user-exec.c: aarch64 initial implementation of cpu_signal_handler

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 51AF4028.5030504@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tcg/aarch64: implement sign/zero extend operations

implement the optional sign/zero extend operations with the dedicated
aarch64 instructions.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-id: 51AC9A58.40502@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tcg/aarch64: implement byte swap operations

implement the optional byte swap operations with the dedicated
aarch64 instructions.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-id: 51AC9A33.9050003@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tcg/aarch64: implement AND/TEST immediate pattern

add functions to AND/TEST registers with immediate patterns.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-id: 51AC9A0C.3090303@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tcg/aarch64: improve arith shifted regs operations

for arith operations, add SUBS, ANDS, ADDS and add a shift parameter
so that all arith instructions can make use of shifted registers.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-id: 51AC998B.7070506@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tcg/aarch64: implement new TCG target for aarch64

add preliminary support for TCG target aarch64.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 51A5C596.3090108@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

include/elf.h: add aarch64 ELF machine and relocs

we will use the 26bit relative relocs in the aarch64 tcg target.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-id: 51A5C52A.4080001@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

configure: Drop CONFIG_ATFILE test

Nobody uses the CONFIG_ATFILE test now, so just drop it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Tested-by: Claudio Fontana <claudio.fontana@huawei.com>
Message-id: 1370126121-22975-3-git-send-email-peter.maydell@linaro.org

linux-user: Drop direct use of openat etc syscalls

The linux-user syscall emulation layer currently supports the
openat family of syscalls via two mechanisms: simply calling
the corresponding libc functions, and making direct syscalls.
Since glibc has supported these functions since at least glibc
2.5, there's no real need to retain the (essentially untested)
direct syscall fallback code, so simply delete it. This allows
us to remove some ifdeffery that was attempting to disable
provision of some of the syscalls if the host didn't seem to
support them, which in some cases was actually wrong (eg where
there are several flavours of the syscall and we only need
one of them, not necessarily the exact one the guest has,
as with the fstatat* calls).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Tested-by: Claudio Fontana <claudio.fontana@huawei.com>
Message-id: 1370126121-22975-2-git-send-email-peter.maydell@linaro.org

linux-user: Allow getdents to be provided by getdents64

Newer architectures may only implement the getdents64 syscall, not
getdents. Provide an implementation of getdents in terms of getdents64
so that we can run getdents-using targets on a getdents64-only host.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Tested-by: Claudio Fontana <claudio.fontana@huawei.com>
Message-id: 1370344377-27445-1-git-send-email-peter.maydell@linaro.org
Message-id: 1370193044-24535-1-git-send-email-peter.maydell@linaro.org

qemu-char: don't issue CHR_EVENT_OPEN in a BH

When CHR_EVENT_OPENED was initially added, it was CHR_EVENT_RESET,
and it was issued as a bottom-half:

86e94dea5b740dad65446c857f6959eae43e0ba6

Which we basically used to print out a greeting/prompt for the
monitor.

AFAICT the only reason this was ever done in a BH was because in
some cases we'd modify the chr_write handler for a new chardev
backend *after* the site where we issued the reset (see:
86e94d:qemu_chr_open_stdio())

At some point this event was renamed to CHR_EVENT_OPENED, and we've
maintained the use of this BH ever since.

However, due to 9f939df955a4152aad69a19a77e0898631bb2c18, we schedule
the BH via g_idle_add(), which is causing events to sometimes be
delivered after we've already begun processing data from backends,
leading to:

 known bugs:

  QMP:
    session negotation resets with OPENED event, in some cases this
    is causing new sessions to get sporadically reset

 potential bugs:

  hw/usb/redirect.c:
    can_read handler checks for dev->parser != NULL, which may be
    true if CLOSED BH has not been executed yet. In the past, OPENED
    quiesced outstanding CLOSED events prior to us reading client
    data. If it's delayed, our check may allow reads to occur even
    though we haven't processed the OPENED event yet, and when we
    do finally get the OPENED event, our state may get reset.

  qtest.c:
    can begin session before OPENED event is processed, leading to
    a spurious reset of the system and irq_levels

  gdbstub.c:
    may start a gdb session prior to the machine being paused

To fix these, let's just drop the BH.

