OSDN Git Service

qmiga/qemu.git
7 years agoMerge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20170124b' into...
Peter Maydell [Tue, 24 Jan 2017 19:25:19 +0000 (19:25 +0000)]
Merge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20170124b' into staging

Migration

1 My maintainer change
2 Jianjun's qtailq
3 Ashijeet's only-migratable
4 Zhanghailiang's re-active images
5 Pankaj's change name of migration thread
6 My PCI migration merge
7 Juan's debug to tracing
8 My tracing on save

# gpg: Signature made Tue 24 Jan 2017 18:39:35 GMT
# gpg:                using RSA key 0x0516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <dgilbert@redhat.com>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication 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-20170124b:
  migration/tracing: Add tracing on save
  migration: transform remaining DPRINTF into trace_
  PCI/migration merge vmstate_pci_device and vmstate_pcie_device
  migration: Change name of live migration thread
  migration: re-active images while migration been canceled after inactive them
  migration: Fail migration blocker for --only-migratable
  migration: disallow migrate_add_blocker during migration
  migration: Allow "device add" options to only add migratable devices
  migration: Add a new option to enable only-migratable
  block/vvfat: Remove the undesirable comment
  migration: add error_report
  tests/migration: Add test for QTAILQ migration
  migration: migrate QTAILQ
  migration: extend VMStateInfo
  MAINTAINERS: Add myself as a migration submaintainer

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agomigration/tracing: Add tracing on save
Dr. David Alan Gilbert [Mon, 12 Dec 2016 12:58:38 +0000 (12:58 +0000)]
migration/tracing: Add tracing on save

Add some tracing to vmstate_subsection_save and vmstate_save_state
to help in debugging when you're not sure if a conditional piece
of data is being saved.

In vmstate_subsection_save I renamed the inner vmsd to avoid the aliasing
and be able to print both names.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20161212125838.14425-1-dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: transform remaining DPRINTF into trace_
Juan Quintela [Mon, 23 Jan 2017 21:32:05 +0000 (22:32 +0100)]
migration: transform remaining DPRINTF into trace_

So we can remove DPRINTF() macro

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <1485207141-1941-2-git-send-email-quintela@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
  Fixed up 'remained/remaining' as requested by Eric

7 years agoPCI/migration merge vmstate_pci_device and vmstate_pcie_device
Dr. David Alan Gilbert [Wed, 14 Dec 2016 19:58:29 +0000 (19:58 +0000)]
PCI/migration merge vmstate_pci_device and vmstate_pcie_device

The vmstate_pci_device and vmstate_pcie_devices differ
just in the size of one buffer; combine the two using a _TEST
macro.

I think this is safe as long as everywhere which currently
uses either of these two uses the right type.

One thing that concerns me is that some places use pci_device_load/save
which does some irq mangling, but others just use the VMSTATE_PCI_DEVICE
macro - how are they getting the same irq mangling?

This passes a smoke test migrate of:
./x86_64-softmmu/qemu-system-x86_64 -M pc,accel=kvm -m 1024
./littlefed20.img -device e1000e -device virtio-net -device
e1000 -device virtio-rng -device megasas -device megasas-gen2 -device
ioh3420 -device nec-usb-xhci

to an unmodified qemu.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20161214195829.18241-1-dgilbert@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: Change name of live migration thread
Pankaj Gupta [Mon, 23 Jan 2017 13:42:56 +0000 (19:12 +0530)]
migration: Change name of live migration thread

Change the name of live migration thread from 'migration'
to 'live_migration' to identify it clearly. 'migration'
is a generic word and kernel also has  tasks for process
migration with the name 'migration/cpu#'.

Signed-off-by: Pankaj Gupta <pagupta@redhat.com>
Message-Id: <1485178976-15225-1-git-send-email-pagupta@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: re-active images while migration been canceled after inactive them
zhanghailiang [Tue, 24 Jan 2017 07:59:52 +0000 (15:59 +0800)]
migration: re-active images while migration been canceled after inactive them

commit fe904ea8242cbae2d7e69c052c754b8f5f1ba1d6 fixed a case
which migration aborted QEMU because it didn't regain the control
of images while some errors happened.

Actually, there are another two cases can trigger the same error reports:
" bdrv_co_do_pwritev: Assertion `!(bs->open_flags & 0x0800)' failed",

Case 1, codes path:
migration_thread()
    migration_completion()
        bdrv_inactivate_all() ----------------> inactivate images
        qemu_savevm_state_complete_precopy()
            socket_writev_buffer() --------> error because destination fails
                qemu_fflush() ----------------> set error on migration stream
-> qmp_migrate_cancel() ----------------> user cancelled migration concurrently
    -> migrate_set_state() ------------------> set migrate CANCELLIN
    migration_completion() -----------------> go on to fail_invalidate
if (s->state == MIGRATION_STATUS_ACTIVE) -> Jump this branch

Case 2, codes path:
migration_thread()
    migration_completion()
        bdrv_inactivate_all() ----------------> inactivate images
    migreation_completion() finished
-> qmp_migrate_cancel() ---------------> user cancelled migration concurrently
    qemu_mutex_lock_iothread();
    qemu_bh_schedule (s->cleanup_bh);

As we can see from above, qmp_migrate_cancel can slip in whenever
migration_thread does not hold the global lock. If this happens after
bdrv_inactive_all() been called, the above error reports will appear.

To prevent this, we can call bdrv_invalidate_cache_all() in qmp_migrate_cancel()
directly if we find images become inactive.

Besides, bdrv_invalidate_cache_all() in migration_completion() doesn't have the
protection of big lock, fix it by add the missing qemu_mutex_lock_iothread();

Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Message-Id: <1485244792-11248-1-git-send-email-zhang.zhanghailiang@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: Fail migration blocker for --only-migratable
Ashijeet Acharya [Mon, 16 Jan 2017 11:31:54 +0000 (17:01 +0530)]
migration: Fail migration blocker for --only-migratable

migrate_add_blocker should rightly fail if the '--only-migratable'
option was specified and the device in use should not be able to
perform the action which results in an unmigratable VM.

Make migrate_add_blocker return -EACCES in this case.

Signed-off-by: Ashijeet Acharya <ashijeetacharya@gmail.com>
Message-Id: <1484566314-3987-6-git-send-email-ashijeetacharya@gmail.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: disallow migrate_add_blocker during migration
Ashijeet Acharya [Mon, 16 Jan 2017 11:31:53 +0000 (17:01 +0530)]
migration: disallow migrate_add_blocker during migration

If a migration is already in progress and somebody attempts
to add a migration blocker, this should rightly fail.

Add an errp parameter and a retcode return value to migrate_add_blocker.

Signed-off-by: John Snow <jsnow@redhat.com>
Signed-off-by: Ashijeet Acharya <ashijeetacharya@gmail.com>
Message-Id: <1484566314-3987-5-git-send-email-ashijeetacharya@gmail.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Acked-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
  Merged with recent 'Allow invtsc migration' change

7 years agomigration: Allow "device add" options to only add migratable devices
Ashijeet Acharya [Mon, 16 Jan 2017 11:31:52 +0000 (17:01 +0530)]
migration: Allow "device add" options to only add migratable devices

Introduce checks for the unmigratable flag in the VMStateDescription
structs of respective devices when user attempts to add them. If the
"--only-migratable" was specified, all unmigratable devices will
rightly fail to add. This feature is made compatible for both "-device"
and "-usbdevice" command line options and covers their hmp and qmp
counterparts as well.

Signed-off-by: Ashijeet Acharya <ashijeetacharya@gmail.com>
Message-Id: <1484566314-3987-4-git-send-email-ashijeetacharya@gmail.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: Add a new option to enable only-migratable
Ashijeet Acharya [Mon, 16 Jan 2017 11:31:51 +0000 (17:01 +0530)]
migration: Add a new option to enable only-migratable

Add a new option "--only-migratable" in qemu which will allow to add
only those devices which will not fail qemu after migration. Devices
set with the flag 'unmigratable' cannot be added when this option will
be used.

