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

pc: fix event_idx compatibility for virtio devices

event_idx was introduced in 0.15 and must be disabled for all virtio-pci devices
(including virtio-balloon-pci).

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

pc: add pc-0.15

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

phys_page_find_alloc: Use correct initial region_offset.

This fixes a common bug with initial region_offset value.
Usually, the pages are re-assigned afterwards, so the bug
has a very small effect on regular QEMU use flows.

Signed-off-by: Alex Rozenman <Alex_Rozenman@mentor.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

network scripts: don't block SIGCHLD before forking

This patch fixes a bug where child processes of launch_script() can
misbehave due to SIGCHLD being blocked. In the case of `sudo`, this
causes a permanent hang.

Previously a SIGCHLD handler was added to reap fork_exec()'d zombie
processes by calling waitpid(-1, ...). This required other
fork()/waitpid() callers to temporarilly block SIGCHILD to avoid
having the final wait status being intercepted by the SIGCHLD
handler:

7c3370d4fe3fa6cda8655f109e4659afc8ca4269

Since then, the qemu_add_child_watch() interface was added to allow
registration of such processes and reap only from that specific set
of PIDs:

4d54ec7898bd951007cb6122d5315584bd41d0c4

As a result, we can now avoid blocking SIGCHLD in launch_script(), so
drop that behavior.

Reviewed-by: Jan Kiszka <jan.kiszka@siemens.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

HACKING: clarify allocation/free recommendations

Clarify the allocation/free recommendations; this is mostly
just tidying up following the global-search-and-replace done
with the conversion to the GLib g_malloc and friends.

Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

usb: fix usb_qdev_init() error handling again

Commit f462141f18ffdd75847f6459ef83d90b831d12c0 introduced clean up code
when usb_qdev_init() fails.  Unfortunately it calls .handle_destroy()
when .init() was never invoked or failed.  This can lead to crashes when
.handle_destroy() tries to clean up things that were never initialized.

This patch is careful to undo only those steps that completed along the
usb_qdev_init() code path.  It's not as pretty as the unified error
handling in f462141f18ffdd75847f6459ef83d90b831d12c0 but it's necessary.

Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

exec.c: Fix subpage memory access to RAM MemoryRegion

Commit 95c318f5e1f88d7e5bcc6deac17330fd4806a2d3 (Fix segfault in mmio
subpage handling code.) prevented a segfault by making all subpage
registrations over an existing memory page perform an unassigned access.
Symptoms were writes not taking effect and reads returning zero.

Very small page sizes are not currently supported either,
so subpage memory areas cannot fully be avoided.

Therefore change the previous fix to use a new IO_MEM_SUBPAGE_RAM
instead of IO_MEM_UNASSIGNED. Suggested by Avi.

Reviewed-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Cc: Avi Kivity <avi@redhat.com>
Cc: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

vl.c: Print the actual program name in help output

In help() we do what boils down to:

  printf("%s", "qemu");

This seems to be an artifact of be995c27640a82c7056b6f53d02ec823570114e5
("removed unused code"), which removed some ifdef'ery that used to print
a different name depending on CONFIG_SOFTMMU.

Instead print the actual program name, originally from argv[0].

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

error: Add an accessor for progname

We'd like to get the progname for help output, so add an accessor.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Reviewed-by: Andreas F=E4rber <afaerber@suse.de>
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

fix win32 build

On Windows, cpus.c needs access to the hThread.  Add a Windows-specific
function to grab it.  This requires changing the CPU threads to
joinable.  There is no substantial change because the threads run
in an infinite loop.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qdev: add a qdev_get_type() function and expose as a 'type' property

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

qom: add string property type

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

qom: add vga node to the pc composition tree

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

qom: optimize qdev_get_canonical_path using a parent link

The full tree search was a bit unreasonable.

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

rtc: add a dynamic property for retrieving the date

This really shows the power of dynamic object properties compared to qdev
static properties.

