Introduce sPAPRMachineClass.dr_cpu_enabled to indicate support for
CPU core hotplug. Initialize boot time CPUs as core deivces and prevent
topologies that result in partially filled cores. Both of these are done
only if CPU core hotplug is supported.
Note: An unrelated change in the call to xics_system_init() is done
in this patch as it makes sense to use the local variable smt introduced
in this patch instead of kvmppc_smt_threads() call here.
TODO: We derive sPAPR core type by looking at -cpu <model>. However
we don't take care of "compat=" feature yet for boot time as well
as hotplug CPUs.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Start consolidating CPU init related routines in spapr_cpu_core.c. As
part of this, move spapr_cpu_init() and its dependencies from spapr.c
to spapr_cpu_core.c
No functionality change in this patch.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
[dwg: Rename TIMEBASE_FREQ to SPAPR_TIMEBASE_FREQ, since it's now in a
public(ish) header]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Add sPAPR specific abastract CPU core device that is based on generic
CPU core device. Use this as base type to create sPAPR CPU specific core
devices.
TODO:
- Add core types for other remaining CPU types
- Handle CPU model alias correctly
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Add an API object_type_get_size(const char *typename) that returns the
instance_size of the give typename.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
If a CPU is hot removed while hotplug of the same is still in progress,
the guest crashes. Prevent this by ensuring that detach is done only
after attach has completed.
The existing code already prevents such race for PCI hotplug. However
given that CPU is a logical DR unlike PCI and starts with ISOLATED
state, we need a logic that works for CPU too.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
[Don't set awaiting_attach for PCI devices]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
XICS is setup for each CPU during initialization. Provide a routine
to undo the same when CPU is unplugged. While here, move ss->cs management
into xics from xics_kvm since there is nothing KVM specific in it.
Also ensure xics reset doesn't set irq for CPUs that are already unplugged.
This allows reboot of a VM that has undergone CPU hotplug and unplug
to work correctly.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Add an abstract CPU core type that could be used by machines that want
to define and hotplug CPUs in core granularity.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
[Integer core property]
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
[dwg: changed property names to 'core-id' and 'nr-threads']
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
pre_plug callback is to be called before device.realize() is executed.
This would allow to check/set device's properties from HotplugHandler.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
While doing DMA read into ESP command buffer 's->cmdbuf', it could
write past the 's->cmdbuf' area, if it was transferring more than 16
bytes. Increase the command buffer size to 32, which is maximum when
's->do_cmd' is set, and add a check on 'len' to avoid OOB access.
Reported-by: Li Qiang <liqiang6-s@360.cn>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Declare a constant and use that when determining if an export
name fits within the constraints we are willing to support.
Note that upstream NBD recently documented that clients MUST
support export names of 256 bytes (not including trailing NUL),
and SHOULD support names up to 4096 bytes. 4096 is a bit big
(we would lose benefits of stack-allocation of a name array),
and we already have other limits in place (for example, qcow2
snapshot names are clamped around 1024). So for now, just
stick to the required minimum, as that's easier to audit than
a full-scale support for larger names.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463006384-7734-12-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
These structs are never used to represent the bytes that go over the
network. The big-endian network data is built into a uint8_t array
in nbd_{receive,send}_{request,reply}. Remove the unused magic field,
reorder the struct to avoid holes, and remove the packed attribute.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
qemu/osdep.h checks whether MAP_ANONYMOUS is defined, but this check
is bogus without a previous inclusion of sys/mman.h. Include it in
sysemu/os-posix.h and remove it from everywhere else.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Currently, we are trying to move the backing BDS from the source to the
target in bdrv_replace_in_backing_chain() which is called from
mirror_exit(). However, mirror_complete() already tries to open the
target's backing chain with a call to bdrv_open_backing_file().
First, we should only set the target's backing BDS once. Second, the
mirroring block job has a better idea of what to set it to than the
generic code in bdrv_replace_in_backing_chain() (in fact, the latter's
conditions on when to move the backing BDS from source to target are not
really correct).
