Previously, qapi-types and qapi-visit filtered out implicit
objects during visit_object_type() by using 'info' (works since
implicit objects do not [yet] have associated info); meanwhile
qapi-introspect filtered out all schema types on the first pass
by returning a python type from visit_begin(), which was then
used at a distance in QAPISchema.visit() to do the filtering.

Rather than keeping these ad hoc approaches, add a new visitor
callback visit_needed() which returns False to skip a given
entity, and which defaults to True unless overridden.  Use the
new mechanism to simplify all three filtering visitors.

No change to the generated code.
Suggested-by: NMarkus Armbruster <armbru@redhat.com>
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1444710158-8723-2-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Commit e36c714e causes 'qemu -netdev help' to dump core, because the
call to visit_end_union() is no longer conditional on whether *obj was
allocated.

Reported by Marc-André Lureau <marcandre.lureau@gmail.com>
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1444861825-19256-1-git-send-email-eblake@redhat.com>
[Commit message tweaked to say 'help' instead of '?']
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

QAPI patches

# gpg: Signature made Mon 12 Oct 2015 18:56:35 BST using RSA key ID EB918653
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>"
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>"

* remotes/armbru/tags/pull-qapi-2015-10-12:
  qapi: Simplify gen_visit_fields() error handling
  qapi: Share gen_visit_fields()
  qapi: Share gen_err_check()
  qapi: Consistent generated code: minimize push_indent() usage
  qapi: Consistent generated code: prefer common indentation
  qapi: Consistent generated code: prefer common labels
  qapi: Consistent generated code: prefer visitor 'v'
  qapi: Consistent generated code: prefer error 'err'
  qapi: Reuse code for flat union base validation
  qapi: Test use of 'number' within alternates
  qapi: Add tests for empty unions
  qapi: Avoid assertion failure on union 'type' collision
  qapi: Test for various name collisions
  qapi: Clean up qapi.py per pep8
  qapi: Invoke exception superclass initializer
  qapi: Improve 'include' error message
  qapi: Sort qapi-schema tests
  MAINTAINERS: Specify QAPI include and test files
  MAINTAINERS: Specify QObject include and test files
  docs: Move files from docs/qmp/ to docs/
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Since we have consolidated all generated code to use 'err' as
the name of the local variable for error detection, we can
simplify the decision on whether to skip error detection (useful
for deallocation paths) to be a boolean.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-18-git-send-email-eblake@redhat.com>
[Change to gen_visit_fields() simplified]
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Consolidate the code between visit, command marshalling, and
event generation that iterates over the members of a struct.
It reduces code duplication in the generator, so that a future
patch can reduce the size of generated code while touching only
one instead of three locations.

There are no changes to the generated marshal code.

The visitor code becomes slightly more verbose, but remains
semantically equivalent, and is actually easier to read as
it follows a more common idiom:

|     visit_optional(v, &(*obj)->has_device, "device", &err);
|-    if (!err && (*obj)->has_device) {
|-        visit_type_str(v, &(*obj)->device, "device", &err);
|-    }
|     if (err) {
|         goto out;
|     }
|+    if ((*obj)->has_device) {
|+        visit_type_str(v, &(*obj)->device, "device", &err);
|+        if (err) {
|+            goto out;
|+        }
|+    }

The event code becomes slightly more verbose, but this is
arguably a bug fix: although the visitors are not well
documented, use of an optional member should not be attempted
unless guarded by a prior call to visit_optional().  Works only
because the output qmp visitor has a no-op visit_optional():

|+    visit_optional(v, &has_offset, "offset", &err);
|+    if (err) {
|+        goto out;
|+    }
|     if (has_offset) {
|         visit_type_int(v, &offset, "offset", &err);
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-17-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

qapi-commands has a nice helper gen_err_check(), but did not
use it everywhere. In fact, using it in more places makes it
easier to reduce the lines of code used for generating error
checks.  This in turn will make it easier for later patches
to consolidate another common pattern among the generators.

The generated code has fewer blank lines in qapi-event.c functions,
but has no semantic difference.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-16-git-send-email-eblake@redhat.com>
[Drop another blank line for symmetry]
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

We had some pointless differences in the generated code for visit,
command marshalling, and events; unifying them makes it easier for
future patches to consolidate to common helper functions.
This is one patch of a series to clean up these differences.