Signed-off-by: Ashijeet Acharya <ashijeetacharya@gmail.com>
Message-Id: <1484566314-3987-3-git-send-email-ashijeetacharya@gmail.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agoblock/vvfat: Remove the undesirable comment
Ashijeet Acharya [Mon, 16 Jan 2017 11:31:50 +0000 (17:01 +0530)]
block/vvfat: Remove the undesirable comment

Remove the "// assert(is_consistent(s))" comment in block/vvfat.c

Signed-off-by: Ashijeet Acharya <ashijeetacharya@gmail.com>
Message-Id: <1484566314-3987-2-git-send-email-ashijeetacharya@gmail.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: add error_report
Jianjun Duan [Thu, 19 Jan 2017 19:00:53 +0000 (11:00 -0800)]
migration: add error_report

Added error_report where version_ids do not match in vmstate_load_state.

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Jianjun Duan <duanj@linux.vnet.ibm.com>
Message-Id: <1484852453-12728-5-git-send-email-duanj@linux.vnet.ibm.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agotests/migration: Add test for QTAILQ migration
Jianjun Duan [Thu, 19 Jan 2017 19:00:52 +0000 (11:00 -0800)]
tests/migration: Add test for QTAILQ migration

Add a test for QTAILQ migration to tests/test-vmstate.c.

Signed-off-by: Jianjun Duan <duanj@linux.vnet.ibm.com>
Message-Id: <1484852453-12728-4-git-send-email-duanj@linux.vnet.ibm.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: migrate QTAILQ
Jianjun Duan [Thu, 19 Jan 2017 19:00:51 +0000 (11:00 -0800)]
migration: migrate QTAILQ

Currently we cannot directly transfer a QTAILQ instance because of the
limitation in the migration code. Here we introduce an approach to
transfer such structures. We created VMStateInfo vmstate_info_qtailq
for QTAILQ. Similar VMStateInfo can be created for other data structures
such as list.

When a QTAILQ is migrated from source to target, it is appended to the
corresponding QTAILQ structure, which is assumed to have been properly
initialized.

This approach will be used to transfer pending_events and ccs_list in spapr
state.

We also create some macros in qemu/queue.h to access a QTAILQ using pointer
arithmetic. This ensures that we do not depend on the implementation
details about QTAILQ in the migration code.

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Jianjun Duan <duanj@linux.vnet.ibm.com>
Message-Id: <1484852453-12728-3-git-send-email-duanj@linux.vnet.ibm.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: extend VMStateInfo
Jianjun Duan [Thu, 19 Jan 2017 19:00:50 +0000 (11:00 -0800)]
migration: extend VMStateInfo

Current migration code cannot handle some data structures such as
QTAILQ in qemu/queue.h. Here we extend the signatures of put/get
in VMStateInfo so that customized handling is supported. put now
will return int type.

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Jianjun Duan <duanj@linux.vnet.ibm.com>
Message-Id: <1484852453-12728-2-git-send-email-duanj@linux.vnet.ibm.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agoMAINTAINERS: Add myself as a migration submaintainer
Dr. David Alan Gilbert [Tue, 24 Jan 2017 10:04:37 +0000 (10:04 +0000)]
MAINTAINERS: Add myself as a migration submaintainer

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20170124100437.18200-1-dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/cohuck/tags/s390x-20170124' into staging
Peter Maydell [Tue, 24 Jan 2017 17:26:26 +0000 (17:26 +0000)]
Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20170124' into staging

Two s390x fixes: One for the kvm.c build failure, and one for a bug
that might cause random guest crashes with zeroed out pages on host
kernels with working cmma (< 4.6 and likely >= 4.10).

# gpg: Signature made Tue 24 Jan 2017 15:00:50 GMT
# gpg:                using RSA key 0xDECF6B93C6F02FAF
# gpg: Good signature from "Cornelia Huck <huckc@linux.vnet.ibm.com>"
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>"
# Primary key fingerprint: C3D0 D66D C362 4FF6 A8C0  18CE DECF 6B93 C6F0 2FAF

* remotes/cohuck/tags/s390x-20170124:
  s390x/kvm: fix cmma reset for KVM
  s390x/kvm: include hw_accel.h instead of kvm.h

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/ehabkost/tags/x86-and-machine-pull-request...
Peter Maydell [Tue, 24 Jan 2017 15:39:09 +0000 (15:39 +0000)]
Merge remote-tracking branch 'remotes/ehabkost/tags/x86-and-machine-pull-request' into staging

x86, machine, numa queue (2017-01-23)

# gpg: Signature made Mon 23 Jan 2017 23:26:59 GMT
# gpg:                using RSA key 0x2807936F984DC5A6
# gpg: Good signature from "Eduardo Habkost <ehabkost@redhat.com>"
# Primary key fingerprint: 5A32 2FD5 ABC4 D3DB ACCF  D1AA 2807 936F 984D C5A6

* remotes/ehabkost/tags/x86-and-machine-pull-request:
  kvm: Allow invtsc migration if tsc-khz is set explicitly
  kvm: Simplify invtsc check
  hw/core/null-machine: Add the possibility to instantiate a CPU and RAM
  qemu-options: Rename variables on the -numa "cpus" option
  MAINTAINERS: Add an entry for hw/core/null-machine.c
  machine: Make possible_cpu_arch_ids() return const pointer
  pc: don't return cpu pointer from pc_new_cpu() as it's not needed anymore
  pc: cleanup: move smbios_set_cpuid() into pc_build_smbios()
  arch_init: Remove unnecessary default_config_files table
  vl: Ensure the numa_post_machine_init func in the appropriate location
  i386: Return migration-safe field on query-cpu-definitions
  i386: Remove AMD feature flag aliases from Opteron models
  x86: add AVX512_VPOPCNTDQ features

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agos390x/kvm: fix cmma reset for KVM
Christian Borntraeger [Tue, 24 Jan 2017 13:28:13 +0000 (14:28 +0100)]
s390x/kvm: fix cmma reset for KVM

We must reset the CMMA states for normal memory (when not on mem path),
but the current code does the opposite. This was unnoticed for some time
as the kernel since 4.6 also had a bug which mostly disabled the paging
optimizations.

Fixes: 07059effd14e ("s390x/kvm: let the CPU model control CMM(A)")
Cc: qemu-stable@nongnu.org # v2.8
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
7 years agos390x/kvm: include hw_accel.h instead of kvm.h
Cornelia Huck [Tue, 24 Jan 2017 09:18:02 +0000 (10:18 +0100)]
s390x/kvm: include hw_accel.h instead of kvm.h

Commit b394662 ("kvm: move cpu synchronization code") switched
to hw_accel.h instead of kvm.h, but missed s390x, resulting in

  CC      s390x-softmmu/target/s390x/kvm.o
/home/cohuck/git/qemu/target/s390x/kvm.c: In function ‘kvm_sclp_service_call’:
/home/cohuck/git/qemu/target/s390x/kvm.c:1034:5: error: implicit declaration of function ‘cpu_synchronize_state’ [-Werror=implicit-function-declaration]
     cpu_synchronize_state(CPU(cpu));
     ^
/home/cohuck/git/qemu/target/s390x/kvm.c:1034:5: error: nested extern declaration of ‘cpu_synchronize_state’ [-Werror=nested-externs]
/home/cohuck/git/qemu/target/s390x/kvm.c: In function ‘sigp_initial_cpu_reset’:
/home/cohuck/git/qemu/target/s390x/kvm.c:1628:5: error: implicit declaration of function ‘cpu_synchronize_post_reset’ [-Werror=implicit-function-declaration]
     cpu_synchronize_post_reset(cs);
     ^
/home/cohuck/git/qemu/target/s390x/kvm.c:1628:5: error: nested extern declaration of ‘cpu_synchronize_post_reset’ [-Werror=nested-externs]
/home/cohuck/git/qemu/target/s390x/kvm.c: In function ‘sigp_set_prefix’:
/home/cohuck/git/qemu/target/s390x/kvm.c:1665:5: error: implicit declaration of function ‘cpu_synchronize_post_init’ [-Werror=implicit-function-declaration]
     cpu_synchronize_post_init(cs);
     ^
/home/cohuck/git/qemu/target/s390x/kvm.c:1665:5: error: nested extern declaration of ‘cpu_synchronize_post_init’ [-Werror=nested-externs]
cc1: all warnings being treated as errors
/home/cohuck/git/qemu/rules.mak:64: recipe for target 'target/s390x/kvm.o' failed

Fix this.