Therefore, remove that code from bdrv_replace_in_backing_chain() and
leave it to mirror_complete().
Depending on what kind of mirroring is performed, we furthermore want to
use different strategies to open the target's backing chain:
- If blockdev-mirror is used, we can assume the user made sure that the
target already has the correct backing chain. In particular, we should
not try to open a backing file if the target does not have any yet.
- If drive-mirror with mode=absolute-paths is used, we can and should
reuse the already existing chain of nodes that the source BDS is in.
In case of sync=full, no backing BDS is required; with sync=top, we
just link the source's backing BDS to the target, and with sync=none,
we use the source BDS as the target's backing BDS.
We should not try to open these backing files anew because this would
lead to two BDSs existing per physical file in the backing chain, and
we would like to avoid such concurrent access.
- If drive-mirror with mode=existing is used, we have to use the
information provided in the physical image file which means opening
the target's backing chain completely anew, just as it has been done
already.
If the target's backing chain shares images with the source, this may
lead to multiple BDSs per physical image file. But since we cannot
reliably ascertain this case, there is nothing we can do about it.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Message-id: 20160610185750.30956-3-mreitz@redhat.com
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
It is always true for open images now.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
This allows drivers to share code between normal I/O and vmstate
accesses.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
This brings it in line with .bdrv_save_vmstate().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
We already have a byte-based bdrv_pwritev(), but the read counterpart
was still missing. This commit adds it.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
In a first step to convert the common I/O path to work on bytes rather
than sectors, this converts the copy-on-read logic that is used by
bdrv_aligned_preadv().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Add a new BDRV_REQ_MASK constant, and use it to make sure that
caller flags are always valid.
Tested with 'make check' and with qemu-iotests on both '-raw'
and '-qcow2'; the only failure turned up was fixed in the
previous commit.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Apply the following renames for starting incoming migration:
process_incoming_migration -> migration_fd_process_incoming
migration_set_incoming_channel -> migration_channel_process_incoming
migration_tls_set_incoming_channel -> migration_tls_channel_process_incoming
and for starting outgoing migration:
migration_set_outgoing_channel -> migration_channel_connect
migration_tls_set_outgoing_channel -> migration_tls_channel_connect
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1464776234-9910-3-git-send-email-berrange@redhat.com
Message-Id: <1464776234-9910-3-git-send-email-berrange@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
On the source, add a count of page requests received from the
destination.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Message-id: 1465816605-29488-4-git-send-email-dgilbert@redhat.com
Message-Id: <1465816605-29488-4-git-send-email-dgilbert@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
I looked at a dozen Intel CPU that have this CPUID and all of them
always had Core offset as 1 (a wasted bit when hyperthreading is
disabled) and Package offset at least 4 (wasted bits at <= 4 cores).
QEMU uses more compact IDs and it doesn't make much sense to change it
now. I keep the SMT and Core sub-leaves even if there is just one
thread/core; it makes the code simpler and there should be no harm.
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
This adds the DP and the DPDMA to the Zynq MP platform.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Tested-By: Hyun Kwon <hyun.kwon@xilinx.com>
Message-id: 1465833014-21982-10-git-send-email-fred.konrad@greensocs.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This is the implementation of the DisplayPort.
It has an aux-bus to access dpcd and edid.
Graphic plane is connected to the channel 3.
Video plane is connected to the channel 0.
Audio stream are connected to the channels 4 and 5.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Tested-By: Hyun Kwon <hyun.kwon@xilinx.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1465833014-21982-9-git-send-email-fred.konrad@greensocs.com
[PMM: fixed format strings]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This is the implementation of the DPDMA.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Tested-By: Hyun Kwon <hyun.kwon@xilinx.com>
Message-id: 1465833014-21982-8-git-send-email-fred.konrad@greensocs.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Implement an I2C slave which implements DDC and returns the
EDID data for an attached monitor.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Tested-by: Hyun Kwon <hyun.kwon@xilinx.com>
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Message-id: 1465833014-21982-7-git-send-email-fred.konrad@greensocs.com
- Rebased on the current master.