This patch reduces the number of push_indent()/pop_indent() pairs
so that generated code is typically already at its natural output
indentation in the python files.  It is easier to reason about
generated code if the reader does not have to track how much
spacing will be inserted alongside the code, and moreso when all
of the generators use the same patterns (qapi-type and qapi-event
were already using in-place indentation).

Arguably, the resulting python may be a bit harder to read with C
code at the same indentation as python; on the other hand, not
having to think about push_indent() is a win, and most decent
editors provide syntax highlighting that makes it easier to
visually distinguish python code from string literals that will
become C code.

There is no change to the generated output.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-15-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

We had some pointless differences in the generated code for visit,
command marshalling, and events; unifying them makes it easier for
future patches to consolidate to common helper functions.
This is one patch of a series to clean up these differences.

This patch adjusts gen_visit_union() to use the same indentation
as other functions, namely, by jumping early to the error label
if the object was not set rather than placing the rest of the
body inside an if for when it is set.

No change in semantics to the generated code.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-14-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

We had some pointless differences in the generated code for visit,
command marshalling, and events; unifying them makes it easier for
future patches to consolidate to common helper functions.
This is one patch of a series to clean up these differences.

This patch names the goto labels 'out' (not 'clean') and 'out_obj'
(not 'out_end').  Additionally, the generator was inconsistent on
whether labels had a leading space [our HACKING is silent; while
emacs 'gnu' style adds the space to avoid littering column 1].
For minimal churn, prefer no leading space; this also matches
the style that is more prevalent in current qemu.git.

No change in semantics to the generated code.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-13-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

We had some pointless differences in the generated code for visit,
command marshalling, and events; unifying them makes it easier for
future patches to consolidate to common helper functions.
This is one patch of a series to clean up these differences.

This patch names the local visitor variable 'v' rather than 'm'.
Related objects, such as 'QapiDeallocVisitor', are also named by
their initials instead of an unrelated leading m.

No change in semantics to the generated code.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-12-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

We had some pointless differences in the generated code for visit,
command marshalling, and events; unifying them makes it easier for
future patches to consolidate to common helper functions.
This is one patch of a series to clean up these differences.

This patch consistently names the local error variable 'err' rather
than 'local_err'.

No change in semantics to the generated code.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-11-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Rather than open-code the check for a valid base type, we
should reuse the common functionality. This allows for
consistent error messages, and also makes it easier for a
later patch to turn on support for inline anonymous base
structures.

Test flat-union-inline is updated to test only one feature
(anonymous branch dictionaries), which can be implemented
independently (test flat-union-bad-base already covers the
idea of an anonymous base dictionary).
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-10-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Add some testsuite exposure for use of a 'number' as part of
an alternate.  The current state of the tree has a few bugs
exposed by this: our input parser depends on the ordering of
how the qapi schema declared the alternate, and the parser
does not accept integers for a 'number' in an alternate even
though it does for numbers outside of an alternate.

Mixing 'int' and 'number' in the same alternate is unusual,
since both are supplied by json-numbers, but there does not
seem to be a technical reason to forbid it given that our
json lexer distinguishes between json-numbers that can be
represented as an int vs. those that cannot.

Improve the existing test_visitor_in_alternate() to match the
style of the new test_visitor_in_alternate_number(), and to
ensure full coverage of all possible qtype parsing.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-9-git-send-email-eblake@redhat.com>
[Eric's follow-up fixes squashed in]
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

The documentation claims that alternates are useful for
allowing two or more types, although nothing enforces this.
Meanwhile, it is silent on whether empty unions are allowed.
In practice, the generated code will compile, in part because
we have a 'void *data' branch; but attempting to visit such a
type will cause an abort().  While there's no technical reason
that a degenerate union could not be made to work, it's harder
to justify the time spent in chasing known (the current
abort() during visit) and unknown corner cases, than it would
be to just outlaw them.  A future patch will probably take the
approach of forbidding them; in the meantime, we can at least
add testsuite coverage to make it obvious where things stand.

In addition to adding tests to expose the problems, we also
need to adjust existing tests that are meant to test something
else, but which could fail for the wrong reason if we reject
degenerate alternates/unions.