Since the initial reasoning for using it still applies to an extent,
work around that by deferring the delivery of CHR_EVENT_OPENED until
after the chardevs have been fully initialized, toward the end of
qmp_chardev_add() (or some cases, qemu_chr_new_from_opts()). This
defers delivery long enough that we can be assured a CharDriverState
is fully initialized before CHR_EVENT_OPENED is sent.

Also, rather than requiring each chardev to do an explicit open, do it
automatically, and allow the small few who don't desire such behavior to
suppress the OPENED-on-init behavior by setting a 'explicit_be_open'
flag.

We additionally add missing OPENED events for stdio backends on w32,
which were previously not being issued, causing us to not recieve the
banner and initial prompts for qmp/hmp.

Reported-by: Stefan Priebe <s.priebe@profihost.ag>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Message-id: 1370636393-21044-1-git-send-email-mdroth@linux.vnet.ibm.com
Cc: qemu-stable@nongnu.org
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Merge remote-tracking branch 'luiz/queue/qmp' into staging

# By Luiz Capitulino (2) and Marcelo Tosatti (1)
# Via Luiz Capitulino
* luiz/queue/qmp:
  correct RTC_CHANGE_EVENT description (v2)
  MAINTAINERS: split Monitor (QMP/HMP) entry
  MAINTAINERS: new maintainers for qapi-schema.json

Message-id: 1370634855-18337-1-git-send-email-lcapitulino@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Merge remote-tracking branch 'stefanha/net' into staging

# By Jason Wang (1) and Stefan Hajnoczi (1)
# Via Stefan Hajnoczi
* stefanha/net:
  tap: fix NULL dereference when passing invalid parameters to tap
  vmxnet3: fix NICState cleanup

Message-id: 1370613288-14933-1-git-send-email-stefanha@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

gtk: use better icon

The current icon looks pretty terrible rendered in Gnome.  This
switches to a transparent SVG which looks much nicer.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

softfloat: Fix shift128Right for shift counts 64..127

shift128Right would give the wrong result for a shift count
between 64 and 127. This was never noticed because all of
our uses of this function are guaranteed not to use shift
counts in this range.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1370186269-24353-1-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

gdbstub: let the debugger resume from guest panicked state

While in general we forbid a "continue" from the guest panicked
state, it makes sense to have an exception for that when continuing
in the debugger.  Perhaps the guest entered that state due to a bug,
for example, and we want to continue no matter what.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Luiz Capitulino <lcapitulino@redhat.com>
Message-id: 1370272015-9659-3-git-send-email-pbonzini@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

gdbstub: fix for commit 87f25c12bfeaaa0c41fb857713bbc7e8a9b757dc

This commit used the wrong check to prevent an assertion failure.
After this commit, you need to start a guest in the monitor, you
cannot use anymore the "c" command in the debugger.  This is
undesirable.  The commit's aim was to prevent a restart
after a KVM internal error or something like that; use
runstate_needs_reset() for that.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Luiz Capitulino <lcapitulino@redhat.com>
Message-id: 1370272015-9659-2-git-send-email-pbonzini@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

xilinx_axidma: Do not set DMA .notify to NULL after notify

If a stream notify function is not ready, it may re-populate the notify call-
back to indicate it should be re-polled later. This break in this usage, as
immediately following the notify() call, .notify is set to NULL. reverse the
ordering of the notify call and NULL assignment accordingly.

[PC: Reworked commit message]

Signed-off-by: Wendy Liang <jliang@xilinx.com>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

xilinx_axienet: Fix bit mask code

Obviously the code wanted to mask the lower bits but failed to do so
because of a missing "<".

cppcheck detected a conditional expression which was always true (1 < 7).

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

correct RTC_CHANGE_EVENT description (v2)

Fix RTC_CHANGE event description to match implementation.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>

MAINTAINERS: split Monitor (QMP/HMP) entry

This entry doesn't reflect reality for a few years now. This commit
splits it into Human Monitor (HMP), QAPI and QMP. Markus is dropped
as a maintainer.

