sdk/emulator/qemu.git
8 years agoMerge remote-tracking branch 'remotes/riku/tags/pull-linux-user-20160628' into staging
Peter Maydell [Wed, 29 Jun 2016 09:43:07 +0000 (10:43 +0100)]
Merge remote-tracking branch 'remotes/riku/tags/pull-linux-user-20160628' into staging

Drop building linux-user targets on HPPA or m68k host systems
and add safe_syscall support for i386, aarch64, arm, ppc64 and
s390x.

# gpg: Signature made Tue 28 Jun 2016 19:31:16 BST
# gpg:                using RSA key 0xB44890DEDE3C9BC0
# gpg: Good signature from "Riku Voipio <riku.voipio@iki.fi>"
# gpg:                 aka "Riku Voipio <riku.voipio@linaro.org>"
# Primary key fingerprint: FF82 03C8 C391 98AE 0581  41EF B448 90DE DE3C 9BC0

* remotes/riku/tags/pull-linux-user-20160628: (24 commits)
  linux-user: Provide safe_syscall for ppc64
  linux-user: Provide safe_syscall for s390x
  linux-user: Provide safe_syscall for aarch64
  linux-user: Provide safe_syscall for arm
  linux-user: Provide safe_syscall for i386
  linux-user: fix x86_64 safe_syscall
  linux-user: don't swap NLMSG_DATA() fields
  linux-user: fd_trans_host_to_target_data() must process only received data
  linux-user: add missing return in netlink switch statement
  linux-user: update get_thread_area/set_thread_area strace
  linux-user: fix clone() strace
  linux-user: add socket() strace
  linux-user: add socketcall() strace
  linux-user: Support F_GETPIPE_SZ and F_SETPIPE_SZ fcntls
  linux-user: Fix wrong type used for argument to rt_sigqueueinfo
  linux-user: Create a hostdep.h for each host architecture
  user-exec: Remove unused code for OSX hosts
  user-exec: Delete now-unused hppa and m68k cpu_signal_handler() code
  configure: Don't allow user-only targets for unknown CPU architectures
  configure: Don't override ARCH=unknown if enabling TCI
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agocputlb: don't cpu_abort() if guest tries to execute outside RAM or RAM
Peter Maydell [Mon, 20 Jun 2016 17:07:05 +0000 (18:07 +0100)]
cputlb: don't cpu_abort() if guest tries to execute outside RAM or RAM

In get_page_addr_code(), if the guest program counter turns out not to
be in ROM or RAM, we can't handle executing from it, and we call
cpu_abort(). This results in the message
  qemu: fatal: Trying to execute code outside RAM or ROM at 0x08000000
followed by a guest register dump, and then QEMU dumps core.

This situation happens in one of two cases:
 (1) a guest kernel bug, where it jumped off into nowhere
 (2) a user command line mistake, where they tried to run an image for
     board A on a QEMU model of board B, or where they didn't provide
     an image at all, and QEMU executed through a ROM or RAM full of
     NOP instructions and then fell off the end

In either case, a core dump of QEMU itself is entirely useless, and
only confuses users into thinking that this is a bug in QEMU rather
than a bug in the guest or a problem with their command line. (This
is a variation on the general idea that we shouldn't assert() on
something the user can accidentally provoke.)

Replace the cpu_abort() with something that explains the situation
a bit better and exits QEMU without dumping core.

(See LP:1062220 for several examples of confused users.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-id: 1466442425-11885-1-git-send-email-peter.maydell@linaro.org

8 years agoconfigure: Make AVX2 test robust to non-ELF systems
Peter Maydell [Sat, 18 Jun 2016 22:05:02 +0000 (23:05 +0100)]
configure: Make AVX2 test robust to non-ELF systems

The AVX2 optimization test assumes that the object format
is ELF and the system has the readelf utility. If this isn't
true then configure might fail or emit a warning (since in
a pipe "foo | bar >/dev/null 2>&1" does not redirect the
stderr of foo, only of bar). Adjust the check so that if
we don't have readelf or don't have an ELF object then we
just don't enable the AVX2 optimization.

Reported-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Message-id: 1466287502-18730-3-git-send-email-pmaydell@chiark.greenend.org.uk

8 years agoconfigure: Improve usermode relocation linker option probe
Peter Maydell [Sat, 18 Jun 2016 22:05:01 +0000 (23:05 +0100)]
configure: Improve usermode relocation linker option probe

The probe we do to determine what flags to use to make the usermode
executables use a non-default text address has some flaws:
 * we run it even if we're not building the user binaries
 * we don't expect "ld --verbose" to fail

The combination of these two results in a harmless but
ugly "ld: unknown option: --verbose" message when running
configure on OSX.

Improve the probe to only run when we need it and to fail
nicely when even the backstop 'ld --verbose' approach fails.

Reported-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Message-id: 1466287502-18730-2-git-send-email-pmaydell@chiark.greenend.org.uk

8 years agohw/sh4/sh_pci.c: Use ldl_le_p() and stl_le_p()
Peter Maydell [Fri, 10 Jun 2016 16:10:21 +0000 (17:10 +0100)]
hw/sh4/sh_pci.c: Use ldl_le_p() and stl_le_p()

Use ldl_le_p() and stl_le_p() instead of le32_to_cpup() and
cpu_to_le32w(); the former handle misaligned addresses and don't
need casts, and the latter are deprecated.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Acked-by: Aurelien Jarno <aurelien@aurel32.net>
Message-id: 1465575021-3774-1-git-send-email-peter.maydell@linaro.org

8 years agoMerge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging
Peter Maydell [Tue, 28 Jun 2016 13:27:21 +0000 (14:27 +0100)]
Merge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging

# gpg: Signature made Tue 28 Jun 2016 14:23:24 BST
# gpg:                using RSA key 0x9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>"
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>"
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35  775A 9CA4 ABB3 81AB 73C8

* remotes/stefanha/tags/block-pull-request:
  virtio-blk: add num-queues device property
  virtio-blk: dataplane multiqueue support
  virtio-blk: live migrate s->rq with multiqueue
  virtio-blk: associate request with a virtqueue
  virtio-blk: tell dataplane which vq to notify
  virtio-blk: multiqueue batch notify
  virtio-blk: add VirtIOBlockConf->num_queues
  dma-helpers: dma_blk_io() cancel support
  Revert "virtio: sync the dataplane vring state to the virtqueue before virtio_save"

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agovirtio-blk: add num-queues device property
Stefan Hajnoczi [Tue, 21 Jun 2016 12:13:16 +0000 (13:13 +0100)]
virtio-blk: add num-queues device property

Multiqueue virtio-blk can be enabled as follows:

  qemu -device virtio-blk-pci,num-queues=8

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1466511196-12612-8-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agovirtio-blk: dataplane multiqueue support
Stefan Hajnoczi [Tue, 21 Jun 2016 12:13:15 +0000 (13:13 +0100)]
virtio-blk: dataplane multiqueue support

Monitor ioeventfds for all virtqueues in the device's AioContext.  This
is not true multiqueue because requests from all virtqueues are
processed in a single IOThread.  In the future it will be possible to
use multiple IOThreads when the QEMU block layer supports multiqueue.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1466511196-12612-7-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agovirtio-blk: live migrate s->rq with multiqueue
Stefan Hajnoczi [Tue, 21 Jun 2016 12:13:14 +0000 (13:13 +0100)]
virtio-blk: live migrate s->rq with multiqueue

Add a field for the virtqueue index when migrating the s->rq request
list.  The new field is only needed when num_queues > 1.  Existing QEMUs
are unaffected by this change and therefore virtio-blk migration stays
compatible.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1466511196-12612-6-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agovirtio-blk: associate request with a virtqueue
Stefan Hajnoczi [Tue, 21 Jun 2016 12:13:13 +0000 (13:13 +0100)]
virtio-blk: associate request with a virtqueue

Multiqueue requires that each request knows to which virtqueue it
belongs.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1466511196-12612-5-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agovirtio-blk: tell dataplane which vq to notify
Stefan Hajnoczi [Tue, 21 Jun 2016 12:13:12 +0000 (13:13 +0100)]
virtio-blk: tell dataplane which vq to notify

Let the virtio_blk_data_plane_notify() caller decide which virtqueue to
notify.  This will allow the function to be used with multiqueue.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1466511196-12612-4-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agovirtio-blk: multiqueue batch notify
Stefan Hajnoczi [Tue, 21 Jun 2016 12:13:11 +0000 (13:13 +0100)]
virtio-blk: multiqueue batch notify

The batch notification BH needs to know which virtqueues to notify when
multiqueue is enabled.  Use a bitmap to track the virtqueues with
pending notifications.

