Halil Pasic [Thu, 21 Sep 2017 18:08:41 +0000 (20:08 +0200)]
s390x/css: support ccw IDA
Let's add indirect data addressing support for our virtual channel
subsystem. This implementation does not bother with any kind of
prefetching. We simply step through the IDAL on demand.
Halil Pasic [Thu, 21 Sep 2017 18:08:40 +0000 (20:08 +0200)]
390x/css: introduce maximum data address checking
The architecture mandates the addresses to be accessed on the first
indirection level (that is, the data addresses without IDA, and the
(M)IDAW addresses with (M)IDA) to be checked against an CCW format
dependent limit maximum address. If a violation is detected, the storage
access is not to be performed and a channel program check needs to be
generated. As of today, we fail to do this check.
Let us stick even closer to the architecture specification.
Halil Pasic [Thu, 21 Sep 2017 18:08:39 +0000 (20:08 +0200)]
virtio-ccw: use ccw data stream
Replace direct access which implicitly assumes no IDA
or MIDA with the new ccw data stream interface which should
cope with these transparently in the future.
Note that checking the return code for ccw_dstream_* will be
done in a follow-on patch.
Halil Pasic [Thu, 21 Sep 2017 18:08:38 +0000 (20:08 +0200)]
s390x/css: use ccw data stream
Replace direct access which implicitly assumes no IDA
or MIDA with the new ccw data stream interface which should
cope with these transparently in the future.
Note that checking the return code for ccw_dstream_* will be
done in a follow-on patch.
Halil Pasic [Thu, 21 Sep 2017 18:08:37 +0000 (20:08 +0200)]
s390x/css: introduce css data stream
This is a preparation for introducing handling for indirect data
addressing and modified indirect data addressing (CCW). Here we introduce
an interface which should make the addressing scheme transparent for the
client code. Here we implement only the basic scheme (no IDA or MIDA).
env->psa is a 64bit value, while we copy 4 bytes into the save area,
resulting always in 0 getting stored.
Let's try to reduce such errors by using a proper structure. While at
it, use correct cpu->be conversion (and get_psw_mask()), as we will be
reusing this code for TCG soon.
The STFLE bits for the MSA (extension) facilities simply indicate that
the respective instructions can be executed. The QUERY subfunction can then
be used to identify which features exactly are available.
Availability of subfunctions can also vary on real hardware. For now, we
simply implement a CPU model without any available subfunctions except
QUERY (which is always around).
As all MSA functions behave quite similarly, we can use one translation
handler for now. Prepare the code for implementation of actual subfunctions.
At least MSA is helpful for now, as older Linux kernels require this
facility when compiled for a z9 model. Allow to enable the facilities
for the qemu cpu model.
Peter Maydell [Thu, 5 Oct 2017 15:13:46 +0000 (16:13 +0100)]
Merge remote-tracking branch 'remotes/dgilbert/tags/pull-hmp-20171005' into staging
HMP pull 2017-10-05
# gpg: Signature made Thu 05 Oct 2017 11:50:13 BST
# gpg: using RSA key 0x0516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <[email protected]>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 45F5 C71B 4A0C B7FB 977A 9FA9 0516 331E BC5B FDE7
* remotes/dgilbert/tags/pull-hmp-20171005:
hmp-commands-info: Change "@findex FOO" to "@findex info FOO"
hmp-commands-info: Move Texinfo stanzas to conventional place
hmp-commands-info: Fix "info rocker-FOO" misspellings
hmp: Fix unknown command for subtable
hmp: Missing handle_errors
Peter Maydell [Thu, 5 Oct 2017 13:44:12 +0000 (14:44 +0100)]
Merge remote-tracking branch 'remotes/berrange/tags/pull-qio-2017-10-04-1' into staging
Merge qio 2017/10/04 v1
# gpg: Signature made Wed 04 Oct 2017 13:23:04 BST
# gpg: using RSA key 0xBE86EBB415104FDF
# gpg: Good signature from "Daniel P. Berrange <[email protected]>"
# gpg: aka "Daniel P. Berrange <[email protected]>"
# Primary key fingerprint: DAF3 A6FD B26B 6291 2D0E 8E3F BE86 EBB4 1510 4FDF
* remotes/berrange/tags/pull-qio-2017-10-04-1:
io: add trace events for websockets frame handling
io: Attempt to send websocket close messages to client
io: Reply to ping frames
io: Ignore websocket PING and PONG frames
io: Allow empty websocket payload
io: Add support for fragmented websocket binary frames
io: Small updates in preparation for websocket changes
