OSDN Git Service

qmiga/qemu.git
6 years agoarm/translate-a64: mark path as unreachable to eliminate warning
Emilio G. Cota [Mon, 13 Nov 2017 13:55:24 +0000 (13:55 +0000)]
arm/translate-a64: mark path as unreachable to eliminate warning

Fixes the following warning when compiling with gcc 5.4.0 with -O1
optimizations and --enable-debug:

target/arm/translate-a64.c: In function ‘aarch64_tr_translate_insn’:
target/arm/translate-a64.c:2361:8: error: ‘post_index’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
     if (!post_index) {
        ^
target/arm/translate-a64.c:2307:10: note: ‘post_index’ was declared here
     bool post_index;
          ^
target/arm/translate-a64.c:2386:8: error: ‘writeback’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
     if (writeback) {
        ^
target/arm/translate-a64.c:2308:10: note: ‘writeback’ was declared here
     bool writeback;
          ^

Note that idx comes from selecting 2 bits, and therefore its value
can be at most 3.

Signed-off-by: Emilio G. Cota <cota@braap.org>
Acked-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 1510087611-1851-1-git-send-email-cota@braap.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2017-11-09' into staging
Peter Maydell [Mon, 13 Nov 2017 13:13:12 +0000 (13:13 +0000)]
Merge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2017-11-09' into staging

nbd patches for 2017-11-09

- Vladimir Sementsov-Ogievskiy: nbd/server: fix nbd_negotiate_handle_info
- Eric Blake: 0/7 various NBD fixes for 2.11

# gpg: Signature made Thu 09 Nov 2017 16:56:58 GMT
# gpg:                using RSA key 0xA7A16B4A2527436A
# gpg: Good signature from "Eric Blake <eblake@redhat.com>"
# gpg:                 aka "Eric Blake (Free Software Programmer) <ebb9@byu.net>"
# gpg:                 aka "[jpeg image of size 6874]"
# Primary key fingerprint: 71C2 CC22 B1C4 6029 27D2  F3AA A7A1 6B4A 2527 436A

* remotes/ericb/tags/pull-nbd-2017-11-09:
  nbd/server: Fix structured read of length 0
  nbd-client: Stricter enforcing of structured reply spec
  nbd-client: Short-circuit 0-length operations
  nbd: Fix struct name for structured reads
  nbd/client: Nicer trace of structured reply
  nbd-client: Refuse read-only client with BDRV_O_RDWR
  nbd-client: Fix error message typos
  nbd/server: fix nbd_negotiate_handle_info

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/cohuck/tags/s390x-20171109' into staging
Peter Maydell [Mon, 13 Nov 2017 11:41:47 +0000 (11:41 +0000)]
Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20171109' into staging

s390x changes: let pci devices start out in a usable state, and make
RISBGN work in tcg.

# gpg: Signature made Thu 09 Nov 2017 15:27:21 GMT
# gpg:                using RSA key 0xDECF6B93C6F02FAF
# gpg: Good signature from "Cornelia Huck <conny@cornelia-huck.de>"
# gpg:                 aka "Cornelia Huck <huckc@linux.vnet.ibm.com>"
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>"
# gpg:                 aka "Cornelia Huck <cohuck@kernel.org>"
# gpg:                 aka "Cornelia Huck <cohuck@redhat.com>"
# Primary key fingerprint: C3D0 D66D C362 4FF6 A8C0  18CE DECF 6B93 C6F0 2FAF

* remotes/cohuck/tags/s390x-20171109:
  target/s390x: Finish implementing RISBGN
  s390x/pci: let pci devices start in configured mode

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/rth/tags/pull-cap-20171109' into staging
Peter Maydell [Mon, 13 Nov 2017 10:05:18 +0000 (10:05 +0000)]
Merge remote-tracking branch 'remotes/rth/tags/pull-cap-20171109' into staging

Capstone fixes for 2.11

# gpg: Signature made Thu 09 Nov 2017 07:49:25 GMT
# gpg:                using RSA key 0x64DF38E8AF7E215F
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>"
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* remotes/rth/tags/pull-cap-20171109:
  Makefile: Capstone: Add support for cross compile ranlib
  disas: Dump insn bytes along with capstone disassembly

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging
Peter Maydell [Fri, 10 Nov 2017 17:25:15 +0000 (17:25 +0000)]
Merge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging

Pull request

v2:
 * v1 emails 2/3 and 3/3 weren't sent due to an email failure
 * Included Sergio's updated wording in the commit description

# gpg: Signature made Wed 08 Nov 2017 19:12:01 GMT
# gpg:                using RSA key 0x9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>"
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>"
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35  775A 9CA4 ABB3 81AB 73C8

* remotes/stefanha/tags/block-pull-request:
  util/async: use atomic_mb_set in qemu_bh_cancel
  tests-aio-multithread: fix /aio/multi/schedule race condition

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/berrange/tags/pull-qcrypto-2017-11-08-1' into...
Peter Maydell [Fri, 10 Nov 2017 16:01:35 +0000 (16:01 +0000)]
Merge remote-tracking branch 'remotes/berrange/tags/pull-qcrypto-2017-11-08-1' into staging

Merge qcrypto 2017/11/08 v1

# gpg: Signature made Wed 08 Nov 2017 11:06:38 GMT
# gpg:                using RSA key 0xBE86EBB415104FDF
# gpg: Good signature from "Daniel P. Berrange <dan@berrange.com>"
# gpg:                 aka "Daniel P. Berrange <berrange@redhat.com>"
# Primary key fingerprint: DAF3 A6FD B26B 6291 2D0E  8E3F BE86 EBB4 1510 4FDF

* remotes/berrange/tags/pull-qcrypto-2017-11-08-1:
  crypto: afalg: fix a NULL pointer dereference
  tests: Run the luks tests in test-crypto-block only if encryption is available

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.11-20171108' into staging
Peter Maydell [Fri, 10 Nov 2017 15:05:56 +0000 (15:05 +0000)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.11-20171108' into staging

ppc patch queue 2017-11-08

Here's the current set of accumulated ppc patches for qemu-2.11.
Since we're now in hard freeze these are all bugfixes (although some
fix a bug by way of a cleanup).

# gpg: Signature made Wed 08 Nov 2017 08:10:38 GMT
# gpg:                using RSA key 0x6C38CACA20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>"
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>"
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>"
# gpg:                 aka "David Gibson (kernel.org) <dwg@kernel.org>"
# Primary key fingerprint: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* remotes/dgibson/tags/ppc-for-2.11-20171108:
  e500: ppce500_init_mpic() return device instead of IRQ array
  hw/display/sm501: Fix comment in sm501_sysbus_class_init()
  ppc: fix setting of compat mode

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agonbd/server: Fix structured read of length 0
Eric Blake [Wed, 8 Nov 2017 21:57:03 +0000 (15:57 -0600)]
nbd/server: Fix structured read of length 0

The NBD spec was recently clarified to state that a read of length 0
should not be attempted by a compliant client; but that a server must
still handle it correctly in an unspecified manner (that is, either
a successful no-op or an error reply, but not a crash) [1].  However,
it also implies that NBD_REPLY_TYPE_OFFSET_DATA must have a non-zero
payload length, but our existing code was replying with a chunk
that a picky client could reject as invalid because it was missing
a payload (our own client implementation was recently patched to be
that picky, after first fixing it to not send 0-length requests).

We are already doing successful no-ops for 0-length writes and for
non-structured reads; so for consistency, we want structured reply
reads to also be a no-op.  The easiest way to do this is to return
a NBD_REPLY_TYPE_NONE chunk; this is best done via a new helper
function (especially since future patches for other structured
replies may benefit from using the same helper).

[1] https://github.com/NetworkBlockDevice/nbd/commit/ee926037

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171108215703.9295-8-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6 years agonbd-client: Stricter enforcing of structured reply spec
Eric Blake [Wed, 8 Nov 2017 21:57:02 +0000 (15:57 -0600)]
nbd-client: Stricter enforcing of structured reply spec

Ensure that the server is not sending unexpected chunk lengths
for either the NONE or the OFFSET_DATA chunk, nor unexpected
hole length for OFFSET_HOLE.  This will flag any server as
broken that responds to a zero-length read with an OFFSET_DATA
(what our server currently does, but that's about to be fixed)
or with OFFSET_HOLE, even though we previously fixed our client
to never be able to send such a request over the wire.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171108215703.9295-7-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6 years agonbd-client: Short-circuit 0-length operations
Eric Blake [Wed, 8 Nov 2017 21:57:01 +0000 (15:57 -0600)]
nbd-client: Short-circuit 0-length operations

The NBD spec was recently clarified to state that clients should
not send 0-length requests to the server, as the server behavior
is undefined [1].  We know that qemu-nbd's behavior is a successful
no-op (once it has filtered for read-only exports), but other NBD
implementations might return an error.  To avoid any questionable
server implementations, it is better to just short-circuit such
requests on the client side (we are relying on the block layer to
already filter out requests such as invalid offset, write to a
read-only volume, and so forth); do the short-circuit as late as
possible to still benefit from protections from assertions that
the block layer is not violating our assumptions.