This is what we have been for the last few years. Also, it's going
to help me to offload some of this work to someone else in the near
future.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>

MAINTAINERS: new maintainers for qapi-schema.json

I'm facing two problems lately wrt QMP patch review: increasingly
lack of bandwidth and lack of background in so many different areas
that are getting new QMP commands almost every week.

In order to help me mitigate this problem, I'm adding Eric and Markus
(besides me) as maintainers of the qapi-schema.json file.

Markus has been an old timer reviewer. Eric is being the most active
and prolific reviewer of QMP patches for some time now.

I believe Markus and Eric will keep doing their work as before, but
starting now I'll require the ACK of at least one of them before
appling a patch/series that touches the qapi-schema.json file.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>

tap: fix NULL dereference when passing invalid parameters to tap

This patch forbid the following invalid parameters to tap:

1) fd and vhostfds were specified but vhostfd were not specified
2) vhostfds were specified but fds were not specified
3) fds and vhostfd were specified

For 1 and 2, net_init_tap_one() will still pass NULL as vhostfdname to
monitor_handle_fd_param(), which may crash the qemu.

Also remove the unnecessary has_fd check.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Stefan Hajnoczi <shajnocz@redhat.com>
Cc: Laszlo Ersek <lersek@redhat.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge remote-tracking branch 'stefanha/block' into staging

# By Kevin Wolf (19) and others
# Via Stefan Hajnoczi
* stefanha/block: (26 commits)
  hmp: add parameters device and -v for info block
  hmp: show ImageInfo in 'info block'
  qmp: add ImageInfo in BlockDeviceInfo used by query-block
  block: add image info query function bdrv_query_image_info()
  block: add snapshot info query function bdrv_query_snapshot_info_list()
  ide-test: Add FLUSH CACHE test case
  ide: Set BSY bit during FLUSH
  ide-test: Add enum value for DEV
  blkdebug: Add BLKDBG_FLUSH_TO_OS/DISK events
  Make qemu-io commands available in HMP
  qemu-io: Use the qemu version for -V
  qemu-io: Interface cleanup
  qemu-io: Move remaining helpers from cmd.c
  qemu-io: Move command_loop() and friends
  qemu-io: Move functions for registering and running commands
  qemu-io: Move qemu_strsep() to cutils.c
  qemu-io: Move 'quit' function
  qemu-io: Move 'help' function
  qemu-io: Factor out qemuio_command
  qemu-io: Split off commands to qemu-io-cmds.c
  ...

Message-id: 1370606325-10680-1-git-send-email-stefanha@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Merge remote-tracking branch 'cohuck/virtio-ccw-upstr' into staging

# By Cornelia Huck
# Via Cornelia Huck
* cohuck/virtio-ccw-upstr:
  virtio-ccw: Fix unsetting of indicators.
  s390x/css: Fix concurrent sense.

Message-id: 1370592676-22532-1-git-send-email-cornelia.huck@de.ibm.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

hmp: add parameters device and -v for info block

With these parameters, user can choose the information to be showed,
to avoid message flood in the monitor.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

hmp: show ImageInfo in 'info block'

Now human monitor can show image details, include internal
snapshot and backing chain info for every block device.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qmp: add ImageInfo in BlockDeviceInfo used by query-block

Now image info will be retrieved as an embbed json object inside
BlockDeviceInfo, backing chain info and all related internal snapshot
info can be got in the enhanced recursive structure of ImageInfo. New
recursive member *backing-image is added to reflect the backing chain
status.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block: add image info query function bdrv_query_image_info()

This patch adds function bdrv_query_image_info(), which will
retrieve image info in qmp object format. The implementation is
based on the code moved from qemu-img.c, but uses block layer
function to get snapshot info.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block: add snapshot info query function bdrv_query_snapshot_info_list()

This patch adds function bdrv_query_snapshot_info_list(), which will
retrieve snapshot info of an image in qmp object format. The implementation
is based on the code moved from qemu-img.c with modification to fit more
for qmp based block layer API.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

ide-test: Add FLUSH CACHE test case

This checks in particular that BSY is set while the flush request is in
flight.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

ide: Set BSY bit during FLUSH

The implementation of the ATA FLUSH command invokes a flush at the block
layer, which may on raw files on POSIX entail a synchronous fdatasync().
This may in some cases take so long that the SLES 11 SP1 guest driver
reports I/O errors and filesystems get corrupted or remounted read-only.

