job: Add lifecycle QMP commands

This adds QMP commands that control the transition between states of the
job lifecycle.

Signed-off-by: Kevin Wolf <[email protected]>

job: Add JOB_STATUS_CHANGE QMP event

This adds a QMP event that is emitted whenever a job transitions from
one status to another.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Eric Blake <[email protected]>

job: Introduce qapi/job.json

This adds a separate schema file for all job-related definitions that
aren't tied to the block layer.

For a start, move the enums JobType, JobStatus and JobVerb.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Eric Blake <[email protected]>

job: Move progress fields to Job

BlockJob has fields .offset and .len, which are actually misnomers today
because they are no longer tied to block device sizes, but just progress
counters. As such they make a lot of sense in generic Jobs.

This patch moves the fields to Job and renames them to .progress_current
and .progress_total to describe their function better.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Add job_transition_to_ready()

The transition to the READY state was still performed in the BlockJob
layer, in the same function that sent the BLOCK_JOB_READY QMP event.

This patch brings the state transition to the Job layer and implements
the QMP event using a notifier called from the Job layer, like we
already do for other events related to state transitions.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Add job_is_ready()

Instead of having a 'bool ready' in BlockJob, add a function that
derives its value from the job status.

At the same time, this fixes the behaviour to match what the QAPI
documentation promises for query-block-job: 'true if the job may be
completed'. When the ready flag was introduced in commit ef6dbf1e46e,
the flag never had to be reset to match the description because after
being ready, the jobs would immediately complete and disappear.

Job transactions and manual job finalisation were introduced only later.
With these changes, jobs may stay around even after having completed
(and they are not ready to be completed a second time), however their
patches forgot to reset the ready flag.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Add job_dismiss()

This moves block_job_dismiss() to the Job layer.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Add job_yield()

This moves block_job_yield() to the Job layer.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

block: Cancel job in bdrv_close_all() callers

Now that we cancel all jobs and not only block jobs on shutdown, doing
that in bdrv_close_all() isn't really appropriate any more. Move the
job_cancel_sync_all() call to the callers, and only assert that there
are no job running in bdrv_close_all().

Signed-off-by: Kevin Wolf <[email protected]>

job: Move completion and cancellation to Job

This moves the top-level job completion and cancellation functions from
BlockJob to Job.

Signed-off-by: Kevin Wolf <[email protected]>

job: Move transactions to Job

This moves the logic that implements job transactions from BlockJob to
Job.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Switch transactions to JobTxn

This doesn't actually move any transaction code to Job yet, but it
renames the type for transactions from BlockJobTxn to JobTxn and makes
them contain Jobs rather than BlockJobs

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Move job_finish_sync() to Job

block_job_finish_sync() doesn't contain anything block job specific any
more, so it can be moved to Job.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Move .complete callback to Job

This moves the .complete callback that tells a READY job to complete
from BlockJobDriver to JobDriver. The wrapper function job_complete()
doesn't require anything block job specific any more and can be moved
to Job.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Add job_drain()

block_job_drain() contains a blk_drain() call which cannot be moved to
Job, so add a new JobDriver callback JobDriver.drain which has a common
implementation for all BlockJobs. In addition to this we keep the
existing BlockJobDriver.drain callback that is called by the common
drain implementation for all block jobs.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Convert block_job_cancel_async() to Job

block_job_cancel_async() did two things that were still block job
specific:

* Setting job->force. This field makes sense on the Job level, so we can
  just move it. While at it, rename it to job->force_cancel to make its
  purpose more obvious.

* Resetting the I/O status. This can't be moved because generic Jobs
  don't have an I/O status. What the function really implements is a
  user resume, except without entering the coroutine. Consequently, it
  makes sense to call the .user_resume driver callback here which
  already resets the I/O status.

  The old block_job_cancel_async() has two separate if statements that
  check job->iostatus != BLOCK_DEVICE_IO_STATUS_OK and job->user_paused.
  However, the former condition always implies the latter (as is
  asserted in block_job_iostatus_reset()), so changing the explicit call
  of block_job_iostatus_reset() on the former condition with the
  .user_resume callback on the latter condition is equivalent and
  doesn't need to access any BlockJob specific state.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Move single job finalisation to Job

This moves the finalisation of a single job from BlockJob to Job.

