qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH v4 11/20] qcow2: Switch to .bdrv_co_block_status


From: Eric Blake
Subject: Re: [Qemu-devel] [PATCH v4 11/20] qcow2: Switch to .bdrv_co_block_status()
Date: Thu, 30 Nov 2017 16:38:05 -0600
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.4.0

On 11/30/2017 03:51 AM, Vladimir Sementsov-Ogievskiy wrote:
12.10.2017 21:59, Eric Blake wrote:
We are gradually moving away from sector-based interfaces, towards
byte-based.  Update the qcow2 driver accordingly.

For now, we are ignoring the 'want_zero' hint.  However, it should
be relatively straightforward to honor the hint as a way to return
larger *pnum values when we have consecutive clusters with the same
data/zero status but which differ only in having non-consecutive
mappings.

useful thing (for example, to get several compressed clusters in one chunk,
but do not include following zero clusters). but should not the hint be called want_mapping for it?
or may be we will move to "int hint_flags", which will include status flags
we a interested in?

Right now, there are only two public interfaces: bdrv_is_allocated (want_zero is false), and bdrv_block_status (want_zero is true). You may have a point that there could be further optimizations possible if the caller has more control over what flags it is interested in, but that should be a patch series for a later day; I'm already cramming enough into this series, and it's already stretched out (first patches went into 2.10), so I want it done.



Signed-off-by: Eric Blake <address@hidden>

Reviewed-by: Vladimir Sementsov-Ogievskiy <address@hidden>


-static int64_t coroutine_fn qcow2_co_get_block_status(BlockDriverState *bs, -        int64_t sector_num, int nb_sectors, int *pnum, BlockDriverState **file)
+static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
+                                              bool want_zero,
+                                              int64_t offset, int64_t count,

'count' is used instead of 'bytes' because of qcow2_get_cluster_offset which wants int*..

Yes,...


+                                              int64_t *pnum, int64_t *map,
+                                              BlockDriverState **file)
  {
      BDRVQcow2State *s = bs->opaque;
      uint64_t cluster_offset;
      int index_in_cluster, ret;
      unsigned int bytes;

...because of that.

-    int64_t status = 0;
+    int status = 0;

-    bytes = MIN(INT_MAX, nb_sectors * BDRV_SECTOR_SIZE);
+    bytes = MIN(INT_MAX, count);
      qemu_co_mutex_lock(&s->lock);
-    ret = qcow2_get_cluster_offset(bs, sector_num << BDRV_SECTOR_BITS, &bytes,
-                                   &cluster_offset);
+    ret = qcow2_get_cluster_offset(bs, offset, &bytes, &cluster_offset);

--
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org



reply via email to

[Prev in Thread] Current Thread [Next in Thread]