ui: Always remove an old VNC channel watch before adding a new one
io: use case insensitive check for Connection & Upgrade websock headers
io: include full error message in websocket handshake trace
io: send proper HTTP response for websocket errors
* remotes/awilliam/tags/vfio-updates-20171003.0:
vfio/pci: Add NVIDIA GPUDirect Cliques support
vfio/pci: Add virtual capabilities quirk infrastructure
vfio/pci: Do not unwind on error
Peter Maydell [Thu, 5 Oct 2017 11:02:21 +0000 (12:02 +0100)]
Merge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging
# gpg: Signature made Tue 03 Oct 2017 19:53:34 BST
# gpg: using RSA key 0x9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <[email protected]>"
# gpg: aka "Stefan Hajnoczi <[email protected]>"
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35 775A 9CA4 ABB3 81AB 73C8
* remotes/stefanha/tags/block-pull-request:
aio: fix assert when remove poll during destroy
iothread: delay the context release to finalize
iothread: export iothread_stop()
iothread: provide helpers for internal use
qom: provide root container for internal objs
io: Attempt to send websocket close messages to client
Make a best effort attempt to close websocket connections according to
the RFC. Sends the close message, as room permits in the socket buffer,
and immediately closes the socket.
io: Add support for fragmented websocket binary frames
Allows fragmented binary frames by saving the previous opcode. Handles
the case where an intermediary (i.e., web proxy) fragments frames
originally sent unfragmented by the client.
io: Small updates in preparation for websocket changes
Gets rid of unnecessary bit shifting and performs proper EOF checking to
avoid a large number of repeated calls to recvmsg() when a client
abruptly terminates a connection (bug fix).
io: use case insensitive check for Connection & Upgrade websock headers
When checking the value of the Connection and Upgrade HTTP headers
the websock RFC (6455) requires the comparison to be case insensitive.
The Connection value should be an exact match not a substring.
io: send proper HTTP response for websocket errors
When any error occurs while processing the websockets handshake,
QEMU just terminates the connection abruptly. This is in violation
of the HTTP specs and does not help the client understand what they
did wrong. This is particularly bad when the client gives the wrong
path, as a "404 Not Found" would be very helpful.
Refactor the handshake code so that it always sends a response to
the client unless there was an I/O error.
Alex Williamson [Tue, 29 Aug 2017 22:05:47 +0000 (16:05 -0600)]
vfio/pci: Add NVIDIA GPUDirect Cliques support
NVIDIA has defined a specification for creating GPUDirect "cliques",
where devices with the same clique ID support direct peer-to-peer DMA.
When running on bare-metal, tools like NVIDIA's p2pBandwidthLatencyTest
(part of cuda-samples) determine which GPUs can support peer-to-peer
based on chipset and topology. When running in a VM, these tools have
no visibility to the physical hardware support or topology. This
option allows the user to specify hints via a vendor defined
capability. For instance:
This enables two cliques. The first is a singleton clique with ID 0,
for the first hostdev defined in the XML (note that since cliques
define peer-to-peer sets, singleton clique offer no benefit). The
subsequent two hostdevs are both added to clique ID 1, indicating
peer-to-peer is possible between these devices.
QEMU only provides validation that the clique ID is valid and applied
to an NVIDIA graphics device, any validation that the resulting
cliques are functional and valid is the user's responsibility. The
NVIDIA specification allows a 4-bit clique ID, thus valid values are
0-15.
If the hypervisor needs to add purely virtual capabilties, give us a
hook through quirks to do that. Note that we determine the maximum
size for a capability based on the physical device, if we insert a
virtual capability, that can change. Therefore if maximum size is
smaller after added virt capabilities, use that.
Alex Williamson [Tue, 29 Aug 2017 22:05:32 +0000 (16:05 -0600)]
vfio/pci: Do not unwind on error
If vfio_add_std_cap() errors then going to out prepends irrelevant
errors for capabilities we haven't attempted to add as we unwind our
recursive stack. Just return error.
