On 01/26/2015 09:02 AM, Max Reitz wrote:
BlockAcctStats contains statistics about the data transferred from and
to the device; wr_highest_offset does not fit in with the rest.
Furthermore, those statistics are supposed to be specific for a certain
device and not necessarily for a BDS (see the comment above
bdrv_get_stats()); on the other hand, wr_highest_offset may be a rather
important information to know for each BDS. When BlockAcctStats is
finally removed from the BDS, we will want to keep wr_highest_offset in
the BDS.
Yes, I recently did work in libvirt to expose wr_highest_offset of
backing images during block commit (qemu still isn't populating it on
images opened only for read, but the point remains that it is a
statistic tied to the BDS, not the BB).
On the other hand, even the other statistics might make sense on both
BDS and BB level (at the BB level, how many bytes has the guest
read/written; at the BDS level, how many bytes were serviced by the
active layer vs. delegated to a backing layer).
I'm not sure if we are set up for that fine of a level of reporting yet,
but we shouldn't make it hard to implement later. But for now, I agree
with separating the definite BDS-only stat, leaving the rest of the
struct usable for either BDS or BB.
Signed-off-by: Max Reitz <address@hidden>
---
block.c | 4 +++-
block/accounting.c | 9 ---------
block/qapi.c | 4 ++--
include/block/accounting.h | 3 ---
include/block/block_int.h | 3 +++
5 files changed, 8 insertions(+), 15 deletions(-)
diff --git a/block.c b/block.c
+++ b/include/block/block_int.h
@@ -366,6 +366,9 @@ struct BlockDriverState {
/* I/O stats (display with "info blockstats"). */
BlockAcctStats stats;
+ /* Highest sector index written to */
+ uint64_t wr_highest_sector;
Umm, now would be a great time to track this in bytes instead of
sectors, if that is not too difficult to do.