- Modified for QOM.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Tested-By: Hyun Kwon <hyun.kwon@xilinx.com>
[PMM: actually wire up the vmstate to dc->vmsd]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This introduces dpcd module.
It wires on a aux-bus and can be accessed by the driver to get lane-speed, etc.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Reviewed-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Tested-By: Hyun Kwon <hyun.kwon@xilinx.com>
Message-id: 1465833014-21982-6-git-send-email-fred.konrad@greensocs.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This introduces a new bus: aux-bus.
It contains an address space for aux slaves devices and a bridge to an I2C bus
for I2C through AUX transactions.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Tested-By: Hyun Kwon <hyun.kwon@xilinx.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1465833014-21982-5-git-send-email-fred.konrad@greensocs.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Most of the control flow logic between send and recv (error checking
etc) is the same. Factor this out into a common send_recv() API.
This is then usable by clients, where the control logic for send
and receive differs only by a boolean. E.g.
if (send)
i2c_send(...):
else
i2c_recv(...);
becomes:
i2c_send_recv(... , send);
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Message-id: 1465833014-21982-4-git-send-email-fred.konrad@greensocs.com
Changes from FK:
* Rebased on master.
* Rebased on my i2c broadcast patch.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Add a virtual PMU device for virt machine while use PPI 7 for PMU
overflow interrupt number.
Signed-off-by: Shannon Zhao <shannon.zhao@linaro.org>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 1465267577-1808-3-git-send-email-zhaoshenglong@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Clean up includes so that osdep.h is included first and headers
which it implies are not included manually.
This commit was created with scripts/clean-includes.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>
Let's introduce a CssDevId to handle device ids of the xx.x.xxxx
type used for channel devices. This has some benefits:
- We can use them in virtio-ccw and split the validity checks for
a channel device id in general from the constraint checking
within the virtio-ccw scope.
- We can reuse the device id type for future non-virtio channel
devices.
While we're at it, improve the validity checks and disallow e.g.
trailing characters.
Suggested-by: Dong Jia Shi <bjsdjshi@linux.vnet.ibm.com>
Acked-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Reviewed-by: Dong Jia Shi <bjsdjshi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
According to the platform specification, under certain conditions,
pending IO interruptions have to be cleared. Let's add an interface
for that.
Signed-off-by: Halil Pasic <pasic@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Memory hotplug can fail for some combinations of RAM and maxmem when
DDW is enabled in the presence of devices like nec-usb-xhci. DDW depends
on maximum addressable memory returned by guest and this value is currently
being calculated wrongly by the guest kernel routine memory_hotplug_max().
While there is an attempt to fix the guest kernel, this patch works
around the problem within QEMU itself.
memory_hotplug_max() routine in the guest kernel arrives at max
addressable memory by multiplying lmb-size with the lmb-count obtained
from ibm,dynamic-memory property. There are two assumptions here:
- All LMBs are part of ibm,dynamic memory: This is not true for PowerKVM
where only hot-pluggable LMBs are present in this property.
- The memory area comprising of RAM and hotplug region is contiguous: This
needn't be true always for PowerKVM as there can be gap between
boot time RAM and hotplug region.
To work around this guest kernel bug, ensure that ibm,dynamic-memory
has information about all the LMBs (RMA, boot-time LMBs, future
hotpluggable LMBs, and dummy LMBs to cover the gap between RAM and
hotpluggable region).
RMA is represented separately by memory@0 node. Hence mark RMA LMBs
and also the LMBs for the gap b/n RAM and hotpluggable region as
reserved and as having no valid DRC so that these LMBs are not considered
by the guest.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Nathan Fontenot <nfont@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This ensures that the underlying memory is marked dirty once the transfer
is complete and resolves cache coherency problems under MacOS 9.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
We need the PPC_FEATURE2_HAS_HTM bit in a subsequent patch, so
add the PowerPC AT_HWCAP2 definitions.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
OpenSSL's libcrypto always defines AES symbols with the same names as
qemu's local aes code. This is problematic when enabling at least curl
as that frequently also uses libcrypto. It might not be noticed when
running, but if you try to statically link, everything falls down.