Note that empty structs are explicitly supported (for example,
right now they are the only way to specify that one branch of a
flat union adds no additional members), and empty enums are
covered by the testsuite as working (even if they do not seem
to have much use).
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-8-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

The previous commit added two tests that triggered an assertion
failure. It's fairly straightforward to avoid the failure by
just outright forbidding the collision between a union's tag
values and its discriminator name (including the implicit name
'kind' supplied for simple unions [*]).  Ultimately, we'd like
to move the collision detection into QAPISchema*.check(), but
for now it is easier just to enhance the existing checks.

[*] Of course, down the road, we have plans to rename the simple
union tag name to 'type' to match the QMP wire name, but the
idea of the collision will still be present even then.

Technically, we could avoid the collision by naming the C union
members representing each enum value as '_case_value' rather
than 'value'; but until we have an actual qapi client (and not
just our testsuite) that has a legitimate reason to match a
case label to the name of a QMP key and needs the name munging
to satisfy the compiler, it's easier to just reject the qapi
as invalid.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-7-git-send-email-eblake@redhat.com>
[Polished a few comments]
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Expose some weaknesses in the generator: we don't always forbid
the generation of structs that contain multiple members that map
to the same C or QMP name.  This has already been marked FIXME in
qapi.py in commit d90675fa, but having more tests will make sure
future patches produce desired behavior; and updating existing
patches to better document things doesn't hurt, either.  Some of
these collisions are already caught in the old-style parser
checks, but ultimately we want all collisions to be caught in the
new-style QAPISchema*.check() methods.

This patch focuses on C struct members, and does not consider
collisions between commands and events (affecting C function
names), or even collisions between generated C type names with
user type names (for things like automatic FOOList struct
representing array types or FOOKind for an implicit enum).

There are two types of struct collisions we want to catch:
 1) Collision between two keys in a JSON object. qapi.py prevents
    that within a single struct (see test duplicate-key), but it is
    possible to have collisions between a type's members and its
    base type's members (existing tests struct-base-clash,
    struct-base-clash-deep), and its flat union variant members
    (renamed test flat-union-clash-member).
 2) Collision between two members of the C struct that is generated
    for a given QAPI type:
    a) Multiple QAPI names map to the same C name (new test
       args-name-clash)
    b) A QAPI name maps to a C name that is used for another purpose
       (new tests flat-union-clash-branch, struct-base-clash-base,
       union-clash-data). We already fixed some such cases in commit
       0f61af3e and 1e6c1616, but more remain.
    c) Two C names generated for other purposes clash
       (updated test alternate-clash, new test union-clash-branches,
       union-clash-type, flat-union-clash-type)

Ultimately, if we need to have a flat union where a tag value
clashes with a base member name, we could change the generator to
name the union (using 'foo.u.value' rather than 'foo.value') or
otherwise munge the C name corresponding to tag values.  But
unless such a need arises, it will probably be easier to just
forbid these collisions.

Some of these negative tests will be deleted later, and positive
tests added to qapi-schema-test.json in their place, when the
generator code is reworked to avoid particular code generation
collisions in class 2).

[Note that viewing this patch with git rename detection enabled
may see some confusion due to renaming some tests while adding
others, but where the content is similar enough that git picks
the wrong pre- and post-patch files to associate]
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-6-git-send-email-eblake@redhat.com>
[Improve commit message and comments a bit, drop an unrelated test]
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Silence pep8, and make pylint a bit happier.  Just style cleanups,
plus killing a useless comment in camel_to_upper(); no semantic
changes.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-5-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

pylint recommends that every exception class should explicitly
invoke the superclass __init__, even though things seem to work
fine without it.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-4-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Use of '"...%s" % include' to print non-strings can lead to
ugly messages, such as this (if the .json change is applied
without the qapi.py change):
 Expected a file name (string), got: OrderedDict()

Better is to just omit the actual non-string value in the
message.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-3-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Recent changes to qapi have provided quite a bit of churn in
the makefile, because we are inconsistent on what order test
names appear in, and on whether to re-wrap the list of tests or
just add arbitrary line lengths.  Writing the list in a sorted
fashion, one test per line, will make future patches easier
to see what tests are being added or removed by a patch.