Avoid this by setting BUSY_STAT, so that the guest is made aware we are
in the middle of an operation and no ATA commands are attempted to be
processed concurrently.

Addresses BNC#637297.

Suggested-by: Gonglei (Arei) <arei.gonglei@huawei.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

ide-test: Add enum value for DEV

Get rid of the magic number.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

blkdebug: Add BLKDBG_FLUSH_TO_OS/DISK events

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Make qemu-io commands available in HMP

It was decided to not make this command available in QMP in order to
make clear that this is not supposed to be a stable API and should be
used only for testing and debugging purposes.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Use the qemu version for -V

Always printing 0.0.1 and never updating the version number wasn't very
useful. qemu-io is released with qemu, so using the same version number
makes most sense.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Interface cleanup

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Move remaining helpers from cmd.c

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Move command_loop() and friends

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Move functions for registering and running commands

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Move qemu_strsep() to cutils.c

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Move 'quit' function

This one only makes sense in the context of the qemu-io tool, so move it
to qemu-io.c. Adapt coding style and register it like other commands.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Move 'help' function

No reason to treat it different from other commands. Move it to
qemu-io-cmds.c, adapt the coding style and register it like any other
command.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Factor out qemuio_command

It's duplicated code. Move it to qemu-io-cmds.c because it's not
dependent on any static data of the qemu-io tool.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Split off commands to qemu-io-cmds.c

This is the implementation of all qemu-io commands that make sense to be
called from the qemu monitor, i.e. everything except open, close and
quit.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Don't use global bs in command implementations

Pass in the BlockDriverState to the command handlers instead of using
the global variable. This is an important step to make the commands
usable outside of qemu-io.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Handle cvtnum() errors in 'alloc'

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Make cvtnum() a wrapper around strtosz_suffix()

No reason to implement the same thing multiple times. A nice side effect
is that fractional numbers like 0.5M can be used in qemu-io now.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

cutils: Support 'P' and 'E' suffixes in strtosz()

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-io: Remove unused args_command

The original intention seems to be something with handling multiple
images at once, but this has never been implemented and the only
function ever registered is implemented to make everything behave like a
"global" command. Just do that unconditionally now.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

virtio-ccw: Fix unsetting of indicators.

Interpretation of the ccws to register (configuration) indicators contained
a thinko: We want to disallow reading from 0, but setting the indicator
pointer to 0 is fine.

Let's fix the handling for CCW_CMD_SET{,_CONF}_IND.

Cc: qemu-stable@nongnu.org
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

s390x/css: Fix concurrent sense.

Fix an off-by-one error when indicating availablity of concurrent
sense data.

Cc: qemu-stable@nongnu.org
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

blockdev: reset werror/rerror on drive_del

Paolo Bonzini <pbonzini@redhat.com> suggested the following test case:

1. Launch a guest and wait at the GRUB boot menu:

  qemu-system-x86_64 -enable-kvm -m 1024 \
   -drive if=none,cache=none,file=test.img,id=foo,werror=stop,rerror=stop
   -device virtio-blk-pci,drive=foo,id=virtio0,addr=4

2. Hot unplug the device:

  (qemu) drive_del foo

3. Select the first boot menu entry

Without this patch the guest pauses due to ENOMEDIUM.  The guest is
stuck in a continuous pause loop since the I/O request is retried and
fails immediately again when the guest is resumed.

With this patch the error is reported to the guest.

Note that this scenario actually happens sometimes during libvirt disk
hot unplug, where device_del is followed by drive_del.  I/O may still be
submitted to the drive after drive_del if the guest does not process the
PCI hot unplug notification.

Reported-by: Dafna Ron <dron@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>

vmxnet3: fix NICState cleanup

Use qemu_del_nic() instead of qemu_del_net_client() to correctly free
the entire NICState.