Fixes: 7ef165b9a8d9 ("vfio/pci: Pass an error object to vfio_add_capabilities") Signed-off-by: Alex Williamson <[email protected]>
The problem is that g_source_remove_poll() does not allow to remove one
source from array if the source is detached from its owner
context. (peterx: which IMHO does not make much sense)
Fix it on QEMU side by avoid calling g_source_remove_poll() if we know
the object is during destruction, and we won't leak anything after all
since the array will be gone soon cleanly even with that fd.
Peter Xu [Thu, 28 Sep 2017 02:59:57 +0000 (10:59 +0800)]
iothread: delay the context release to finalize
When gcontext is used with iothread, the context will be destroyed
during iothread_stop(). That's not good since sometimes we would like
to keep the resources until iothread is destroyed, but we may want to
stop the thread before that point.
Delay the destruction of gcontext to iothread finalize. Then we can do:
We may need this patch if we want to run chardev IOs in iothreads and
hopefully clean them up correctly. For more specific information,
please see 2b316774f6 ("qemu-char: do not operate on sources from
finalize callbacks").
Peter Xu [Thu, 28 Sep 2017 02:59:56 +0000 (10:59 +0800)]
iothread: export iothread_stop()
So that internal iothread users can explicitly stop one iothread without
destroying it.
Since at it, fix iothread_stop() to allow it to be called multiple
times. Before this patch we may call iothread_stop() more than once on
single iothread, while that may not be correct since qemu_thread_join()
is not allowed to run twice. From manual of pthread_join():
Joining with a thread that has previously been joined results in
undefined behavior.
Peter Xu [Thu, 28 Sep 2017 02:59:55 +0000 (10:59 +0800)]
iothread: provide helpers for internal use
IOThread is a general framework that contains IO loop environment and a
real thread behind. It's also good to be used internally inside qemu.
Provide some helpers for it to create iothreads to be used internally.
Put all the internal used iothreads into the internal object container.
Peter Xu [Thu, 28 Sep 2017 02:59:54 +0000 (10:59 +0800)]
qom: provide root container for internal objs
We have object_get_objects_root() to keep user created objects, however
no place for objects that will be used internally. Create such a
container for internal objects.
* remotes/bonzini/tags/for-upstream:
kvmclock: use the updated system_timer_msr
kvm: check KVM_CAP_NR_VCPUS with kvm_vm_check_extension()
kvm: check KVM_CAP_SYNC_MMU with kvm_vm_check_extension()
linux-headers: sync against v4.14-rc1
iothread: Make iothread_stop() idempotent
scsi: Ignore executable for in-tree builds
Peter Maydell [Tue, 3 Oct 2017 14:11:00 +0000 (15:11 +0100)]
Merge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2017-10-02' into staging
QAPI patches for 2017-10-02
# gpg: Signature made Mon 02 Oct 2017 12:09:32 BST
# gpg: using RSA key 0x3870B400EB918653
# gpg: Good signature from "Markus Armbruster <[email protected]>"
# gpg: aka "Markus Armbruster <[email protected]>"
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867 4E5F 3870 B400 EB91 8653
* remotes/armbru/tags/pull-qapi-2017-10-02:
watchdog: Allow setting action on the fly
watchdog.h: Drop local redefinition of actions enum
qapi: Rename WatchdogExpirationAction enum
Peter Maydell [Tue, 3 Oct 2017 11:43:03 +0000 (12:43 +0100)]
Merge remote-tracking branch 'remotes/famz/tags/docker-testing-pull-request' into staging
# gpg: Signature made Fri 29 Sep 2017 04:17:37 BST
# gpg: using RSA key 0xCA35624C6A9171C6
# gpg: Good signature from "Fam Zheng <[email protected]>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 5003 7CB7 9706 0F76 F021 AD56 CA35 624C 6A91 71C6
* remotes/famz/tags/docker-testing-pull-request:
docker: Don't mount ccache db if NOUSER=1
docker: test-block: Don't continue if build fails
tests/docker/run: don't source /etc/profile
docker: Fix test-mingw
docker: add installation to build tests
Jim Somerville [Fri, 29 Sep 2017 16:00:19 +0000 (12:00 -0400)]
kvmclock: use the updated system_timer_msr
Fixes e2b6c17 (kvmclock: update system_time_msr address forcibly)
which makes a call to get the latest value of the address
stored in system_timer_msr, but then uses the old address anyway.