At this point there is only one virtqueue so hard-code virtqueue index
0.  A later patch will switch to real virtqueue indices.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1466511196-12612-3-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agovirtio-blk: add VirtIOBlockConf->num_queues
Stefan Hajnoczi [Tue, 21 Jun 2016 12:13:10 +0000 (13:13 +0100)]
virtio-blk: add VirtIOBlockConf->num_queues

The num_queues field is always 1 for the time being.  A later patch will
make it a configurable device property so that multiqueue can be
enabled.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1466511196-12612-2-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agodma-helpers: dma_blk_io() cancel support
Stefan Hajnoczi [Mon, 20 Jun 2016 19:36:57 +0000 (20:36 +0100)]
dma-helpers: dma_blk_io() cancel support

Attempting to cancel a dma_blk_io() request causes an abort(3):

  void bdrv_aio_cancel(BlockAIOCB *acb)
  {
      ...
      while (acb->refcnt > 1) {
          if (acb->aiocb_info->get_aio_context) {
              aio_poll(acb->aiocb_info->get_aio_context(acb), true);
          } else if (acb->bs) {
              aio_poll(bdrv_get_aio_context(acb->bs), true);
          } else {
              abort();
          }
      }
      ...
  }

This happens because DMAAIOCB->bs is NULL and
dma_aiocb_info.get_aio_context() is also NULL.

This patch trivially implements dma_aiocb_info.get_aio_context() by
fetching the DMAAIOCB->ctx field.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1466451417-27988-1-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoRevert "virtio: sync the dataplane vring state to the virtqueue before virtio_save"
Stefan Hajnoczi [Tue, 21 Jun 2016 10:02:11 +0000 (11:02 +0100)]
Revert "virtio: sync the dataplane vring state to the virtqueue before virtio_save"

This reverts commit 10a06fd65f667a972848ebbbcac11bdba931b544.

Dataplane has used the same virtqueue code as non-dataplane since
commits e24a47c5b73e04f94030e2daa356c7582aebfca2 ("virtio-scsi: do not
use vring in dataplane") and 03de2f527499ae0c6d16a379665d072345254f2c
("virtio-blk: do not use vring in dataplane").  It is no longer
necessary to stop dataplane in order to sync state since there is no
duplicated virtqueue state.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Message-id: 1466503331-9831-1-git-send-email-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging
Peter Maydell [Tue, 28 Jun 2016 09:32:13 +0000 (10:32 +0100)]
Merge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging

# gpg: Signature made Tue 28 Jun 2016 04:29:53 BST
# 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:
  vmxnet3: Fix reading/writing guest memory specially when behind an IOMMU
  rtl8139: save/load RxMulOk counter (again)
  Change net/socket.c to use socket_*() functions
  net: mipsnet: check transmit buffer size before sending
  net: fix qemu_announce_self not emitting packets

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agomisc/aspeed_scu: Fix build error caused by missing header
Pranith Kumar [Mon, 27 Jun 2016 21:53:04 +0000 (17:53 -0400)]
misc/aspeed_scu: Fix build error caused by missing header

Tracing configurations error out currently as follows:

/home/travis/build/pranith/qemu/hw/misc/aspeed_scu.c: In function ‘aspeed_scu_read’:
/home/travis/build/pranith/qemu/hw/misc/aspeed_scu.c:130:9: error: implicit declaration of function ‘qemu_log_mask’ [-Werror=implicit-function-declaration]
/home/travis/build/pranith/qemu/hw/misc/aspeed_scu.c:130:9: error: nested extern declaration of ‘qemu_log_mask’ [-Werror=nested-externs]
/home/travis/build/pranith/qemu/hw/misc/aspeed_scu.c:130:23: error: ‘LOG_GUEST_ERROR’ undeclared (first use in this function)
/home/travis/build/pranith/qemu/hw/misc/aspeed_scu.c:130:23: note: each undeclared identifier is reported only once for each function it appears in
/home/travis/build/pranith/qemu/hw/misc/aspeed_scu.c: In function ‘aspeed_scu_write’:
/home/travis/build/pranith/qemu/hw/misc/aspeed_scu.c:154:23: error: ‘LOG_GUEST_ERROR’ undeclared (first use in this function)

This is caused by a missing header file. Fix it.

Signed-off-by: Pranith Kumar <bobby.prani@gmail.com>
Reviewed-by: Andrew Jeffery <andrew@aj.id.au>
Message-id: 20160627215304.821-1-bobby.prani@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/jnsnow/tags/ide-pull-request' into staging
Peter Maydell [Tue, 28 Jun 2016 08:04:05 +0000 (09:04 +0100)]
Merge remote-tracking branch 'remotes/jnsnow/tags/ide-pull-request' into staging

# gpg: Signature made Mon 27 Jun 2016 20:23:19 BST
# gpg:                using RSA key 0x7DEF8106AAFC390E
# gpg: Good signature from "John Snow (John Huston) <jsnow@redhat.com>"
# Primary key fingerprint: FAEB 9711 A12C F475 812F  18F2 88A9 064D 1835 61EB
#      Subkey fingerprint: F9B7 ABDB BCAC DF95 BE76  CBD0 7DEF 8106 AAFC 390E

* remotes/jnsnow/tags/ide-pull-request:
  macio: Use blk_drain instead of blk_drain_all

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agovmxnet3: Fix reading/writing guest memory specially when behind an IOMMU
KarimAllah Ahmed [Mon, 20 Jun 2016 13:50:40 +0000 (15:50 +0200)]
vmxnet3: Fix reading/writing guest memory specially when behind an IOMMU

When a PCI device lives behind an IOMMU, it should use 'pci_dma_*' family of
functions when any transfer from/to guest memory is required while
'cpu_physical_memory_*' family of functions completely bypass any MMU/IOMMU in
the system.

vmxnet3 in some places was using 'cpu_physical_memory_*' family of functions
which works fine with the default QEMU setup where IOMMU is not enabled but
fails miserably when IOMMU is enabled. This commit converts all such instances
in favor of 'pci_dma_*'

Cc: Dmitry Fleytman <dmitry@daynix.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: qemu-devel@nongnu.org
Cc: Anthony Liguori <aliguori@amazon.com>
Signed-off-by: KarimAllah Ahmed <karahmed@amazon.de>
Acked-by: Dmitry Fleytman <dmitry@daynix.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agortl8139: save/load RxMulOk counter (again)
David Vrabel [Mon, 20 Jun 2016 17:53:46 +0000 (18:53 +0100)]
rtl8139: save/load RxMulOk counter (again)

Commit 9d29cdeaaca3a0383af764000b71492c4fc67c6e (rtl8139: port
TallyCounters to vmstate) introduced in incompatibility in the v4
format as it omitted the RxOkMul counter.

There are presumably no users that were impacted by the v4 to v4'
breakage, so increase the save version to 5 and re-add the field,
keeping backward compatibility with v4'.

We can't have a field conditional on the section version in
vmstate_tally_counters since this version checked would not be the
section version (but the version defined in this structure).  So, move
all the fields into the main state structure.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agoChange net/socket.c to use socket_*() functions
Ashijeet Acharya [Sat, 18 Jun 2016 07:54:02 +0000 (13:24 +0530)]
Change net/socket.c to use socket_*() functions

Use socket_*() functions from include/qemu/sockets.h instead of
listen()/bind()/connect()/parse_host_port(). socket_*() fucntions are
QAPI based and this patch  performs this api conversion since
everything will be using QAPI based sockets in the future. Also add a
helper function socket_address_to_string() in util/qemu-sockets.c
which returns the string representation of socket address. Thetask was
listed on http://wiki.qemu.org/BiteSizedTasks page.