This property represents a complex structure who's format is preserved over the
wire.  This is enabled by visitors.

It also shows an entirely synthetic property that is not tied to device state.

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

rtc: make piix3 set the rtc as a child (v2)

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

dev: add an anonymous peripheral container

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

qdev: add explicitly named devices to the root complex

We first add a 'peripheral' container to the root device that we add user
created devices to.  This provides all user created devices with a unique and
isolated namespace.

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

qom: qom_{get, set} monitor commands (v2)

This allows clients to read and write device model properties through QMP.  QAPI
doesn't support Visitor types yet and these commands are special in that they
don't work with fixed types.

I've added a documentation stub to qapi-schema.json so we can keep consistency
there.

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

qmp: add qom-list command

This can be used to list properties in the device model.

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

qapi: allow a 'gen' key to suppress code generation

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

qom: add link properties (v2)

Links represent an ephemeral relationship between devices.  They are meant to
replace the qdev concept of busses by allowing more informal relationships
between devices.

Links are fairly limited in their usefulness without implementing QOM-style
subclassing and interfaces.

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

qom: add child properties (composition) (v3)

Child properties express a relationship of composition.

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

qdev: provide a path resolution (v2)

There are two types of supported paths--absolute paths and partial paths.

Absolute paths are derived from the root device and can follow child<> or
link<> properties.  Since they can follow link<> properties, they can be
arbitrarily long.  Absolute paths look like absolute filenames and are prefixed
with a leading slash.

Partial paths are look like relative filenames.  They do not begin with a
prefix.  The matching rules for partial paths are subtle but designed to make
specifying devices easy.  At each level of the composition tree, the partial
path is matched as an absolute path.  The first match is not returned.  At
least two matches are searched for.  A successful result is only returned if
only one match is founded.  If more than one match is found, a flag is returned
to indicate that the match was ambiguous.

At the end of the day, partial path support means that if you create a device
called 'ide0', you can just say 'ide0' as the path name and it will Just Work.
If we internally create a device called 'i440fx', you can just say 'i440fx' and
it will Just Work and long as you don't do anything silly.

A management tool should probably always use absolute paths since then they
don't have to deal with the possibility of ambiguity.

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

qdev: provide an interface to return canonical path from root (v2)

The canonical path is the path in the composition tree from the root to the
device.  This is effectively the name of the device.

This is an incredibly unefficient implementation that will be optimized in
a future patch.

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

qom: introduce root device

This is based on Jan's suggestion for how to do unique naming.  The root device
is the root of composition.  All devices are reachable via child<> links from
this device.

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

qom: register legacy properties as new style properties (v2)

Expose all legacy properties through the new QOM property mechanism.  The qdev
property types are exposed through the 'legacy<>' namespace.  They are always
visited as strings since they do their own string parsing.

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

qom: add new dynamic property infrastructure based on Visitors (v2)

qdev properties are settable only during construction and static to classes.
This isn't flexible enough for QOM.

This patch introduces a property interface for qdev that provides dynamic
properties that are tied to objects, instead of classes.  These properties are
Visitor based instead of string based too.

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

qom: add a reference count to qdev objects

To ensure that a device isn't removed from the graph until all of its links are
broken.

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

qiov: prevent double free or use-after-free

qemu_iovec_destroy does not clear the QEMUIOVector fully, and the data
could thus be used after free or freed again.  While I do not know any
example in the tree, I observed this using virtio-scsi (and SCSI
scatter/gather) when canceling DMA requests.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/cow: Return real error code

Signed-off-by: Li Zhi Hui <zhihuili@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

coroutine: switch per-thread free pool to a global pool

ucontext-based coroutines use a free pool to reduce allocations and
deallocations of coroutine objects.  The pool is per-thread, presumably
to improve locality.  However, as coroutines are usually allocated in
a vcpu thread and freed in the I/O thread, the pool accounting gets
screwed up and we end allocating and freeing a coroutine for every I/O
request.  This is expensive since large objects are allocated via the
kernel, and are not cached by the C runtime.