Greg Kurz [Thu, 21 Sep 2017 16:01:02 +0000 (18:01 +0200)]
kvm: check KVM_CAP_NR_VCPUS with kvm_vm_check_extension()
On a modern server-class ppc host with the following CPU topology:
Architecture: ppc64le
Byte Order: Little Endian
CPU(s): 32
On-line CPU(s) list: 0,8,16,24
Off-line CPU(s) list: 1-7,9-15,17-23,25-31
Thread(s) per core: 1
If both KVM PR and KVM HV loaded and we pass:
-machine pseries,accel=kvm,kvm-type=PR -smp 8
We expect QEMU to warn that this exceeds the number of online CPUs:
Warning: Number of SMP cpus requested (8) exceeds the recommended
cpus supported by KVM (4)
Warning: Number of hotpluggable cpus requested (8) exceeds the
recommended cpus supported by KVM (4)
but nothing is printed...
This happens because on ppc the KVM_CAP_NR_VCPUS capability is VM
specific ndreally depends on the KVM type, but we currently use it
as a global capability. And KVM returns a fallback value based on
KVM HV being present. Maybe KVM on POWER shouldn't presume anything
as long as it doesn't have a VM, but in all cases, we should call
KVM_CREATE_VM first and use KVM_CAP_NR_VCPUS as a VM capability.
This patch hence changes kvm_recommended_vcpus() accordingly and
moves the sanity checking of smp_cpus after the VM creation.
It is okay for the other archs that also implement KVM_CAP_NR_VCPUS,
ie, mips, s390, x86 and arm, because they don't depend on the VM
being created or not.
Greg Kurz [Thu, 21 Sep 2017 16:00:53 +0000 (18:00 +0200)]
kvm: check KVM_CAP_SYNC_MMU with kvm_vm_check_extension()
On a server-class ppc host, this capability depends on the KVM type,
ie, HV or PR. If both KVM are present in the kernel, we will always
get the HV specific value, even if we explicitely requested PR on
the command line.
This can have an impact if we're using hugepages or a balloon device.
Since we've already created the VM at the time any user calls
kvm_has_sync_mmu(), switching to kvm_vm_check_extension() is
enough to fix any potential issue.
It is okay for the other archs that also implement KVM_CAP_SYNC_MMU,
ie, mips, s390, x86 and arm, because they don't depend on the VM being
created or not.
While here, let's cache the state of this extension in a bool variable,
since it has several users in the code, as suggested by Thomas Huth.
Currently, the only time that users can set watchdog action is at
the start as all we expose is this -watchdog-action command line
argument. This is suboptimal when users want to plug the device
later via monitor. Alternatively, they might want to change the
action for already existing device on the fly.
watchdog.h: Drop local redefinition of actions enum
We already have enum that enumerates all the actions that a
watchdog can take when hitting its timeout: WatchdogAction.
Use that instead of inventing our own.
Thomas Huth [Wed, 27 Sep 2017 15:28:26 +0000 (17:28 +0200)]
hw/usb/bus: Remove bad object_unparent() from usb_try_create_simple()
Valgrind detects an invalid read operation when hot-plugging of an
USB device fails:
$ valgrind x86_64-softmmu/qemu-system-x86_64 -device usb-ehci -nographic -S
==30598== Memcheck, a memory error detector
==30598== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==30598== Using Valgrind-3.12.0 and LibVEX; rerun with -h for copyright info
==30598== Command: x86_64-softmmu/qemu-system-x86_64 -device usb-ehci -nographic -S
==30598==
QEMU 2.10.50 monitor - type 'help' for more information
(qemu) device_add usb-tablet
(qemu) device_add usb-tablet
(qemu) device_add usb-tablet
(qemu) device_add usb-tablet
(qemu) device_add usb-tablet
(qemu) device_add usb-tablet
==30598== Invalid read of size 8
==30598== at 0x60EF50: object_unparent (object.c:445)
==30598== by 0x580F0D: usb_try_create_simple (bus.c:346)
==30598== by 0x581BEB: usb_claim_port (bus.c:451)
==30598== by 0x582310: usb_qdev_realize (bus.c:257)
==30598== by 0x4CB399: device_set_realized (qdev.c:914)
==30598== by 0x60E26D: property_set_bool (object.c:1886)
==30598== by 0x61235E: object_property_set_qobject (qom-qobject.c:27)
==30598== by 0x61000F: object_property_set_bool (object.c:1162)
==30598== by 0x4567C3: qdev_device_add (qdev-monitor.c:630)
==30598== by 0x456D52: qmp_device_add (qdev-monitor.c:807)
==30598== by 0x470A99: hmp_device_add (hmp.c:1933)
==30598== by 0x3679C3: handle_hmp_command (monitor.c:3123)
The object_unparent() here is not necessary anymore since commit 69382d8b3e8600b3 ("qdev: Fix object reference leak in case device.realize()
fails"), so let's remove it now.