Some part of this code depends on job transactions, and job transactions
call this code, we introduce some temporary calls from Job functions to
BlockJob ones. This will be fixed once transactions move to Job, too.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Add job_event_*()

Go through the Job layer in order to send QMP events. For the moment,
these functions only call a notifier in the BlockJob layer that sends
the existing commands.

This uses notifiers rather than JobDriver callbacks because internal
users of jobs won't receive QMP events, but might still be interested
in getting notified for the events.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

blockjob: Split block_job_event_pending()

block_job_event_pending() doesn't only send a QMP event, but it also
transitions to the PENDING state. Split the function so that we get one
part only sending the event (like other block_job_event_* functions) and
another part that does the state transition.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Move BlockJobCreateFlags to Job

This renames the BlockJobCreateFlags constants, moves a few JOB_INTERNAL
checks to job_create() and the auto_{finalize,dismiss} fields from
BlockJob to Job.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Replace BlockJob.completed with job_is_completed()

Since we introduced an explicit status to block job, BlockJob.completed
is redundant because it can be derived from the status. Remove the field
from BlockJob and add a function to derive it from the status at the Job
level.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Move pause/resume functions to Job

While we already moved the state related to job pausing to Job, the
functions to do were still BlockJob only. This commit moves them over to
Job.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Add job_sleep_ns()

There is nothing block layer specific about block_job_sleep_ns(), so
move the function to Job.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: John Snow <[email protected]>
Reviewed-by: Max Reitz <[email protected]>

job: Move coroutine and related code to Job

This commit moves some core functions for dealing with the job coroutine
from BlockJob to Job. This includes primarily entering the coroutine
(both for the first and reentering) and yielding explicitly and at pause
points.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Move defer_to_main_loop to Job

Move the defer_to_main_loop functionality from BlockJob to Job.

The code can be simplified because we can use job->aio_context in
job_defer_to_main_loop_bh() now, instead of having to access the
BlockDriverState.

Probably taking the data->aio_context lock in addition was already
unnecessary in the old code because we didn't actually make use of
anything protected by the old AioContext except getting the new
AioContext, in case it changed between scheduling the BH and running it.
But it's certainly unnecessary now that the BDS isn't accessed at all
any more.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Add Job.aio_context

When block jobs need an AioContext, they just take it from their main
block node. Generic jobs don't have a main block node, so we need to
assign them an AioContext explicitly.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Move cancelled to Job

We cannot yet move the whole logic around job cancelling to Job because
it depends on quite a few other things that are still only in BlockJob,
but we can move the cancelled field at least.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Add reference counting

This moves reference counting from BlockJob to Job.

In order to keep calling the BlockJob cleanup code when the job is
deleted via job_unref(), introduce a new JobDriver.free callback. Every
block job must use block_job_free() for this callback, this is asserted
in block_job_create().

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Move state transitions to Job

This moves BlockJob.status and the closely related functions
(block_)job_state_transition() and (block_)job_apply_verb to Job. The
two QAPI enums are renamed to JobStatus and JobVerb.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>
Reviewed-by: Eric Blake <[email protected]>

job: Maintain a list of all jobs

This moves the job list from BlockJob to Job. Now we can check for
duplicate IDs in job_create().

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Add job_delete()

This moves freeing the Job object and its fields from block_job_unref()
to job_delete().

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Add JobDriver.job_type

This moves the job_type field from BlockJobDriver to JobDriver.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Rename BlockJobType into JobType

QAPI types aren't externally visible, so we can rename them without
causing problems. Before we add a job type to Job, rename the enum
so it can be used for more than just block jobs.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Eric Blake <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

job: Create Job, JobDriver and job_create()

This is the first step towards creating an infrastructure for generic
background jobs that aren't tied to a block device. For now, Job only
stores its ID and JobDriver, the rest stays in BlockJob.

The following patches will move over more parts of BlockJob to Job if
they are meaningful outside the context of a block job.

