 23, 2019 at 03:49:24PM -0400, Dmitry Fomichev wrote:
> +uint8_t bdrv_is_zoned(BlockDriverState *bs)
> +{
> + /*
> + * Host Aware zone devices are supposed to be able to work
> + * just like regular block devices. Thus, we only consider
> + * Host Managed devices to be zoned here.
> + */
> + return bdrv_get_zoned_model(bs) == BLK_ZONED_MODEL_HM;
This function doesn't say whether bs is a zoned device in general, it
says whether it's Host Managed. Please rename to
bdrv_is_host_managed_zoned(), bdrv_is_zoned_hm(), or similar.
> +}
> +
> bool bdrv_is_sg(BlockDriverState *bs)
> {
> return bs->sg;
> diff --git a/include/block/block.h b/include/block/block.h
> index 124ad40809..238c0f5ed7 100644
> --- a/include/block/block.h
> +++ b/include/block/block.h
> @@ -271,18 +271,35 @@ enum {
> */
> BLK_PERM_GRAPH_MOD = 0x10,
>
> + /** This permission is required to open zoned block devices. */
> + BLK_PERM_SUPPORT_ZONED = 0x20,
Maybe BLK_PERM_SUPPORT_ZONED_HM is clearer? Let's keep the distinction
between general zoned devices and host-managed zoned devices clear,
otherwise the code will get confusing if we ever want to treat
host-aware specially.
> +
> BLK_PERM_ALL = 0x1f,
>
> DEFAULT_PERM_PASSTHROUGH = BLK_PERM_CONSISTENT_READ
> | BLK_PERM_WRITE
> | BLK_PERM_WRITE_UNCHANGED
> - | BLK_PERM_RESIZE,
> + | BLK_PERM_RESIZE
> + | BLK_PERM_SUPPORT_ZONED,
>
> DEFAULT_PERM_UNCHANGED = BLK_PERM_ALL & ~DEFAULT_PERM_PASSTHROUGH,
> };
>
> char *bdrv_perm_names(uint64_t perm);
>
> +/*
> + * Known zoned device models.
> + *
> + * TODO For a Linux host, it could be preferrable to include
> + * /usr/include/linux/blkzoned.h instead of defining ZBD-specific
> + * values here.
It depends. If it's necessary to convert back to the Linux enum often
then I agree. Otherwise the code is cleaner a QEMU enum is defined here
and we can avoid #ifdef __linux__.
> + */
> +enum blk_zoned_model {
> + BLK_ZONED_MODEL_NONE, /* Regular block device */
> + BLK_ZONED_MODEL_HA, /* Host-aware zoned block device */
> + BLK_ZONED_MODEL_HM, /* Host-managed zoned block device */
> +};
Please use the same typedef enum approach as the rest of block.h:
typedef enum {
BDRV_ZONED_MODEL_NONE, /* Regular block device */
BDRV_ZONED_MODEL_HA, /* Host-aware zoned block device */
BDRV_ZONED_MODEL_HM, /* Host-managed zoned block device */
} BdrvZonedModel;
> +
> /* disk I/O throttling */
> void bdrv_init(void);
> void bdrv_init_with_whitelist(void);
> @@ -359,6 +376,8 @@ int64_t bdrv_get_allocated_file_size(BlockDriverState
> *bs);
> BlockMeasureInfo *bdrv_measure(BlockDriver *drv, QemuOpts *opts,
> BlockDriverState *in_bs, Error **errp);
> void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr);
> +uint8_t bdrv_get_zoned_model(BlockDriverState *bs);
Please use the enum:
BdrvZonedModel bdrv_get_zoned_model(BlockDriverState *bs)
> +uint8_t bdrv_is_zoned(BlockDriverState *bs);
Please use bool instead of uint8_t.
> void bdrv_refresh_limits(BlockDriverState *bs, Error **errp);
> int bdrv_commit(BlockDriverState *bs);
> int bdrv_change_backing_file(BlockDriverState *bs,
> diff --git a/include/block/block_int.h b/include/block/block_int.h
> index ceec8c2f56..91496e8149 100644
> --- a/include/block/block_int.h
> +++ b/include/block/block_int.h
> @@ -415,6 +415,7 @@ struct BlockDriver {
> bool (*bdrv_debug_is_suspended)(BlockDriverState *bs, const char *tag);
>
> void (*bdrv_refresh_limits)(BlockDriverState *bs, Error **errp);
> + void (*bdrv_get_zoned_info)(BlockDriverState *bs);
"get" is strange since this function returns void. What is it supposed
to do, refresh just the BlockLimits::zoned_model field? Should this be
called bdrv_refresh_zoned_model() instead?
>
> /*
> * Returns 1 if newly created images are guaranteed to contain only
> @@ -620,6 +621,9 @@ typedef struct BlockLimits {
>
> /* maximum number of iovec elements */
> int max_iov;
> +
> + /* Zoned device model. Zero value indicates a regular block device */
> + uint8_t zoned_model;
Please use the enum.
signature.asc
Description: PGP signature
