migration: move migration_fd_put_ready()

Put it near its use and un-export it.

Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: add XFER_LIMIT_RATIO

The "magic" divisions by 10 are there because of the value of BUFFER_DELAY.
Introduce a constant to explain them better.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: move buffered_file.c code into migration.c

This only moves the code (also from buffered_file.h to migration.h).
Fix whitespace until checkpatch is happy.

Signed-off-by: Juan Quintela <quintela@redhat.com>

savevm: New save live migration method: pending

Code just now does (simplified for clarity)

    if (qemu_savevm_state_iterate(s->file) == 1) {
       vm_stop_force_state(RUN_STATE_FINISH_MIGRATE);
       qemu_savevm_state_complete(s->file);
    }

Problem here is that qemu_savevm_state_iterate() returns 1 when it
knows that remaining memory to sent takes less than max downtime.

But this means that we could end spending 2x max_downtime, one
downtime in qemu_savevm_iterate, and the other in
qemu_savevm_state_complete.

Changed code to:

    pending_size = qemu_savevm_state_pending(s->file, max_size);
    DPRINTF("pending size %lu max %lu\n", pending_size, max_size);
    if (pending_size >= max_size) {
        ret = qemu_savevm_state_iterate(s->file);
     } else {
        vm_stop_force_state(RUN_STATE_FINISH_MIGRATE);
        qemu_savevm_state_complete(s->file);
     }

So what we do is: at current network speed, we calculate the maximum
number of bytes we can sent: max_size.

Then we ask every save_live section how much they have pending.  If
they are less than max_size, we move to complete phase, otherwise we
do an iterate one.

This makes things much simpler, because now individual sections don't
have to caluclate the bandwidth (it was implossible to do right from
there).

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

buffered_file: unfold buffered_append in buffered_put_buffer

It was the only user, and now buffered_put_buffer just do the append

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

buffered_file: don't flush on put buffer

We call buffered_put_buffer with iothread held, and buffered_flush() does
synchronous writes.  We only want to do the synchronous writes outside.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

buffered_file: Unfold the trick to restart generating migration data

This was needed before due to the way that the callbacks worked.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

migration: just lock migrate_fd_put_ready

Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: remove unfreeze logic

Now that we have a thread, and blocking writes, we don't need it.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

migration: make writes blocking

Move all the writes to the migration_thread, and make writings
blocking.  Notice that are still using the iothread for everything
that we do.

Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: move migration thread init code to migrate_fd_put_ready

This way everything related with migration is run on the migration
thread and no locking is needed.

Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: make qemu_fopen_ops_buffered() return void

We want the file assignment to happen before the thread is created to
avoid locking, so we just do it before creating the thread.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>

buffered_file: Move from using a timer to use a thread

We still protect everything except the wait with the iothread lock.
But we moved from a timer to a thread.  Steps one by one.

We also need to detect when we have finished with a variable "complete".

Signed-off-by: Juan Quintela <quintela@redhat.com>

protect the ramlist with a separate mutex

Add the new mutex that protects shared state between ram_save_live
and the iothread.  If the iothread mutex has to be taken together
with the ramlist mutex, the iothread shall always be _outside_.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Umesh Deshpande <udeshpan@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>

add a version number to ram_list

This will be used to detect if last_block might have become invalid
across different calls to ram_save_live.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Umesh Deshpande <udeshpan@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>

exec: sort the memory from biggest to smallest

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

exec: change RAM list to a TAILQ

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

exec: change ramlist from MRU order to a 1-item cache

Most of the time, only 2 items will be active (from/to for a string operation,
or code/data).  But TCG guests likely won't have gigabytes of memory, so
this actually goes down to 1 item.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration-fd: remove duplicate include

Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: include qemu-file.h

They don't use/know anything about buffered-file.

Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: remove double call to migrate_fd_close

The call in buffered_close is enough, because buffered_close is called
already by migrate_fd_cleanup.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

buffered_file: do not send more than s->bytes_xfer bytes per tick

Sending more was possible if the buffer was large.

Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: fix migration_bitmap leak

Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Merge remote-tracking branch 'bonzini/header-dirs' into staging

* bonzini/header-dirs: (45 commits)
  janitor: move remaining public headers to include/
  hw: move executable format header files to hw/
  fpu: move public header file to include/fpu
  softmmu: move remaining include files to include/ subdirectories
  softmmu: move include files to include/sysemu/
  misc: move include files to include/qemu/
  qom: move include files to include/qom/
  migration: move include files to include/migration/
  monitor: move include files to include/monitor/
  exec: move include files to include/exec/
  block: move include files to include/block/
  qapi: move include files to include/qobject/
  janitor: add guards to headers
  qapi: make struct Visitor opaque
  qapi: remove qapi/qapi-types-core.h
  qapi: move inclusions of qemu-common.h from headers to .c files
  ui: move files to ui/ and include/ui/
  qemu-ga: move qemu-ga files to qga/
  net: reorganize headers
  net: move net.c to net/
  ...