Fixes: b394662 ("kvm: move cpu synchronization code")
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: Vincent Palatin <vpalatin@chromium.org>
7 years agoMakefile: Add qemu-doc.txt to distclean and dependencies
Peter Maydell [Tue, 24 Jan 2017 14:02:18 +0000 (14:02 +0000)]
Makefile: Add qemu-doc.txt to distclean and dependencies

When qemu-doc.txt was added as a new output format in
commit f8bab10b4c3f, it was not added to either the
list of files to remove in distclean or to the dependency
line that forces qemu-options.texi to be built before
attempting to build qemu-doc.*.

In particular, the missing dependency meant that on
some platforms (notably OSX hosts) we would try to
build qemu-doc.txt before qemu-options.texi had been
fully written out, and then makeinfo would complain
about missing cross-reference targets.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1485266538-10119-1-git-send-email-peter.maydell@linaro.org
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/rth/tags/pull-hppa-20170123' into staging
Peter Maydell [Tue, 24 Jan 2017 09:52:42 +0000 (09:52 +0000)]
Merge remote-tracking branch 'remotes/rth/tags/pull-hppa-20170123' into staging

hppa-linux target support

# gpg: Signature made Mon 23 Jan 2017 17:54:09 GMT
# gpg:                using RSA key 0xAD1270CC4DD0279B
# gpg: Good signature from "Richard Henderson <rth7680@gmail.com>"
# gpg:                 aka "Richard Henderson <rth@redhat.com>"
# gpg:                 aka "Richard Henderson <rth@twiddle.net>"
# Primary key fingerprint: 9CB1 8DDA F8E8 49AD 2AFC  16A4 AD12 70CC 4DD0 279B

* remotes/rth/tags/pull-hppa-20170123: (25 commits)
  target-hppa: Implement floating-point insns
  target-hppa: Implement system and memory-management insns
  target-hppa: Implement loads and stores
  target-hppa: Implement shifts and deposits
  target-hppa: Implement linux-user gateway page
  target-hppa: Implement branches
  target-hppa: Implement basic arithmetic
  target-hppa: Add nullification framework
  target-hppa: Add framework and enable compilation
  target-hppa: Add softfloat specializations
  linux-user: Add HPPA startup and main loop
  linux-user: Add HPPA signal handling
  linux-user: Add HPPA target_signal.h and target_cpu.h
  linux-user: Add HPPA target_structs.h
  linux-user: Add HPPA definitions to syscall_defs.h
  linux-user: Add HPPA target_syscall.h
  linux-user: Add HPPA termbits.h
  linux-user: Add HPPA syscall numbers
  linux-user: Add HPPA socket.h definitions
  linux-user: Add some hppa ioctls
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agokvm: Allow invtsc migration if tsc-khz is set explicitly
Eduardo Habkost [Sun, 8 Jan 2017 17:32:34 +0000 (15:32 -0200)]
kvm: Allow invtsc migration if tsc-khz is set explicitly

We can safely allow a VM to be migrated with invtsc enabled if
tsc-khz is set explicitly, because:
* QEMU already refuses to start if it can't set the TSC frequency
  to the configured value.
* Management software is already required to keep device
  configuration (including CPU configuration) the same on
  migration source and destination.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170108173234.25721-3-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agokvm: Simplify invtsc check
Eduardo Habkost [Sun, 8 Jan 2017 17:32:33 +0000 (15:32 -0200)]
kvm: Simplify invtsc check

Instead of searching the table we have just built, we can check
the env->features field directly.

Reviewed-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170108173234.25721-2-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agohw/core/null-machine: Add the possibility to instantiate a CPU and RAM
Thomas Huth [Wed, 18 Jan 2017 12:44:50 +0000 (13:44 +0100)]
hw/core/null-machine: Add the possibility to instantiate a CPU and RAM

Sometimes it is useful to have just a machine with CPU and RAM, without
any further hardware in it, e.g. if you just want to do some instruction
debugging for TCG with a remote GDB attached to QEMU, or run some embedded
code with the "-semihosting" QEMU parameter. qemu-system-m68k already
features a "dummy" machine, and xtensa a "sim" machine for exactly this
purpose.
All target architectures have nowadays also a "none" machine, which would
be a perfect match for this, too - but it currently does not allow to add
CPU and RAM yet. Thus let's add these possibilities in a generic way to the
"none" machine, too, so that we hopefully do not need additional "dummy"
machines in the future anymore (and maybe can also get rid of the already
existing "dummy"/"sim" machines one day).
Note that the default behaviour of the "none" machine is not changed, i.e.
no CPU and no RAM is instantiated by default. You have explicitely got to
specify the CPU model with "-cpu" and the amount of RAM with "-m" to get
these new features.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <1484743490-24721-1-git-send-email-thuth@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqemu-options: Rename variables on the -numa "cpus" option
Eduardo Habkost [Mon, 23 Jan 2017 18:06:31 +0000 (16:06 -0200)]
qemu-options: Rename variables on the -numa "cpus" option

Use @var{firstcpu} and @var{lastcpu} to make the metasyntatic
variables a bit clearer. While doing this, use @var only around
the metasyntatic variables, not including the square brackets and
hyphen.

The semantics of the "cpus" option will be clarified by rewriting
the whole -numa documentation in a follow-up patch.

Reported-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170123180632.28942-2-ehabkost@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoMAINTAINERS: Add an entry for hw/core/null-machine.c
Thomas Huth [Mon, 23 Jan 2017 05:54:55 +0000 (06:54 +0100)]
MAINTAINERS: Add an entry for hw/core/null-machine.c

The "Machine core" section sounds like a good match for this file.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <1485150895-19753-1-git-send-email-thuth@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agomachine: Make possible_cpu_arch_ids() return const pointer
Igor Mammedov [Wed, 18 Jan 2017 17:13:20 +0000 (18:13 +0100)]
machine: Make possible_cpu_arch_ids() return const pointer

make sure that external callers won't try to modify
possible_cpus and owner of possible_cpus can access
it directly when it modifies it.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <1484759609-264075-5-git-send-email-imammedo@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agopc: don't return cpu pointer from pc_new_cpu() as it's not needed anymore
Igor Mammedov [Wed, 18 Jan 2017 17:13:19 +0000 (18:13 +0100)]
pc: don't return cpu pointer from pc_new_cpu() as it's not needed anymore

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <1484759609-264075-4-git-send-email-imammedo@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agopc: cleanup: move smbios_set_cpuid() into pc_build_smbios()
Igor Mammedov [Wed, 18 Jan 2017 17:13:18 +0000 (18:13 +0100)]
pc: cleanup: move smbios_set_cpuid() into pc_build_smbios()

move smbios_set_cpuid() close to the rest of smbios init code
where it belongs to instead of calling it from pc_cpus_init().