BlockJob.driver is now redundant, but this patch leaves it around to
avoid unnecessary churn. The next patches will get rid of almost all of
its uses anyway so that it can be removed later with much less churn.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Max Reitz <[email protected]>
Reviewed-by: John Snow <[email protected]>

blockjob: Improve BlockJobInfo.offset/len documentation

Clarify that len is just an estimation of the end value of offset, and
that offset increases monotonically while len can change arbitrarily.

While touching the documentation of offset, move it directly after len
to match the order of the declaration below.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Eric Blake <[email protected]>
Reviewed-by: John Snow <[email protected]>

blockjob: Update block-job-pause/resume documentation

Commit 0ec4dfb8d changed block-job_pause/resume so that they return an
error if they don't do anything because the job is already
paused/running. It forgot to update the documentation, so do that now.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Eric Blake <[email protected]>
Reviewed-by: John Snow <[email protected]>

qemu-iotests: Remove MIG_SOCKET from non-migration tests

185 and 191 define a MIG_SOCKET even though they don't do anything with
migration. Remove the useless variable.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Juan Quintela <[email protected]>

qemu-iotests: Add more tests to "migration" group

grep for "migrate" turns up a few test cases which use migration, but
haven't been in the "migration" group so far. Add them to the group.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Juan Quintela <[email protected]>

sheepdog: Remove unnecessary NULL check in sd_prealloc()

In commit 8b9ad56e9cbfd852a, we removed the code that could result
in our getting to sd_prealloc()'s out_with_err_set label with a
NULL blk pointer. That makes the NULL check in the error-handling
path unnecessary, and Coverity gripes about it (CID 1390636).
Delete the redundant check.

Signed-off-by: Peter Maydell <[email protected]>
Signed-off-by: Kevin Wolf <[email protected]>

qemu-iotests: 086 doesn't work with NFS

The reference output file only works for file. 'qemu-img convert -p'
makes a lot more progress updates for NFS than for file, so disable the
test for NFS.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Fam Zheng <[email protected]>

qemu-iotests: Filter NFS paths

NFS paths were only partially filtered in _filter_img_create, _img_info
and _filter_img_info, resulting in "nfs://127.0.0.1TEST_DIR/t.IMGFMT".
This adds another replacement to the sed calls that matches the test
directory not as a host path, but as an NFS URL (the prefix as used for
$TEST_IMG).

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Fam Zheng <[email protected]>

qemu-iotests: Fix paths for NFS

Test cases were trying to use nfs:// URLs as local filenames, which made
every test fail for NFS. With TEST_IMG and TEST_IMG_FILE set like for
the other protocols, NFS tests can pass again.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Fam Zheng <[email protected]>

Merge remote-tracking branch 'remotes/ehabkost/tags/x86-next-pull-request' into staging

Speculative store buffer bypass mitigation (CVE-2018-3639)

# gpg: Signature made Mon 21 May 2018 23:00:46 BST
# gpg:                using RSA key 2807936F984DC5A6
# gpg: Good signature from "Eduardo Habkost <[email protected]>"
# Primary key fingerprint: 5A32 2FD5 ABC4 D3DB ACCF  D1AA 2807 936F 984D C5A6

* remotes/ehabkost/tags/x86-next-pull-request:
  i386: define the AMD 'virt-ssbd' CPUID feature bit (CVE-2018-3639)
  i386: Define the Virt SSBD MSR and handling of it (CVE-2018-3639)
  i386: define the 'ssbd' CPUID feature bit (CVE-2018-3639)

Signed-off-by: Peter Maydell <[email protected]>

i386: define the AMD 'virt-ssbd' CPUID feature bit (CVE-2018-3639)

AMD Zen expose the Intel equivalant to Speculative Store Bypass Disable
via the 0x80000008_EBX[25] CPUID feature bit.

This needs to be exposed to guest OS to allow them to protect
against CVE-2018-3639.

Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Reviewed-by: Daniel P. Berrangé <[email protected]>
Signed-off-by: Daniel P. Berrangé <[email protected]>
Message-Id: <20180521215424[email protected]>
Signed-off-by: Eduardo Habkost <[email protected]>

i386: Define the Virt SSBD MSR and handling of it (CVE-2018-3639)

"Some AMD processors only support a non-architectural means of enabling
speculative store bypass disable (SSBD).  To allow a simplified view of
this to a guest, an architectural definition has been created through a new
CPUID bit, 0x80000008_EBX[25], and a new MSR, 0xc001011f.  With this, a
hypervisor can virtualize the existence of this definition and provide an
architectural method for using SSBD to a guest.