Although it is tempting to use $(wildcard qapi-schema/*.json)
for a more compact listing, such an approach would risk picking
up leftover garbage .json files in the directory; so keeping
the list explicit is safer for ensuring reproducible tarballs
and test results.
Signed-off-by: NEric Blake <eblake@redhat.com>
Message-Id: <1443565276-4535-2-git-send-email-eblake@redhat.com>
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>

Signed-off-by: NMarkus Armbruster <armbru@redhat.com>
Message-Id: <1443111117-29831-4-git-send-email-armbru@redhat.com>
Reviewed-by: NEric Blake <eblake@redhat.com>
Reviewed-by: NLuiz Capitulino <lcapitulino@redhat.com>

Signed-off-by: NMarkus Armbruster <armbru@redhat.com>
Message-Id: <1443111117-29831-3-git-send-email-armbru@redhat.com>
Reviewed-by: NEric Blake <eblake@redhat.com>
Reviewed-by: NLuiz Capitulino <lcapitulino@redhat.com>

Giving QMP its own subdirectory in docs/ is hardly worthwhile when we
have just four files, and one of them isn't even in the subdirectory.
Move the files from docs/qmp/ to docs/, renaming docs/qmp/README to
docs/qmp-intro.

Update MAINTAINERS.  The new pattern also captures the fourth file
docs/writing-qmp-commands.txt.
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>
Message-Id: <1443111117-29831-2-git-send-email-armbru@redhat.com>
Reviewed-by: NEric Blake <eblake@redhat.com>
Reviewed-by: NLuiz Capitulino <lcapitulino@redhat.com>

Pull request

v2:
 * Fix virtio 16lx -> HWADDR_PRIx format specifier [Peter]

# gpg: Signature made Mon 12 Oct 2015 11:19:06 BST using RSA key ID 81AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>"
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>"

* remotes/stefanha/tags/block-pull-request:
  sdhci.c: Limit the maximum block size
  block: switch from g_slice allocator to malloc
  virtio dataplane: adapt dataplane for virtio Version 1
  virtio-blk: use blk_io_plug/unplug for Linux AIO batching
  sdhci: Pass drive parameter to sdhci-pci via qdev property
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

# gpg: Signature made Mon 12 Oct 2015 08:56:47 BST using RSA key ID 398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <jasowang@redhat.com>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg:          It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 215D 46F4 8246 689E C77F  3562 EF04 965B 398D 6211

* remotes/jasowang/tags/net-pull-request:
  tests: add test cases for netfilter object
  netfilter: add a netbuffer filter
  net/queue: export qemu_net_queue_append_iov
  netfilter: print filter info associate with the netdev
  netfilter: add an API to pass the packet to next filter
  net/queue: introduce NetQueueDeliverFunc
  net: merge qemu_deliver_packet and qemu_deliver_packet_iov
  netfilter: hook packets before net queue send
  init/cleanup of netfilter object
  vl.c: init delayed object after net_init_clients
  vmxnet3: Add support for VMXNET3_CMD_GET_ADAPTIVE_RING_INFO command
  e1000: use alias for default model
  vmxnet3: Support reading IMR registers on bar0
  net/vmxnet3: Refine l2 header validation
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

It is possible for the guest to set an invalid block
size which is larger then the fifo_buffer[] array. This
could cause a buffer overflow.

To avoid this limit the maximum size of the blksize variable.
Signed-off-by: NAlistair Francis <alistair.francis@xilinx.com>
Reported-by: NIntel Security ATR <secure@intel.com>
Reviewed-by: NStefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: NPeter Crosthwaite <crosthwaite.peter@gmail.com>
Message-id: abe4c51f513290bbb85d1ee271cb1a3d463d7561.1444067470.git.alistair.francis@xilinx.com
Suggested-by: NIgor Mitsyanko <i.mitsyanko@gmail.com>
Reported-by: NIntel Security ATR <secure@intel.com>
Reviewed-by: NStefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>

Simplify memory allocation by sticking with a single API.  GSlice
is not that fast anyway (tcmalloc/jemalloc are better).
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>

Let dataplane allocate different region for the desc/avail/used
ring regions.
Take VIRTIO_RING_F_EVENT_IDX into account to increase the used/avail
rings accordingly.