Fix by switching to a global pool.  This is safe since we're protected
by the global mutex.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qcow2: Allow >4 GB VM state

This is a compatible extension to the snapshot header format that allows
saving a 64 bit VM state size.

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

Documentation: Add qemu-img -t parameter in man page

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

qemu-img rebase: Fix for undersized backing files

Backing files may be smaller than the corresponding COW file. When
reading directly from the backing file, qemu-img rebase must consider
this and assume zero sectors after the end of backing files.

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

rbd: always set out parameter in qemu_rbd_snap_list

The caller expects psn_tab to be NULL when there are no snapshots or
an error occurs. This results in calling g_free on an invalid address.

Reported-by: Oliver Francke <Oliver@filoo.de>
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/qcow2.c: call qcow2_free_snapshots in the function of qcow2_close

Signed-off-by: Li Zhi Hui <zhihuili@linux.vnet.ibm.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: avoid useless checks on acb->bh

Coverity is confused by this "if" and reports leaks on acb->bh.
The bottom half is always deleted before releasing the AIOCB,
in either bdrv_aio_cancel_em or bdrv_aio_bh_cb.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: dma_bdrv_* does not return NULL

Initially attempted with the following semantic patch:

@ rule1 @
expression E;
statement S;
@@
  E =
(
   dma_bdrv_io
|  dma_bdrv_read
|  dma_bdrv_write
)
     (...);
(
- if (E == NULL) { ... }
|
- if (E)
    { <... S ...> }
)

which however did not match anything.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

dma: the passed io_func does not return NULL

Initially found with the following semantic patch:

@ type @
BlockDriverAIOCB *x;
expression E;
@@
  x = E;
- if (x == NULL) { ... }

@ acb1 @
expression E, E1;
@@
  E1->acb = E;
- if (E1->acb == NULL) { ... }

@ aiocb1 @
expression E, E1;
@@
  E1->aiocb = E;
- if (E1->aiocb == NULL) { ... }

@ acb @
expression E, E1;
@@
  E1.acb = E;
- if (E1.acb == NULL) { ... }

@ aiocb @
expression E, E1;
@@
  E1.aiocb = E;
- if (E1.aiocb == NULL) { ... }

but changed manually to include an assert.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: qemu_aio_get does not return NULL

Initially done with the following semantic patch:

@ rule1 @
expression E;
statement S;
@@
  E = qemu_aio_get (...);
(
- if (E == NULL) { ... }
|
- if (E)
    { <... S ...> }
)

which however missed occurrences in linux-aio.c and posix-aio-compat.c.
Those were done by hand.

The change in vdi_aio_setup's caller was also done by hand.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: simplify failure handling for bdrv_aio_multiwrite

Now that early failure of bdrv_aio_writev is not possible anymore,
mcb->num_requests can be set before the loop starts.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: bdrv_aio_* do not return NULL

Initially done with the following semantic patch:

@ rule1 @
expression E;
statement S;
@@
  E =
(
   bdrv_aio_readv
|  bdrv_aio_writev
|  bdrv_aio_flush
|  bdrv_aio_discard
|  bdrv_aio_ioctl
)
     (...);
(
- if (E == NULL) { ... }
|
- if (E)
    { <... S ...> }
)

which however missed the occurrence in block/blkverify.c
(as it should have done), and left behind some unused
variables.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

tcg/arm: remove fixed map code buffer restriction

On ARM, don't map the code buffer at a fixed location, and fix up the
call/goto tcg routines to let it do long jumps.

Mapping the code buffer at a fixed address could sometimes result in it being
mapped over the top of the heap with pretty random results.