[1] https://github.com/NetworkBlockDevice/nbd/commit/ee926037

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171108215703.9295-6-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6 years agonbd: Fix struct name for structured reads
Eric Blake [Wed, 8 Nov 2017 21:57:00 +0000 (15:57 -0600)]
nbd: Fix struct name for structured reads

A closer read of the NBD spec shows that a structured reply chunk
for a hole is not quite identical to the prefix of a data chunk,
because the hole has to also send a 32-bit size field.  Although
we do not yet send holes, we should fix the misleading information
in our header and make it easier for a future patch to support
sparse reads.  Messed up in commit bae245d1.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171108215703.9295-5-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6 years agonbd/client: Nicer trace of structured reply
Eric Blake [Wed, 8 Nov 2017 21:56:59 +0000 (15:56 -0600)]
nbd/client: Nicer trace of structured reply

It's useful to know which structured reply chunk is being processed.
Missed in commit d2febedb.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171108215703.9295-4-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6 years agonbd-client: Refuse read-only client with BDRV_O_RDWR
Eric Blake [Wed, 8 Nov 2017 21:56:58 +0000 (15:56 -0600)]
nbd-client: Refuse read-only client with BDRV_O_RDWR

The NBD spec says that clients should not try to write/trim to
an export advertised as read-only by the server.  But we failed
to check that, and would allow the block layer to use NBD with
BDRV_O_RDWR even when the server is read-only, which meant we
were depending on the server sending a proper EPERM failure for
various commands, and also exposes a leaky abstraction: using
qemu-io in read-write mode would succeed on 'w -z 0 0' because
of local short-circuiting logic, but 'w 0 0' would send a
request over the wire (where it then depends on the server, and
fails at least for qemu-nbd but might pass for other NBD
implementations).

With this patch, a client MUST request read-only mode to access
a server that is doing a read-only export, or else it will get
a message like:

can't open device nbd://localhost:10809/foo: request for write access conflicts with read-only export

It is no longer possible to even attempt writes over the wire
(including the corner case of 0-length writes), because the block
layer enforces the explicit read-only request; this matches the
behavior of qcow2 when backed by a read-only POSIX file.

Fix several iotests to comply with the new behavior (since
qemu-nbd of an internal snapshot, as well as nbd-server-add over QMP,
default to a read-only export, we must tell blockdev-add/qemu-io to
set up a read-only client).

CC: qemu-stable@nongnu.org
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171108215703.9295-3-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6 years agonbd-client: Fix error message typos
Eric Blake [Wed, 8 Nov 2017 21:56:57 +0000 (15:56 -0600)]
nbd-client: Fix error message typos

Provide missing spaces that are required when using string
concatenation to break error messages across source lines.
Introduced in commit f140e300.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171108215703.9295-2-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6 years agotarget/s390x: Finish implementing RISBGN
Richard Henderson [Tue, 7 Nov 2017 14:55:46 +0000 (15:55 +0100)]
target/s390x: Finish implementing RISBGN

We added the entry to insn-data.def, but failed to update op_risbg
to match.  No need to special-case the imask inversion, since that
is already ~0 for RISBG (and now RISBGN).

Fixes: 375ee58bedcda359011fe7fa99e0647f66f9ffa0
Fixes: https://bugs.launchpad.net/qemu/+bug/1701798 (s390x part)
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20171107145546.767-1-richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Tested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
6 years agoMakefile: Capstone: Add support for cross compile ranlib
Alistair Francis [Wed, 8 Nov 2017 01:10:46 +0000 (17:10 -0800)]
Makefile: Capstone: Add support for cross compile ranlib

When cross compiling QEMU for Windows we need to specify the cross
version of ranlib to avoid build errors when building capstone. This
patch ensures we use the same cross prefix on ranlib as other toolchain
components.