An example snippet:
LINK qemu-nbd
.../libcrypto.a(aes-x86_64.o): In function 'AES_encrypt':
(.text+0x460): multiple definition of 'AES_encrypt'
crypto/aes.o:aes.c:(.text+0x670): first defined here
.../libcrypto.a(aes-x86_64.o): In function 'AES_decrypt':
(.text+0x9f0): multiple definition of 'AES_decrypt'
crypto/aes.o:aes.c:(.text+0xb30): first defined here
.../libcrypto.a(aes-x86_64.o): In function 'AES_cbc_encrypt':
(.text+0xf90): multiple definition of 'AES_cbc_encrypt'
crypto/aes.o:aes.c:(.text+0xff0): first defined here
collect2: error: ld returned 1 exit status
.../qemu-2.6.0/rules.mak:105: recipe for target 'qemu-nbd' failed
make: *** [qemu-nbd] Error 1
The aes.h header has redefines already for FreeBSD, but go ahead and
enable that for everyone since there's no real good reason to not use
a namespace all the time.
Signed-off-by: Mike Frysinger <vapier@chromium.org>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
This wrapper for machine_usb(current_machine) is not necessary,
replace all usages of usb_enabled() with machine_usb().
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alexander Graf <agraf@suse.de>
Cc: qemu-arm@nongnu.org
Cc: qemu-ppc@nongnu.org
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-id: 1465419025-21519-3-git-send-email-ehabkost@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Having a fixed-size hash table for keeping track of all translation blocks
is suboptimal: some workloads are just too big or too small to get maximum
performance from the hash table. The MRU promotion policy helps improve
performance when the hash table is a little undersized, but it cannot
make up for severely undersized hash tables.
Furthermore, frequent MRU promotions result in writes that are a scalability
bottleneck. For scalability, lookups should only perform reads, not writes.
This is not a big deal for now, but it will become one once MTTCG matures.
The appended fixes these issues by using qht as the implementation of
the TB hash table. This solution is superior to other alternatives considered,
namely:
- master: implementation in QEMU before this patchset
- xxhash: before this patch, i.e. fixed buckets + xxhash hashing + MRU.
- xxhash-rcu: fixed buckets + xxhash + RCU list + MRU.
MRU is implemented here by adding an intermediate struct
that contains the u32 hash and a pointer to the TB; this
allows us, on an MRU promotion, to copy said struct (that is not
at the head), and put this new copy at the head. After a grace
period, the original non-head struct can be eliminated, and
after another grace period, freed.
- qht-fixed-nomru: fixed buckets + xxhash + qht without auto-resize +
no MRU for lookups; MRU for inserts.
The appended solution is the following:
- qht-dyn-nomru: dynamic number of buckets + xxhash + qht w/ auto-resize +
no MRU for lookups; MRU for inserts.
The plots below compare the considered solutions. The Y axis shows the
boot time (in seconds) of a debian jessie image with arm-softmmu; the X axis
sweeps the number of buckets (or initial number of buckets for qht-autoresize).
The plots in PNG format (and with errorbars) can be seen here:
http://imgur.com/a/Awgnq
Each test runs 5 times, and the entire QEMU process is pinned to a
single core for repeatability of results.