Cc: qemu-stable@nongnu.org
Reported-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge remote-tracking branch 'sstabellini/xen_fixes_20130603' into staging

* sstabellini/xen_fixes_20130603:
  xen: use pc_init_pci instead of pc_init_pci_no_kvmclock
  xen: remove xen_vcpu_init
  xen: start PCI hole at 0xe0000000 (same as pc_init1 and qemu-xen-traditional)
  xen_machine_pv: do not create a dummy CPU in machine->init
  main_loop: do not set nonblocking if xen_enabled()
  xen: simplify xen_enabled

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Merge remote-tracking branch 'kwolf/for-anthony' into staging

# By Stefan Hajnoczi (6) and others
# Via Kevin Wolf
* kwolf/for-anthony:
  block: dump snapshot and image info to specified output
  block: move qmp and info dump related code to block/qapi.c
  block: move snapshot code in block.c to block/snapshot.c
  block: drop bs_snapshots global variable
  qemu-iotests: make create_image() common
  qemu-iotests: make compare_images() common
  qemu-iotests: make cancel_and_wait() common
  qemu-iotests: make assert_no_active_block_jobs() common
  block: add block driver read only whitelist
  qemu-iotests: fix 054 cluster size help output

Message-id: 1370349940-4703-1-git-send-email-kwolf@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Merge remote-tracking branch 'mst/tags/for_anthony' into staging

pci: misc cleanups

This includes some pci-related cleanups,
and fw cfg cleanups which will be useful for on-going
pci related work.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Sun 02 Jun 2013 02:46:52 PM CDT using RSA key ID D28D5469
# gpg: Can't check signature: public key not found

# By Michael S. Tsirkin (8) and Laszlo Ersek (1)
# Via Michael S. Tsirkin
* mst/tags/for_anthony:
  pvpanic: use FWCfgState explicitly
  fw_cfg: fw_cfg is a singleton
  fw_cfg: add API to find FW cfg object
  fw_cfg: move typedef to qemu/typedefs.h
  refer to FWCfgState explicitly
  apic: rename apic specific bitopts
  firmware_abi: move to include/hw/nvram/
  dec.c - move to pci-bridge
  q35: set fw_name

Message-id: 1370202787-3712-1-git-send-email-mst@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

block: dump snapshot and image info to specified output

bdrv_snapshot_dump() and bdrv_image_info_dump() do not dump to a buffer now,
some internal buffers are still used for format control, which have no
chance to be truncated. As a result, these two functions have no more issue
of truncation, and they can be used by both qemu and qemu-img with correct
parameter specified.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: move qmp and info dump related code to block/qapi.c

This patch is a pure code move patch, except following modification:
1 get_human_readable_size() is changed to static function.
2 dump_human_image_info() is renamed to bdrv_image_info_dump().
3 in qmp_query_block() and qmp_query_blockstats, use bdrv_next(bs)
instead of direct traverse of global array 'bdrv_states'.
4 collect_snapshots() and collect_image_info() are renamed, unused parameter
*fmt in collect_image_info() is removed.
5 code style fix.

To avoid conflict and tip better, macro in header file is BLOCK_QAPI_H
instead of QAPI_H. Now block.h and snapshot.h are at the same level in
include path, block_int.h and qapi.h will both include them.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: move snapshot code in block.c to block/snapshot.c

All snapshot related code, except bdrv_snapshot_dump() and
bdrv_is_snapshot(), is moved to block/snapshot.c. bdrv_snapshot_dump()
will be moved to another file later. bdrv_is_snapshot() is not related
with internal snapshot. It also fixes small code style errors reported
by check script.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: drop bs_snapshots global variable

The bs_snapshots global variable points to the BlockDriverState which
will be used to save vmstate.  This is really a savevm.c concept but was
moved into block.c:bdrv_snapshots() when it became clear that hotplug
could result in a dangling pointer.

While auditing the block layer's global state I came upon bs_snapshots
and realized that a variable is not necessary here.  Simply find the
first BlockDriverState capable of internal snapshots each time this is
needed.

The behavior of bdrv_snapshots() is preserved across hotplug because new
drives are always appended to the bdrv_states list.  This means that
calling the new find_vmstate_bs() function is idempotent - it returns
the same BlockDriverState unless it was hot-unplugged.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-iotests: make create_image() common

Both 030 and 041 use create_image().  Move it to iotests.py.