Currently, iothread_stop_all() makes all iothread objects unsafe
to be destroyed, because qemu_thread_join() ends up being called
twice.
To fix this, make iothread_stop() idempotent by checking
thread->stopped.
Fixes the following crash:
qemu-system-x86_64 -object iothread,id=iothread0 -monitor stdio -display none
QEMU 2.10.50 monitor - type 'help' for more information
(qemu) quit
qemu: qemu_thread_join: No such process
Aborted (core dumped)
Handle the translation from vga chars to curses chars in curses_update()
instead of console_write_ch(). Purge any curses support bits from
ui/console.h include file.
With NOUSER=1 the container runs code as root, which may create
privileged files that will not be be accssible next time. Skip ccache
dir mount in this case.
Alex Bennée [Tue, 26 Sep 2017 13:36:22 +0000 (14:36 +0100)]
tests/docker/run: don't source /etc/profile
The usual behaviour of /etc/profile is to set the default PATH for
users. This runs into problems when we have updated PATH in our
dockerfile e.g. to access a cross-compiler in a non-standard
location. It shouldn't be needed anyway as we inherit the env from the
image when it was setup.
Feature "dtc" is explicitly required by test-mingw, but is not detected
by the run script since we switched to archive-source.sh in b7f404201e4.
Since it isn't available in the Fedora image which runs this test on
patchew, the way we get dtc is still from submodule.
archive-source.sh takes care of bundling the submodule files already, so
what we need to do is just checking if files are there. Makefile is
chosen because it is one that is unlikely to get renamed in the future.
Peter Maydell [Wed, 27 Sep 2017 21:44:51 +0000 (22:44 +0100)]
Merge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20170927a' into staging
Migration pull 2017-09-27
# gpg: Signature made Wed 27 Sep 2017 14:56:23 BST
# gpg: using RSA key 0x0516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <[email protected]>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 45F5 C71B 4A0C B7FB 977A 9FA9 0516 331E BC5B FDE7
* remotes/dgilbert/tags/pull-migration-20170927a:
migration: Route more error paths
migration: Route errors up through vmstate_save
migration: wire vmstate_save_state errors up to vmstate_subsection_save
migration: Check field save returns
migration: check pre_save return in vmstate_save_state
migration: pre_save return int
migration: disable auto-converge during bulk block migration
Peter Maydell [Wed, 27 Sep 2017 17:20:31 +0000 (18:20 +0100)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.11-20170927' into staging
ppc patch queue 2017-09-27
Contains
* a number of Mac machine type fixes
* a number of embedded machine type fixes (preliminary to adding the
Sam460ex board)
* a important fix for handling of migration with KVM PR
* assorted other minor fixes and cleanups
* remotes/dgibson/tags/ppc-for-2.11-20170927: (26 commits)
macio: use object link between MACIO_IDE and MAC_DBDMA object
macio: pass channel into MACIOIDEState via qdev property
mac_dbdma: remove DBDMA_init() function
mac_dbdma: QOMify
mac_dbdma: remove unused IO fields from DBDMAState
spapr: fix the value of SDR1 in kvmppc_put_books_sregs()
ppc/pnv: check for OPAL firmware file presence
ppc: remove all unused CPU definitions
ppc: remove unused CPU definitions
spapr_pci: make index property mandatory
macio: convert pmac_ide_ops from old_mmio
ppc/pnv: Improve macro parenthesization
spapr: introduce helpers to migrate HPT chunks and the end marker
ppc/kvm: generalize the use of kvmppc_get_htab_fd()
ppc/kvm: change kvmppc_get_htab_fd() to return -errno on error
ppc: Fix OpenPIC model
ppc/ide/macio: Add missing registers
ppc/mac: More rework of the DBDMA emulation
ppc/mac: Advertise a high clock frequency for NewWorld Macs
ppc: QOMify g3beige machine
...