Signed-off-by: Ashijeet Acharya <ashijeetacharya@gmail.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agonet: mipsnet: check transmit buffer size before sending
Prasad J Pandit [Wed, 8 Jun 2016 10:37:04 +0000 (16:07 +0530)]
net: mipsnet: check transmit buffer size before sending

When processing MIPSnet I/O port write operation, it uses a
transmit buffer tx_buffer[MAX_ETH_FRAME_SIZE=1514]. Two indices
's->tx_written' and 's->tx_count' are used to control data written
to this buffer. If the two were to be equal before writing, it'd
lead to an OOB write access beyond tx_buffer. Add check to avoid it.

Reported-by: Li Qiang <qiang6-s@360.cn>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agonet: fix qemu_announce_self not emitting packets
Peter Lieven [Thu, 9 Jun 2016 09:39:27 +0000 (11:39 +0200)]
net: fix qemu_announce_self not emitting packets

commit fefe2a78 accidently dropped the code path for injecting
raw packets. This feature is needed for sending gratuitous ARPs
after an incoming migration has completed. The result is increased
network downtime for vservers where the network card is not virtio-net
with the VIRTIO_NET_F_GUEST_ANNOUNCE feature.

Fixes: fefe2a78abde932e0f340b21bded2c86def1d242
Cc: qemu-stable@nongnu.org
Cc: hongyang.yang@easystack.cn
Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agotarget-alpha: Avoid gcc 6.1 werror for linux-user
Richard Henderson [Fri, 17 Jun 2016 19:46:33 +0000 (12:46 -0700)]
target-alpha: Avoid gcc 6.1 werror for linux-user

Using gcc 6.1 for alpha-linux-user target we see the following build error:

.../target-alpha/translate.c: In function ‘in_superpage’:
.../target-alpha/translate.c:454:52: error: self-comparison always evaluates to true [-Werror=tautological-compare]
             && addr >> TARGET_VIRT_ADDR_SPACE_BITS == addr >> 63);

Reported-by: Pranith Kumar <bobby.prani@gmail.com>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Message-id: 1466192793-2559-1-git-send-email-rth@twiddle.net
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agomacio: Use blk_drain instead of blk_drain_all
Fam Zheng [Mon, 27 Jun 2016 18:28:31 +0000 (14:28 -0400)]
macio: Use blk_drain instead of blk_drain_all

We only care about the associated backend, so blk_drain is more
appropriate here.

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 20160612065603.21911-1-famz@redhat.com
Signed-off-by: John Snow <jsnow@redhat.com>
8 years agohw/net/e1000: Don't use *_to_cpup()
Peter Maydell [Thu, 16 Jun 2016 17:17:26 +0000 (18:17 +0100)]
hw/net/e1000: Don't use *_to_cpup()

Don't use *_to_cpup() to do byte-swapped loads; instead use
ld*_p() which correctly handle misaligned accesses.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Dmitry Fleytman <dmitry@daynix.com <mailto:dmitry@daynix.com>>
Message-id: 1466097446-981-6-git-send-email-peter.maydell@linaro.org

8 years agohw/net/virtio-net.c: Don't use *_to_cpup()
Peter Maydell [Thu, 16 Jun 2016 17:17:25 +0000 (18:17 +0100)]
hw/net/virtio-net.c: Don't use *_to_cpup()

Don't use *_to_cpup() to do byte-swapped loads; instead use
ld*_p() which correctly handle misaligned accesses.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Dmitry Fleytman <dmitry@daynix.com <mailto:dmitry@daynix.com>>
Message-id: 1466097446-981-5-git-send-email-peter.maydell@linaro.org

8 years agohw/net/rocker: Don't use *_to_cpup()
Peter Maydell [Thu, 16 Jun 2016 17:17:24 +0000 (18:17 +0100)]
hw/net/rocker: Don't use *_to_cpup()

Don't use *_to_cpup() to do byte-swapped loads; instead use
ld*_p() which correctly handle misaligned accesses.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Dmitry Fleytman <dmitry@daynix.com <mailto:dmitry@daynix.com>>
Message-id: 1466097446-981-4-git-send-email-peter.maydell@linaro.org

8 years agohw/net/rtl8139.c: Don't use *_to_cpup()
Peter Maydell [Thu, 16 Jun 2016 17:17:23 +0000 (18:17 +0100)]
hw/net/rtl8139.c: Don't use *_to_cpup()

Don't use *_to_cpup() to do byte-swapped loads; instead use
ld*_p() which correctly handle misaligned accesses.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Dmitry Fleytman <dmitry@daynix.com <mailto:dmitry@daynix.com>>
Message-id: 1466097446-981-3-git-send-email-peter.maydell@linaro.org

8 years agohw/net/eepro100.c: Don't use cpu_to_*w() and *_to_cpup()
Peter Maydell [Thu, 16 Jun 2016 17:17:22 +0000 (18:17 +0100)]
hw/net/eepro100.c: Don't use cpu_to_*w() and *_to_cpup()

Don't use cpu_to_*w() and *_to_cpup() to do byte-swapped loads
and stores; instead use ld*_p() and st*_p() which correctly handle
misaligned accesses.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Dmitry Fleytman <dmitry@daynix.com <mailto:dmitry@daynix.com>>
Message-id: 1466097446-981-2-git-send-email-peter.maydell@linaro.org

8 years agoMerge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20160627' into...
Peter Maydell [Mon, 27 Jun 2016 14:46:32 +0000 (15:46 +0100)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20160627' into staging

target-arm queue:
 * arm_gicv3: add missing 'break' statements
 * cadence_uart: protect against transmit errors
 * cadence_gem: avoid infinite loops with misconfigured buffer
 * cadence_gem: set the 'last' bit when 'wrap' is set
 * reenable tmp105 test case
 * palmetto-bmc: add ASPEED system control unit model
 * m25p80: add new 512Mbit and 1Gbit devices

# gpg: Signature made Mon 27 Jun 2016 15:43:42 BST
# 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-20160627:
  m25p80: Fix WINBOND fast read command handling
  m25p80: New flash devices.
  m25p80: Fast read commands family changes.
  m25p80: Introduce configuration registers.
  m25p80: Introduce quad and equad modes.
  m25p80: Add additional flash commands:
  m25p80: Introduce COLLECTING_VAR_LEN_DATA state.
  m25p80: Allow more than four banks.
  m25p80: Make a table for JEDEC ID.
  m25p80: Replace JEDEC ID masking with function.
  palmetto-bmc: Configure the SCU's hardware strapping register
  ast2400: Integrate the SCU model and set silicon revision
  hw/misc: Add a model for the ASPEED System Control Unit
  arm: Re-enable tmp105 test
  cadence_gem: Set the last bit when wrap is set
  cadence_gem: Avoid infinite loops with a misconfigured buffer
  cadence_uart: Protect against transmit errors
  hw/intc/arm_gicv3: Add missing break

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Fix WINBOND fast read command handling
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:34 +0000 (15:37 +0100)]
m25p80: Fix WINBOND fast read command handling

This commit fix obvious bug in WINBOND command handling.
Datasheet states that default dummy cycles is 8 so fix it.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Message-id: 1466755631-25201-11-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: New flash devices.
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:34 +0000 (15:37 +0100)]
m25p80: New flash devices.

Macronix: mx66u51235f and mx66u1g45g
Micron: mt25ql01g and mt25qu01g
Spansion: s25fs512s and s70fs01gs

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466755631-25201-10-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Fast read commands family changes.
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:34 +0000 (15:37 +0100)]
m25p80: Fast read commands family changes.

Support for Spansion and Macronix flashes.
Additionally Numonyx(Micron) moved from default
in fast read commands family. Also moved fast read
command decoding to functions.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466755631-25201-9-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Introduce configuration registers.
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:34 +0000 (15:37 +0100)]
m25p80: Introduce configuration registers.

Configuration registers for Spansion and Macronix devices.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466755631-25201-8-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Introduce quad and equad modes.
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:34 +0000 (15:37 +0100)]
m25p80: Introduce quad and equad modes.

Quad and Equad modes for Spansion and Macronix flash devices.
This commit also includes modification and new command to manipulate
quad mode (status registers and dedicated commands).
This work is based on Pawel Lenkow work.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466755631-25201-7-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Add additional flash commands:
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:33 +0000 (15:37 +0100)]
m25p80: Add additional flash commands:

Page program 4byte/quad and erase 32K sectors 4 bytes.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466755631-25201-6-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Introduce COLLECTING_VAR_LEN_DATA state.
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:33 +0000 (15:37 +0100)]
m25p80: Introduce COLLECTING_VAR_LEN_DATA state.