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

janitor: move remaining public headers to include/

Headers in the root directory are now used only from within that
directory.

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

hw: move executable format header files to hw/

Or delete a.out.h which is unused.

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

fpu: move public header file to include/fpu

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

softmmu: move remaining include files to include/ subdirectories

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

softmmu: move include files to include/sysemu/

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

misc: move include files to include/qemu/

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

qom: move include files to include/qom/

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

migration: move include files to include/migration/

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

monitor: move include files to include/monitor/

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

exec: move include files to include/exec/

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

block: move include files to include/block/

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

qapi: move include files to include/qobject/

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

janitor: add guards to headers

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

qapi: make struct Visitor opaque

Move its definition from qapi-visit-core.h to qapi-visit-impl.h.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

qapi: remove qapi/qapi-types-core.h

The file is only including error.h and qerror.h.  Prefer explicit
inclusion of whatever files are needed.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

qapi: move inclusions of qemu-common.h from headers to .c files

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

ui: move files to ui/ and include/ui/

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

qemu-ga: move qemu-ga files to qga/

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

net: reorganize headers

Move public headers to include/net, and leave private headers in net/.
Put the virtio headers in include/net/tap.h, removing the multiple copies
that existed.  Leave include/net/tap.h as the interface for NICs, and
net/tap_int.h as the interface for OS-specific parts of the tap backend.

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

net: move net.c to net/

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

net: do not include net.h everywhere

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

net: move Bluetooth stuff out of net.h

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

janitor: do not include qemu-char everywhere

Touching char/char.h basically causes the whole of QEMU to
be rebuilt.  Avoid this, it is usually unnecessary.

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

janitor: do not rely on indirect inclusions of or from qemu-char.h

Various header files rely on qemu-char.h including qemu-config.h or
main-loop.h, but they really do not need qemu-char.h at all (particularly
interesting is the case of the block layer!).  Clean this up, and also
add missing inclusions of qemu-char.h itself.

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

build: kill libuser

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

build: kill libdis, move disassemblers to disas/

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

build: create ldscripts/

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

build: move rules from Makefile to */Makefile.objs

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

build: add $(TARGET_DIR) to "GEN config-target.h" lines

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

build: adjust setting of QEMU_INCLUDES

Make it correct for nested directories, and move the static part
from Makefile to configure.

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

vscclient: use per-target variables

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

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

* stefanha/trivial-patches:
  configure: Earlier pkg-config probe
  vmmouse_reset(): remove minimal code duplication
  linux-user/syscall.c: remove wrong forward decl of setgroups()
  fix build error on ARM due to wrong glibc check
  gitignore: Add virtfs-proxy-helper
  arm_gic: Add cpu nr to Raised IRQ message
  zynq_slcr: Compile time warning fixes.
  pflash_cfi0x: Send debug messages to stderr
  pflash_cfi01: qemu_log_mask "unimplemented" msg
  net, hub: fix the indent in the comments

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

Merge remote-tracking branch 'amit/master' into staging

* amit/master:
  virtio-serial-bus: assert port is non-null in remove_port()
  virtio-serial-bus: send_control_msg() should not deal with cpkts
  virtio-serial: delete timer if active during exit
  virtio-serial: allocate post_load only at load-time
  virtio-serial: move active ports loading to separate function
  virtio-serial: use uint32_t to count ports

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

Merge remote-tracking branch 'spice/spice.v66' into staging

* spice/spice.v66:
  docs: add spice-port-fqdn.txt
  spice-qemu-char: register spicevmc ports during qemu_spice_init()
  spice-qemu-char: keep a list of spice chardev
  spice-qemu-char: add spiceport chardev
  spice-qemu-char: factor out CharDriverState creation
  spice-qemu-char: write to chardev whatever amount it can read
  qxl+vnc: register a vm state change handler for dummy spice_server
  qxl: save qemu_create_displaysurface_from result

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

Merge remote-tracking branch 'sstabellini/xen-20121217' into staging

* sstabellini/xen-20121217:
  cpu_ioreq_pio, cpu_ioreq_move: i should be uint32_t rather than int
  cpu_ioreq_pio, cpu_ioreq_move: introduce read_phys_req_item, write_phys_req_item
  Fix compile errors when enabling Xen debug logging.
  xen: fix trivial PCI passthrough MSI-X bug
  xen: implement support for secondary consoles in the console backend

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

configure: Earlier pkg-config probe

Probe pkg-config before it is used for the first time (libseccomp check).