Add the new CPUID feature, the new MSR and update the existing SSBD
support to use this MSR when present." (from x86/speculation: Add virtualized
speculative store bypass disable support in Linux).

Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Reviewed-by: Daniel P. Berrangé <[email protected]>
Signed-off-by: Daniel P. Berrangé <[email protected]>
Message-Id: <20180521215424[email protected]>
Signed-off-by: Eduardo Habkost <[email protected]>

i386: define the 'ssbd' CPUID feature bit (CVE-2018-3639)

New microcode introduces the "Speculative Store Bypass Disable"
CPUID feature bit. This needs to be exposed to guest OS to allow
them to protect against CVE-2018-3639.

Signed-off-by: Daniel P. Berrangé <[email protected]>
Reviewed-by: Konrad Rzeszutek Wilk <[email protected]>
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Message-Id: <20180521215424[email protected]>
Signed-off-by: Eduardo Habkost <[email protected]>

Merge remote-tracking branch 'remotes/mjt/tags/trivial-patches-fetch' into staging

trivial patches for 2018-05-20

# gpg: Signature made Sun 20 May 2018 07:13:20 BST
# gpg:                using RSA key 701B4F6B1A693E59
# gpg: Good signature from "Michael Tokarev <[email protected]>"
# gpg:                 aka "Michael Tokarev <[email protected]>"
# gpg:                 aka "Michael Tokarev <[email protected]>"
# Primary key fingerprint: 6EE1 95D1 886E 8FFB 810D  4324 457C E0A0 8044 65C5
#      Subkey fingerprint: 7B73 BAD6 8BE7 A2C2 8931  4B22 701B 4F6B 1A69 3E59

* remotes/mjt/tags/trivial-patches-fetch: (22 commits)
  acpi: fix a comment about aml_call0()
  qapi/net.json: Fix the version number of the "vlan" removal
  gdbstub: Handle errors in gdb_accept()
  gdbstub: Use qemu_set_cloexec()
  replace functions which are only available in glib-2.24
  typedefs: Remove PcGuestInfo from qemu/typedefs.h
  qemu-options: Allow -no-user-config again
  hw/timer/mt48t59: Fix bit-rotten NVRAM_PRINTF format strings
  Remove unnecessary variables for function return value
  trivial: Do not include pci.h if it is not necessary
  tests: fix tpm-crb tpm-tis tests race
  hw/ide/ahci: Keep ALLWINNER_AHCI() macro internal
  qemu-img-cmds.hx: add passive-aggressive note
  qemu-img: Make documentation between .texi and .hx consistent
  qemu-img: remove references to GEN_DOCS
  qemu-img.texi: fix command ordering
  qemu-img-commands.hx: argument ordering fixups
  HACKING: document preference for g_new instead of g_malloc
  qemu-option-trace: -trace enable= is a pattern, not a file
  slirp/debug: Print IP addresses in human readable form
  ...

Signed-off-by: Peter Maydell <[email protected]>

Merge remote-tracking branch 'remotes/rth/tags/pull-fpu-20180518' into staging

Honor CPU_DUMP_FPU

# gpg: Signature made Fri 18 May 2018 22:56:12 BST
# gpg:                using RSA key 64DF38E8AF7E215F
# gpg: Good signature from "Richard Henderson <[email protected]>"
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* remotes/rth/tags/pull-fpu-20180518:
  target/xtensa: Honor CPU_DUMP_FPU
  target/unicore32: Honor CPU_DUMP_FPU
  target/sparc: Honor CPU_DUMP_FPU
  target/s390x: Honor CPU_DUMP_FPU
  target/riscv: Honor CPU_DUMP_FPU
  target/ppc: Honor CPU_DUMP_FPU
  target/mips: Honor CPU_DUMP_FPU
  target/alpha: Honor CPU_DUMP_FPU

Signed-off-by: Peter Maydell <[email protected]>

acpi: fix a comment about aml_call0()

Signed-off-by: Marc-André Lureau <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

qapi/net.json: Fix the version number of the "vlan" removal

"vlan" will be dropped in 2.13, not in 2.12. And while we're at it,
use the better wording "dropped in" instead of "removed with" (also
for the "dump" removal).