Some flash allows to stop read at any time.
Allow framework to support this.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466755631-25201-5-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Allow more than four banks.
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:33 +0000 (15:37 +0100)]
m25p80: Allow more than four banks.

Allow to have more than four 16MiB regions for bigger flash devices.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466755631-25201-4-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Make a table for JEDEC ID.
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:33 +0000 (15:37 +0100)]
m25p80: Make a table for JEDEC ID.

Since it is now longer than 4. This work based on Pawel Lenkow
changes and the kernel SPI framework.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466755631-25201-3-git-send-email-marcin.krzeminski@nokia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agom25p80: Replace JEDEC ID masking with function.
Marcin Krzeminski [Mon, 27 Jun 2016 14:37:33 +0000 (15:37 +0100)]
m25p80: Replace JEDEC ID masking with function.

Instead of always reading and comparing jededc ID,
replace it by function.

Signed-off-by: Marcin Krzeminski <marcin.krzeminski@nokia.com>
Message-id: 1466755631-25201-2-git-send-email-marcin.krzeminski@nokia.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agopalmetto-bmc: Configure the SCU's hardware strapping register
Andrew Jeffery [Mon, 27 Jun 2016 14:37:33 +0000 (15:37 +0100)]
palmetto-bmc: Configure the SCU's hardware strapping register

The magic constant configures the following options:

* 28:27: Configure DRAM size as 256MB
* 26:24: DDR3 SDRAM with CL = 6, CWL = 5
* 23: Configure 24/48MHz CLKIN
* 22: Disable GPIOE pass-through mode
* 21: Disable GPIOD pass-through mode
* 20: Enable LPC decode of SuperIO 0x2E/0x4E addresses
* 19: Disable ACPI
* 18: Configure 48MHz CLKIN
* 17: Disable BMC 2nd boot watchdog timer
* 16: Decode SuperIO address 0x2E
* 15: VGA Class Code
* 14: Enable LPC dedicated reset pin
* 13:12: Enable SPI Master and SPI Slave to AHB Bridge
* 11:10: Select CPU:AHB ratio = 2:1
* 9:8: Select 384MHz H-PLL
* 7: Configure MAC#2 for RMII/NCSI
* 6: Configure MAC#1 for RMII/NCSI
* 5: No VGA BIOS ROM
* 4: Boot using 32bit SPI address mode
* 3:2: Select 16MB VGA memory
* 1:0: Boot from SPI flash memory

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Tested-by: Cédric Le Goater <clg@kaod.org>
Message-id: 1466744305-23163-4-git-send-email-andrew@aj.id.au
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoast2400: Integrate the SCU model and set silicon revision
Andrew Jeffery [Mon, 27 Jun 2016 14:37:33 +0000 (15:37 +0100)]
ast2400: Integrate the SCU model and set silicon revision

By specifying the silicon revision we select the appropriate reset
values for the SoC.

Additionally, expose hardware strapping properties aliasing those
provided by the SCU for board-specific configuration.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1466744305-23163-3-git-send-email-andrew@aj.id.au
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agohw/misc: Add a model for the ASPEED System Control Unit
Andrew Jeffery [Mon, 27 Jun 2016 14:37:33 +0000 (15:37 +0100)]
hw/misc: Add a model for the ASPEED System Control Unit

The SCU is a collection of chip-level control registers that manage the
various functions supported by ASPEED SoCs. Typically the bits control
interactions with clocks, external hardware or reset behaviour, and we
can largly take a hands-off approach to reads and writes.

Firmware makes heavy use of the state to determine how to boot, but the
reset values vary from SoC to SoC (eg AST2400 vs AST2500). A qdev
property is exposed so that the integrating SoC model can configure the
silicon revision, which in-turn selects the appropriate reset values.
Further qdev properties are exposed so the board model can configure the
board-dependent hardware strapping.

Almost all provided AST2400 reset values are specified by the datasheet.
The notable exception is SOC_SCRATCH1, where we mark the DRAM as
successfully initialised to avoid unnecessary dark corners in the SoC's
u-boot support.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Message-id: 1466744305-23163-2-git-send-email-andrew@aj.id.au
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: drop unnecessary inttypes.h include]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoarm: Re-enable tmp105 test
Thomas Huth [Mon, 27 Jun 2016 14:37:32 +0000 (15:37 +0100)]
arm: Re-enable tmp105 test

The tmp105 test is currently not executed since the following
line in the Makefile overwrites the check-qtest-arm-y variable
instead of extending it.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-id: 1466760306-21849-1-git-send-email-thuth@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agocadence_gem: Set the last bit when wrap is set
Alistair Francis [Mon, 27 Jun 2016 14:37:32 +0000 (15:37 +0100)]
cadence_gem: Set the last bit when wrap is set

The Cadence GEM data sheet says:
"Wrap - marks last descriptor in transmit buffer descriptor list. This
can be set for any buffer within the frame."
which seems to imply that when the wrap bit is set so is the last bit.

Previously if the wrap bit is set, but the last is not then QEMU will
enter an infinite loop.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Reported-by: Li Qiang <liqiang6-s@360.cn>
Reported-by: P J P <ppandit@redhat.com>
Message-id: eb23f15c67989ea6a53609dc66568399dadf52a7.1466539342.git.alistair.francis@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agocadence_gem: Avoid infinite loops with a misconfigured buffer
Alistair Francis [Mon, 27 Jun 2016 14:37:32 +0000 (15:37 +0100)]
cadence_gem: Avoid infinite loops with a misconfigured buffer

A guest can write zero to the DMACFG resulting in an infinite loop when
it reaches the while(bytes_to_copy) loop.

To avoid this issue enforce a minimum size for the RX buffer. Hardware
does not have this enforcement and relies on the guest to set a non-zero
value.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Reported-by: Li Qiang <liqiang6-s@360.cn>
Reported-by: P J P <ppandit@redhat.com>
Message-id: 84bb1c391b833275da3f573d4972920cea34c188.1466539342.git.alistair.francis@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agocadence_uart: Protect against transmit errors
Alistair Francis [Mon, 27 Jun 2016 14:37:32 +0000 (15:37 +0100)]
cadence_uart: Protect against transmit errors

If qemu_chr_fe_write() returns an error (represented by a negative
number) we should skip incrementing the count and initiating a
memmove().

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 667e5dc534d33338fcfc2471e5aa32fe7cbd13dc.1466546703.git.alistair.francis@xilinx.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agohw/intc/arm_gicv3: Add missing break
Shannon Zhao [Mon, 27 Jun 2016 14:37:32 +0000 (15:37 +0100)]
hw/intc/arm_gicv3: Add missing break

These are spotted by coverity 1356936 and 1356937.

Signed-off-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1466387717-13740-1-git-send-email-zhaoshenglong@huawei.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.7-20160627' into staging
Peter Maydell [Mon, 27 Jun 2016 11:54:54 +0000 (12:54 +0100)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.7-20160627' into staging

ppc patch queue for 2016-06-27

Small queue this time.  Main reason for sending it is the pair of
patches to fix up the new cpu hotplug model used on Power to what
should be an actually usable state.  There's also a small BookE bugfix
and a XICS trivial cleanup.

