Re: [Qemu-devel] scsi-hd with discard_granularity and unmap results in Aborted Commands

[Stefan Priebe - Profihost AG <s.priebe@profihost.ag>]

[-- Attachment #1: Type: text/plain, Size: 717 bytes --]

Hi Paolo,

this is my current work status on porting these fixes to rbd. Right now 
the discards get still canceled by the client kernel.

Might you have a look what i have forgotten?

Thanks!

Stefan
Am 19.11.2012 14:06, schrieb Paolo Bonzini:
> Il 19/11/2012 14:01, Stefan Priebe - Profihost AG ha scritto:
>>> The right behavior is to return
>>> only after the target says whether the cancellation was done or not.
>>> For libiscsi, it was implemented by the commits you mention.
>>
>> So the whole bunch of changes is needed for rbd?
>
> Something like the first three:
>
> 1bd075f29ea6d11853475c7c42734595720c3ac6
> cfb3f5064af2d2e29c976e292c9472dfe9d61e31
> 27cbd828c617944c0f9603763fdf4fa87e7ad923
>
> Paolo
>

[-- Attachment #2: 0001-do-not-check-for-cancellation-in-qemu_rbd_complete_a.patch --]
[-- Type: text/x-patch, Size: 1041 bytes --]

>From 486fdb8b18310ff32ca64fbb2e0217c37319cff4 Mon Sep 17 00:00:00 2001
From: Stefan Priebe <s.priebe@profihost.ag>
Date: Mon, 19 Nov 2012 14:31:40 +0100
Subject: [PATCH 1/2] do not check for cancellation in qemu_rbd_complete_aio


Signed-off-by: Stefan Priebe <s.priebe@profhost.ag>
---
 block/rbd.c |    8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/block/rbd.c b/block/rbd.c
index 5a0f79f..583bcc3 100644
--- a/block/rbd.c
+++ b/block/rbd.c
@@ -376,12 +376,6 @@ 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);
-        goto done;
-    }
-
     r = rcb->ret;
 
     if (acb->cmd == RBD_AIO_WRITE ||
@@ -409,7 +403,7 @@ static void qemu_rbd_complete_aio(RADOSCB *rcb)
     /* 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);
 }
 
-- 
1.7.10.4


[-- Attachment #3: 0002-rbd-fix-races-between-io-completition-and-abort.patch --]
[-- Type: text/x-patch, Size: 2758 bytes --]

>From e9eac2c7ed7b98ff102ab7da4573f081ebca32fa Mon Sep 17 00:00:00 2001
From: Stefan Priebe <s.priebe@profihost.ag>
Date: Mon, 19 Nov 2012 15:01:16 +0100
Subject: [PATCH 2/2] rbd: fix races between io completition and abort


Signed-off-by: Stefan Priebe <s.priebe@profhost.ag>
---
 block/rbd.c |   40 +++++++++++++++++++++++++++++++++++++++-
 1 file changed, 39 insertions(+), 1 deletion(-)

diff --git a/block/rbd.c b/block/rbd.c
index 583bcc3..ae1d03b 100644
--- a/block/rbd.c
+++ b/block/rbd.c
@@ -77,6 +77,7 @@ typedef struct RBDAIOCB {
     int error;
     struct BDRVRBDState *s;
     int cancelled;
+    int status;
 } RBDAIOCB;
 
 typedef struct RADOSCB {
@@ -376,6 +377,10 @@ static void qemu_rbd_complete_aio(RADOSCB *rcb)
     RBDAIOCB *acb = rcb->acb;
     int64_t r;
 
+    if (acb->bh) {
+        return;
+    }
+
     r = rcb->ret;
 
     if (acb->cmd == RBD_AIO_WRITE ||
@@ -560,6 +565,20 @@ static void qemu_rbd_close(BlockDriverState *bs)
     rados_shutdown(s->cluster);
 }
 
+static void qemu_rbd_aio_abort(void *private_data)
+{
+    RBDAIOCB *acb = (RBDAIOCB *) private_data;
+
+    acb->status = -ECANCELED;
+
+    if (acb->bh) {
+        return;
+    }
+
+    acb->bh = qemu_bh_new(rbd_aio_bh_cb, acb);
+    qemu_bh_schedule(acb->bh);
+}
+
 /*
  * Cancel aio. Since we don't reference acb in a non qemu threads,
  * it is safe to access it here.
@@ -567,7 +586,22 @@ static void qemu_rbd_close(BlockDriverState *bs)
 static void qemu_rbd_aio_cancel(BlockDriverAIOCB *blockacb)
 {
     RBDAIOCB *acb = (RBDAIOCB *) blockacb;
+
+    if (acb->status != -EINPROGRESS) {
+        return;
+    }
+
     acb->cancelled = 1;
+
+    // TODO / FIXME: send an abort command to rbd
+    // Normally we should call abort librbd and
+    // librbd gets qemu_rbd_aio_abort as a callback function
+    // i wasn't able to find an abort function in librbd at all
+    qemu_rbd_aio_abort(acb);
+
+    while (acb->status == -EINPROGRESS) {
+        qemu_aio_wait();
+    }
 }
 
 static AIOPool rbd_aio_pool = {
@@ -636,10 +670,13 @@ 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;
 
+    if (acb->cancelled == 0) {
+        acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret));
+    }
+
     qemu_aio_release(acb);
 }
 
@@ -685,6 +722,7 @@ static BlockDriverAIOCB *rbd_start_aio(BlockDriverState *bs,
     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