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <1484759609-264075-3-git-send-email-imammedo@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoarch_init: Remove unnecessary default_config_files table
Eduardo Habkost [Tue, 17 Jan 2017 18:00:51 +0000 (16:00 -0200)]
arch_init: Remove unnecessary default_config_files table

The existing default_config_files table in arch_init.c has a
single entry, making it completely unnecessary. The whole code
can be replaced by a single qemu_read_config_file() call in vl.c.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170117180051.11958-1-ehabkost@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agovl: Ensure the numa_post_machine_init func in the appropriate location
Dou Liyang [Tue, 17 Jan 2017 14:42:31 +0000 (22:42 +0800)]
vl: Ensure the numa_post_machine_init func in the appropriate location

In the numa_post_machine_init(), we use CPU_FOREACH macro to set all
CPUs' namu_node. So, we should make sure that we call it after Qemu
has already initialied all the CPUs.

As we all know, the CPUs can be created by "-smp"(pc_new_cpu) or
"-device"(qdev_device_add) command. But, before the device init,
Qemu execute the numa_post_machine_init earlier. It makes the mapping
of NUMA nodes and CPUs incorrect.

The patch move the numa_post_machine_init func in the appropriate
location.

Signed-off-by: Dou Liyang <douly.fnst@cn.fujitsu.com>
Message-Id: <1484664152-24446-2-git-send-email-douly.fnst@cn.fujitsu.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoi386: Return migration-safe field on query-cpu-definitions
Eduardo Habkost [Mon, 16 Jan 2017 18:12:12 +0000 (16:12 -0200)]
i386: Return migration-safe field on query-cpu-definitions

Return the migration-safe field on query-cpu-definitions. All CPU
models in x86 are migration-safe except "host".

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170116181212.31565-1-ehabkost@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoi386: Remove AMD feature flag aliases from Opteron models
Eduardo Habkost [Fri, 13 Jan 2017 19:00:57 +0000 (17:00 -0200)]
i386: Remove AMD feature flag aliases from Opteron models

When CPU vendor is set to AMD, the AMD feature alias bits on
CPUID[0x80000001].EDX are already automatically copied from CPUID[1].EDX
on x86_cpu_realizefn(). When CPU vendor is Intel, those bits are
reserved and should be zero. On either case, those bits shouldn't be set
in the CPU model table.

Commit 726a8ff68677d8d5fba17eb0ffb85076bfb598dc removed those
bits from most CPU models, but the Opteron_* entries still have
them. Remove the alias bits from Opteron_* too.

Add an assert() to x86_register_cpudef_type() to ensure we don't
make the same mistake again.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170113190057.6327-1-ehabkost@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agox86: add AVX512_VPOPCNTDQ features
He Chen [Fri, 13 Jan 2017 01:53:31 +0000 (09:53 +0800)]
x86: add AVX512_VPOPCNTDQ features

AVX512_VPOPCNTDQ: Vector POPCNT instructions for word and qwords.
variable precision.

Signed-off-by: He Chen <he.chen@linux.intel.com>
Message-Id: <1484272411-28073-1-git-send-email-he.chen@linux.intel.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agotarget-hppa: Implement floating-point insns
Richard Henderson [Thu, 15 Dec 2016 23:04:19 +0000 (15:04 -0800)]
target-hppa: Implement floating-point insns

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Implement system and memory-management insns
Richard Henderson [Thu, 15 Dec 2016 22:59:03 +0000 (14:59 -0800)]
target-hppa: Implement system and memory-management insns

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Implement loads and stores
Richard Henderson [Thu, 15 Dec 2016 22:54:51 +0000 (14:54 -0800)]
target-hppa: Implement loads and stores

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Implement shifts and deposits
Richard Henderson [Thu, 15 Dec 2016 22:04:13 +0000 (14:04 -0800)]
target-hppa: Implement shifts and deposits

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Implement linux-user gateway page
Richard Henderson [Thu, 15 Dec 2016 21:55:46 +0000 (13:55 -0800)]
target-hppa: Implement linux-user gateway page

For linux, page 0 is mapped as an execute-only gateway.  A gateway
page is a special bit in the page table that allows a B,GATE insn
within that page to raise processor permissions.  This is how system
calls are implemented for HPPA.

Rather than actually map anything here, or handle permissions at all,
implement the semantics of the actual linux syscall entry points.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Implement branches
Richard Henderson [Thu, 15 Dec 2016 21:50:28 +0000 (13:50 -0800)]
target-hppa: Implement branches

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Implement basic arithmetic
Richard Henderson [Thu, 15 Dec 2016 21:37:23 +0000 (13:37 -0800)]
target-hppa: Implement basic arithmetic

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Add nullification framework
Richard Henderson [Thu, 15 Dec 2016 19:58:17 +0000 (11:58 -0800)]
target-hppa: Add nullification framework

The HPPA cpu has a unique form of predicated execution in which
almost any instruction can set the PSW[N] (or "nullify") bit,
which suppresses execution (and even decoding) of the following
instruction.  Execution of a nullified insn clears the PSW[N] bit.

This adds a generic framework for branching over nullified insns,
or for sufficiently simple insns, transforming the writeback of
the result to a conditional move.  In the process, we want to be
able to represent PSW[N] as a TCG condition, which implies management
of the related tcg temps.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Add framework and enable compilation
Richard Henderson [Thu, 15 Dec 2016 19:26:14 +0000 (11:26 -0800)]
target-hppa: Add framework and enable compilation

This is just about the minimum required to enable compilation
without actually executing any instructions.  This contains the
HPPACPU structure and the required callbacks, the gdbstub, the
basic translation loop, and a translate_one function that always
results in an illegal instruction.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agotarget-hppa: Add softfloat specializations
Richard Henderson [Thu, 15 Dec 2016 18:46:40 +0000 (10:46 -0800)]
target-hppa: Add softfloat specializations

Like the original MIPS, HPPA has the MSB of an SNaN set.
However, it has different rules for silencing an SNaN:
(1) msb is cleared and (2) msb-1 must be set if the fraction
is now zero, and (implementation defined) may be set always.
I haven't checked real hardware but chose the set always
alternative because it's easy and within spec.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add HPPA startup and main loop
Richard Henderson [Thu, 15 Dec 2016 18:01:00 +0000 (10:01 -0800)]
linux-user: Add HPPA startup and main loop

Including support for the atomic memory op syscalls.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add HPPA signal handling
Richard Henderson [Thu, 15 Dec 2016 17:59:01 +0000 (09:59 -0800)]
linux-user: Add HPPA signal handling

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add HPPA target_signal.h and target_cpu.h
Richard Henderson [Thu, 15 Dec 2016 17:58:14 +0000 (09:58 -0800)]
linux-user: Add HPPA target_signal.h and target_cpu.h

The cpu.h structure that these manipulate hasn't been defined
yet, but we haven't enabled compilation yet either.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add HPPA target_structs.h
Richard Henderson [Thu, 15 Dec 2016 17:57:04 +0000 (09:57 -0800)]
linux-user: Add HPPA target_structs.h

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add HPPA definitions to syscall_defs.h
Richard Henderson [Thu, 15 Dec 2016 17:56:41 +0000 (09:56 -0800)]
linux-user: Add HPPA definitions to syscall_defs.h

7 years agolinux-user: Add HPPA target_syscall.h
Richard Henderson [Thu, 15 Dec 2016 17:55:12 +0000 (09:55 -0800)]
linux-user: Add HPPA target_syscall.h

Which is primarily a re-definition of errno numbers.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add HPPA termbits.h
Richard Henderson [Thu, 15 Dec 2016 17:54:13 +0000 (09:54 -0800)]
linux-user: Add HPPA termbits.h

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add HPPA syscall numbers
Richard Henderson [Thu, 15 Dec 2016 17:53:28 +0000 (09:53 -0800)]
linux-user: Add HPPA syscall numbers

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add HPPA socket.h definitions
Richard Henderson [Thu, 15 Dec 2016 17:49:25 +0000 (09:49 -0800)]
linux-user: Add HPPA socket.h definitions

Split this out into a "cpu/sockbits.h" file now,
like we ought to do for all of the other targets.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add some hppa ioctls
Helge Deller [Wed, 7 Dec 2016 19:32:58 +0000 (11:32 -0800)]
linux-user: Add some hppa ioctls

Add F_GETLK, F_SETLK, F_SETLKW, F_GETOWN, F_SETOWN, F_SETSIG, F_GETSIG,
F_GETLK64, F_SETLK64 and F_SETLKW64 for hppa.