# gpg: Signature made Mon 27 Jun 2016 06:28:37 BST
# gpg:                using RSA key 0x6C38CACA20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>"
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>"
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>"
# 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: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* remotes/dgibson/tags/ppc-for-2.7-20160627:
  qapi: keep names in 'CpuInstanceProperties' in sync with struct CPUCore
  qapi: Report support for -device cpu hotplug in query-machines
  ppc/xics: Remove unused xics_set_irq_type()
  target-ppc: ppce500_spin.c uses SPR_PIR, should use SPR_BOOKE_PIR

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/lalrae/tags/mips-20160624' into staging
Peter Maydell [Mon, 27 Jun 2016 10:48:21 +0000 (11:48 +0100)]
Merge remote-tracking branch 'remotes/lalrae/tags/mips-20160624' into staging

MIPS patches 2016-06-24

Changes:
* support IEEE 754-2008 in MIPS CPUs

# gpg: Signature made Fri 24 Jun 2016 16:09:38 BST
# gpg:                using RSA key 0x52118E3C0B29DA6B
# gpg: Good signature from "Leon Alrae <leon.alrae@imgtec.com>"
# Primary key fingerprint: 8DD3 2F98 5495 9D66 35D4  4FC0 5211 8E3C 0B29 DA6B

* remotes/lalrae/tags/mips-20160624:
  target-mips: Add FCR31's FS bit definition
  target-mips: Implement FCR31's R/W bitmask and related functionalities
  target-mips: Add nan2008 flavor of <CEIL|CVT|FLOOR|ROUND|TRUNC>.<L|W>.<S|D>
  target-mips: Add abs2008 flavor of <ABS|NEG>.<S|D>
  target-mips: Activate IEEE 754-2008 signaling NaN bit meaning for MSA
  linux-user: Update preprocessor constants for Mips-specific e_flags bits
  softfloat: Handle snan_bit_is_one == 0 in MIPS pickNaNMulAdd()
  softfloat: For Mips only, correct default NaN values
  softfloat: Clean code format in fpu/softfloat-specialize.h
  softfloat: Implement run-time-configurable meaning of signaling NaN bit

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/mcayland/tags/qemu-sparc-signed' into staging
Peter Maydell [Mon, 27 Jun 2016 09:56:11 +0000 (10:56 +0100)]
Merge remote-tracking branch 'remotes/mcayland/tags/qemu-sparc-signed' into staging

qemu-sparc update

# gpg: Signature made Fri 24 Jun 2016 18:19:36 BST
# gpg:                using RSA key 0x5BC2C56FAE0F321F
# gpg: Good signature from "Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>"
# Primary key fingerprint: CC62 1AB9 8E82 200D 915C  C9C4 5BC2 C56F AE0F 321F

* remotes/mcayland/tags/qemu-sparc-signed:
  target-sparc: fix register corruption in ldstub if there is no write permission

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoqapi: keep names in 'CpuInstanceProperties' in sync with struct CPUCore
Peter Krempa [Thu, 23 Jun 2016 21:23:34 +0000 (23:23 +0200)]
qapi: keep names in 'CpuInstanceProperties' in sync with struct CPUCore

struct CPUCore uses 'id' suffix in the property name. As docs for
query-hotpluggable-cpus state that the cpu core properties should be
passed back to device_add by management in case new members are added
and thus the names for the fields should be kept in sync.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
[dwg: Removed a duplicated word in comment]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoqapi: Report support for -device cpu hotplug in query-machines
Peter Krempa [Thu, 23 Jun 2016 21:23:33 +0000 (23:23 +0200)]
qapi: Report support for -device cpu hotplug in query-machines

For management apps it's very useful to know whether the selected
machine type supports cpu hotplug via the new -device approach. Using
the presence of 'query-hotpluggable-cpus' alone is not enough as a
witness.

Add a property to 'MachineInfo' called 'hotpluggable-cpus' that will
report the presence of this feature.

Example of output:
    {
        "hotpluggable-cpus": false,
        "name": "mac99",
        "cpu-max": 1
    },
    {
        "hotpluggable-cpus": true,
        "name": "pseries-2.7",
        "is-default": true,
        "cpu-max": 255,
        "alias": "pseries"
    },

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoppc/xics: Remove unused xics_set_irq_type()
Benjamin Herrenschmidt [Thu, 23 Jun 2016 17:47:23 +0000 (23:17 +0530)]
ppc/xics: Remove unused xics_set_irq_type()

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
[dwg: Adjusted for context to apply without original series]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agotarget-ppc: ppce500_spin.c uses SPR_PIR, should use SPR_BOOKE_PIR
Aaron Larson [Thu, 23 Jun 2016 22:35:17 +0000 (15:35 -0700)]
target-ppc: ppce500_spin.c uses SPR_PIR, should use SPR_BOOKE_PIR

ppce500_spin.c uses SPR_PIR to initialize the spin table, however on
Book E processors the correct SPR is SPR_BOOKE_PIR.

Signed-off-by: Aaron Larson <alarson@ddci.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agolinux-user: Provide safe_syscall for ppc64
Richard Henderson [Wed, 22 Jun 2016 00:32:13 +0000 (17:32 -0700)]
linux-user: Provide safe_syscall for ppc64

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: Provide safe_syscall for s390x
Richard Henderson [Wed, 22 Jun 2016 00:32:12 +0000 (17:32 -0700)]
linux-user: Provide safe_syscall for s390x

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: Provide safe_syscall for aarch64
Richard Henderson [Wed, 22 Jun 2016 00:32:11 +0000 (17:32 -0700)]
linux-user: Provide safe_syscall for aarch64

Signed-off-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
[RV] Updated syscall argument comment to match code

8 years agolinux-user: Provide safe_syscall for arm
Richard Henderson [Wed, 22 Jun 2016 00:32:10 +0000 (17:32 -0700)]
linux-user: Provide safe_syscall for arm

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: Provide safe_syscall for i386
Richard Henderson [Wed, 22 Jun 2016 00:32:09 +0000 (17:32 -0700)]
linux-user: Provide safe_syscall for i386

Signed-off-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: fix x86_64 safe_syscall
Richard Henderson [Mon, 13 Jun 2016 21:45:21 +0000 (14:45 -0700)]
linux-user: fix x86_64 safe_syscall

Do what the comment says, test for signal_pending non-zero,
rather than the current code which tests for bit 0 non-zero.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: don't swap NLMSG_DATA() fields
Laurent Vivier [Thu, 16 Jun 2016 19:01:37 +0000 (21:01 +0200)]
linux-user: don't swap NLMSG_DATA() fields

If the structure pointed by NLMSG_DATA() is bigger
than the size of NLMSG_DATA(), don't swap its fields
to avoid memory corruption.

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: fd_trans_host_to_target_data() must process only received data
Laurent Vivier [Thu, 16 Jun 2016 19:01:36 +0000 (21:01 +0200)]
linux-user: fd_trans_host_to_target_data() must process only received data

if we process the whole buffer, the netlink helpers can try
to swap invalid data.

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: add missing return in netlink switch statement
Laurent Vivier [Tue, 14 Jun 2016 13:07:09 +0000 (15:07 +0200)]
linux-user: add missing return in netlink switch statement

Reported-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: update get_thread_area/set_thread_area strace
Laurent Vivier [Sat, 11 Jun 2016 00:19:48 +0000 (02:19 +0200)]
linux-user: update get_thread_area/set_thread_area strace

       int get_thread_area(struct user_desc *u_info);
       int set_thread_area(struct user_desc *u_info);

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: fix clone() strace
Laurent Vivier [Sat, 11 Jun 2016 00:19:47 +0000 (02:19 +0200)]
linux-user: fix clone() strace

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: add socket() strace
Laurent Vivier [Sat, 11 Jun 2016 00:19:46 +0000 (02:19 +0200)]
linux-user: add socket() strace

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: add socketcall() strace
Laurent Vivier [Sat, 11 Jun 2016 00:19:45 +0000 (02:19 +0200)]
linux-user: add socketcall() strace

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: Support F_GETPIPE_SZ and F_SETPIPE_SZ fcntls
Peter Maydell [Mon, 20 Jun 2016 14:50:37 +0000 (15:50 +0100)]
linux-user: Support F_GETPIPE_SZ and F_SETPIPE_SZ fcntls

Support the F_GETPIPE_SZ and F_SETPIPE_SZ fcntl operations.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: Fix wrong type used for argument to rt_sigqueueinfo
Peter Maydell [Mon, 20 Jun 2016 14:50:36 +0000 (15:50 +0100)]
linux-user: Fix wrong type used for argument to rt_sigqueueinfo

The third argument to the rt_sigqueueinfo syscall is a pointer to
a siginfo_t, not a pointer to a sigset_t. Fix the error in the
arguments to lock_user(), which meant that we would not have
detected some faults that we should.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: Create a hostdep.h for each host architecture
Peter Maydell [Mon, 13 Jun 2016 10:57:18 +0000 (11:57 +0100)]
linux-user: Create a hostdep.h for each host architecture

In commit 4d330cee37a21 a new hostdep.h file was added, with the intent
that host architectures which needed one could provide it, and the
build system would automatically fall back to a generic version if
there was no version for the host architecture. Although this works,
it has a flaw: if a subsequent commit switches an architecture from
"uses generic/hostdep.h" to "uses its own hostdep.h" nothing in the
makefile dependencies notices this and so doing a rebuild without
a manual 'make clean' will fail.