Signed-off-by: Dr. David Alan Gilbert <david.gilbert@linaro.org>
Signed-off-by: Andrzej Zaborowski <andrew.zaborowski@intel.com>

Merge branch 'target-arm.for-upstream' of git://git.linaro.org/people/pmaydell/qemu-arm

Merge remote-tracking branch 'stefanha/trivial-patches-next' into staging

doc: Remove Symbian Virtual Platform

Commit d396a657baec8c6b7aa0c888746e0e2f78303650 removed the code
for SVP, so the documentation needs this update.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

migration.h: remove incoming_expected declarations

The variable is deleted by 1bcef683bf840a928d633755031ac572d5fdb851
So remove its declaration.

Acked-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

tcg: make tcg_const_ptr actually accept a pointer argument

Make tcg_const_ptr() include a cast so that you can pass it a
pointer. This allows us to drop the casts we had in all the places
that use this macro.

Acked-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

Fix spelling in comments, documentation and messages

accidently->accidentally
annother->another
choosen->chosen
consideres->considers
decriptor->descriptor
developement->development
paramter->parameter
preceed->precede
preceeding->preceding
priviledge->privilege
propogation->propagation
substraction->subtraction
throught->through
upto->up to
usefull->useful

Fix also grammar in posix-aio-compat.c

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

hw: Fix spelling in comments and code

compatiblity->compatibility
transfered->transferred
transfering->transferring

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

hw: Fix spelling in comments

adress->address
advertisment->advertisement
begining->beginning
bondary->boundary
controler->controller
controll->control
convertion->conversion
doesnt->doesn't
existant->existent
instuction->instruction
loosing->losing
managment->management
multipled->multiplied
negotation->negotiation
runing->running
teh->the
unchangable->unchangeable
writen->written
yeild->yield

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

hw: Fix spelling (licenced->licensed)

New code introduced old misspellings.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

fmopl: Fix spelling in code and comments

algorythm->algorithm
rythm->rhythm

I did not try to fix the coding standard, so checkpatch.pl
reports lots of violations.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

target-arm: Infer VFPv3 feature from VFPv4

VFP4 => VFP3

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target-arm: Infer VFP feature from VFPv3

VFP3 => VFP

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target-arm: Infer Thumb division feature from M profile

M => THUMB_DIV

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target-arm: Infer Thumb2 feature from ARMv7

V7 => THUMB2

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target-arm: Infer AUXCR feature from ARMv6

V6 && !M => AUXCR

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target-arm: Infer ARMv6(K) feature from ARMv7

V7 && M => V6
V7 && !M => V6K

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target-arm: Infer ARMv6 feature from v6K

V6K => V6

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target-arm: Infer ARMv5 feature from ARMv6

V6 => V5

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target-arm: Infer ARMv4T feature from ARMv5

V5 => V4T

Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

arm: Fix CP15 FSR (C5) domain setting

Return the correct value in the domain field in the cp15 DFSR
(C5) -- bug noticed during Xvisor development.

Signed-off-by: Jean-Christophe DUBOIS <jcd@tribudubois.net>
[Peter Maydell: reworded commit message]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

ccid: make threads joinable

Destroying a mutex that another thread might have just unlocked
is racy.  It usually works, but you cannot do that in general and
can lead to deadlocks or segfaults.  Change ccid to use joinable
threads instead.

(Also, qemu_mutex_init/qemu_cond_init were missing).

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qemu-thread: implement joinable threads for Win32

Rewrite the handshaking between qemu_thread_create and the
win32_start_routine, so that the thread can be joined without races.
Similar handshaking is done now between qemu_thread_exit and
qemu_thread_join.

This also simplifies how QemuThreads are initialized.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qemu-thread: implement joinable threads for POSIX

Allow to control if a QEMU thread is created joinable or not. Make it
not joinable by default to avoid that we keep the associated resources
around when terminating a thread without joining it (what we couldn't do
so far for obvious reasons).