Also drop ImageStreamingTestCase since the class now has no methods.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-iotests: make compare_images() common

The iotests.compare_images() function returns True if two image files
have the identical data.  Previously this was implemented by converting
images to raw and then comparing their contents using Python.  Since
"qemu-img compare" is now available and is more efficient, switch to it.

This function will be reused by the 'drive-backup' test case.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-iotests: make cancel_and_wait() common

The cancel_and_wait() function has been duplicated in 030 and 041.  Move
it into iotests.py and let it return the event so tests can perform
additional asserts.

Note that 041's cancel_and_wait(wait_ready=True) is replaced by
wait_ready_and_cancel(), which uses the new wait_ready() and
cancel_and_wait() underneath.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-iotests: make assert_no_active_block_jobs() common

Tests 030 and 041 both use query-block-jobs to check whether any block
jobs are active.  Make this code common so that 'drive-backup' and other
new feature tests will be able to reuse it.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: add block driver read only whitelist

We may want to include a driver in the whitelist for read only tasks
such as diagnosing or exporting guest data (with libguestfs as a good
example). This patch introduces a readonly whitelist option, and for
backward compatibility, the old configure option --block-drv-whitelist
is now an alias to rw whitelist.

Drivers in readonly list is only permitted to open file readonly, and
returns -ENOTSUP for RW opening.

E.g. To include vmdk readonly, and others read+write:
    ./configure --target-list=x86_64-softmmu \
                --block-drv-rw-whitelist=qcow2,raw,file,qed \
                --block-drv-ro-whitelist=vmdk

Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-iotests: fix 054 cluster size help output

Commit f3f4d2c09b9cf46903ba38425ec46c44185162bd added a hint to increase
the cluster size when a large image cannot be created.  Test 054 now has
outdated output and fails because the golden output does not match.

This patch updates the 054 golden output.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

gtk: don't use g_object_unref on GdkCursor

It's not a GObject.

Cc: Gerd Hoffman <kraxel@redhat.com>
Reported-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
---
v1 -> v2
 - Fix summary to agree with code (Peter)

gtk: don't resize window when enabling scaling

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

fix double free the memslot in kvm_set_phys_mem

Luiz Capitulino reported that guest refused to boot and qemu
complained with:
kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument

It is caused by commit 235e8982ad that did double free for the memslot
so that the second one raises the -EINVAL error

Fix it by reset memory size only if it is needed

Reported-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

configure: Report unknown target names more helpfully

If the user specifies a target list themselves, check each entry
to make sure it's a target we recognise. This allows us to print
a helpful error message, rather than falling through (where we
would probably eventually end up hitting the uninformative
"ERROR: Unsupported target CPU").

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1369062976-301-3-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

configure: Autogenerate default target list

Autogenerate the default target list based on what files exist
in default-configs; this allows us to remove one of the places
that has to be kept up to date with a complete list of every
target we support.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1369062976-301-2-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Merge remote-tracking branch 'pmaydell/arm-devs.next' into staging

# By Peter Crosthwaite (20) and others
# Via Peter Maydell
* pmaydell/arm-devs.next: (24 commits)
  i.MX: Improve EPIT timer code.
  exynos4210.c: register rom_mem for memory migration
  hw/arm/exynos4210.c: convert chipid_and_omr to an mmio region
  i.MX: split GPT and EPIT timer implementation
  sd/sd.c: Fix "inquiry" ACMD41
  sd/sdhci:ADMA: fix interrupt
  sd/sdhci.c: Fix bdata_read DPRINT message
  sd/sdhci: Fix Buffer Write Ready interrupt
  sd/sdhci.c: Only reset data_count on new commands
  xilinx_spips: lqspi: Fix byte/misaligned access
  xilinx_spips: lqspi: Push more data to tx-fifo
  xilinx_spips: Multiple debug verbosity levels
  xilinx_spips: Debug msgs for Snoop state
  xilinx_spips: Fix striping behaviour
  xilinx_spips: Fix CTRL register RW bits
  xilinx_spips: lqspi: Dont touch config register
  xilinx_spips: Implement automatic CS
  xilinx_spips: Add automatic start support
  xilinx_spips: Trash LQ page cache on mode change
  xilinx_spips: Fix QSPI FIFO size
  ...