So we drop the idea of having a 'generic' version in favour of
every architecture we support having its own hostdep.h, even if
it doesn't have anything in it. (There are only thirteen of these.)

If the dependency files claim that an object file depends on a
nonexistent file, our dependency system means that make will
rebuild the object file, and regenerate the dependencies in
the process. So moving between trees prior to this commit and
trees after this commit works without requiring a 'make clean'.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agouser-exec: Remove unused code for OSX hosts
Peter Maydell [Mon, 13 Jun 2016 10:48:27 +0000 (11:48 +0100)]
user-exec: Remove unused code for OSX hosts

Since we dropped darwin-user support many years ago, the code in
user-exec to support hosts which define __APPLE__ is unused; delete it.

Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agouser-exec: Delete now-unused hppa and m68k cpu_signal_handler() code
Peter Maydell [Mon, 13 Jun 2016 10:43:18 +0000 (11:43 +0100)]
user-exec: Delete now-unused hppa and m68k cpu_signal_handler() code

Now that configure blocks attempts to build user-mode code on hppa
and m68k hosts, we can delete the cpu_signal_handler() implementations
for those architectures.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agoconfigure: Don't allow user-only targets for unknown CPU architectures
Peter Maydell [Mon, 13 Jun 2016 10:32:24 +0000 (11:32 +0100)]
configure: Don't allow user-only targets for unknown CPU architectures

For the user-only targets, we need to know something about the host CPU
architecture even if we are using the TCI interpreter rather than TCG.
(In particular user-exec.c has code for handling signals that needs
to know about that host's context structures.)

Specifically forbid building the user-only targets on unknown CPU
architectures, rather than allowing them to configure but then fail
when building user-exec.c.

This change drops supports for two configurations which were theoretically
possible before:
 * linux-user targets on M68K hosts using TCI
 * linux-user targets on HPPA hosts using TCI

We don't think anybody is actually trying to use these in practice, though:
 * interpreted TCG on a slow host CPU would be unusably slow
 * the m68k user-exec.c support is missing is_write detection so guest
   code which writes to the same page it is executing from was broken
   (will include any guest program using signals)
 * HPPA TCG backend support was dropped two and a half years ago
   with no complaints

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agoconfigure: Don't override ARCH=unknown if enabling TCI
Peter Maydell [Mon, 13 Jun 2016 10:22:56 +0000 (11:22 +0100)]
configure: Don't override ARCH=unknown if enabling TCI

At the moment if configure finds an unknown CPU it will set
ARCH to 'unknown', and then later either bail out or set it
to 'tci' (depending on whether the user passed configure the
--enable-tcg-interpreter switch). This is unnecessarily
confusing, because we could be using TCI in two cases:
 * a known host architecture (in which case ARCH is set to
   the actual host architecture, like 'i386')
 * an unknown host architecture (in which case ARCH is
   set to 'tci')
so nothing can rely on ARCH=tci to mean "using TCI".
Remove the line setting ARCH, so we leave it as "unknown",
which is what the actual situation is.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: Don't use sigfillset() on uc->uc_sigmask
Peter Maydell [Tue, 14 Jun 2016 11:49:18 +0000 (12:49 +0100)]
linux-user: Don't use sigfillset() on uc->uc_sigmask

The kernel and libc have different ideas about what a sigset_t
is -- for the kernel it is only _NSIG / 8 bytes in size (usually
8 bytes), but for libc it is much larger, 128 bytes. In most
situations the difference doesn't matter, because if you pass a
pointer to a libc sigset_t to the kernel it just acts on the first
8 bytes of it, but for the ucontext_t* argument to a signal handler
it trips us up. The kernel allocates this ucontext_t on the stack
according to its idea of the sigset_t type, but the type of the
ucontext_t defined by the libc headers uses the libc type, and
so do the manipulator functions like sigfillset(). This means that
 (1) sizeof(uc->uc_sigmask) is much larger than the actual
     space used on the stack
 (2) sigfillset(&uc->uc_sigmask) will write garbage 0xff bytes
     off the end of the structure, which can trash data that
     was on the stack before the signal handler was invoked,
     and may result in a crash after the handler returns

To avoid this, we use a memset() of the correct size to fill
the signal mask rather than using the libc function.

This fixes a problem where we would crash at least some of the
time on an i386 host when a signal was taken.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: Use safe_syscall wrapper for fcntl
Peter Maydell [Mon, 13 Jun 2016 10:22:05 +0000 (11:22 +0100)]
linux-user: Use safe_syscall wrapper for fcntl

Use the safe_syscall wrapper for fcntl. This is straightforward now
that we always use 'struct fcntl64' on the host, as we don't need
to select whether to call the host's fcntl64 or fcntl syscall
(a detail that the libc previously hid for us).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agolinux-user: Use __get_user() and __put_user() to handle structs in do_fcntl()
Peter Maydell [Mon, 13 Jun 2016 10:22:05 +0000 (11:22 +0100)]
linux-user: Use __get_user() and __put_user() to handle structs in do_fcntl()

Use the __get_user() and __put_user() to handle reading and writing the
guest structures in do_ioctl(). This has two benefits:
 * avoids possible errors due to misaligned guest pointers
 * correctly sign extends signed fields (like l_start in struct flock)
   which might be different sizes between guest and host

To do this we abstract out into copy_from/to_user functions. We
also standardize on always using host flock64 and the F_GETLK64
etc flock commands, as this means we always have 64 bit offsets
whether the host is 64-bit or 32-bit and we don't need to support
conversion to both host struct flock and struct flock64.

In passing we fix errors in converting l_type from the host to
the target (where we were doing a byteswap of the host value
before trying to do the convert-bitmasks operation rather than
otherwise, and inexplicably shifting left by 1); these were
accidentally left over when the original simple "just shift by 1"
arm<->x86 conversion of commit 43f238d was changed to the more
general scheme of using target_to_host_bitmask() functions in 2ba7f73.

[RV: fixed ifdef guard for eabi functions]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
8 years agotarget-sparc: fix register corruption in ldstub if there is no write permission
Artyom Tarasenko [Fri, 24 Jun 2016 12:34:30 +0000 (14:34 +0200)]
target-sparc: fix register corruption in ldstub if there is no write permission

Signed-off-by: Artyom Tarasenko <atar4qemu@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
8 years agotarget-mips: Add FCR31's FS bit definition
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:37 +0000 (11:57 +0200)]
target-mips: Add FCR31's FS bit definition

Add preprocessor definition of FCR31's FS bit, and update related
code for setting this bit.

Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agotarget-mips: Implement FCR31's R/W bitmask and related functionalities
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:36 +0000 (11:57 +0200)]
target-mips: Implement FCR31's R/W bitmask and related functionalities

This patch implements read and write access rules for Mips floating
point control and status register (FCR31). The change can be divided
into following parts:

- Add fields that will keep FCR31's R/W bitmask in procesor
  definitions and processor float_status structure.

- Add appropriate value for FCR31's R/W bitmask for each supported
  processor.

- Add function for setting snan_bit_is_one, and integrate it in
  appropriate places.

- Modify handling of CTC1 (case 31) instruction to use FCR31's R/W
  bitmask.

- Modify handling user mode executables for Mips, in relation to the
  bit EF_MIPS_NAN2008 from ELF header, that is in turn related to
  reading and writing to FCR31.

- Modify gdb behavior in relation to FCR31.

Signed-off-by: Thomas Schwinge <thomas@codesourcery.com>
Signed-off-by: Maciej W. Rozycki <macro@codesourcery.com>
Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agotarget-mips: Add nan2008 flavor of <CEIL|CVT|FLOOR|ROUND|TRUNC>.<L|W>.<S|D>
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:35 +0000 (11:57 +0200)]
target-mips: Add nan2008 flavor of <CEIL|CVT|FLOOR|ROUND|TRUNC>.<L|W>.<S|D>

New set of helpers for handling nan2008-syle versions of instructions
<CEIL|CVT|FLOOR|ROUND|TRUNC>.<L|W>.<S|D>, for Mips R6.