Signed-off-by: Helge Deller <deller@gmx.de>
Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agoMerge remote-tracking branch 'remotes/berrange/tags/pull-qio-2017-01-23-2' into staging
Peter Maydell [Mon, 23 Jan 2017 15:59:09 +0000 (15:59 +0000)]
Merge remote-tracking branch 'remotes/berrange/tags/pull-qio-2017-01-23-2' into staging

Merge io/ 2017-01-23

# gpg: Signature made Mon 23 Jan 2017 15:56:14 GMT
# gpg:                using RSA key 0xBE86EBB415104FDF
# gpg: Good signature from "Daniel P. Berrange <dan@berrange.com>"
# gpg:                 aka "Daniel P. Berrange <berrange@redhat.com>"
# Primary key fingerprint: DAF3 A6FD B26B 6291 2D0E  8E3F BE86 EBB4 1510 4FDF

* remotes/berrange/tags/pull-qio-2017-01-23-2:
  io: introduce a DNS resolver API
  io: remove Error parameter from QIOTask thread worker
  io: change the QIOTask callback signature
  io: add ability to associate an error with a task
  io: add ability to associate an opaque "result" with with a task
  io: fix typo in docs for QIOTask
  io: stop incrementing reference in qio_task_get_source
  sockets: add ability to disable DNS resolution for InetSocketAddress

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoio: introduce a DNS resolver API
Daniel P. Berrange [Tue, 19 Jul 2016 11:54:47 +0000 (12:54 +0100)]
io: introduce a DNS resolver API

Currently DNS resolution is done automatically as part
of the creation of a QIOChannelSocket object instance.
This works ok for network clients where you just end
up a single network socket, but for servers, the results
of DNS resolution may require creation of multiple
sockets.

Introducing a DNS resolver API allows DNS resolution
to be separated from the socket object creation. This
will make it practical to create multiple QIOChannelSocket
instances for servers.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
7 years agoio: remove Error parameter from QIOTask thread worker
Daniel P. Berrange [Thu, 11 Aug 2016 16:38:07 +0000 (17:38 +0100)]
io: remove Error parameter from QIOTask thread worker

Now that task objects have a directly associated error,
there's no need for an an Error **errp parameter to
the QIOTask thread worker function. It already has a
QIOTask object, so can directly set the error on it.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
7 years agoio: change the QIOTask callback signature
Daniel P. Berrange [Thu, 11 Aug 2016 14:20:58 +0000 (15:20 +0100)]
io: change the QIOTask callback signature

Currently the QIOTaskFunc signature takes an Object * for
the source, and an Error * for any error. We also need to
be able to provide a result pointer. Rather than continue
to add parameters to QIOTaskFunc, remove the existing
ones and simply pass the QIOTask object instead. This
has methods to access all the other data items required
in the callback impl.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
7 years agoio: add ability to associate an error with a task
Daniel P. Berrange [Thu, 11 Aug 2016 13:40:44 +0000 (14:40 +0100)]
io: add ability to associate an error with a task

Currently when a task fails, the error is never explicitly
associated with the task object, it is just passed along
through the completion callback. This adds the ability to
explicitly associate an error with the task.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
7 years agoio: add ability to associate an opaque "result" with with a task
Daniel P. Berrange [Thu, 11 Aug 2016 13:36:21 +0000 (14:36 +0100)]
io: add ability to associate an opaque "result" with with a task

Currently there is no data associated with a successful
task completion. This adds an opaque pointer to the task
to store an arbitrary result.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
7 years agoio: fix typo in docs for QIOTask
Daniel P. Berrange [Thu, 11 Aug 2016 13:25:30 +0000 (14:25 +0100)]
io: fix typo in docs for QIOTask

The GDestroyNotify parameter is already a pointer, so does
not need a '*' suffix on the type.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
7 years agoio: stop incrementing reference in qio_task_get_source
Daniel P. Berrange [Thu, 11 Aug 2016 17:11:04 +0000 (18:11 +0100)]
io: stop incrementing reference in qio_task_get_source

Incrementing the reference in qio_task_get_source is
not necessary, since we're not running concurrently
with any other code touching the QIOTask. This
minimizes chances of further memory leaks.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
7 years agosockets: add ability to disable DNS resolution for InetSocketAddress
Daniel P. Berrange [Tue, 19 Jul 2016 11:58:52 +0000 (12:58 +0100)]
sockets: add ability to disable DNS resolution for InetSocketAddress

Add a 'numeric' flag to the InetSocketAddress struct to allow the
caller to indicate that DNS should be skipped for the host/port
fields. This is useful if the caller knows the address is already
numeric and wants to guarantee no (potentially blocking) DNS
lookups are attempted.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
7 years agolinux-user: Handle ERFKILL and EHWPOISON
Richard Henderson [Mon, 5 Dec 2016 21:08:12 +0000 (13:08 -0800)]
linux-user: Handle ERFKILL and EHWPOISON

With definitions for generic, alpha and mips taken from 4.9-rc2.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Add SIOCGPGRP, SIOCGSTAMP, SIOCGSTAMPNS
Helge Deller [Wed, 7 Dec 2016 19:31:46 +0000 (11:31 -0800)]
linux-user: Add SIOCGPGRP, SIOCGSTAMP, SIOCGSTAMPNS

Mirror syscall_defs.h for the element type of struct timeval
and struct timespec, even though that's not 100% accurate for
each guest.

Signed-off-by: Helge Deller <deller@gmx.de>
[rth: Changed the MK_ARRAY types as per above; added ioctl.h entries.]
Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Handle TIOCSTART and TIOCSTOP
Helge Deller [Tue, 6 Dec 2016 15:24:03 +0000 (16:24 +0100)]
linux-user: Handle TIOCSTART and TIOCSTOP

Some architectures (ppc, alpha, sparc, parisc, sh and xtensa) define the
BSD TIOCSTART and TIOCSTOP ioctls in their kernel headers to provide
compatibility to other operating systems.

Those ioctls are not implemented in Linux, nevertheless, bash will use
this ioctl if it's available on those architectures.

To avoid false warnings, add code to simply ignore those ioctls.

Signed-off-by: Helge Deller <deller@gmx.de>
Message-Id: <20161206152403.GA6651@ls3530>
Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agolinux-user: Support stack-grows-up in elfload.c
Richard Henderson [Thu, 15 Dec 2016 17:38:11 +0000 (09:38 -0800)]
linux-user: Support stack-grows-up in elfload.c

HPPA is a (the) stack-grows-up target, and supporting that requires
rearranging how we compute addresses while laying out the initial
program stack.  In addition, hppa32 requires 64-byte stack alignment
so parameterize that as well.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agoRevert "Remove remainders of HPPA backend"
Richard Henderson [Thu, 29 Sep 2016 17:55:53 +0000 (10:55 -0700)]
Revert "Remove remainders of HPPA backend"

This reverts commit d41f3c3cc7a5fb9de144cc4022da14a9ff010671.