Reported-by: Stefan Hajnoczi <[email protected]>
Reported-by: Eric Blake <[email protected]>
Signed-off-by: Thomas Huth <[email protected]>
Reviewed-by: Eric Blake <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

gdbstub: Handle errors in gdb_accept()

In gdb_accept(), we both fail to check all errors (notably
that from socket_set_nodelay(), as Coverity notes in CID 1005666),
and fail to return an error status back to our caller. Correct
both of these things, so that errors in accept() result in our
stopping with a useful error message rather than ignoring it.

Signed-off-by: Peter Maydell <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Reviewed-by: Thomas Huth <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

gdbstub: Use qemu_set_cloexec()

Use the utility routine qemu_set_cloexec() rather than
manually calling fcntl(). This lets us drop the #ifndef _WIN32
guards and also means Coverity doesn't complain that we're
ignoring the fcntl error return (CID 1005665, CID 1005667).

Signed-off-by: Peter Maydell <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Reviewed-by: Thomas Huth <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

replace functions which are only available in glib-2.24

Currently the minimal supported version of glib is 2.22.
Since testing is done with a glib that claims to be 2.22, but in fact
has APIs from newer version of glib, this bug was not caught during
submit of the patch referenced below.

Replace g_realloc_n, which is available only since 2.24, with g_renew.

Fixes commit 418026ca43 ("util: Introduce vfio helpers")

Signed-off-by: Olaf Hering <[email protected]>
Reviewed-by: Thomas Huth <[email protected]>
Reviewed-by: Cornelia Huck <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>
CC: [email protected]

typedefs: Remove PcGuestInfo from qemu/typedefs.h

It is long gone since e4e8ba04c2007 ...

Signed-off-by: Philippe Mathieu-Daudé <[email protected]>
Reviewed-by: Thomas Huth <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

qemu-options: Allow -no-user-config again

After 1217d6ca2bf28c0febe1bd7d5b3fa912bbf6af2a we error out
explicitly if an unknown -option was passed on the command line.
However, we are doing two pass command line option parsing. In
the first pass we just look for -no-user-config or -nodefconfig
being present which determines whether we load user config or
not. Then in the second pass we finally parse everything else
throwing an error if an unsupported -option was found. Problem is
that in the second pass -no-user-config and -nodefconfig are not
handled explicitly which makes us throw the unsupported option
error.

Signed-off-by: Michal Privoznik <[email protected]>
Reviewed-by: Thomas Huth <[email protected]>
Reviewed-by: Marc-André Lureau <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

hw/timer/mt48t59: Fix bit-rotten NVRAM_PRINTF format strings

When compiling with NVRAM_PRINTF enabled, gcc currently bails out with:

  CC      hw/timer/m48t59.o
  CC      hw/timer/m48t59-isa.o
hw/timer/m48t59.c: In function ‘NVRAM_writeb’:
hw/timer/m48t59.c:460:5: error: format ‘%x’ expects argument of type ‘unsigned int’, but argument 3 has type ‘hwaddr’ [-Werror=format=]
     NVRAM_PRINTF("%s: 0x%08x => 0x%08x\n", __func__, addr, val);
     ^
hw/timer/m48t59.c:460:5: error: format ‘%x’ expects argument of type ‘unsigned int’, but argument 4 has type ‘uint64_t’ [-Werror=format=]
hw/timer/m48t59.c: In function ‘NVRAM_readb’:
hw/timer/m48t59.c:492:5: error: format ‘%x’ expects argument of type ‘unsigned int’, but argument 3 has type ‘hwaddr’ [-Werror=format=]
     NVRAM_PRINTF("%s: 0x%08x <= 0x%08x\n", __func__, addr, retval);

Fix it by using the correct format strings and while we're at it,
also change the definition of NVRAM_PRINTF so that this can not
bit-rot so easily again.