All involved instructions have float operand and integer result. Their
core functionality is implemented via invocations of appropriate SoftFloat
functions. The problematic cases are when the operand is a NaN, and also
when the operand (float) is out of the range of the result.

Here one can distinguish three cases:

CASE MIPS-A: (FCR31.NAN2008 == 1)

   1. Operand is a NaN, result should be 0;
   2. Operand is larger than INT_MAX, result should be INT_MAX;
   3. Operand is smaller than INT_MIN, result should be INT_MIN.

CASE MIPS-B: (FCR31.NAN2008 == 0)

   1. Operand is a NaN, result should be INT_MAX;
   2. Operand is larger than INT_MAX, result should be INT_MAX;
   3. Operand is smaller than INT_MIN, result should be INT_MAX.

CASE SoftFloat:

   1. Operand is a NaN, result is INT_MAX;
   2. Operand is larger than INT_MAX, result is INT_MAX;
   3. Operand is smaller than INT_MIN, result is INT_MIN.

Current implementation of <CEIL|CVT|FLOOR|ROUND|TRUNC>.<L|W>.<S|D>
implements case MIPS-B. This patch relates to case MIPS-A. For case
MIPS-A, only return value for NaN-operands should be corrected after
appropriate SoftFloat library function is called.

Related MSA instructions FTRUNC_S and FTINT_S already handle well
all cases, in the fashion similar to the code from this patch.

Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
[leon.alrae@imgtec.com:
 * removed a statement from the description which caused slight confusion]
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agotarget-mips: Add abs2008 flavor of <ABS|NEG>.<S|D>
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:34 +0000 (11:57 +0200)]
target-mips: Add abs2008 flavor of <ABS|NEG>.<S|D>

Updated handling of instructions <ABS|NEG>.<S|D>. Note that legacy
(pre-abs2008) ABS and NEG instructions are arithmetic (and, therefore,
any NaN operand causes signaling invalid operation), while abs2008
ones are non-arithmetic, always and only changing the sign bit, even
for NaN-like operands. Details on these instructions are documented
in [1] p. 35 and 359.

Implementation-wise, abs2008 versions are implemented without helpers,
for simplicity and performance sake.

[1] "MIPS Architecture For Programmers Volume II-A:
    The MIPS64 Instruction Set Reference Manual",
    Imagination Technologies LTD, Revision 6.04, November 13, 2015

Signed-off-by: Thomas Schwinge <thomas@codesourcery.com>
Signed-off-by: Maciej W. Rozycki <macro@codesourcery.com>
Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agotarget-mips: Activate IEEE 754-2008 signaling NaN bit meaning for MSA
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:33 +0000 (11:57 +0200)]
target-mips: Activate IEEE 754-2008 signaling NaN bit meaning for MSA

Function msa_reset() is updated so that flag snan_bit_is_one is
properly set to 0.

By applying this patch, a number of incorrect MSA behaviors that
require IEEE 754-2008 compliance will be fixed. Those are behaviors
that (up to the moment of applying this patch) did not get the desired
functionality from SoftFloat library with respect to distinguishing
between quiet and signaling NaN, getting default NaN values (both
quiet and signaling), establishing if a floating point number is NaN
or not, etc.

Two examples:

* FMAX, FMIN will now correctly detect and propagate NaNs.
* FCLASS.D ans FCLASS.S will now correcty detect NaN flavors.

Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agolinux-user: Update preprocessor constants for Mips-specific e_flags bits
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:32 +0000 (11:57 +0200)]
linux-user: Update preprocessor constants for Mips-specific e_flags bits

Missing values EF_MIPS_FP64 and EF_MIPS_NAN2008 added.

Signed-off-by: Thomas Schwinge <thomas@codesourcery.com>
Signed-off-by: Maciej W. Rozycki <macro@codesourcery.com>
Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agosoftfloat: Handle snan_bit_is_one == 0 in MIPS pickNaNMulAdd()
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:31 +0000 (11:57 +0200)]
softfloat: Handle snan_bit_is_one == 0 in MIPS pickNaNMulAdd()

Only for Mips platform, and only for cases when snan_bit_is_one is 0,
correct the order of argument comparisons in pickNaNMulAdd().

For more info, see [1], page 53, section "3.5.3 NaN Propagation".

[1] "MIPS Architecture for Programmers Volume IV-j:
    The MIPS32 SIMD Architecture Module",
    Imagination Technologies LTD, Revision 1.12, February 3, 2016

Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[leon.alrae@imgtec.com:
 * reworded the subject of the patch
 * swapped if/else code blocks to match the commit description]
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agosoftfloat: For Mips only, correct default NaN values
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:30 +0000 (11:57 +0200)]
softfloat: For Mips only, correct default NaN values

Only for Mips platform, and only for cases when snan_bit_is_one is 0,
correct default NaN values (in their 16-, 32-, and 64-bit flavors).

For more info, see [1], page 84, Table 6.3 "Value Supplied When a New
Quiet NaN Is Created", and [2], page 52, Table 3.7 "Default NaN
Encodings".

[1] "MIPS Architecture For Programmers Volume II-A:
    The MIPS64 Instruction Set Reference Manual",
    Imagination Technologies LTD, Revision 6.04, November 13, 2015

[2] "MIPS Architecture for Programmers Volume IV-j:
    The MIPS32 SIMD Architecture Module",
    Imagination Technologies LTD, Revision 1.12, February 3, 2016

Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agosoftfloat: Clean code format in fpu/softfloat-specialize.h
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:29 +0000 (11:57 +0200)]
softfloat: Clean code format in fpu/softfloat-specialize.h

fpu/softfloat-specialize.h is the most critical file in SoftFloat
library, since it handles numerous differences between platforms in
relation to floating point arithmetics. This patch makes the code
in this file more consistent format-wise, and hopefully easier to
debug and maintain.

Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agosoftfloat: Implement run-time-configurable meaning of signaling NaN bit
Aleksandar Markovic [Fri, 10 Jun 2016 09:57:28 +0000 (11:57 +0200)]
softfloat: Implement run-time-configurable meaning of signaling NaN bit

This patch modifies SoftFloat library so that it can be configured in
run-time in relation to the meaning of signaling NaN bit, while, at the
same time, strictly preserving its behavior on all existing platforms.

Background:

In floating-point calculations, there is a need for denoting undefined or
unrepresentable values. This is achieved by defining certain floating-point
numerical values to be NaNs (which stands for "not a number"). For additional
reasons, virtually all modern floating-point unit implementations use two
kinds of NaNs: quiet and signaling. The binary representations of these two
kinds of NaNs, as a rule, differ only in one bit (that bit is, traditionally,
the first bit of mantissa).

Up to 2008, standards for floating-point did not specify all details about
binary representation of NaNs. More specifically, the meaning of the bit
that is used for distinguishing between signaling and quiet NaNs was not
strictly prescribed. (IEEE 754-2008 was the first floating-point standard
that defined that meaning clearly, see [1], p. 35) As a result, different
platforms took different approaches, and that presented considerable
challenge for multi-platform emulators like QEMU.

Mips platform represents the most complex case among QEMU-supported
platforms regarding signaling NaN bit. Up to the Release 6 of Mips
architecture, "1" in signaling NaN bit denoted signaling NaN, which is
opposite to IEEE 754-2008 standard. From Release 6 on, Mips architecture
adopted IEEE standard prescription, and "0" denotes signaling NaN. On top of
that, Mips architecture for SIMD (also known as MSA, or vector instructions)
also specifies signaling bit in accordance to IEEE standard. MSA unit can be
implemented with both pre-Release 6 and Release 6 main processor units.

QEMU uses SoftFloat library to implement various floating-point-related
instructions on all platforms. The current QEMU implementation allows for
defining meaning of signaling NaN bit during build time, and is implemented
via preprocessor macro called SNAN_BIT_IS_ONE.

On the other hand, the change in this patch enables SoftFloat library to be
configured in run-time. This configuration is meant to occur during CPU
initialization, at the moment when it is definitely known what desired
behavior for particular CPU (or any additional FPUs) is.

The change is implemented so that it is consistent with existing
implementation of similar cases. This means that structure float_status is
used for passing the information about desired signaling NaN bit on each
invocation of SoftFloat functions. The additional field in float_status is
called snan_bit_is_one, which supersedes macro SNAN_BIT_IS_ONE.