Signed-off-by: Richard Henderson <rth@twiddle.net>
7 years agoMerge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
Peter Maydell [Fri, 20 Jan 2017 16:42:07 +0000 (16:42 +0000)]
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging

* QOM interface fix (Eduardo)
* RTC fixes (Gaohuai, Igor)
* Memory leak fixes (Li Qiang, me)
* Ctrl-a b regression (Marc-André)
* Stubs cleanups and fixes (Leif, me)
* hxtool tweak (me)
* HAX support (Vincent)
* QemuThread, exec.c and SCSI fixes (Roman, Xinhua, me)
* PC_COMPAT_2_8 fix (Marcelo)
* stronger bitmap assertions (Peter)

# gpg: Signature made Fri 20 Jan 2017 12:49:01 GMT
# gpg:                using RSA key 0xBFFBD25F78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>"
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* remotes/bonzini/tags/for-upstream: (35 commits)
  pc.h: move x-mach-use-reliable-get-clock compat entry to PC_COMPAT_2_8
  bitmap: assert that start and nr are non negative
  Revert "win32: don't run subprocess tests on Mingw32 platform"
  hax: add Darwin support
  Plumb the HAXM-based hardware acceleration support
  target/i386: Add Intel HAX files
  kvm: move cpu synchronization code
  KVM: PPC: eliminate unnecessary duplicate constants
  ramblock-notifier: new
  char: fix ctrl-a b not working
  exec: Add missing rcu_read_unlock
  x86: ioapic: fix fail migration when irqchip=split
  x86: ioapic: dump version for "info ioapic"
  x86: ioapic: add traces for ioapic
  hxtool: emit Texinfo headings as @subsection
  qemu-thread: fix qemu_thread_set_name() race in qemu_thread_create()
  serial: fix memory leak in serial exit
  scsi-block: fix direction of BYTCHK test for VERIFY commands
  pc: fix crash in rtc_set_memory() if initial cpu is marked as hotplugged
  acpi: filter based on CONFIG_ACPI_X86 rather than TARGET
  ...

# Conflicts:
# include/hw/i386/pc.h

7 years agoMerge remote-tracking branch 'remotes/cohuck/tags/s390x-20170120-v2' into staging
Peter Maydell [Fri, 20 Jan 2017 15:53:58 +0000 (15:53 +0000)]
Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20170120-v2' into staging

First set of s390x patches for 2.9:
- rework of the zpci code, giving us proper multibus support
- introduction of the 2.9 machine
- fixes and improvements

# gpg: Signature made Fri 20 Jan 2017 09:11:58 GMT
# gpg:                using RSA key 0xDECF6B93C6F02FAF
# gpg: Good signature from "Cornelia Huck <huckc@linux.vnet.ibm.com>"
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>"
# Primary key fingerprint: C3D0 D66D C362 4FF6 A8C0  18CE DECF 6B93 C6F0 2FAF

* remotes/cohuck/tags/s390x-20170120-v2:
  virtio-ccw: fix ring sizing
  s390x/pci: merge msix init functions
  s390x/pci: handle PCIBridge bus number
  s390x/pci: use hashtable to look up zpci via fh
  s390x/pci: PCI multibus bridge handling
  s390x/pci: optimize calling s390_get_phb()
  s390x/pci: change the device array to a list
  s390x/pci: dynamically allocate iommu
  s390x/pci: make S390PCIIOMMU inherit Object
  s390x/kvm: use kvm_gsi_routing_enabled in flic
  s390x: add compat machine for 2.9
  s390x: remove double compat statement

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging
Peter Maydell [Fri, 20 Jan 2017 14:56:40 +0000 (14:56 +0000)]
Merge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging

# gpg: Signature made Fri 20 Jan 2017 02:58:57 GMT
# gpg:                using RSA key 0xEF04965B398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <jasowang@redhat.com>"
# 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: 215D 46F4 8246 689E C77F  3562 EF04 965B 398D 6211

* remotes/jasowang/tags/net-pull-request:
  tap: fix memory leak on failure in net_init_tap()
  hw/pci: use-after-free in pci_nic_init_nofail when nic device fails to initialize
  hw/net/dp8393x: Avoid unintentional sign extensions on addresses
  m68k: QOMify the MCF Fast Ethernet Controller device
  net: optimize checksum computation
  docs: Fix description of the sentence

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
Peter Maydell [Fri, 20 Jan 2017 14:05:44 +0000 (14:05 +0000)]
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

virtio, vhost, pc: fixes, features

writeable fw cfg blobs which will be used for guest to host
communication
fixes and cleanups all over the place

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Thu 19 Jan 2017 21:08:04 GMT
# gpg:                using RSA key 0x281F0DB8D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>"
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* remotes/mst/tags/for_upstream:
  virtio: force VIRTIO_F_IOMMU_PLATFORM
  virtio: fix up max size checks
  vhost: drop VHOST_F_DEVICE_IOTLB
  update-linux-headers.sh: support __bitwise
  virtio_crypto: header update
  pci_regs: update to latest linux
  virtio-mmio: switch to linux headers
  virtio_mmio: add standard header file
  virtio: drop an obsolete comment
  fw-cfg: bump "x-file-slots" to 0x20 for 2.9+ machine types
  pc: Add 2.9 machine-types
  fw-cfg: turn FW_CFG_FILE_SLOTS into a device property
  fw-cfg: support writeable blobs
  vhost_net: device IOTLB support
  virtio: disable notifications again after poll succeeded
  Revert "virtio: turn vq->notification into a nested counter"
  virtio-net: enable ioeventfd even if vhost=off

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/armbru/tags/pull-error-2017-01-19' into staging
Peter Maydell [Fri, 20 Jan 2017 12:35:10 +0000 (12:35 +0000)]
Merge remote-tracking branch 'remotes/armbru/tags/pull-error-2017-01-19' into staging

Error reporting patches for 2017-01-19

# gpg: Signature made Thu 19 Jan 2017 14:51:17 GMT
# gpg:                using RSA key 0x3870B400EB918653
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>"
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>"
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* remotes/armbru/tags/pull-error-2017-01-19:
  error: Report certain hints on stderr when no monitor
  error: error_setg_errno(): errno gets preserved

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agopc.h: move x-mach-use-reliable-get-clock compat entry to PC_COMPAT_2_8
Marcelo Tosatti [Wed, 18 Jan 2017 17:53:45 +0000 (15:53 -0200)]
pc.h: move x-mach-use-reliable-get-clock compat entry to PC_COMPAT_2_8

As noticed by David Gilbert, commit 6053a86 'kvmclock: reduce kvmclock
differences on migration' added 'x-mach-use-reliable-get-clock' and a
compatibility entry that turns it off; however it got merged after 2.8.0
was released but the entry has gone into PC_COMPAT_2_7 where it should
have gone into PC_COMPAT_2_8.

Fix it by moving the entry to PC_COMPAT_2_8.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20170118175343.GA26873@amt.cnet>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
7 years agobitmap: assert that start and nr are non negative
Peter Lieven [Thu, 19 Jan 2017 16:43:50 +0000 (17:43 +0100)]
bitmap: assert that start and nr are non negative

commit e1123a3b introduced a data corruption regression
in the iscsi driver because it passed -1 as nr to bitmap_set
and bitmap_clear. Add an assertion to catch such flaws earlier.