Peter Maydell [Wed, 27 Sep 2017 15:48:39 +0000 (16:48 +0100)]
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Block layer patches
# gpg: Signature made Tue 26 Sep 2017 14:52:32 BST
# gpg: using RSA key 0x7F09B272C88F2FD6
# gpg: Good signature from "Kevin Wolf <[email protected]>"
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74 56FE 7F09 B272 C88F 2FD6
* remotes/kevin/tags/for-upstream: (24 commits)
block/qcow2-bitmap: fix use of uninitialized pointer
qemu-iotests: add shrinking image test
qcow2: add shrink image support
qcow2: add qcow2_cache_discard
qemu-img: add --shrink flag for resize
iotests: fix 181: enable postcopy-ram capability on target
qemu-iotests: Test change-backing-file command
block: Fix permissions after bdrv_reopen()
block: reopen: Queue children after their parents
block: Base permissions on rw state after reopen
block: Add reopen queue to bdrv_check_perm()
block: Add reopen_queue to bdrv_child_perm()
qemu-io: Drop write permissions before read-only reopen
block: Clean up some bad code in the vvfat driver
block/throttle-groups.c: allocate RestartData on the heap
throttle: Assert that bkt->max is valid in throttle_compute_wait()
iotests: Print full path of bad output if mismatch
iotests: use virtio aliases for 067
iotests: use -ccw on s390x for 051
iotests: use -ccw on s390x for 040, 139, and 182
...
vmstate_save_state is called in lots of places.
Route error returns from the easier cases back up; there are lots
of more complex cases where their own error paths need fixing.
Modify the pre_save method on VMStateDescription to return an int
rather than void so that it potentially can fail.
Changed zillions of devices to make them return 0; the only
case I've made it return non-0 is hw/intc/s390_flic_kvm.c that already
had an error_report/return case.
Note: If you add an error exit in your pre_save you must emit
an error_report to say why.
Peter Lieven [Tue, 26 Sep 2017 10:33:16 +0000 (12:33 +0200)]
migration: disable auto-converge during bulk block migration
auto-converge and block migration currently do not play well together.
During block migration the auto-converge logic detects that ram
migration makes no progress and thus throttles down the vm until
it nearly stalls completely. Avoid this by disabling the throttling
logic during the bulk phase of the block migration.
Mark Cave-Ayland [Sun, 24 Sep 2017 14:47:44 +0000 (15:47 +0100)]
macio: use object link between MACIO_IDE and MAC_DBDMA object
Using a standard QOM object link we can pass a reference to the MAC_DBDMA
controller to the MACIO_IDE object which removes the last external parameter
to macio_ide_register_dma().
Mark Cave-Ayland [Sun, 24 Sep 2017 14:47:43 +0000 (15:47 +0100)]
macio: pass channel into MACIOIDEState via qdev property
One of the reasons macio_ide_register_dma() needs to exist is because the
channel id isn't passed into the MACIO_IDE object. Pass in the channel id
using a qdev property to remove this requirement.
Mark Cave-Ayland [Sun, 24 Sep 2017 14:47:42 +0000 (15:47 +0100)]
mac_dbdma: remove DBDMA_init() function
Instead we can now instantiate the MAC_DBDMA object directly within the
macio device. We also add the DBDMA device as a child property so that
it is possible to retrieve later.
Mark Cave-Ayland [Sun, 24 Sep 2017 14:47:40 +0000 (15:47 +0100)]
mac_dbdma: remove unused IO fields from DBDMAState
These fields were used to manually handle IO requests that weren't aligned
to a sector boundary before this feature was supported by the block API.
Once the block API changed to support byte-aligned IO requests, the macio
controller was switched over to use it in commit be1e343 but these fields
were accidentally left behind. Remove them, including the initialisation
in DBDMA_init().