IMPORTANT:

This change is not meant to create any change in emulator behavior or
functionality on any platform. It just provides the means for SoftFloat
library to be used in a more flexible way - in other words, it will just
prepare SoftFloat library for usage related to Mips platform and its
specifics regarding signaling bit meaning, which is done in some of
subsequent patches from this series.

Further break down of changes:

  1) Added field snan_bit_is_one to the structure float_status, and
     correspondent setter function set_snan_bit_is_one().

  2) Constants <float16|float32|float64|floatx80|float128>_default_nan
     (used both internally and externally) converted to functions
     <float16|float32|float64|floatx80|float128>_default_nan(float_status*).
     This is necessary since they are dependent on signaling bit meaning.
     At the same time, for the sake of code cleanup and simplicity, constants
     <floatx80|float128>_default_nan_<low|high> (used only internally within
     SoftFloat library) are removed, as not needed.

  3) Added a float_status* argument to SoftFloat library functions
     XXX_is_quiet_nan(XXX a_), XXX_is_signaling_nan(XXX a_),
     XXX_maybe_silence_nan(XXX a_). This argument must be present in
     order to enable correct invocation of new version of functions
     XXX_default_nan(). (XXX is <float16|float32|float64|floatx80|float128>
     here)

  4) Updated code for all platforms to reflect changes in SoftFloat library.
     This change is twofolds: it includes modifications of SoftFloat library
     functions invocations, and an addition of invocation of function
     set_snan_bit_is_one() during CPU initialization, with arguments that
     are appropriate for each particular platform. It was established that
     all platforms zero their main CPU data structures, so snan_bit_is_one(0)
     in appropriate places is not added, as it is not needed.

[1] "IEEE Standard for Floating-Point Arithmetic",
    IEEE Computer Society, August 29, 2008.

Signed-off-by: Thomas Schwinge <thomas@codesourcery.com>
Signed-off-by: Maciej W. Rozycki <macro@codesourcery.com>
Signed-off-by: Aleksandar Markovic <aleksandar.markovic@imgtec.com>
Tested-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Leon Alrae <leon.alrae@imgtec.com>
Tested-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[leon.alrae@imgtec.com:
 * cherry-picked 2 chunks from patch #2 to fix compilation warnings]
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
8 years agoMerge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
Peter Maydell [Fri, 24 Jun 2016 10:00:15 +0000 (11:00 +0100)]
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

pc, pci, virtio: new features, cleanups, fixes

nvdimm label support
cpu acpi hotplug rework
virtio rework
misc cleanups and fixes

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Fri 24 Jun 2016 06:50:32 BST
# 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: (34 commits)
  virtio-bus: remove old set_host_notifier callback
  virtio-mmio: convert to ioeventfd callbacks
  virtio-pci: convert to ioeventfd callbacks
  virtio-ccw: convert to ioeventfd callbacks
  virtio-bus: have callers tolerate new host notifier api
  virtio-bus: common ioeventfd infrastructure
  pc: acpi: drop intermediate PCMachineState.node_cpu
  acpi-test-data: update expected
  pc: use new CPU hotplug interface since 2.7 machine type
  acpi: cpuhp: add cpu._OST handling
  acpi: cpuhp: implement hot-remove parts of CPU hotplug interface
  acpi: cpuhp: implement hot-add parts of CPU hotplug interface
  pc: acpi: introduce AcpiDeviceIfClass.madt_cpu hook
  acpi: cpuhp: add CPU devices AML with _STA method
  pc: piix4/ich9: add 'cpu-hotplug-legacy' property
  docs: update ACPI CPU hotplug spec with new protocol
  i386: pci-assign: Fix MSI-X table size
  docs: add NVDIMM ACPI documentation
  nvdimm acpi: support Set Namespace Label Data function
  nvdimm acpi: support Get Namespace Label Data function
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: Avoid possible misalignment in host_to_target_siginfo()
Peter Maydell [Mon, 13 Jun 2016 10:22:05 +0000 (11:22 +0100)]
linux-user: Avoid possible misalignment in host_to_target_siginfo()

host_to_target_siginfo() is implemented by a combination of
host_to_target_siginfo_noswap() followed by tswap_siginfo().
The first of these two functions assumes that the target_siginfo_t
it is writing to is correctly aligned, but the pointer passed
into host_to_target_siginfo() is directly from the guest and
might be misaligned. Use a local variable to avoid this problem.
(tswap_siginfo() does now correctly handle a misaligned destination.)

We have to add a memset() to host_to_target_siginfo_noswap()
to avoid some false positive "may be used uninitialized" warnings
from gcc about subfields of the _sifields union if it chooses to
inline both tswap_siginfo() and host_to_target_siginfo_noswap()
into host_to_target_siginfo().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Peter Maydell <riku.voipio@linaro.org>
8 years agovirtio-bus: remove old set_host_notifier callback
Cornelia Huck [Fri, 10 Jun 2016 09:04:14 +0000 (11:04 +0200)]
virtio-bus: remove old set_host_notifier callback

All users have been converted to the new ioevent callbacks.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovirtio-mmio: convert to ioeventfd callbacks
Cornelia Huck [Fri, 10 Jun 2016 09:04:13 +0000 (11:04 +0200)]
virtio-mmio: convert to ioeventfd callbacks

Convert to the new interface.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovirtio-pci: convert to ioeventfd callbacks
Cornelia Huck [Fri, 10 Jun 2016 09:04:12 +0000 (11:04 +0200)]
virtio-pci: convert to ioeventfd callbacks

Convert to new interface.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovirtio-ccw: convert to ioeventfd callbacks
Cornelia Huck [Fri, 10 Jun 2016 09:04:11 +0000 (11:04 +0200)]
virtio-ccw: convert to ioeventfd callbacks

Use the new interface.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovirtio-bus: have callers tolerate new host notifier api
Cornelia Huck [Fri, 10 Jun 2016 09:04:10 +0000 (11:04 +0200)]
virtio-bus: have callers tolerate new host notifier api

Have vhost and dataplane use the new api for transports that
have been converted.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovirtio-bus: common ioeventfd infrastructure
Cornelia Huck [Fri, 10 Jun 2016 09:04:09 +0000 (11:04 +0200)]
virtio-bus: common ioeventfd infrastructure

Introduce a set of ioeventfd callbacks on the virtio-bus level
that can be implemented by the individual transports. At the
virtio-bus level, do common handling for host notifiers (which
is actually most of it).

Two things of note:
- When setting the host notifier, we only switch from/to the
  generic ioeventfd handler. This fixes a latent bug where we
  had no ioeventfd assigned for a certain window.
- We always iterate over all possible virtio queues, even though
  ccw (currently) has a lower limit. It does not really matter
  here.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agopc: acpi: drop intermediate PCMachineState.node_cpu
Igor Mammedov [Thu, 16 Jun 2016 12:23:48 +0000 (14:23 +0200)]
pc: acpi: drop intermediate PCMachineState.node_cpu

PCMachineState.node_cpu was used for mapping APIC ID
to numa node id as CPU entries in SRAT used to be
built on sparse APIC ID bitmap (up to apic_id_limit).
However since commit
  5803fce pc: acpi: SRAT: create only valid processor lapic entries
CPU entries in SRAT aren't build using apic bitmap
but using 0..maxcpus index instead which is also used
for creating numa_info[x].node_cpu map.
So instead of doing useless intermediate conversion from
  1. node by cpu index -> node by apic id
       i.e. numa_info[x].node_cpu -> PCMachineState.node_cpu
  2. apic id -> srat entry PMX
       PCMachineState.node_cpu[apic id] -> PMX value
use numa_info[x].node_cpu map directly like ARM does and do
  1. numa_info[x].node_cpu -> PMX value using index
     in range 0..maxcpus
and drop not necessary PCMachineState.node_cpu and related
code.

That also removes the last (not counting legacy hotplug)
dependency of ACPI code on apic_id_limit and need to allocate
huge sparse PCMachineState.node_cpu array in case of 32-bit
APIC IDs.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agoacpi-test-data: update expected
Michael S. Tsirkin [Fri, 24 Jun 2016 04:45:28 +0000 (07:45 +0300)]
acpi-test-data: update expected

switched to new cpu hotplug interface, aml changed.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>