Message-id: 1370277021-26129-1-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

i.MX: Improve EPIT timer code.

* Unify function and type naming
* use dynamic cast whenever possible
* simplify Debug printf.
* use new style device intialization.

Signed-off-by: Jean-Christophe DUBOIS <jcd@tribudubois.net>
Reviewed-by: Peter Chubb <peter.chubb@nicta.com.au>
Message-id: 1369839656-24466-1-git-send-email-jcd@tribudubois.net
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

exynos4210.c: register rom_mem for memory migration

Even if we do not register newly created RAM MemoryRegion for migration with
vmstate_register_ram_global() function, ram_save_setup() still saves this region
to snapshot file with empty idstr=="". Consequently this results in error during
VM loading in ram_load().
Register rom_mem for migration.

Signed-off-by: Igor Mitsyanko <i.mitsyanko@samsung.com>
Message-id: 1368199981-45292-3-git-send-email-i.mitsyanko@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/exynos4210.c: convert chipid_and_omr to an mmio region

Exynos SoC was misusing memory_region_init_ram_ptr(): this interface can safely
be used only for memory regions which size is a multiple of target page size.
Change chipid_and_omr memory to an mmio region to fix this.

Signed-off-by: Igor Mitsyanko <i.mitsyanko@samsung.com>
Message-id: 1368199981-45292-2-git-send-email-i.mitsyanko@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

i.MX: split GPT and EPIT timer implementation

There is no common code between these 2 timer implementation.
So it is better to split them.

Signed-off-by: Jean-Christophe DUBOIS <jcd@tribudubois.net>
Message-id: 1368990197-19694-1-git-send-email-jcd@tribudubois.net
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sd/sd.c: Fix "inquiry" ACMD41

QEMU models two (of the three) ACMD41 has two modes, "inquiry" and
"first". The selection logic for which of the two is incorrect - it
compares != 0 for the entire argument value rather than only bits 23:0
as per the spec. Fix.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 3ef0a7fd1b2f3ebb23b4fdeabcc14caf3fad6d71.1369622254.git.peter.crosthwaite@xilinx.com
Reviewed-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sd/sdhci:ADMA: fix interrupt

The end of transfer check was occurring and potentially returning before
the interrupt flag was checked. This means the interrupt will be missed
if it occurs on the last packet. Fix by checking for the interrupt
before checking for the end of transfer.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Message-id: 9969ec154777957ec738fc4e539d68e7494d0081.1369370934.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sd/sdhci.c: Fix bdata_read DPRINT message

This message was printing out the data in decimal only, which is not
very friendly to the debugging developer. Add hex variant in
parenthesis to make it consistent with other similar messages in this
module.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: d624179649137832eaa8caa263ef9589b4395d5e.1369370934.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sd/sdhci: Fix Buffer Write Ready interrupt

This interrupt is not risen after the last block is written to sd. It
is mutually exclusive with the end of transfer conditions. Fix.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Message-id: 7ca9fd3e03ce1bec94aff08f607c15a0ec3d3371.1369370934.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sd/sdhci.c: Only reset data_count on new commands

The data_count variable was being reset on every transfer, including
DMA transfer resumptions. This is incorrect, it should only be set
on a new command.

Manifests as a bug when using ADMA and there is a timer delay between
ADMA frames where the fifo is left in a non empty state.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Message-id: 15a98609cc32315211b0963091a8efd67522e160.1369370934.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: lqspi: Fix byte/misaligned access

The LQSPI bus attachment supports byte/halfword and misaligned
accesses. Fixed. Refactored the LQSPI cache to be byte-wise
instead of word wise accordingly.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 5ec47b13563ad2d22105a1f26186d7756718394b.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: lqspi: Push more data to tx-fifo

Do 16 words per fifo flush. Increases performance and decreases
debug verbosity. This data depth has no real hardware analogue,
so just go with something that has reasonable performance.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 5621ee4621941d3639b5cacfdec26bd3148f31d5.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Multiple debug verbosity levels

