yeahuh / rpms / qemu-kvm

Forked from rpms/qemu-kvm 2 years ago
Clone

Blame SOURCES/kvm-virtio-blk-Increase-in_flight-for-request-restart-BH.patch

b38b0f
From 2f11a1952881d9e2dcc8acb3549441dbd9ade1e3 Mon Sep 17 00:00:00 2001
b38b0f
From: Kevin Wolf <kwolf@redhat.com>
b38b0f
Date: Wed, 14 Aug 2019 08:42:26 +0100
b38b0f
Subject: [PATCH 03/10] virtio-blk: Increase in_flight for request restart BH
b38b0f
b38b0f
RH-Author: Kevin Wolf <kwolf@redhat.com>
b38b0f
Message-id: <20190814084229.6458-3-kwolf@redhat.com>
b38b0f
Patchwork-id: 89967
b38b0f
O-Subject: [RHEL-8.1.0 qemu-kvm PATCH 2/5] virtio-blk: Increase in_flight for request restart BH
b38b0f
Bugzilla: 1716349
b38b0f
RH-Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
b38b0f
RH-Acked-by: Max Reitz <mreitz@redhat.com>
b38b0f
RH-Acked-by: Paolo Bonzini <pbonzini@redhat.com>
b38b0f
b38b0f
virtio_blk_dma_restart_bh() submits new requests, so in order to make
b38b0f
sure that these requests are not started inside a drained section of the
b38b0f
attached BlockBackend, we need to make sure that draining the
b38b0f
BlockBackend waits for the BH to be executed.
b38b0f
b38b0f
This BH is still questionable because its scheduled in the main thread
b38b0f
instead of the configured iothread. Leave a FIXME comment for this.
b38b0f
b38b0f
But with this fix, enabling the data plane at least waits for these
b38b0f
requests (in bdrv_set_aio_context()) instead of changing the AioContext
b38b0f
under their feet and making them run in the wrong thread, causing
b38b0f
crashes and failures (e.g. due to missing locking).
b38b0f
b38b0f
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
b38b0f
(cherry picked from commit 680f200217748e0920b79ec1d524717c2f50935b)
b38b0f
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
b38b0f
Signed-off-by: Danilo C. L. de Paula <ddepaula@redhat.com>
b38b0f
---
b38b0f
 hw/block/virtio-blk.c | 4 ++++
b38b0f
 1 file changed, 4 insertions(+)
b38b0f
b38b0f
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
b38b0f
index b1532e4..691ade4 100644
b38b0f
--- a/hw/block/virtio-blk.c
b38b0f
+++ b/hw/block/virtio-blk.c
b38b0f
@@ -678,6 +678,7 @@ static void virtio_blk_dma_restart_bh(void *opaque)
b38b0f
     if (mrb.num_reqs) {
b38b0f
         virtio_blk_submit_multireq(s->blk, &mrb);
b38b0f
     }
b38b0f
+    blk_dec_in_flight(s->conf.conf.blk);
b38b0f
     aio_context_release(blk_get_aio_context(s->conf.conf.blk));
b38b0f
 }
b38b0f
 
b38b0f
@@ -691,8 +692,11 @@ static void virtio_blk_dma_restart_cb(void *opaque, int running,
b38b0f
     }
b38b0f
 
b38b0f
     if (!s->bh) {
b38b0f
+        /* FIXME The data plane is not started yet, so these requests are
b38b0f
+         * processed in the main thread. */
b38b0f
         s->bh = aio_bh_new(blk_get_aio_context(s->conf.conf.blk),
b38b0f
                            virtio_blk_dma_restart_bh, s);
b38b0f
+        blk_inc_in_flight(s->conf.conf.blk);
b38b0f
         qemu_bh_schedule(s->bh);
b38b0f
     }
b38b0f
 }
b38b0f
-- 
b38b0f
1.8.3.1
b38b0f