Blame SOURCES/kvm-Revert-spapr-Don-t-allow-memory-hotplug-to-memory-le.patch

357786
From 55cc729e37b945bcbb9135fe5553d1bb9a293424 Mon Sep 17 00:00:00 2001
357786
From: Serhii Popovych <spopovyc@redhat.com>
357786
Date: Wed, 2 May 2018 18:52:28 +0200
357786
Subject: [PATCH 05/13] Revert "spapr: Don't allow memory hotplug to memory
357786
 less nodes"
357786
357786
RH-Author: Serhii Popovych <spopovyc@redhat.com>
357786
Message-id: <1525287148-92715-3-git-send-email-spopovyc@redhat.com>
357786
Patchwork-id: 80012
357786
O-Subject: [RHEL-7.6 qemu-kvm-rhev PATCH 2/2] Revert "spapr: Don't allow memory hotplug to memory less nodes"
357786
Bugzilla: 1570525
357786
RH-Acked-by: David Gibson <dgibson@redhat.com>
357786
RH-Acked-by: Thomas Huth <thuth@redhat.com>
357786
RH-Acked-by: Laurent Vivier <lvivier@redhat.com>
357786
357786
This reverts commit b556854bd8524c26b8be98ab1bfdf0826831e793.
357786
357786
Leave change @node type from uint32_t to to int from reverted commit
357786
because node < 0 is always false.
357786
357786
Note that implementing capability or some trick to detect if guest
357786
kernel does not support hot-add to memory: this returns previous
357786
behavour where memory added to first non-empty node.
357786
357786
Signed-off-by: Serhii Popovych <spopovyc@redhat.com>
357786
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
357786
(cherry picked from commit e47f1d2786c3d01a7894a493aafe0efa6b64453c)
357786
Signed-off-by: Serhii Popovych <spopovyc@redhat.com>
357786
Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>
357786
---
357786
 hw/ppc/spapr.c | 22 ----------------------
357786
 1 file changed, 22 deletions(-)
357786
357786
diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
357786
index c9561e1..6f005a0 100644
357786
--- a/hw/ppc/spapr.c
357786
+++ b/hw/ppc/spapr.c
357786
@@ -3501,28 +3501,6 @@ static void spapr_machine_device_plug(HotplugHandler *hotplug_dev,
357786
             return;
357786
         }
357786
 
357786
-        /*
357786
-         * Currently PowerPC kernel doesn't allow hot-adding memory to
357786
-         * memory-less node, but instead will silently add the memory
357786
-         * to the first node that has some memory. This causes two
357786
-         * unexpected behaviours for the user.
357786
-         *
357786
-         * - Memory gets hotplugged to a different node than what the user
357786
-         *   specified.
357786
-         * - Since pc-dimm subsystem in QEMU still thinks that memory belongs
357786
-         *   to memory-less node, a reboot will set things accordingly
357786
-         *   and the previously hotplugged memory now ends in the right node.
357786
-         *   This appears as if some memory moved from one node to another.
357786
-         *
357786
-         * So until kernel starts supporting memory hotplug to memory-less
357786
-         * nodes, just prevent such attempts upfront in QEMU.
357786
-         */
357786
-        if (nb_numa_nodes && !numa_info[node].node_mem) {
357786
-            error_setg(errp, "Can't hotplug memory to memory-less node %d",
357786
-                       node);
357786
-            return;
357786
-        }
357786
-
357786
         spapr_memory_plug(hotplug_dev, dev, node, errp);
357786
     } else if (object_dynamic_cast(OBJECT(dev), TYPE_SPAPR_CPU_CORE)) {
357786
         spapr_core_plug(hotplug_dev, dev, errp);
357786
-- 
357786
1.8.3.1
357786