Changelog:
v1->v2:
* take into account Paolo's review comments
* do not overload ram_addr
* ifdef linux specific code
* reseve HVA using API from exec.c instead of calling
mmap() dircely from memory.c
* support unmapping of HVA remapped region
Series still is RFC due to VHOST issues with small memory
amount (i.e. last patch 6/6 isn't usable yet), so I'm posting
series mainly for revewing memory API bits.
--
When more than ~50 pc-dimm devices are hotplugged with
vhost enabled, QEMU will assert in vhost vhost_commit()
due to backend refusing to accept too many memory ranges.
Series introduces Reserved HVA MemoryRegion container
where all hotplugged memory is remapped and passes
the single container range to vhost instead of multiple
memory ranges for each hotlugged pc-dimm device.
It's alternative approach to increasing backend supported
memory regions limit since what I've come up with
backend side approach is quite a bit more code so far.
With this approach it would be possible to extend it to
initial memory later and provide a single range for all
RAM to vhost, which should speed up its hot-path by replacing
current GPA<->HVA lookup loop with offset calculation.
Igor Mammedov (6):
memory: get rid of memory_region_destructor_ram_from_ptr()
memory: introduce MemoryRegion container with reserved HVA range
memory: support unmapping of MemoryRegion mapped into HVA parent
hostmem: return recreated MemoryRegion if current can't be reused
pc: reserve hotpluggable memory range with
memory_region_init_hva_range()
pc: fix QEMU crashing when more than ~50 memory hotplugged
backends/hostmem.c | 6 ++++
exec.c | 67 +++++++++++++++++++++++++++++-------------
hw/i386/pc.c | 4 +--
hw/virtio/vhost.c | 15 ++++++++--
include/exec/cpu-common.h | 3 ++
include/exec/memory.h | 57 +++++++++++++++++++++++++++++++++--
include/exec/ram_addr.h | 1 -
memory.c | 75 +++++++++++++++++++++++++++++++++++++++++++----
8 files changed, 194 insertions(+), 34 deletions(-)
--
1.8.3.1