[Fix 32-bit builds by changing 16lx format specifier to HWADDR_PRIx.
--Stefan]
Signed-off-by: NPierre Morel <pmorel@linux.vnet.ibm.com>
Tested-by: NGreg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: NGreg Kurz <gkurz@linux.vnet.ibm.com>
Message-id: 1441625636-23773-1-git-send-email-pmorel@linux.vnet.ibm.com
(changed __virtio16 into uint16_t,
 map descriptor table and available ring read-only)
Signed-off-by: NGreg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>

Fix device introspection regressions

# gpg: Signature made Fri 09 Oct 2015 14:43:41 BST using RSA key ID EB918653
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>"
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>"

* remotes/armbru/tags/pull-monitor-2015-10-09:
  Revert "qdev: Use qdev_get_device_class() for -device <type>,help"
  qdev: Protect device-list-properties against broken devices
  qmp: Fix device-list-properties not to crash for abstract device
  device-introspect-test: New, covering device introspection
  libqtest: New hmp() & friends
  libqtest: Clean up unused QTestState member sigact_old
  tests: Fix how qom-test is run
  macio: move DBDMA_init from instance_init to realize
  hw: do not pass NULL to memory_region_init from instance_init
  memory: allow destroying a non-empty MemoryRegion
  virtio-input: Fix device introspection on non-Linux hosts
  update-linux-headers: Rename SW_MAX to SW_MAX_
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

The raw-posix block driver implements Linux AIO batching so multiple
requests can be submitted with a single io_submit(2) system call.
Batching is currently only used by virtio-scsi and
virtio-blk-data-plane.

Enable batching for regular virtio-blk so the number of io_submit(2)
system calls is reduced for workloads with queue depth > 1.

In 4KB random read performance tests with queue depth 32, the CPU
utilization on the host is reduced by 9.4%.  The fio job is as follows:

  [global]
  bs=4k
  ioengine=libaio
  iodepth=32
  direct=1
  sync=0
  time_based=1
  runtime=30
  clocksource=gettimeofday
  ramp_time=5

  [job1]
  rw=randread
  filename=/dev/vdb
  size=4096M
  write_bw_log=fio
  write_iops_log=fio
  write_lat_log=fio
  log_avg_msec=1000

This benchmark was run on an raw image on LVM.  The disk was an SSD
drive and -drive cache=none,aio=native was used.
Tested-by: NPradeep Surisetty <psuriset@redhat.com>
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: NFam Zheng <famz@redhat.com>

Commit 19109131 disabled the sdhci-pci support because it used
drive_get_next().  This patch reenables sdhci-pci and changes it to
pass the drive via a qdev property - for example:
 -device sdhci-pci,drive=drive0 -drive id=drive0,if=sd,file=myimage
Reviewed-by: NStefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: NKevin O'Connor <kevin@koconnor.net>
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>

Using qtest qmp interface to implement following cases:
1) add/remove netfilter
2) add a netfilter then delete the netdev
3) add/remove more than one netfilters
4) add more than one netfilters and then delete the netdev
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

This filter is to buffer/release packets. Can be used when using
MicroCheckpointing or other Remus like VM FT solutions.
You can also use it to crudely simulate network delay.  Doesn't
actually delay individual packets, but batches them together, which is
a delay of sorts.

Usage:
 -netdev tap,id=bn0
 -object filter-buffer,id=f0,netdev=bn0,queue=rx,interval=1000

NOTE:
 Interval is in microseconds, it can't be omitted currently, and can't be 0.
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Reviewed-by: NMarkus Armbruster <armbru@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

This will be used by buffer filter implementation later to
queue packets.
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Reviewed-by: NThomas Huth <thuth@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

When execute "info network", print filter info also.
add a info_str member to NetFilterState, store specific filters
info.
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

add an API qemu_netfilter_pass_to_next() to pass the packet
to next filter.
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Reviewed-by: NThomas Huth <thuth@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

net/queue.c has logic to send/queue/flush packets but a
qemu_deliver_packet_iov() call is hardcoded. Abstract this
func so that we can use our own deliver function in netfilter.
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

qemu_deliver_packet_iov already have the compat delivery, we
can drop qemu_deliver_packet.
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Capture packets that will be sent.
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Reviewed-by: NThomas Huth <thuth@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Add a netfilter object based on QOM.

A netfilter is attached to a netdev, captures all network packets
that pass through the netdev. When we delete the netdev, we also
delete the netfilter object attached to it, because if the netdev is
removed, the filter which attached to it is useless.
Signed-off-by: NYang Hongyang <yanghy@cn.fujitsu.com>
Reviewed-by: NMarkus Armbruster <armbru@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>