The audio subsystem will need the join feature when converting it to
QEMU threading/locking abstractions, so provide that service.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qemu-thread: add API for joinable threads

Split from Jan's original qemu-thread-posix.c patch.  No semantic change,
just introduce the new API that POSIX and Win32 implementations will
conform to.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

syborg: drop support for Symbian Virtual Platform

The Symbian Virtual Platform was an ARM-based development and debugging
board.  Since Symbian has been disbanded and the code is no longer being
used it can now be removed.

Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

net: take ownership of fd in socket init functions

Today net/socket.c has no consistent policy for closing the socket file
descriptor when initialization fails.  This means we leak the file
descriptor in some cases or we could also try to close it twice.

Make error paths consistent by taking ownership of the file descriptor
and closing it on error.

Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

net: expand tabs in net/socket.c

In order to make later patches sane, expand the tab characters and
conform to QEMU coding style now.

Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

guest agent: add supported command list to guest-info RPC

Not that there is blacklisting functionality we can no longer infer
the agent's capabilities via version. This patch extends the current
guest-info RPC to also return a list of dictionaries containing the name
of each supported RPC, along with a boolean indicating whether or not
the command has been disabled by a guest administrator/distro.

Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

guest agent: add RPC blacklist command-line option

This adds a command-line option, -b/--blacklist, that accepts a
comma-seperated list of RPCs to disable, or prints a list of
available RPCs if passed "?".

In consequence this also adds general blacklisting and RPC listing
facilities to the new QMP dispatch/registry facilities, should the
QMP monitor ever have a need for such a thing.

Ideally, to avoid support/compatability issues in the future,
blacklisting guest agent functionality will be the exceptional
case, but we add the functionality here to handle guest administrators
with specific requirements.

Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Add a .mailmap to map pre-git-conversion authors to friendly names

Add a .mailmap file so 'git shortlog' can map the unfriendly
pre-git-conversion author entries to real names.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

unix_close(): check for close() errors too (v2)

In case close() fails, we want to report the error back.

Changes v1 -> v2:
 - Use braces on if statement to match coding style

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

tcp_close(): check for close() errors too (v2)

In case close() fails, we want to report the error back.

Changes v1 -> v2:
 - Use braces on if statement to match coding style

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

exec_close(): return -errno on errors (v2)

All qemu_fclose() callers were already changed to accept any negative
value as error, so we now can change it to return -errno.

When the process exits with a non-zero exit code, we return -EIO to as a
fake errno value.

Changes v1 -> v2:
 - Don't use "//" comments, to make checkpatch.pl happy

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

stdio_fclose: return -errno on errors (v2)

This is what qemu_fclose() expects.

Changes v1 -> v2:
 - Add braces to if statement to match coding style

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

stdio_pclose: return -errno on error (v3)

This is what qemu_fclose() expects.

Changes v1 -> v2:
 - On success, keep returning pclose() return value, instead of always 0.

Changes v2 -> v3:
 - Add braces on if statements to match coding style

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qemu_fclose: return last_error if set (v3)

This will make sure no error will be missed as long as callers always
check for qemu_fclose() return value. For reference, this is the
complete list of qemu_fclose() callers:

 - exec_close(): already fixed to check for negative values, not -1
 - migrate_fd_cleanup(): already fixed to consider only negative values
   as error, not any non-zero value
 - exec_accept_incoming_migration(): no return value check (yet)
 - fd_accept_incoming_migration(): no return value check (yet)
 - tcp_accept_incoming_migration(): no return value check (yet)
 - unix_accept_incoming_migration(): no return value check (yet)
 - do_savevm(): no return value check (yet)
 - load_vmstate(): no return value check (yet)

Changes v1 -> v2:
 - Add small comment about the need to return previously-spotted errors

Changes v2 -> v3:
 - Add braces to "if" statements to match coding style

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

migrate_fd_cleanup: accept any negative qemu_fclose() value as error

Also, we now return the qemu_fclose() value unchanged to the caller. For
reference, the migrate_fd_cleanup() callers are the following:

- migrate_fd_completed(): any negative value is considered an
  error, so the change is OK.
- migrate_fd_error(): doesn't check the migrate_fd_cleanup() return value
- migrate_fd_cancel(): doesn't check the migrate_fd_cleanup() return
  value

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

exec_close(): accept any negative value as qemu_fclose() error

Note that we don't return the unchanged return value back yet, because
we need to change all qemu_fclose() callers to accept any positive value
as success.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

QEMUFileCloseFunc: add return value documentation (v2)

qemu_fclose() and QEMUFile->close will return -errno on error, and any
positive value on success.