Signed-off-by: Thomas Huth <[email protected]>
Reviewed-by: Mark Cave-Ayland <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

Remove unnecessary variables for function return value

Re-run Coccinelle script scripts/coccinelle/return_directly.cocci

Signed-off-by: Laurent Vivier <[email protected]>
ppc part
Acked-by: David Gibson <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

trivial: Do not include pci.h if it is not necessary

There is no need to include pci.h in these files.

Signed-off-by: Thomas Huth <[email protected]>
Reviewed-by: Michael S. Tsirkin <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

tests: fix tpm-crb tpm-tis tests race

No need to close the TPM data socket on the emulator end, qemu will
close it after a SHUTDOWN. This avoids a race between close() and
read() in the TPM data thread.

Reported-by: Peter Maydell <[email protected]>
Signed-off-by: Marc-André Lureau <[email protected]>
Signed-off-by: Stefan Berger <[email protected]>
Reviewed-by: Daniel P. Berrangé <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

hw/ide/ahci: Keep ALLWINNER_AHCI() macro internal

The ALLWINNER_AHCI() macro is only used in ahci-allwinner.c.

Signed-off-by: Philippe Mathieu-Daudé <[email protected]>
Reviewed-by: John Snow <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

qemu-img-cmds.hx: add passive-aggressive note

I'm kidding. It's very easy to forget there are per-command sections
in the texi, and insane that we don't autogenerate those, too.

Until then, leave a little post-it note in this .hx file until I
find a way to delete it.

Signed-off-by: John Snow <[email protected]>
Reviewed-by: Jeff Cody <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

qemu-img: Make documentation between .texi and .hx consistent

These are also different and out of order for whatever reason.
I'd like to automate this in the future, but for now let's put
on the band-aid.

In the case of resize, there were options missing from all
three docstrings; the new string is based on the code.

Signed-off-by: John Snow <[email protected]>
Reviewed-by: Jeff Cody <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

qemu-img: remove references to GEN_DOCS

Nothing seemingly uses this.
(jcody: commit 77bd1119ba even mentions that it appears unused)

Signed-off-by: John Snow <[email protected]>
Reviewed-by: Jeff Cody <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

qemu-img.texi: fix command ordering

This should match the summary ordering, which is alphabetical.

Signed-off-by: John Snow <[email protected]>
Reviewed-by: Jeff Cody <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

qemu-img-commands.hx: argument ordering fixups

The TEXI and string versions are actually identical, except for markup.
We can probably automate this... but make the ordering the same until
then.

Signed-off-by: John Snow <[email protected]>
Reviewed-by: Jeff Cody <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

HACKING: document preference for g_new instead of g_malloc

This patch documents the preference for g_new instead of g_malloc. The
reasons were adapted from commit b45c03f585ea9bb1af76c73e82195418c294919d.

Discussion in QEMU's mailing list:
  http://lists.nongnu.org/archive/html/qemu-devel/2018-05/msg03238.html

Cc: [email protected]
Cc: David Hildenbrand <[email protected]>
Cc: Eduardo Habkost <[email protected]>
Cc: Markus Armbruster <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Signed-off-by: Murilo Opsfelder Araujo <[email protected]>
Reviewed-by: Eric Blake <[email protected]>
Reviewed-by: Alex Bennée <[email protected]>
Reviewed-by: David Hildenbrand <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

qemu-option-trace: -trace enable= is a pattern, not a file

Signed-off-by: Michael Tokarev <[email protected]>

slirp/debug: Print IP addresses in human readable form

Signed-off-by: Alexey Kardashevskiy <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

misc, ide: remove use of HWADDR_PRIx in trace events

The trace events all use a uint64_t data type, so should be using the
corresponding PRIx64 format, not HWADDR_PRIx which is intended for use
with the 'hwaddr' type.

