pci, pc fixes, features
A bunch of bugfixes - these will make sense for 2.1.1
Initial Intel IOMMU support.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Tue 02 Sep 2014 16:05:04 BST using RSA key ID D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg: aka "Michael S. Tsirkin <mst@redhat.com>"
* remotes/mst/tags/for_upstream:
vhost_net: start/stop guest notifiers properly
pci: avoid losing config updates to MSI/MSIX cap regs
virtio-net: don't run bh on vm stopped
ioh3420: remove unused ioh3420_init() declaration
vhost_net: cleanup start/stop condition
intel-iommu: add IOTLB using hash table
intel-iommu: add context-cache to cache context-entry
intel-iommu: add supports for queued invalidation interface
intel-iommu: fix coding style issues around in q35.c and machine.c
intel-iommu: add Intel IOMMU emulation to q35 and add a machine option "iommu" as a switch
intel-iommu: add DMAR table to ACPI tables
intel-iommu: introduce Intel IOMMU (VT-d) emulation
iommu: add is_write as a parameter to the translate function of MemoryRegionIOMMUOps
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
QEMU system mode page table walks are expensive. Taken by running QEMU
qemu-system-x86_64 system mode on Intel PIN , a TLB miss and walking a
4-level page tables in guest Linux OS takes ~450 X86 instructions on
average.
QEMU system mode TLB is implemented using a directly-mapped hashtable.
This structure suffers from conflict misses. Increasing the
associativity of the TLB may not be the solution to conflict misses as
all the ways may have to be walked in serial.
A victim TLB is a TLB used to hold translations evicted from the
primary TLB upon replacement. The victim TLB lies between the main TLB
and its refill path. Victim TLB is of greater associativity (fully
associative in this patch). It takes longer to lookup the victim TLB,
but its likely better than a full page table walk. The memory
translation path is changed as follows :
Before Victim TLB:
1. Inline TLB lookup
2. Exit code cache on TLB miss.
3. Check for unaligned, IO accesses
4. TLB refill.
5. Do the memory access.
6. Return to code cache.
After Victim TLB:
1. Inline TLB lookup
2. Exit code cache on TLB miss.
3. Check for unaligned, IO accesses
4. Victim TLB lookup.
5. If victim TLB misses, TLB refill
6. Do the memory access.
7. Return to code cache
The advantage is that victim TLB can offer more associativity to a
directly mapped TLB and thus potentially fewer page table walks while
still keeping the time taken to flush within reasonable limits.
However, placing a victim TLB before the refill path increase TLB
refill path as the victim TLB is consulted before the TLB refill. The
performance results demonstrate that the pros outweigh the cons.
some performance results taken on SPECINT2006 train
datasets and kernel boot and qemu configure script on an
Intel(R) Xeon(R) CPU E5620 @ 2.40GHz Linux machine are shown in the
Google Doc link below.
https://docs.google.com/spreadsheets/d/1eiItzekZwNQOal_h-5iJmC4tMDi051m9qidi5_nwvH4/edit?usp=sharing
In summary, victim TLB improves the performance of qemu-system-x86_64 by
11% on average on SPECINT2006, kernelboot and qemu configscript and with
highest improvement of in 26% in 456.hmmer. And victim TLB does not result
in any performance degradation in any of the measured benchmarks. Furthermore,
the implemented victim TLB is architecture independent and is expected to
benefit other architectures in QEMU as well.
Although there are measurement fluctuations, the performance
improvement is very significant and by no means in the range of
noises.
Signed-off-by: Xin Tong <trent.tong@gmail.com>
Message-id: 1407202523-23553-1-git-send-email-trent.tong@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Add a bool variable is_write as a parameter to the translate function of
MemoryRegionIOMMUOps to indicate the operation of the access. It can be
used for correct fault reporting from within the callback.
Change the interface of related functions.
Signed-off-by: Le Tan <tamlokveer@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Build /proc/self/maps doing a match against guest memory translation table.
Output only that map records which are valid for guest memory layout.
Signed-off-by: Mikhail Ilyin <m.ilin@samsung.com>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
SCSI changes that enable sending vendor-specific commands via virtio-scsi.
Memory changes for QOMification and automatic tracking of MR lifetime.