Greg Kurz [Mon, 25 Sep 2017 11:00:02 +0000 (13:00 +0200)]
spapr: fix the value of SDR1 in kvmppc_put_books_sregs()
When running with KVM PR, if a new HPT is allocated we need to inform
KVM about the HPT address and size. This is currently done by hacking
the value of SDR1 and pushing it to KVM in several places.
Also, migration breaks the guest since it is very unlikely the HPT has
the same address in source and destination, but we push the incoming
value of SDR1 to KVM anyway.
This patch introduces a new virtual hypervisor hook so that the spapr
code can provide the correct value of SDR1 to be pushed to KVM each
time kvmppc_put_books_sregs() is called.
It allows to get rid of all the hacking in the spapr/kvmppc code and
it fixes migration of nested KVM PR.
Greg Kurz [Wed, 20 Sep 2017 14:46:20 +0000 (16:46 +0200)]
spapr_pci: make index property mandatory
PHBs can be created with an index property, in which case the machine
code automatically sets all the MMIO windows at addresses derived from
the index. Alternatively, they can be manually created without index,
but the user has to provide addresses for all MMIO windows.
The non-index way happens to be more trouble than it's worth: it's
difficult to use, keeps requiring (potentially incompatible) changes
when some new parameter needs adding, and is awkward to check for
collisions. It currently even has a bug that prevents to use two
non-index PHBs because their child DRCs are all derived from the
same index == -1 value, and, thus, collide.
This patch hence makes the index property mandatory. As a consequence,
the PHB's memory regions and BUID are now always configured according
to the index, and it is no longer possible to set them from the command
line.
This DOES BREAK backwards compat, but we don't think the non-index
PHB feature was used in practice (at least libvirt doesn't) and the
simplification is worth it.
Eric Blake [Tue, 19 Sep 2017 14:16:03 +0000 (09:16 -0500)]
ppc/pnv: Improve macro parenthesization
Although none of the existing macro call-sites were broken,
it's always better to write macros that properly parenthesize
arguments that can be complex expressions, so that the intended
order of operations is not broken.
Greg Kurz [Fri, 15 Sep 2017 13:16:20 +0000 (15:16 +0200)]
ppc/kvm: generalize the use of kvmppc_get_htab_fd()
The use of KVM_PPC_GET_HTAB_FD is open-coded in kvmppc_read_hptes()
and kvmppc_write_hpte().
This patch modifies kvmppc_get_htab_fd() so that it can be used
everywhere we need to access the in-kernel htab:
- add an index argument
=> only kvmppc_read_hptes() passes an actual index, all other users
pass 0
- add an errp argument to propagate error messages to the caller.
=> spapr migration code prints the error
=> hpte helpers pass &error_abort to keep the current behavior
of hw_error()
While here, this also fixes a bug in kvmppc_write_hpte() so that it
opens the htab fd for writing instead of reading as it currently does.
This never broke anything because we currently never call this code,
as explained in the changelog of commit c1385933804bb:
"This support updating htab managed by the hypervisor. Currently
we don't have any user for this feature. This actually bring the
store_hpte interface in-line with the load_hpte one. We may want
to use this when we want to emulate henter hcall in qemu for HV
kvm."
Greg Kurz [Fri, 15 Sep 2017 13:16:10 +0000 (15:16 +0200)]
ppc/kvm: change kvmppc_get_htab_fd() to return -errno on error
When kvmppc_get_htab_fd() fails, its return value is propagated up to
qemu_savevm_state_iterate() or to qemu_savevm_state_complete_precopy().
All savevm handlers expect to receive a negative errno on error.
Let's patch kvmppc_get_htab_fd() accordingly.
While here, let's change htab_load() in the spapr code to also
propagate the error, since it doesn't make sense to abort() if
we couldn't get the htab fd from KVM.
The timing register exists on all variants of MacIO IDE, we just
store and return its value.
The interrupts register only exists on KeyLargo but it doesn't
hurt to have it. The lack of this register causes MacOS X to
hangs under some circumstances.
Both are 32-bit only. The HW might support smaller access sizes
but no known OS uses them.
Because the core IDE subsystem doesn't provide us with a way
to query the main (level) interrupt state, nor do we have a way
to know that DBDMA issued a (edge) interrupt, we reflect both
through a private pair of qirq's in order to maintain the
register state.
projects / qemu.git / log