Host: Intel Xeon E5-2690
28 ++------------+-------------+-------------+-------------+------------++
A***** + + + master **A*** +
27 ++ * xxhash ##B###++
| A******A****** xxhash-rcu $$C$$$ |
26 C$$ A******A****** qht-fixed-nomru*%%D%%%++
D%%$$ A******A******A*qht-dyn-mru A*E****A
25 ++ %%$$ qht-dyn-nomru &&F&&&++
B#####% |
24 ++ #C$$$$$ ++
| B### $ |
| ## C$$$$$$ |
23 ++ # C$$$$$$ ++
| B###### C$$$$$$ %%%D
22 ++ %B###### C$$$$$$C$$$$$$C$$$$$$C$$$$$$C$$$$$$C
| D%%%%%%B###### @E@@@@@@ %%%D%%%@@@E@@@@@@E
21 E@@@@@@E@@@@@@F&&&@@@E@@@&&&D%%%%%%B######B######B######B######B######B
+ E@@@ F&&& + E@ + F&&& + +
20 ++------------+-------------+-------------+-------------+------------++
14 16 18 20 22 24
log2 number of buckets
Host: Intel i7-4790K
14.5 ++------------+------------+-------------+------------+------------++
A** + + + master **A*** +
14 ++ ** xxhash ##B###++
13.5 ++ ** xxhash-rcu $$C$$$++
| qht-fixed-nomru %%D%%% |
13 ++ A****** qht-dyn-mru @@E@@@++
| A*****A******A****** qht-dyn-nomru &&F&&& |
12.5 C$$ A******A******A*****A****** ***A
12 ++ $$ A*** ++
D%%% $$ |
11.5 ++ %% ++
B### %C$$$$$$ |
11 ++ ## D%%%%% C$$$$$ ++
| # % C$$$$$$ |
10.5 F&&&&&&B######D%%%%% C$$$$$$C$$$$$$C$$$$$$C$$$$$C$$$$$$ $$$C
10 E@@@@@@E@@@@@@B#####B######B######E@@@@@@E@@@%%%D%%%%%D%%%###B######B
+ F&& D%%%%%%B######B######B#####B###@@@D%%% +
9.5 ++------------+------------+-------------+------------+------------++
14 16 18 20 22 24
log2 number of buckets
Note that the original point before this patch series is X=15 for "master";
the little sensitivity to the increased number of buckets is due to the
poor hashing function in master.
xxhash-rcu has significant overhead due to the constant churn of allocating
and deallocating intermediate structs for implementing MRU. An alternative
would be do consider failed lookups as "maybe not there", and then
acquire the external lock (tb_lock in this case) to really confirm that
there was indeed a failed lookup. This, however, would not be enough
to implement dynamic resizing--this is more complex: see
"Resizable, Scalable, Concurrent Hash Tables via Relativistic
Programming" by Triplett, McKenney and Walpole. This solution was
discarded due to the very coarse RCU read critical sections that we have
in MTTCG; resizing requires waiting for readers after every pointer update,
and resizes require many pointer updates, so this would quickly become
prohibitive.
qht-fixed-nomru shows that MRU promotion is advisable for undersized
hash tables.
However, qht-dyn-mru shows that MRU promotion is not important if the
hash table is properly sized: there is virtually no difference in
performance between qht-dyn-nomru and qht-dyn-mru.
Before this patch, we're at X=15 on "xxhash"; after this patch, we're at
X=15 @ qht-dyn-nomru. This patch thus matches the best performance that we
can achieve with optimum sizing of the hash table, while keeping the hash
table scalable for readers.
The improvement we get before and after this patch for booting debian jessie
with arm-softmmu is:
- Intel Xeon E5-2690: 10.5% less time
- Intel i7-4790K: 5.2% less time
We could get this same improvement _for this particular workload_ by
statically increasing the size of the hash table. But this would hurt
workloads that do not need a large hash table. The dynamic (upward)
resizing allows us to start small and enlarge the hash table as needed.
A quick note on downsizing: the table is resized back to 2**15 buckets
on every tb_flush; this makes sense because it is not guaranteed that the
table will reach the same number of TBs later on (e.g. most bootup code is
thrown away after boot); it makes sense to grow the hash table as
more code blocks are translated. This also avoids the complication of
having to build downsizing hysteresis logic into qht.
Reviewed-by: Sergey Fedorov <serge.fedorov@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1465412133-3029-15-git-send-email-cota@braap.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
This is a fast, scalable chained hash table with optional auto-resizing, allowing
reads that are concurrent with reads, and reads/writes that are concurrent
with writes to separate buckets.