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

vmmouse_reset(): remove minimal code duplication

Commit 069ab0eb added a vmmouse_disable() call to vmmouse_reset().
vmmouse_disable() resets the status already.

Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

linux-user/syscall.c: remove wrong forward decl of setgroups()

this declaration is wrong:
the correct prototype on linux is:
int setgroups(size_t size, const gid_t *list);

since by default musl libc exposes this symbol in unistd.h
additionally to grp.h, the wrong declaration causes a build error.

the proper fix is to simply include the correct header.

Signed-off-by: John Spencer <maillist-qemu@barfooze.de>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

fix build error on ARM due to wrong glibc check

the test for glibc < 2 "succeeds" wrongly for any non-glibc C library,
and breaks the build on musl libc.
we must first test if __GLIBC__ is defined at all, before using it
unconditionally.

Signed-off-by: John Spencer <maillist-qemu@barfooze.de>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

gitignore: Add virtfs-proxy-helper

Signed-off-by: Cole Robinson <crobinso@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

arm_gic: Add cpu nr to Raised IRQ message

Add the relevant CPU nr to this debug message to make IRQ debugging more
informative.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

zynq_slcr: Compile time warning fixes.

Few warnings when compiled with debug printfs enabled. Fixed all.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

pflash_cfi0x: Send debug messages to stderr

These debug info messages should go to stderr rather than stdout.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

pflash_cfi01: qemu_log_mask "unimplemented" msg

This printf is informing the user of unimplemented functionality. It should be
re-directed to qemu_log(LOG_UNIMP, ...) accordingly.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

net, hub: fix the indent in the comments

  Remove some redundant blanks in the comments of
net_hub_id_for_client().

Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

virtio-serial-bus: assert port is non-null in remove_port()

remove_port() is called from qdev's unplug callback, and we're certain
the port will be found in our list of ports.  Adding an assert()
documents this.

This was flagged by Coverity, fix suggested by Markus.

CC: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>

virtio-serial-bus: send_control_msg() should not deal with cpkts

Stuff the cpkt before calling send_control_msg().  This function should
not be concerned about contents of the buffer it receives.

A few code refactorings recently have made making this change easier
than earlier.

Coverity and clang have flagged this code several times in the past
(cpkt->id not set before send_control_event() passed it on to
send_control_msg()).  This will finally eliminate the false-positive.

CC: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>

libcacard: simplify rules for recursive build

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

Merge commit '1dd3a74d2ee2d873cde0b390b536e45420b3fe05' into HEAD

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

docs: add spice-port-fqdn.txt

Start a simple org.qemu.* registry of well known name.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

spice-qemu-char: register spicevmc ports during qemu_spice_init()

Do the delayed registration of spicevmc ports after Spice server is
initialized.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

spice-qemu-char: keep a list of spice chardev

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

spice-qemu-char: add spiceport chardev

Add a new spice chardev to allow arbitrary communication between the
host and the Spice client via the spice server.

Examples:

This allows the Spice client to have a special port for the qemu
monitor:

... -chardev spiceport,name=org.qemu.monitor,id=monitorport
    -mon chardev=monitorport

v2:
- remove support for chardev to chardev linking
- conditionnaly compile with SPICE_SERVER_VERSION

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

spice-qemu-char: factor out CharDriverState creation

Make the CharDriverState creation code reusable by spicevmc port.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

spice-qemu-char: write to chardev whatever amount it can read

The current code waits until the chardev can read MIN(len, VMC_MAX)
But some chardev may never reach than amount, in fact some of them
will only ever accept write of 1. Fix the min computation and remove
the VMC_MAX constant.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

qxl+vnc: register a vm state change handler for dummy spice_server

When qxl + vnc are used, a dummy spice_server is initialized.
The spice_server has to be told when the VM runstate changes,
which is what this patch does.

Without it, from qxl_send_events(), the following error message is shown:
  qxl_send_events: spice-server bug: guest stopped, ignoring

Cc: qemu-stable@nongnu.org
Signed-off-by: Uri Lublin <uril@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

qxl: save qemu_create_displaysurface_from result

Spotted by Coverity.

https://bugzilla.redhat.com/show_bug.cgi?id=885644

Cc: qemu-stable@nongnu.org
Reported-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

cpu_ioreq_pio, cpu_ioreq_move: i should be uint32_t rather than int

The current code compare i (int) with req->count (uint32_t) in a for
loop, risking an infinite loop if req->count is equal to UINT_MAX.

Also i is only used in comparisons or multiplications with unsigned
integers.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Cc: Dongxiao Xu <dongxiao.xu@intel.com>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>

cpu_ioreq_pio, cpu_ioreq_move: introduce read_phys_req_item, write_phys_req_item

Replace a lot of formulaic multiplications (containing casts, no less)
with calls to a pair of functions.  This encapsulates in a single
place the operations which require care relating to integer overflow.

