From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:56190) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TaStm-0003qi-N8 for qemu-devel@nongnu.org; Mon, 19 Nov 2012 10:04:44 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1TaStY-0006FZ-MQ for qemu-devel@nongnu.org; Mon, 19 Nov 2012 10:04:34 -0500 Received: from mail.profihost.ag ([85.158.179.208]:54365) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TaStY-0006FC-Bv for qemu-devel@nongnu.org; Mon, 19 Nov 2012 10:04:20 -0500 Message-ID: <50AA4A70.1090806@profihost.ag> Date: Mon, 19 Nov 2012 16:04:16 +0100 From: Stefan Priebe - Profihost AG MIME-Version: 1.0 References: <589082222.7272963.1352241778123.JavaMail.root@redhat.com> <50A95A6A.4010109@profihost.ag> <50A9E991.2070609@redhat.com> <50A9FD86.3050400@profihost.ag> <50AA01C0.8030709@redhat.com> <50AA0314.6090208@profihost.ag> <50AA04A7.4080200@redhat.com> <50AA065F.5010602@profihost.ag> <50AA089B.7010400@redhat.com> <50AA0A2E.2090506@profihost.ag> <50AA0B90.6060702@redhat.com> <50AA1085.6000106@profihost.ag> <50AA151A.7030008@redhat.com> <50AA1CE7.7090100@profihost.ag> <50AA24E7.7010009@redhat.com> <50AA2DB0.6010107@profihost.ag> <50AA2EDF.6070409@redhat.com> <50AA420D.6070806@profihost.ag> <50AA4507.4080401@redhat.com> In-Reply-To: <50AA4507.4080401@redhat.com> Content-Type: multipart/mixed; boundary="------------010907050101050005000005" Subject: Re: [Qemu-devel] scsi-hd with discard_granularity and unmap results in Aborted Commands List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: josh.durgin@inktank.com, qemu-devel@nongnu.org, sage@newdream.net This is a multi-part message in MIME format. --------------010907050101050005000005 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Hi Paolo, new patch attached. Desciption is still wrong. > I think this is all unneeded. Just store rcb->ret into > rcb->acb->status, and your version of qemu_rbd_aio_cancel should just > work. > > Also, I think the acb->cancelled field is not necessary anymore after > these changes. 1.) It removes cancelled 2.) It adds status variable 3.) aio cancel now just waits for io completetion This should fix the write race you mentioned. But it still does not help with discard the kernel starts to cancel as the reply takes too long. See: [ 49.183366] sd 2:0:0:1: [sdb] [ 49.183366] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE [ 49.183366] sd 2:0:0:1: [sdb] [ 49.183366] Sense Key : Aborted Command [current] [ 49.183366] sd 2:0:0:1: [sdb] [ 49.183366] Add. Sense: I/O process terminated [ 49.183366] sd 2:0:0:1: [sdb] CDB: [ 49.183366] Write same(16): 93 08 00 00 00 00 03 ff ff f8 00 7f ff ff 00 00 [ 49.183366] end_request: I/O error, dev sdb, sector 67108856 [ 49.183366] sd 2:0:0:1: [sdb] [ 49.183366] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE [ 49.183366] sd 2:0:0:1: [sdb] [ 49.183366] Sense Key : Aborted Command [current] [ 49.183366] sd 2:0:0:1: [sdb] [ 49.183366] Add. Sense: I/O process terminated [ 49.183366] sd 2:0:0:1: [sdb] CDB: [ 49.183366] Write same(16): 93 08 00 00 00 00 04 7f ff f7 00 62 00 09 00 00 [ 49.183366] end_request: I/O error, dev sdb, sector 75497463 Greets, Stefan --------------010907050101050005000005 Content-Type: text/x-patch; name="0001-fix-cancel-rbd-race.patch" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="0001-fix-cancel-rbd-race.patch" >>From d65f2c2ba8c81842992953dd772355898e702968 Mon Sep 17 00:00:00 2001 From: Stefan Priebe Date: Mon, 19 Nov 2012 15:54:05 +0100 Subject: [PATCH] fix cancel rbd race Signed-off-by: Stefan Priebe --- block/rbd.c | 19 ++++++++++++------- 1 file changed, 12 insertions(+), 7 deletions(-) diff --git a/block/rbd.c b/block/rbd.c index 5a0f79f..7b3bcbb 100644 --- a/block/rbd.c +++ b/block/rbd.c @@ -76,7 +76,7 @@ typedef struct RBDAIOCB { int64_t sector_num; int error; struct BDRVRBDState *s; - int cancelled; + int status; } RBDAIOCB; typedef struct RADOSCB { @@ -376,9 +376,7 @@ static void qemu_rbd_complete_aio(RADOSCB *rcb) RBDAIOCB *acb = rcb->acb; int64_t r; - if (acb->cancelled) { - qemu_vfree(acb->bounce); - qemu_aio_release(acb); + if (acb->bh) { goto done; } @@ -406,9 +404,12 @@ static void qemu_rbd_complete_aio(RADOSCB *rcb) acb->ret = r; } } + acb->status = acb->ret; + /* Note that acb->bh can be NULL in case where the aio was cancelled */ acb->bh = qemu_bh_new(rbd_aio_bh_cb, acb); qemu_bh_schedule(acb->bh); + done: g_free(rcb); } @@ -573,7 +574,10 @@ static void qemu_rbd_close(BlockDriverState *bs) static void qemu_rbd_aio_cancel(BlockDriverAIOCB *blockacb) { RBDAIOCB *acb = (RBDAIOCB *) blockacb; - acb->cancelled = 1; + + while (acb->status == -EINPROGRESS) { + qemu_aio_wait(); + } } static AIOPool rbd_aio_pool = { @@ -642,10 +646,11 @@ static void rbd_aio_bh_cb(void *opaque) qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size); } qemu_vfree(acb->bounce); - acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret)); qemu_bh_delete(acb->bh); acb->bh = NULL; + acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret)); + qemu_aio_release(acb); } @@ -689,8 +694,8 @@ static BlockDriverAIOCB *rbd_start_aio(BlockDriverState *bs, acb->ret = 0; acb->error = 0; acb->s = s; - acb->cancelled = 0; acb->bh = NULL; + acb->status = -EINPROGRESS; if (cmd == RBD_AIO_WRITE) { qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size); -- 1.7.10.4 --------------010907050101050005000005--