[PULL 0/3] Host Memory Backends and Memory devices queue 2024-02-6

[PULL 0/3] Host Memory Backends and Memory devices queue 2024-02-6

[David Hildenbrand]

The following changes since commit 39a6e4f87e7b75a45b08d6dc8b8b7c2954c87440:

  Merge tag 'pull-qapi-2024-02-03' of https://repo.or.cz/qemu/armbru into staging (2024-02-03 13:31:58 +0000)

are available in the Git repository at:

  https://github.com/davidhildenbrand/qemu.git tags/mem-2024-02-06

for you to fetch changes up to a32c31979a852a07d590effa1586f6ab4fa4d784:

  oslib-posix: initialize backend memory objects in parallel (2024-02-04 17:51:13 +0100)

----------------------------------------------------------------
Hi,

"Host Memory Backends" and "Memory devices" queue ("mem"):
- Reintroduce memory region size checks for memory devices; the removal
  lead to some undesired side effects
- Preallocate memory of memory backends in selected configurations
  asynchronously (so we preallocate concurrently), to speed up QEMU
  startup time.

----------------------------------------------------------------
David Hildenbrand (2):
      hv-balloon: use get_min_alignment() to express 32 GiB alignment
      memory-device: reintroduce memory region size check

Mark Kanda (1):
      oslib-posix: initialize backend memory objects in parallel

 backends/hostmem.c     |   7 ++-
 hw/hyperv/hv-balloon.c |  37 +++++++-----
 hw/mem/memory-device.c |  14 +++++
 hw/virtio/virtio-mem.c |   4 +-
 include/hw/qdev-core.h |   5 ++
 include/qemu/osdep.h   |  18 +++++-
 system/vl.c            |   9 +++
 util/oslib-posix.c     | 131 +++++++++++++++++++++++++++++++----------
 util/oslib-win32.c     |   8 ++-
 9 files changed, 180 insertions(+), 53 deletions(-)

-- 
2.43.0

[PULL 1/3] hv-balloon: use get_min_alignment() to express 32 GiB alignment

[David Hildenbrand]

Let's implement the get_min_alignment() callback for memory devices, and
copy for the device memory region the alignment of the host memory
region. This mimics what virtio-mem does, and allows for re-introducing
proper alignment checks for the memory region size (where we don't care
about additional device requirements) in memory device core.

Message-ID: <20240117135554.787344-2-david@redhat.com>
Reviewed-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 hw/hyperv/hv-balloon.c | 37 +++++++++++++++++++++----------------
 1 file changed, 21 insertions(+), 16 deletions(-)

diff --git a/hw/hyperv/hv-balloon.c b/hw/hyperv/hv-balloon.c
index 0238365712..ade283335a 100644
--- a/hw/hyperv/hv-balloon.c
+++ b/hw/hyperv/hv-balloon.c
@@ -1477,22 +1477,7 @@ static void hv_balloon_ensure_mr(HvBalloon *balloon)
     balloon->mr = g_new0(MemoryRegion, 1);
     memory_region_init(balloon->mr, OBJECT(balloon), TYPE_HV_BALLOON,
                        memory_region_size(hostmem_mr));
-
-    /*
-     * The VM can indicate an alignment up to 32 GiB. Memory device core can
-     * usually only handle/guarantee 1 GiB alignment. The user will have to
-     * specify a larger maxmem eventually.
-     *
-     * The memory device core will warn the user in case maxmem might have to be
-     * increased and will fail plugging the device if there is not sufficient
-     * space after alignment.
-     *
-     * TODO: we could do the alignment ourselves in a slightly bigger region.
-     * But this feels better, although the warning might be annoying. Maybe
-     * we can optimize that in the future (e.g., with such a device on the
-     * cmdline place/size the device memory region differently.
-     */
-    balloon->mr->align = MAX(32 * GiB, memory_region_get_alignment(hostmem_mr));
+    balloon->mr->align = memory_region_get_alignment(hostmem_mr);
 }
 
 static void hv_balloon_free_mr(HvBalloon *balloon)
@@ -1654,6 +1639,25 @@ static MemoryRegion *hv_balloon_md_get_memory_region(MemoryDeviceState *md,
     return balloon->mr;
 }
 
+static uint64_t hv_balloon_md_get_min_alignment(const MemoryDeviceState *md)
+{
+    /*
+     * The VM can indicate an alignment up to 32 GiB. Memory device core can
+     * usually only handle/guarantee 1 GiB alignment. The user will have to
+     * specify a larger maxmem eventually.
+     *
+     * The memory device core will warn the user in case maxmem might have to be
+     * increased and will fail plugging the device if there is not sufficient
+     * space after alignment.
+     *
+     * TODO: we could do the alignment ourselves in a slightly bigger region.
+     * But this feels better, although the warning might be annoying. Maybe
+     * we can optimize that in the future (e.g., with such a device on the
+     * cmdline place/size the device memory region differently.
+     */
+    return 32 * GiB;
+}
+
 static void hv_balloon_md_fill_device_info(const MemoryDeviceState *md,
                                            MemoryDeviceInfo *info)
 {
@@ -1766,5 +1770,6 @@ static void hv_balloon_class_init(ObjectClass *klass, void *data)
     mdc->get_memory_region = hv_balloon_md_get_memory_region;
     mdc->decide_memslots = hv_balloon_decide_memslots;
     mdc->get_memslots = hv_balloon_get_memslots;
+    mdc->get_min_alignment = hv_balloon_md_get_min_alignment;
     mdc->fill_device_info = hv_balloon_md_fill_device_info;
 }
-- 
2.43.0

Re: [PULL 0/3] Host Memory Backends and Memory devices queue 2024-02-6

[David Hildenbrand]

On 06.02.24 08:02, David Hildenbrand wrote:
> The following changes since commit 39a6e4f87e7b75a45b08d6dc8b8b7c2954c87440:
> 
>    Merge tag 'pull-qapi-2024-02-03' of https://repo.or.cz/qemu/armbru into staging (2024-02-03 13:31:58 +0000)
> 
> are available in the Git repository at:
> 
>    https://github.com/davidhildenbrand/qemu.git tags/mem-2024-02-06
> 
> for you to fetch changes up to a32c31979a852a07d590effa1586f6ab4fa4d784:
> 
>    oslib-posix: initialize backend memory objects in parallel (2024-02-04 17:51:13 +0100)
> 
> ----------------------------------------------------------------
> Hi,
> 
> "Host Memory Backends" and "Memory devices" queue ("mem"):
> - Reintroduce memory region size checks for memory devices; the removal
>    lead to some undesired side effects
> - Preallocate memory of memory backends in selected configurations
>    asynchronously (so we preallocate concurrently), to speed up QEMU
>    startup time.

A mail address in the second patch is wrong. Will send a v2.

-- 
Cheers,

David / dhildenb