A hash table with these features will be necessary for the scalability
of the ongoing MTTCG work; before those changes arrive we can already
benefit from the single-threaded speedup that qht also provides.
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1465412133-3029-11-git-send-email-cota@braap.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Sometimes it is useful to have a quick histogram to represent a certain
distribution -- for example, when investigating a performance regression
in a hash table due to inadequate hashing.
The appended allows us to easily represent a distribution using Unicode
characters. Further, the data structure keeping track of the distribution
is so simple that obtaining its values for off-line processing is trivial.
Example, taking the last 10 commits to QEMU:
Characters in commit title Count
-----------------------------------
39 1
48 1
53 1
54 2
57 1
61 1
67 1
78 1
80 1
qdist_init(&dist);
qdist_inc(&dist, 39);
[...]
qdist_inc(&dist, 80);
char *str = qdist_pr(&dist, 9, QDIST_PR_LABELS);
// -> [39.0,43.6)▂▂ █▂ ▂ ▄[75.4,80.0]
g_free(str);
char *str = qdist_pr(&dist, 4, QDIST_PR_LABELS);
// -> [39.0,49.2)▁█▁▁[69.8,80.0]
g_free(str);
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1465412133-3029-9-git-send-email-cota@braap.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
For some workloads such as arm bootup, tb_phys_hash is performance-critical.
The is due to the high frequency of accesses to the hash table, originated
by (frequent) TLB flushes that wipe out the cpu-private tb_jmp_cache's.
More info:
https://lists.nongnu.org/archive/html/qemu-devel/2016-03/msg05098.html
To dig further into this I modified an arm image booting debian jessie to
immediately shut down after boot. Analysis revealed that quite a bit of time
is unnecessarily spent in tb_phys_hash: the cause is poor hashing that
results in very uneven loading of chains in the hash table's buckets;
the longest observed chain had ~550 elements.
The appended addresses this with two changes:
1) Use xxhash as the hash table's hash function. xxhash is a fast,
high-quality hashing function.
2) Feed the hashing function with not just tb_phys, but also pc and flags.
This improves performance over using just tb_phys for hashing, since that
resulted in some hash buckets having many TB's, while others getting very few;
with these changes, the longest observed chain on a single hash bucket is
brought down from ~550 to ~40.
Tests show that the other element checked for in tb_find_physical,
cs_base, is always a match when tb_phys+pc+flags are a match,
so hashing cs_base is wasteful. It could be that this is an ARM-only
thing, though. UPDATE:
On Tue, Apr 05, 2016 at 08:41:43 -0700, Richard Henderson wrote:
> The cs_base field is only used by i386 (in 16-bit modes), and sparc (for a TB
> consisting of only a delay slot).
> It may well still turn out to be reasonable to ignore cs_base for hashing.
BTW, after this change the hash table should not be called "tb_hash_phys"
anymore; this is addressed later in this series.
This change gives consistent bootup time improvements. I tested two
host machines:
- Intel Xeon E5-2690: 11.6% less time
- Intel i7-4790K: 19.2% less time
Increasing the number of hash buckets yields further improvements. However,
using a larger, fixed number of buckets can degrade performance for other
workloads that do not translate as many blocks (600K+ for debian-jessie arm
bootup). This is dealt with later in this series.
Reviewed-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1465412133-3029-8-git-send-email-cota@braap.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
This will be used by upcoming changes for hashing the tb hash.
Add this into a separate file to include the copyright notice from
xxhash.
Reviewed-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1465412133-3029-7-git-send-email-cota@braap.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Signed-off-by: Guillaume Delbergue <guillaume.delbergue@greensocs.com>
[Rewritten. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[Emilio's additions: use TAS instead of atomic_xchg; emit acquire/release
barriers; return bool from trylock; call cpu_relax() while spinning;
optimize for uncontended locks by acquiring the lock with TAS instead
of TATAS; add qemu_spin_locked().]
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1465412133-3029-6-git-send-email-cota@braap.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
