[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 3/8] block: Support to keep track of I/O status
From: |
Luiz Capitulino |
Subject: |
Re: [Qemu-devel] [PATCH 3/8] block: Support to keep track of I/O status |
Date: |
Tue, 12 Jul 2011 11:38:06 -0300 |
On Tue, 12 Jul 2011 11:12:04 +0200
Markus Armbruster <address@hidden> wrote:
> Kevin Wolf <address@hidden> writes:
>
> > Am 12.07.2011 09:45, schrieb Markus Armbruster:
> >> Luiz Capitulino <address@hidden> writes:
> >>
> >>> This commit adds support to the BlockDriverState type to keep track
> >>> of the last I/O status. That is, at every I/O operation we update
> >>> a status field in the BlockDriverState instance. Valid statuses are:
> >>> OK, FAILED and ENOSPC.
> >>>
> >>> ENOSPC is distinguished from FAILED because an management application
> >>> can use it to implement thin-provisioning.
> >>>
> >>> This feature has to be explicit enabled by buses/devices supporting it.
> >>
> >> buses?
I think I should have called it 'interface'.
> >>
> >>>
> >>> Signed-off-by: Luiz Capitulino <address@hidden>
> >>> ---
> >>> block.c | 18 ++++++++++++++++++
> >>> block.h | 7 +++++++
> >>> block_int.h | 2 ++
> >>> 3 files changed, 27 insertions(+), 0 deletions(-)
> >>>
> >>> diff --git a/block.c b/block.c
> >>> index 24a25d5..cc0a34e 100644
> >>> --- a/block.c
> >>> +++ b/block.c
> >>> @@ -195,6 +195,7 @@ BlockDriverState *bdrv_new(const char *device_name)
> >>> if (device_name[0] != '\0') {
> >>> QTAILQ_INSERT_TAIL(&bdrv_states, bs, list);
> >>> }
> >>> + bs->iostatus_enabled = false;
> >>> return bs;
> >>> }
> >>>
> >>> @@ -2876,6 +2877,23 @@ int bdrv_in_use(BlockDriverState *bs)
> >>> return bs->in_use;
> >>> }
> >>>
> >>> +void bdrv_enable_iostatus(BlockDriverState *bs)
> >>> +{
> >>> + bs->iostatus_enabled = true;
> >>> +}
> >>> +
> >>> +void bdrv_iostatus_update(BlockDriverState *bs, int error)
> >>> +{
> >>> + error = abs(error);
> >>> +
> >>> + if (!error) {
> >>> + bs->iostatus = BDRV_IOS_OK;
> >>> + } else {
> >>> + bs->iostatus = (error == ENOSPC) ? BDRV_IOS_ENOSPC :
> >>> + BDRV_IOS_FAILED;
> >>> + }
> >>> +}
> >>> +
> >>> int bdrv_img_create(const char *filename, const char *fmt,
> >>> const char *base_filename, const char *base_fmt,
> >>> char *options, uint64_t img_size, int flags)
> >>> diff --git a/block.h b/block.h
> >>> index 859d1d9..0dca1bb 100644
> >>> --- a/block.h
> >>> +++ b/block.h
> >>> @@ -50,6 +50,13 @@ typedef enum {
> >>> BDRV_ACTION_REPORT, BDRV_ACTION_IGNORE, BDRV_ACTION_STOP
> >>> } BlockMonEventAction;
> >>>
> >>> +typedef enum {
> >>> + BDRV_IOS_OK, BDRV_IOS_FAILED, BDRV_IOS_ENOSPC
> >>> +} BlockIOStatus;
> >>> +
> >>> +void bdrv_iostatus_update(BlockDriverState *bs, int error);
> >>> +void bdrv_enable_iostatus(BlockDriverState *bs);
> >>> +void bdrv_enable_io_status(BlockDriverState *bs);
> >>> void bdrv_mon_event(const BlockDriverState *bdrv,
> >>> BlockMonEventAction action, int is_read);
> >>> void bdrv_info_print(Monitor *mon, const QObject *data);
> >>> diff --git a/block_int.h b/block_int.h
> >>> index 1e265d2..09f038d 100644
> >>> --- a/block_int.h
> >>> +++ b/block_int.h
> >>> @@ -195,6 +195,8 @@ struct BlockDriverState {
> >>> drivers. They are not used by the block driver */
> >>> int cyls, heads, secs, translation;
> >>> BlockErrorAction on_read_error, on_write_error;
> >>> + bool iostatus_enabled;
> >>> + BlockIOStatus iostatus;
> >>> char device_name[32];
> >>> unsigned long *dirty_bitmap;
> >>> int64_t dirty_count;
> >>
> >> Okay, let's see what we got here.