Signed-off-by: Daniel P. Berrangé <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

tcg: fix s/compliment/complement/ typos

Signed-off-by: Emilio G. Cota <[email protected]>
Signed-off-by: Michael Tokarev <[email protected]>

target/xtensa: Honor CPU_DUMP_FPU

Acked-by: Max Filippov <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>

target/unicore32: Honor CPU_DUMP_FPU

Cc: Guan Xuetao <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>

target/sparc: Honor CPU_DUMP_FPU

Cc: Mark Cave-Ayland <[email protected]>
Cc: Artyom Tarasenko <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>

target/s390x: Honor CPU_DUMP_FPU

Also do not dump both "fpu" and "vector" registers
as the former overlaps the latter.

Cc: Alexander Graf <[email protected]>
Reviewed-by: David Hildenbrand <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>

target/riscv: Honor CPU_DUMP_FPU

Cc: Palmer Dabbelt <[email protected]>
Cc: Sagar Karandikar <[email protected]>
Cc: Bastian Koppelmann <[email protected]>
Reviewed-by: Michael Clark <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>

target/ppc: Honor CPU_DUMP_FPU

Cc: Alexander Graf <[email protected]>
Cc: David Gibson <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>

target/mips: Honor CPU_DUMP_FPU

Cc: Aurelien Jarno <[email protected]>
Cc: Yongbok Kim <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>

target/alpha: Honor CPU_DUMP_FPU

Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>

Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20180518' into staging

target-arm queue:
 * Initial part of SVE implementation (currently disabled)
 * smmuv3: fix some minor Coverity issues
 * add model of Xilinx ZynqMP generic DMA controller
 * expose (most) Arm coprocessor/system registers to
   gdb via QEMU's gdbstub, for reads only

# gpg: Signature made Fri 18 May 2018 18:18:27 BST
# gpg:                using RSA key 3C2525ED14360CDE
# gpg: Good signature from "Peter Maydell <[email protected]>"
# gpg:                 aka "Peter Maydell <[email protected]>"
# gpg:                 aka "Peter Maydell <[email protected]>"
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* remotes/pmaydell/tags/pull-target-arm-20180518: (32 commits)
  target/arm: Implement SVE Permute - Extract Group
  target/arm: Implement SVE Integer Wide Immediate - Predicated Group
  target/arm: Implement SVE Bitwise Immediate Group
  target/arm: Implement SVE Element Count Group
  target/arm: Implement SVE floating-point trig select coefficient
  target/arm: Implement SVE floating-point exponential accelerator
  target/arm: Implement SVE Compute Vector Address Group
  target/arm: Implement SVE Bitwise Shift - Unpredicated Group
  target/arm: Implement SVE Stack Allocation Group
  target/arm: Implement SVE Index Generation Group
  target/arm: Implement SVE Integer Arithmetic - Unpredicated Group
  target/arm: Implement SVE Integer Multiply-Add Group
  target/arm: Implement SVE Integer Arithmetic - Unary Predicated Group
  target/arm: Implement SVE bitwise shift by wide elements (predicated)
  target/arm: Implement SVE bitwise shift by vector (predicated)
  target/arm: Implement SVE bitwise shift by immediate (predicated)
  target/arm: Implement SVE Integer Reduction Group
  target/arm: Implement SVE Integer Binary Arithmetic - Predicated Group
  target/arm: Implement SVE Predicate Misc Group
  target/arm: Implement SVE Predicate Logical Operations Group
  ...

Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Permute - Extract Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Integer Wide Immediate - Predicated Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Bitwise Immediate Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Element Count Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE floating-point trig select coefficient

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE floating-point exponential accelerator

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Compute Vector Address Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Bitwise Shift - Unpredicated Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Stack Allocation Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Index Generation Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Integer Arithmetic - Unpredicated Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Integer Multiply-Add Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Integer Arithmetic - Unary Predicated Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE bitwise shift by wide elements (predicated)

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE bitwise shift by vector (predicated)

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE bitwise shift by immediate (predicated)

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Integer Reduction Group

Excepting MOVPRFX, which isn't a reduction.  Presumably it is
placed within the group because of its encoding.

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Integer Binary Arithmetic - Predicated Group

Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Predicate Misc Group

Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE Predicate Logical Operations Group

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>

target/arm: Implement SVE predicate test

Reviewed-by: Peter Maydell <[email protected]>
Signed-off-by: Richard Henderson <[email protected]>
Message-id: 20180516223007[email protected]
Signed-off-by: Peter Maydell <[email protected]>