Cc: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Fix compile errors when enabling Xen debug logging.

Signed-off-by: Sander Eikelenboom <linux@eikelenboom.it>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

xen: fix trivial PCI passthrough MSI-X bug

We are currently passing entry->data as address parameter. Pass
entry->addr instead.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Tested-by: Sander Eikelenboom <linux@eikelenboom.it>
Xen-devel: http://marc.info/?l=xen-devel&m=135515462613715

xen: implement support for secondary consoles in the console backend

This patch corresponds to commit
840184a106bc24e745beda5c77e392f6cecd2bc9 from
git://xenbits.xensource.com/qemu-xen-unstable.git.

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

kvm: do not flush after deleting gsi

Deleting a GSI isn't necessary: it is enough
to stop using it. Delay flush until an entry is used.

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

pci_bus.h: tweak include guards

Now that header has been renamed, tweak include guards
to match.

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

pci_bus: update comment

Don't ask everyone to desist from including this header,
simply recommend using accessors.

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

pci: rename pci_internals.h pci_bus.h

There are lots of external users of pci_internals.h,
apparently making it an internal interface only didn't
work out. Let's stop pretending it's an internal header.

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

Revert "pci: prepare makefiles for pci code reorganization"

This reverts commit 475d67c3bcd6ba9fef917b6e59d96ae69eb1a9b4.

Now that all users have been updated, we don't need the
makefile hack or the softlink anymore.

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

pci: fix path for local includes

Include dependencies from pci core using the correct path.
This is required now that it's in the separate directory.
Need to check whether they can be minimized, for now,
keep the code as is.

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

pci: update all users to look in pci/

update all users so we can remove the makefile hack.

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

pci: move pci core code to hw/pci

Move files and modify makefiles to pick them at the
new location.

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

exec: refactor cpu_restore_state

Refactor common code around calls to cpu_restore_state().

tb_find_pc() has now no external users, make it static.

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

exec: move TB handling to translate-all.c

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

exec: extract TB watchpoint check

Will be moved by the next patch.

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

exec: fix coding style

Fix coding style in areas to be moved by later patches.

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Merge branch 'ppc-for-upstream' of git://repo.or.cz/qemu/agraf

* 'ppc-for-upstream' of git://repo.or.cz/qemu/agraf: (40 commits)
  pseries: Increase default NVRAM size
  target-ppc: Don't use hwaddr to represent hardware state
  PPC: e500: pci: Export slot2irq calculation
  PPC: E500plat: Make a lot of PCI slots available
  PPC: E500: Move PCI slot information into params
  PPC: E500: Generate dt pci irq map dynamically
  PPC: E500: PCI: Make IRQ calculation more generic
  PPC: E500: PCI: Make first slot qdev settable
  openpic: Accelerate pending irq search
  openpic: fix minor coding style issues
  MSI-X: Fix endianness
  PPC: e500: Declare pci bridge as bridge
  PPC: e500: Add MSI support
  openpic: add Shared MSI support
  openpic: make brr1 model specific
  openpic: convert to qdev
  openpic: remove irq_out
  openpic: rename openpic_t to OpenPICState
  openpic: convert simple reg operations to builtin bitops
  openpic: remove unused type variable
  ...

target-xtensa: fix ITLB/DTLB page protection flags

With MMU option xtensa architecture has two TLBs: ITLB and DTLB. ITLB is
only used for code access, DTLB is only for data. However TLB entries in
both TLBs have attribute field controlling write and exec access. These
bits need to be properly masked off depending on TLB type before being
used as tlb_set_page prot argument. Otherwise the following happens:

(1) ITLB entry for some PFN gets invalidated
(2) DTLB entry for the same PFN gets updated, attributes allow code
    execution
(3) code at the page with that PFN is executed (possible due to step 2),
    entry for the TB is written into the jump cache
(4) QEMU TLB entry for the PFN gets replaced with an entry for some
    other PFN
(5) code in the TB from step 3 is executed (possible due to jump cache)
    and it accesses data, for which there's no DTLB entry, causing DTLB
    miss exception
(6) re-translation of the TB from step 5 is attempted, but there's no
    QEMU TLB entry nor xtensa ITLB entry for that PFN, which causes ITLB
    miss exception at the TB start address
(7) ITLB miss exception is handled by the guest, but execution is
    resumed from the beginning of the faulting TB (the point where ITLB
    miss occured), not from the point where DTLB miss occured, which is
    wrong.

With that fix the above scenario causes ITLB miss exception (that used
to be step 7) at step 3, right at the beginning of the TB.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>