The debug printfs on every SPI operation is extremely verbose. Add
a second level of debug for this.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: f63478b8e5b29cc011cdc10e29f8537bb2fc2b5e.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Debug msgs for Snoop state

This is worth keeping track of when debugging the device model.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: d3b44ecf23d671798b062eee5dc362c716ea54cd.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Fix striping behaviour

The QSPI controller was using byte-wide stripes when striping across
the two flashes in dual parallel mode. The real hardware however uses
individual bit striping. QEMU misbehaves in the (corner) case where
data is written/read in dual-parallel mode and read/written back in
single mode.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 718a61df1bf746ec06f6da44d12f8317af7b08ce.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Fix CTRL register RW bits

The CTRL register was RAZ/WI on some of the RW bits. Even though the
function behind these bits is invalid in QEMU, they should still be
guest accessible. Fix.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: b7aaad93163ce4af0c428635804ac7b77a567b25.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: lqspi: Dont touch config register

The LQSPI mode is supposed to work via the automatic CS mode feature
rather than manipulate CS lines itself. Now that auto CS is implemented
remove LQSPIs CS mode override logic. There is still a need to
manipulate the U_PAGE bit in LQSPI config register to implement
dual-stack mode however.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 000c8dd54df09523f17052638100722ef0f5a3af.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Implement automatic CS

Implement the automatic CS control feature. If the MANUAL_CS bit is
cleared then the chip select stay de-asserted as long as the tx FIFO
is empty.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 1d67383adc42761af715a93f161344b9284dfc9a.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Add automatic start support

SPI has a mode where it automatically starts based on tx fifo
occupancy. Implemented.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: f4e9accb5de87b526fff6ed937f63278db76533b.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Trash LQ page cache on mode change

Invalidate the LQSPI cached page when transitioning into LQSPI mode.
Otherwise there is a possibility that the controller will return stale
data to the guest when transitioning back to LQ_MODE after a page
program.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 677490a6ee1953fe5d366e599d665de645ac84db.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Fix QSPI FIFO size

QSPI has a bigger FIFO than the regular SPI controller. Differentiate
between the two with correct FIFO sizes for each.

This is the first piece of class data for SPIPS, so this patch sees
the creation of the XilinxSPIPSClass definition and assoicated QOM
constructs.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: acee25dd5e203215cbc15ca5d3cb5d5b2efebe7b.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Add verbose LQSPI debug output

You really need this is you want to track a guest banging on LQSPI.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 26e508da4af11058d37daa777064c9e5c2a69abb.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Inhibit interrupts in LQSPI mode

The real hardware does not produce interrupts in LQSPI mode. Inhibit
generation of interrupts when the LQ_MODE bit is set.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: dff794a06872009ea7e5733ce6adcff94d18bbd0.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: Make interrupts clear on read

By default these interrupts are clear on read.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 999ff0091ed3cc3969a431bf55c00ef934cecc8e.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xilinx_spips: seperate SPI and QSPI as two classes

Make SPI and QSPI different classes. QSPIPS is setup as a child of SPIPS.
Only QSPI has the LQSPI functionality, so move all that to the child class.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Message-id: 2cdd0cadb5ba77ca02fde5cae627852dc9a64c71.1369117359.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xen: use pc_init_pci instead of pc_init_pci_no_kvmclock

Call kvmclock_create only if kvmclock_enabled.
Use pc_init_pci on Xen rather than pc_init_pci_no_kvmclock.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

xen: remove xen_vcpu_init

No need for xen_vcpu_init anymore:

- the RTC emulator doesn't have any periodic timers continuously running
even in absence of guest interactions anymore;

- qemu_dummy_start_vcpu takes care of disabling TCG for us, so we don't
need to do it manually here.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Reviewed-by: Andreas Färber <afaerber@suse.de>

xen: start PCI hole at 0xe0000000 (same as pc_init1 and qemu-xen-traditional)

We are currently setting the PCI hole to start at HVM_BELOW_4G_RAM_END,
that is 0xf0000000.
Start the PCI hole at 0xe0000000 instead, that is the same value used by
pc_init1 and qemu-xen-traditional.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
CC: qemu-stable@nongnu.org