# gpg: Signature made Mon 18 Aug 2014 13:03:09 BST using RSA key ID 9B4D86F2
# gpg: Good signature from "Paolo Bonzini <pbonzini@redhat.com>"
# gpg: aka "Paolo Bonzini <bonzini@gnu.org>"
* remotes/bonzini/tags/for-upstream:
mtree: remove write-only field
memory: Use canonical path component as the name
memory: Use memory_region_name for name access
memory: constify memory_region_name
exec: Abstract away ref to memory region names
loader: Abstract away ref to memory region names
tpm_tis: remove instance_finalize callback
memory: remove memory_region_destroy
memory: convert memory_region_destroy to object_unparent
ioport: split deletion and destruction
nic: do not destroy memory regions in cleanup functions
vga: do not dynamically allocate chain4_alias
sysbus: remove unused function sysbus_del_io
qom: object: move unparenting to the child property's release callback
qom: object: delete properties before calling instance_finalize
virtio-scsi: implement parse_cdb
scsi-block, scsi-generic: implement parse_cdb
scsi-block: extract scsi_block_is_passthrough
scsi-bus: introduce parse_cdb in SCSIDeviceClass and SCSIBusInfo
scsi-bus: prepare scsi_req_new for introduction of parse_cdb
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Rather than having the name as separate state. This prepares support
for creating a MemoryRegion dynamically (i.e. without
memory_region_init() and friends) and the MemoryRegion still getting
a usable name.
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
It doesn't change the MR and some prospective call sites will have
const MRs at hand.
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Devices that use address_space_rw to write large areas to memory
(as opposed to address_space_map/unmap) were broken with respect
to migration since fe680d0 (exec: Limit translation limiting in
address_space_translate to xen, 2014-05-07). Such devices include
IDE CD-ROMs.
The reason is that invalidate_and_set_dirty (called by address_space_rw
but not address_space_map/unmap) was only setting the dirty bit for
the first page in the translation.
To fix this, introduce cpu_physical_memory_set_dirty_range_nocode that
is the same as cpu_physical_memory_set_dirty_range except it does not
muck with the DIRTY_MEMORY_CODE bitmap. This function can be used if
the caller invalidates translations with tb_invalidate_phys_page_range.
There is another difference between cpu_physical_memory_set_dirty_range
and cpu_physical_memory_set_dirty_flag; the former includes a call
to xen_modified_memory. This is handled separately in
invalidate_and_set_dirty, and is not needed in other callers of
cpu_physical_memory_set_dirty_range_nocode, so leave it alone.
Just one nit: now that invalidate_and_set_dirty takes care of handling
multiple pages, there is no need for address_space_unmap to wrap it
in a loop. In fact that loop would now be O(n^2).
Reported-by: Dave Gilbert <dgilbert@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
QOM propertyify the .may-overlap and .priority fields. The setters
will re-add the memory as a subregion if needed (i.e. the values change
when the memory region is already contained).
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
[Remove setters. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
QOMify memory regions as an Object. The former init() and destroy()
routines become instance_init() and instance_finalize() resp.
memory_region_init() is re-implemented to be:
object_initialize() + set fields
memory_region_destroy() is re-implemented to call unparent().
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
[Add newly-created MR as child, unparent on destruction. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Old code was affected by memory gaps which resulted in buffer pointers
pointing to address outside of the mapped regions.
Here we are introducing following changes:
- new function qemu_get_ram_block_host_ptr() returns host pointer
to the ram block, it is needed to calculate offset of specific
region in the host memory
- new field mmap_offset is added to the VhostUserMemoryRegion. It
contains offset where specific region starts in the mapped memory.
As there is stil no wider adoption of vhost-user agreement was made
that we will not bump version number due to this change
- other fileds in VhostUserMemoryRegion struct are not changed, as
they are all needed for usermode app implementation
- region data is not taken from ram_list.blocks anymore, instead we
use region data which is alredy calculated for use in vhost-net
- Now multiple regions can have same FD and user applicaton can call
mmap() multiple times with the same FD but with different offset
(user needs to take care for offset page alignment)
Signed-off-by: Damjan Marion <damarion@cisco.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Damjan Marion <damarion@cisco.com>
A new "share" property can be used with the "memory-file" backend to
map memory with MAP_SHARED instead of MAP_PRIVATE.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
And allow preallocation of file-based memory even without -mem-prealloc.
Some care is necessary because -mem-prealloc does not allow disabling
preallocation for hostmem-file.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Right now, -mem-path will fall back to RAM-based allocation in some
cases. This should never happen with "-object memory-file", prepare
the code by adding correct error propagation.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
MST: drop \n at end of error messages
Like the previous patch did in exec.c, split memory_region_init_ram and
memory_region_init_ram_from_file, and push mem_path one step further up.
Other RAM regions than system memory will now be backed by regular RAM.
Also, boards that do not use memory_region_allocate_system_memory will
not support -mem-path anymore. This can be changed before the patches
are merged by migrating boards to use the function.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Split the internal interface in exec.c to a separate function, and
push the check on mem_path up to memory_region_init_ram.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
which allows to check if MemoryRegion is already mapped.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
