From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:58166) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1URm3y-0007OQ-Dg for qemu-devel@nongnu.org; Mon, 15 Apr 2013 12:15:28 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1URm3u-00054V-Ln for qemu-devel@nongnu.org; Mon, 15 Apr 2013 12:15:26 -0400 Received: from mail-pd0-f173.google.com ([209.85.192.173]:59955) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1URm3u-00054Q-CU for qemu-devel@nongnu.org; Mon, 15 Apr 2013 12:15:22 -0400 Received: by mail-pd0-f173.google.com with SMTP id v14so2581427pde.32 for ; Mon, 15 Apr 2013 09:15:21 -0700 (PDT) From: Liu Yuan Date: Tue, 16 Apr 2013 00:15:04 +0800 Message-Id: <1366042504-18354-1-git-send-email-namei.unix@gmail.com> In-Reply-To: <1365852442-28941-5-git-send-email-namei.unix@gmail.com> References: <1365852442-28941-5-git-send-email-namei.unix@gmail.com> Subject: [Qemu-devel] [PATCH v5] sheepdog: add discard/trim support for sheepdog List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: sheepdog@lists.wpkg.org, qemu-devel@nongnu.org Cc: Kevin Wolf , Paolo Bonzini , Stefan Hajnoczi , MORITA Kazutaka From: Liu Yuan The 'TRIM' command from VM that is to release underlying data storage for better thin-provision is already supported by the Sheepdog. This patch adds the TRIM support at QEMU part. For older Sheepdog that doesn't support it, we return 0(success) to upper layer. Cc: MORITA Kazutaka Cc: Kevin Wolf Cc: Stefan Hajnoczi Cc: Paolo Bonzini Signed-off-by: Liu Yuan --- v5: - adjust macro numbering v4: - adjust discard macro - return success when operation is not supported by sheep - add coroutine_fn marker v3: - fix a silly accidental deletion of 'default' in switch clause. v2: - skip the object when it is not allocated block/sheepdog.c | 56 +++++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 55 insertions(+), 1 deletion(-) diff --git a/block/sheepdog.c b/block/sheepdog.c index 987018e..3b64690 100644 --- a/block/sheepdog.c +++ b/block/sheepdog.c @@ -27,6 +27,8 @@ #define SD_OP_CREATE_AND_WRITE_OBJ 0x01 #define SD_OP_READ_OBJ 0x02 #define SD_OP_WRITE_OBJ 0x03 +/* 0x04 is used internally by Sheepdog */ +#define SD_OP_DISCARD_OBJ 0x05 #define SD_OP_NEW_VDI 0x11 #define SD_OP_LOCK_VDI 0x12 @@ -269,6 +271,7 @@ enum AIOCBState { AIOCB_WRITE_UDATA, AIOCB_READ_UDATA, AIOCB_FLUSH_CACHE, + AIOCB_DISCARD_OBJ, }; struct SheepdogAIOCB { @@ -656,7 +659,7 @@ static void coroutine_fn aio_read_response(void *opaque) int ret; AIOReq *aio_req = NULL; SheepdogAIOCB *acb; - unsigned long idx; + uint64_t idx; if (QLIST_EMPTY(&s->inflight_aio_head)) { goto out; @@ -727,6 +730,20 @@ static void coroutine_fn aio_read_response(void *opaque) rsp.result = SD_RES_SUCCESS; } break; + case AIOCB_DISCARD_OBJ: + switch (rsp.result) { + case SD_RES_INVALID_PARMS: + error_report("you are running the old sheep that doesn't support " + "discard command.\n"); + rsp.result = SD_RES_SUCCESS; + break; + case SD_RES_SUCCESS: + idx = data_oid_to_idx(aio_req->oid); + s->inode.data_vdi_id[idx] = 0; + break; + default: + break; + } } if (rsp.result != SD_RES_SUCCESS) { @@ -1016,6 +1033,9 @@ static int coroutine_fn add_aio_request(BDRVSheepdogState *s, AIOReq *aio_req, wlen = datalen; hdr.flags = SD_FLAG_CMD_WRITE | flags; break; + case AIOCB_DISCARD_OBJ: + hdr.opcode = SD_OP_DISCARD_OBJ; + break; } if (s->cache_flags) { @@ -1633,6 +1653,15 @@ static int coroutine_fn sd_co_rw_vector(void *p) flags = SD_FLAG_CMD_COW; } break; + case AIOCB_DISCARD_OBJ: + /* + * We discard the object only when the whole object is + * 1) allocated 2) trimmed. Otherwise, simply skip it. + */ + if (len != SD_DATA_OBJ_SIZE || inode->data_vdi_id[idx] == 0) { + goto done; + } + break; default: break; } @@ -2071,6 +2100,28 @@ static int sd_load_vmstate(BlockDriverState *bs, uint8_t *data, } +static coroutine_fn int sd_co_discard(BlockDriverState *bs, int64_t sector_num, + int nb_sectors) +{ + SheepdogAIOCB *acb; + QEMUIOVector dummy; + int ret; + + acb = sd_aio_setup(bs, &dummy, sector_num, nb_sectors); + acb->aiocb_type = AIOCB_DISCARD_OBJ; + acb->aio_done_func = sd_finish_aiocb; + + ret = sd_co_rw_vector(acb); + if (ret <= 0) { + qemu_aio_release(acb); + return ret; + } + + qemu_coroutine_yield(); + + return acb->ret; +} + static QEMUOptionParameter sd_create_options[] = { { .name = BLOCK_OPT_SIZE, @@ -2103,6 +2154,7 @@ static BlockDriver bdrv_sheepdog = { .bdrv_co_readv = sd_co_readv, .bdrv_co_writev = sd_co_writev, .bdrv_co_flush_to_disk = sd_co_flush_to_disk, + .bdrv_co_discard = sd_co_discard, .bdrv_snapshot_create = sd_snapshot_create, .bdrv_snapshot_goto = sd_snapshot_goto, @@ -2128,6 +2180,7 @@ static BlockDriver bdrv_sheepdog_tcp = { .bdrv_co_readv = sd_co_readv, .bdrv_co_writev = sd_co_writev, .bdrv_co_flush_to_disk = sd_co_flush_to_disk, + .bdrv_co_discard = sd_co_discard, .bdrv_snapshot_create = sd_snapshot_create, .bdrv_snapshot_goto = sd_snapshot_goto, @@ -2153,6 +2206,7 @@ static BlockDriver bdrv_sheepdog_unix = { .bdrv_co_readv = sd_co_readv, .bdrv_co_writev = sd_co_writev, .bdrv_co_flush_to_disk = sd_co_flush_to_disk, + .bdrv_co_discard = sd_co_discard, .bdrv_snapshot_create = sd_snapshot_create, .bdrv_snapshot_goto = sd_snapshot_goto, -- 1.7.9.5