Suggested-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Peter Lieven <pl@kamp.de>
Message-Id: <1484844230-24490-1-git-send-email-pl@kamp.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20170120' into...
Peter Maydell [Fri, 20 Jan 2017 11:36:47 +0000 (11:36 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20170120' into staging

target-arm queue:
 * support virtualization in GICv3
 * enable EL2 in AArch64 CPU models
 * allow EL2 to be enabled on 'virt' board via -machine virtualization=on
 * aspeed: SMC improvements
 * m25p80: support die erase command
 * m25p80: Add Quad Page Program 4byte
 * m25p80: Improve 1GiB Micron flash definition
 * arm: Uniquely name imx25 I2C buses

# gpg: Signature made Fri 20 Jan 2017 11:31:53 GMT
# gpg:                using RSA key 0x3C2525ED14360CDE
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>"
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>"
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>"
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* remotes/pmaydell/tags/pull-target-arm-20170120: (36 commits)
  hw/arm/virt: Add board property to enable EL2
  target-arm: Enable EL2 feature bit on A53 and A57
  target/arm/psci.c: If EL2 implemented, start CPUs in EL2
  hw/arm/virt-acpi-build: use SMC if booting in EL2
  hw/arm/virt: Support using SMC for PSCI
  hw/intc/arm_gicv3: Implement EL2 traps for CPU i/f regs
  hw/intc/arm_gicv3: Implement gicv3_cpuif_virt_update()
  hw/intc/arm_gicv3: Implement ICV_ registers EOIR and IAR
  hw/intc/arm_gicv3: Implement ICV_ HPPIR, DIR and RPR registers
  hw/intc/arm_gicv3: Implement ICV_ registers which are just accessors
  hw/intc/arm_gicv3: Add accessors for ICH_ system registers
  hw/intc/gicv3: Add data fields for virtualization support
  hw/intc/gicv3: Add defines for ICH system register fields
  target-arm: Add ARMCPU fields for GIC CPU i/f config
  hw/arm/virt: Wire VIRQ, VFIQ, maintenance irq lines from GIC to CPU
  target-arm: Expose output GPIO line for VCPU maintenance interrupt
  hw/intc/arm_gic: Add external IRQ lines for VIRQ and VFIQ
  hw/intc/arm_gicv3: Add external IRQ lines for VIRQ and VFIQ
  hw/arm/virt-acpi - reserve ECAM space as PNP0C02 device
  arm: virt: Fix segmentation fault when specifying an unsupported CPU
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agohw/arm/virt: Add board property to enable EL2
Peter Maydell [Fri, 20 Jan 2017 11:15:11 +0000 (11:15 +0000)]
hw/arm/virt: Add board property to enable EL2

Add a board level property to the virt board which will
enable EL2 on the CPU if the user asks for it. The
default is not to provide EL2. If EL2 is enabled then
we will use SMC as our PSCI conduit, and report the
virtualization support in the GICv3 device tree node
and the ACPI tables.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Message-id: 1483977924-14522-19-git-send-email-peter.maydell@linaro.org

7 years agotarget-arm: Enable EL2 feature bit on A53 and A57
Peter Maydell [Fri, 20 Jan 2017 11:15:10 +0000 (11:15 +0000)]
target-arm: Enable EL2 feature bit on A53 and A57

Enable the ARM_FEATURE_EL2 bit on Cortex-A52 and
Cortex-A57, since this is all now sufficiently implemented
to work with the GICv3. We provide the usual CPU property
to disable it for backwards compatibility with the older
virt boards.

In this commit, we disable the EL2 feature on the
virt and ZynpMP boards, so there is no overall effect.
Another commit will expose a board-level property to
allow the user to enable EL2.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1483977924-14522-18-git-send-email-peter.maydell@linaro.org

7 years agotarget/arm/psci.c: If EL2 implemented, start CPUs in EL2
Peter Maydell [Fri, 20 Jan 2017 11:15:10 +0000 (11:15 +0000)]
target/arm/psci.c: If EL2 implemented, start CPUs in EL2

The PSCI spec states that a CPU_ON call should cause the new
CPU to be started in the highest implemented Non-secure
exception level. We were incorrectly starting it at the
exception level of the caller, which happens to be correct
if EL2 is not implemented. Implement the correct logic
as described in the PSCI 1.0 spec section 6.4:
 * if EL2 exists and SCR_EL3.HCE is set: start in EL2
 * otherwise start in EL1

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Tested-by: Andrew Jones <drjones@redhat.com>
Message-id: 1483977924-14522-17-git-send-email-peter.maydell@linaro.org

7 years agohw/arm/virt-acpi-build: use SMC if booting in EL2
Andrew Jones [Fri, 20 Jan 2017 11:15:10 +0000 (11:15 +0000)]
hw/arm/virt-acpi-build: use SMC if booting in EL2

Signed-off-by: Andrew Jones <drjones@redhat.com>
Acked-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1483977924-14522-16-git-send-email-peter.maydell@linaro.org
[PMM: look at vms->psci_conduit rather than vms->virt
 to decide whether to use HVC or SMC, and report no
 PSCI support at all for the 'PSCI disabled' case]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agohw/arm/virt: Support using SMC for PSCI
Peter Maydell [Fri, 20 Jan 2017 11:15:10 +0000 (11:15 +0000)]
hw/arm/virt: Support using SMC for PSCI

If we are giving the guest a CPU with EL2, it is likely to
want to use the HVC instruction itself, for instance for
providing PSCI to inner guest VMs. This makes using HVC
as the PSCI conduit for the outer QEMU a bad idea. We will
want to use SMC instead is this case: this makes sense
because QEMU's PSCI implementation is effectively an
emulation of functionality provided by EL3 firmware.

Add code to support selecting the PSCI conduit to use,
rather than hardcoding use of HVC.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 1483977924-14522-15-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/arm_gicv3: Implement EL2 traps for CPU i/f regs
Peter Maydell [Fri, 20 Jan 2017 11:15:10 +0000 (11:15 +0000)]
hw/intc/arm_gicv3: Implement EL2 traps for CPU i/f regs

Implement the architecturally required traps from NS EL1
to EL2 for the CPU interface registers. These fall into
several different groups:
 * group-0-only registers all trap if ICH_HRC_EL2.TALL0 is set
   (exactly the registers covered by gicv3_fiq_access())
 * group-1-only registers all trap if ICH_HRC_EL2.TALL1 is set
   (exactly the registers covered by gicv3_irq_access())
 * DIR traps if ICH_HCR_EL2.TC or ICH_HCR_EL2.TDIR are set
 * PMR, RPR, CTLR trap if ICH_HCR_EL2.TC is set
 * SGI0R, SGI1R, ASGI1R trap if ICH_HCR_EL2.TC is set or
   if HCR_EL2.IMO or HCR_EL2.FMO are set

We split DIR and the SGI registers out into their own access
functions, leaving the existing gicv3_irqfiq_access() just
handling PMR, RPR and CTLR.

This commit doesn't implement support for trapping on
HSTR_EL2.T12 for the 32-bit registers, as we don't implement
any of those per-coprocessor trap bits currently and
probably will want to do those in some more centralized way.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1483977924-14522-14-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/arm_gicv3: Implement gicv3_cpuif_virt_update()
Peter Maydell [Fri, 20 Jan 2017 11:15:10 +0000 (11:15 +0000)]
hw/intc/arm_gicv3: Implement gicv3_cpuif_virt_update()

Implement the function which signals virtual interrupts to the
CPU as appropriate following CPU interface state changes.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1483977924-14522-13-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/arm_gicv3: Implement ICV_ registers EOIR and IAR
Peter Maydell [Fri, 20 Jan 2017 11:15:10 +0000 (11:15 +0000)]
hw/intc/arm_gicv3: Implement ICV_ registers EOIR and IAR

Implement the two remaining ICV_ registers: EOIR and IAR.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1483977924-14522-12-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/arm_gicv3: Implement ICV_ HPPIR, DIR and RPR registers
Peter Maydell [Fri, 20 Jan 2017 11:15:10 +0000 (11:15 +0000)]
hw/intc/arm_gicv3: Implement ICV_ HPPIR, DIR and RPR registers

Implement the the ICV_ registers HPPIR, DIR and RPR.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1483977924-14522-11-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/arm_gicv3: Implement ICV_ registers which are just accessors
Peter Maydell [Fri, 20 Jan 2017 11:15:09 +0000 (11:15 +0000)]
hw/intc/arm_gicv3: Implement ICV_ registers which are just accessors

If the HCR_EL2.IMO or FMO bits are set, accesses to ICC_
system registers are redirected to be accesses to ICV_
registers (the guest-visible interface to the virtual
interrupt controller). Implement this behaviour for the
ICV_ registers which are simple accessors to the underlying
register state.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1483977924-14522-10-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/arm_gicv3: Add accessors for ICH_ system registers
Peter Maydell [Fri, 20 Jan 2017 11:15:09 +0000 (11:15 +0000)]
hw/intc/arm_gicv3: Add accessors for ICH_ system registers

The GICv3 virtualization interface includes system registers
accessible only to the hypervisor which form the control
interface for interrupt virtualization. Implement these
registers.

