[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 09/11] raw-posix: implement write_zeroes with MA
From: |
Paolo Bonzini |
Subject: |
Re: [Qemu-devel] [PATCH 09/11] raw-posix: implement write_zeroes with MAY_UNMAP for block devices |
Date: |
Wed, 13 Nov 2013 10:44:23 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/20130923 Thunderbird/17.0.9 |
Il 13/11/2013 07:29, Peter Lieven ha scritto:
> Wouldn't it be good to add bdi->can_write_zeroes_with_unmap here as well?
We do:
> + bdi->unallocated_blocks_are_zero = s->discard_zeroes;
> + bdi->can_write_zeroes_with_unmap = s->discard_zeroes;
> This would automatically avoid full allocation when converting something to a
> host device
> supporting BLKDISCARDZEROES.
Yes, that's (part of) the point of this patch.
Regarding the question you posed in the previous patch:
> does BLKDISCARDZEROES ioctl guarantee that a device is
> zero initialized or does it just guarantee that a discard may not
> fail and that it reads as zeroes afterwards?
Only the latter. ".bdrv_has_zero_init" is only present in the bdrv_file
BlockDriver.
Paolo
> Peter
>
> Am 12.11.2013 um 16:49 schrieb Paolo Bonzini <address@hidden>:
>
>> See the next commit for the description of the Linux kernel problem
>> that is worked around in raw_open_common.
>>
>> Signed-off-by: Paolo Bonzini <address@hidden>
>> ---
>> block/raw-posix.c | 38 ++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 38 insertions(+)
>>
>> diff --git a/block/raw-posix.c b/block/raw-posix.c
>> index 830e109..5cb46f1 100644
>> --- a/block/raw-posix.c
>> +++ b/block/raw-posix.c
>> @@ -335,6 +335,23 @@ static int raw_open_common(BlockDriverState *bs, QDict
>> *options,
>> if (S_ISREG(st.st_mode)) {
>> s->discard_zeroes = true;
>> }
>> +#ifdef BLKDISCARDZEROES
>> + if (S_ISBLK(st.st_mode)) {
>> + unsigned int arg;
>> + if (ioctl(s->fd, BLKDISCARDZEROES, &arg) == 0 && arg) {
>> + s->discard_zeroes = true;
>> + }
>> + }
>> +#endif
>> +#ifdef CONFIG_LINUX
>> + /* On Linux 3.10, BLKDISCARD leaves stale data in the page cache. Do
>> + * not rely on the contents of discarded blocks unless using O_DIRECT.
>> + */
>> + if (!(bs->open_flags & BDRV_O_NOCACHE)) {
>> + s->discard_zeroes = false;
>> + }
>> +#endif
>> + }
>>
>> #ifdef CONFIG_XFS
>> if (platform_test_xfs_fd(s->fd)) {
>> @@ -1587,6 +1604,26 @@ static coroutine_fn BlockDriverAIOCB
>> *hdev_aio_discard(BlockDriverState *bs,
>> cb, opaque, QEMU_AIO_DISCARD|QEMU_AIO_BLKDEV);
>> }
>>
>> +static coroutine_fn int hdev_co_write_zeroes(BlockDriverState *bs,
>> + int64_t sector_num, int nb_sectors, BdrvRequestFlags flags)
>> +{
>> + BDRVRawState *s = bs->opaque;
>> + int rc;
>> +
>> + rc = fd_open(bs);
>> + if (rc < 0) {
>> + return rc;
>> + }
>> + if (!(flags & BDRV_REQ_MAY_UNMAP)) {
>> + return -ENOTSUP;
>> + }
>> + if (!s->discard_zeroes) {
>> + return -ENOTSUP;
>> + }
>> + return paio_submit_co(bs, s->fd, sector_num, NULL, nb_sectors,
>> + QEMU_AIO_DISCARD|QEMU_AIO_BLKDEV);
>> +}
>> +
>> static int hdev_create(const char *filename, QEMUOptionParameter *options,
>> Error **errp)
>> {
>> @@ -1639,6 +1676,7 @@ static BlockDriver bdrv_host_device = {
>> .bdrv_reopen_abort = raw_reopen_abort,
>> .bdrv_create = hdev_create,
>> .create_options = raw_create_options,
>> + .bdrv_co_write_zeroes = hdev_co_write_zeroes,
>>
>> .bdrv_aio_readv = raw_aio_readv,
>> .bdrv_aio_writev = raw_aio_writev,
>> --
>> 1.8.4.2
>>
>>
>
>
>
- [Qemu-devel] [PATCH 06/11] scsi-disk: correctly implement WRITE SAME, (continued)
- [Qemu-devel] [PATCH 06/11] scsi-disk: correctly implement WRITE SAME, Paolo Bonzini, 2013/11/12
- [Qemu-devel] [PATCH 07/11] block: handle ENOTSUP from discard in generic code, Paolo Bonzini, 2013/11/12
- [Qemu-devel] [PATCH 08/11] raw-posix: implement write_zeroes with MAY_UNMAP for files, Paolo Bonzini, 2013/11/12
- [Qemu-devel] [PATCH 10/11] raw-posix: add support for write_zeroes on XFS and block devices, Paolo Bonzini, 2013/11/12
- [Qemu-devel] [PATCH 09/11] raw-posix: implement write_zeroes with MAY_UNMAP for block devices, Paolo Bonzini, 2013/11/12
[Qemu-devel] [PATCH 11/11] qemu-iotests: 033 is fast, Paolo Bonzini, 2013/11/12
[Qemu-devel] [PATCH 05/11] scsi-disk: reject ANCHOR=1 for UNMAP and WRITE SAME commands, Paolo Bonzini, 2013/11/12