We need the positive non-zero success values because
migration-exec.c:exec_close() relies on non-zero return values to get
the process exit code.

Changes v1 -> v2:
 - Cosmetic spelling change on comment text

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

savevm: use qemu_file_set_error() instead of setting last_error directly

Some code uses qemu_file_set_error() already, so use it everywhere
when setting last_error, for consistency.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

hw/usb-net.c: Fix precedence bug when checking rndis_state

"!X == 2" is always false (spotted by Coverity), so the checks
for whether rndis is in the correct state would never fire.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Remove unnecessary casts from PCI DMA code in usb-uhci

This patch removes some unnecessary casts in the usb-uhci device,
introduced by commit fff23ee9a5de74ab111b3cea9eec56782e7d7c50
'usb-uhci: Use PCI DMA stub functions'.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Remove unnecessary casts from PCI DMA code in usb-ehci

This patch removes some unnecessary casts in the usb-ehci device,
introduced by commit 68d553587c0aa271c3eb2902921b503740d775b6
'usb-ehci: Use PCI DMA stub functions'.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Remove unnecessary casts from PCI DMA code in rtl8139

This patch removes some unnecessary casts in the rtl8139 device,
introduced by commit 3ada003aee2004d24f23b9cd6f4eda87d9601ddb
'rtl8139: Use PCI DMA stub functions'.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Remove unnecessary casts from PCI DMA code in lsi53c895a

This patch removes some unnecessary casts in the lsi53c895a device,
introduced by commit 9ba4524cda1348cbe741535f77815dca6a57da05
'lsi53c895a: Use PCI DMA stub functions'.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Remove unnecessary casts from PCI DMA code in PCI IDE

This patch removes some unnecessary casts in the PCI IDE device,
introduced by commit 552908fef5b67ad9d96b76d7cb8371ebc26c9bc8
'PCI IDE: Use PCI DMA stub functions'.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Remove unnecessary casts from PCI DMA code in e1000

This patch removes some unnecessary casts in the e1000 device,
introduced by commit 62ecbd353d25e62c4a6c327ea88ba5404e13507a 'e1000:
Use PCI DMA stub functions'.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Remove unnecessary casts from PCI DMA code in eepro100

This patch removes some unnecessary casts in the eepro100 device,
introduced by commit 16ef60c9a8eeee269f7cbc95219a431b1d7cbf29
'eepro100: Use PCI DMA stub functions'.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

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

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

cris: Handle conditional stores on CRISv10

Signed-off-by: Stefan Sandstrom <Stefan.Sandstrom@axis.com>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

hw/mpcore.c: Merge with hw/arm11mpcore.c

hw/mpcore.c is now implementing only ARM11MPCore specific peripherals,
and is #included only from hw/arm11mpcore.c, so just merge it into that
file.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/a9mpcore.c: Implement A9MP peripherals rather than 11MPcore ones

Implement the A9MP private peripheral region correctly, rather
than piggybacking on the 11MPCore code; the two CPUs are not the
same in this area.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/mpcore: Clean up mpcore_priv_read/write as they are now SCU only

The only code left in mpcore_priv_read and mpcore_priv_write is now
the implementation of the SCU registers. Clean up by renaming functions
and removing some unnecessary conditionals to make this clearer.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/realview_gic: Use GIC memory region for the CPU interface

Use the GIC provided memory region for the CPU interface rather
than implementing our own.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>