> >>
> >> The block layer merely holds I/O status, device models set it.
> >>
> >> Device I/O status is not migrated. Why?
Bug. :)
> >>
> >> bdrv_new() creates the BDS with I/O status tracking disabled. Devices
> >> that do tracking enable it in their qdev init method. If a device gets
> >> hot unplugged, tracking remains enabled. If the BDS then gets reused
> >> with a device that doesn't do tracking, I/O status becomes incorrect.
> >> Can't happen right now, because we automatically delete the BDS on hot
> >> unplug, but it's a trap. Suggest to disable tracking in bdrv_detach().
> >>
> >> Actually, this is a symptom of the midlayer disease. I suspect things
> >> would be simpler if we hold the status in its rightful owner, the device
> >> model. Need a getter for it. I'm working on a patch series that moves
> >> misplaced state out of the block layer into device models and block
> >> drivers, and a I/O status getter will fit in easily there.
Excellent.
> >
> > This is host state, so the device is not the rightful owner. Devices
> > should not even be involved with enabling it.
>
> They are because they do the tracking, and thus the tracking only works
> for device models that do it. Could it be done entirely within the
> block layer?
- Re: [Qemu-devel] [PATCH 1/8] Introduce the VMStatus type, (continued)
- Re: [Qemu-devel] [PATCH 1/8] Introduce the VMStatus type, Kevin Wolf, 2011/07/12
- Re: [Qemu-devel] [PATCH 1/8] Introduce the VMStatus type, Luiz Capitulino, 2011/07/12
- Re: [Qemu-devel] [PATCH 1/8] Introduce the VMStatus type, Luiz Capitulino, 2011/07/12
- Re: [Qemu-devel] [PATCH 1/8] Introduce the VMStatus type, Kevin Wolf, 2011/07/12
- Re: [Qemu-devel] [PATCH 1/8] Introduce the VMStatus type, Luiz Capitulino, 2011/07/12
[Qemu-devel] [PATCH 2/8] QMP: query-status: Introduce 'status' key, Luiz Capitulino, 2011/07/05
[Qemu-devel] [PATCH 3/8] block: Support to keep track of I/O status, Luiz Capitulino, 2011/07/05
Re: [Qemu-devel] [PATCH 3/8] block: Support to keep track of I/O status, Kevin Wolf, 2011/07/12
[Qemu-devel] [PATCH 6/8] scsi: Support I/O status, Luiz Capitulino, 2011/07/05
[Qemu-devel] [PATCH 4/8] ide: Support I/O status, Luiz Capitulino, 2011/07/05
[Qemu-devel] [PATCH 7/8] QMP: query-status: Add 'io-status' key, Luiz Capitulino, 2011/07/05
[Qemu-devel] [PATCH 5/8] virtio: Support I/O status, Luiz Capitulino, 2011/07/05
[Qemu-devel] [PATCH 8/8] HMP: Print 'io-status' information, Luiz Capitulino, 2011/07/05
Re: [Qemu-devel] [PATCH v1 0/8]: QMP: Thin provisioning support, Luiz Capitulino, 2011/07/11