The function gicv3_cpuif_virt_update() which determines
whether it needs to signal vIRQ, vFIQ or a maintenance
interrupt is introduced here as a stub function -- its
implementation will be added in a subsequent commit.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1483977924-14522-9-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/gicv3: Add data fields for virtualization support
Peter Maydell [Fri, 20 Jan 2017 11:15:09 +0000 (11:15 +0000)]
hw/intc/gicv3: Add data fields for virtualization support

As the first step in adding support for the virtualization
extensions to the GICv3 emulation:
 * add the necessary data fields to the state structures
 * add the fields to the migration state, as a subsection
   which is only present if virtualization is enabled

The use of a subsection means we retain migration
compatibility as EL2 is not enabled on any CPUs currently.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1483977924-14522-8-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/gicv3: Add defines for ICH system register fields
Peter Maydell [Fri, 20 Jan 2017 11:15:09 +0000 (11:15 +0000)]
hw/intc/gicv3: Add defines for ICH system register fields

Add defines to gicv3_internal.h for fields in the ICH_*
system registers which form the GIC virtualization control
interface.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1483977924-14522-7-git-send-email-peter.maydell@linaro.org

7 years agotarget-arm: Add ARMCPU fields for GIC CPU i/f config
Peter Maydell [Fri, 20 Jan 2017 11:15:09 +0000 (11:15 +0000)]
target-arm: Add ARMCPU fields for GIC CPU i/f config

Add fields to the ARMCPU structure to allow CPU classes to
specify the configurable aspects of their GIC CPU interface.
In particular, the virtualization support allows different
values for number of list registers, priority bits and
preemption bits.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1483977924-14522-6-git-send-email-peter.maydell@linaro.org

7 years agohw/arm/virt: Wire VIRQ, VFIQ, maintenance irq lines from GIC to CPU
Peter Maydell [Fri, 20 Jan 2017 11:15:09 +0000 (11:15 +0000)]
hw/arm/virt: Wire VIRQ, VFIQ, maintenance irq lines from GIC to CPU

Wire the new VIRQ, VFIQ and maintenance interrupt lines from the
GIC to each CPU.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Message-id: 1483977924-14522-5-git-send-email-peter.maydell@linaro.org

7 years agotarget-arm: Expose output GPIO line for VCPU maintenance interrupt
Peter Maydell [Fri, 20 Jan 2017 11:15:09 +0000 (11:15 +0000)]
target-arm: Expose output GPIO line for VCPU maintenance interrupt

The GICv3 support for virtualization includes an outbound
maintenance interrupt signal which is asserted when the
CPU interface wants to signal to the hypervisor that it
needs attention. Expose this as an outbound GPIO line from
the CPU object which can be wired up as a physical interrupt
line by the board code (as we do already for the CPU timers).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1483977924-14522-4-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/arm_gic: Add external IRQ lines for VIRQ and VFIQ
Peter Maydell [Fri, 20 Jan 2017 11:15:09 +0000 (11:15 +0000)]
hw/intc/arm_gic: Add external IRQ lines for VIRQ and VFIQ

Augment the GIC's QOM device interface by adding two
new sets of sysbus IRQ lines, to signal VIRQ and VFIQ to
each CPU.

We never use these, but it's helpful to keep the v2-and-earlier
GIC's external interface in line with that of the GICv3 to
avoid board code having to add extra code conditional on which
version of the GIC is in use.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Message-id: 1483977924-14522-3-git-send-email-peter.maydell@linaro.org

7 years agohw/intc/arm_gicv3: Add external IRQ lines for VIRQ and VFIQ
Peter Maydell [Fri, 20 Jan 2017 11:15:08 +0000 (11:15 +0000)]
hw/intc/arm_gicv3: Add external IRQ lines for VIRQ and VFIQ

Augment the GICv3's QOM device interface by adding two
new sets of sysbus IRQ lines, to signal VIRQ and VFIQ to
each CPU.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1483977924-14522-2-git-send-email-peter.maydell@linaro.org

7 years agohw/arm/virt-acpi - reserve ECAM space as PNP0C02 device
Ard Biesheuvel [Fri, 20 Jan 2017 11:15:08 +0000 (11:15 +0000)]
hw/arm/virt-acpi - reserve ECAM space as PNP0C02 device

Linux for arm64 v4.10 and later will complain if the ECAM config space is
not reserved in the ACPI namespace:

  acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0x3f000000-0x3fffffff] not reserved in ACPI namespace

The rationale is that OSes that don't consume the MCFG table should still
be able to infer that the PCI config space MMIO region is occupied.

So update the ACPI table generation routine to add this reservation.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Message-id: 1484328738-21149-1-git-send-email-ard.biesheuvel@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoarm: virt: Fix segmentation fault when specifying an unsupported CPU
Shannon Zhao [Fri, 20 Jan 2017 11:15:08 +0000 (11:15 +0000)]
arm: virt: Fix segmentation fault when specifying an unsupported CPU

Using -cpu cortex-a9 (or any other unsupported CPU) with the virt
board will cause QEMU to segmentation fault.  This bug was introduced
in commit 9ac4ef77, which incorrectly added a NULL terminator when
converting the VirtBoardInfo array into a simple array of strings
defining the valid CPUs. The cpuname_valid() loop already has
a termination condition based on ARRAY_SIZE, so the NULL is
spurious and causes the strcmp() to segfault if we reach it.
Delete the NULL.

Signed-off-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1484619334-10488-1-git-send-email-zhaoshenglong@huawei.com
[PMM: expanded commit message]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoaspeed: use first FMC flash as a boot ROM
Cédric Le Goater [Fri, 20 Jan 2017 11:15:08 +0000 (11:15 +0000)]
aspeed: use first FMC flash as a boot ROM

Create a ROM region, using the default size of the mapping window for
the CE0 FMC flash module, and fill it with the flash content.

This is a little hacky but until we can boot from a MMIO region, it
seems difficult to do anything else.

Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Reviewed-by: Andrew Jeffery <andrew@aj.id.au>
Message-id: 1483979087-32663-11-git-send-email-clg@kaod.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoaspeed/smc: extend tests for Command mode
Cédric Le Goater [Fri, 20 Jan 2017 11:15:08 +0000 (11:15 +0000)]
aspeed/smc: extend tests for Command mode

The Aspeed SMC controllers have a mode (Command mode) in which
accesses to the flash content are no different than doing MMIOs. The
controller generates all the necessary commands to load (or store)
data in memory.

So add a couple of tests doing direct reads and writes on the AHB bus.

Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Andrew Jeffery <andrew@aj.id.au>
Message-id: 1483979087-32663-10-git-send-email-clg@kaod.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoaspeed/smc: reset flash after each test
Cédric Le Goater [Fri, 20 Jan 2017 11:15:08 +0000 (11:15 +0000)]
aspeed/smc: reset flash after each test

Let's make sure when each test is run that the flash object is in an
initial state and did not keep configuration from the previous tests.

Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1483979087-32663-9-git-send-email-clg@kaod.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>