- Fedora23 mingw
- RHEL-7.2 with mingw packages from epel:

   LINK    qemu-img.exe
 build-win64/capstone/capstone.lib: error adding symbols: Archive has no
index; run ranlib to add one
 collect2: error: ld returned 1 exit status

$ x86_64-w64-mingw32-ar --version
GNU ar (GNU Binutils) 2.25

Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <e457d4e906dceea4de6c3431813a06b137c1ab9c.1510103351.git.alistair.francis@xilinx.com>
Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6 years agodisas: Dump insn bytes along with capstone disassembly
Richard Henderson [Tue, 7 Nov 2017 12:19:18 +0000 (13:19 +0100)]
disas: Dump insn bytes along with capstone disassembly

This feature is present for some targets in the bfd disassembler(s).
Implement it generically for all capstone users.

Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6 years agonbd/server: fix nbd_negotiate_handle_info
Vladimir Sementsov-Ogievskiy [Wed, 1 Nov 2017 15:42:03 +0000 (18:42 +0300)]
nbd/server: fix nbd_negotiate_handle_info

namelen should be here, length is unrelated, and always 0 at this
point.  Broken in introduction in commit f37708f6, but mostly
harmless (replying with '' as the name does not violate protocol,
and does not confuse qemu as the nbd client since our implementation
does not ask for the name; but might confuse some other client that
does ask for the name especially if the default export is different
than the export name being queried).

Adding an assert makes it obvious that we are not skipping any bytes
in the client's message, as well as making it obvious that we were
using the wrong variable.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
CC: qemu-stable@nongnu.org
Message-Id: <20171101154204.27146-1-vsementsov@virtuozzo.com>
[eblake: improve commit message, squash in assert addition]
Signed-off-by: Eric Blake <eblake@redhat.com>
6 years agoutil/async: use atomic_mb_set in qemu_bh_cancel
Sergio Lopez [Wed, 8 Nov 2017 06:34:47 +0000 (07:34 +0100)]
util/async: use atomic_mb_set in qemu_bh_cancel

Commit b7a745d added a qemu_bh_cancel call to the completion function
as an optimization to prevent it from unnecessarily rescheduling itself.

This completion function is scheduled from worker_thread, after setting
the state of a ThreadPoolElement to THREAD_DONE.

This was considered to be safe, as the completion function restarts the
loop just after the call to qemu_bh_cancel. But, as this loop lacks a HW
memory barrier, the read of req->state may actually happen _before_ the
call, seeing it still as THREAD_QUEUED, and ending the completion
function without having processed a pending TPE linked at pool->head:

         worker thread             |            I/O thread
------------------------------------------------------------------------
                                   | speculatively read req->state
req->state = THREAD_DONE;          |
qemu_bh_schedule(p->completion_bh) |
  bh->scheduled = 1;               |
                                   | qemu_bh_cancel(p->completion_bh)
                                   |   bh->scheduled = 0;
                                   | if (req->state == THREAD_DONE)
                                   |   // sees THREAD_QUEUED

The source of the misunderstanding was that qemu_bh_cancel is now being
used by the _consumer_ rather than the producer, and therefore now needs
to have acquire semantics just like e.g. aio_bh_poll.

In some situations, if there are no other independent requests in the
same aio context that could eventually trigger the scheduling of the
completion function, the omitted TPE and all operations pending on it
will get stuck forever.

[Added Sergio's updated wording about the HW memory barrier.
--Stefan]

Signed-off-by: Sergio Lopez <slp@redhat.com>
Message-id: 20171108063447.2842-1-slp@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
6 years agocrypto: afalg: fix a NULL pointer dereference
Longpeng [Tue, 7 Nov 2017 11:32:06 +0000 (19:32 +0800)]
crypto: afalg: fix a NULL pointer dereference

Test-crypto-hash calls qcrypto_hash_bytesv/digest/base64 with
errp=NULL, this will cause a NULL pointer dereference if afalg_driver
doesn't support requested algos:

    ret = qcrypto_hash_afalg_driver.hash_bytesv(alg, iov, niov,
                                                result, resultlen,
                                                errp);
    if (ret == 0) {
        return ret;
    }

    error_free(*errp);  // <--- here

Because the error message is thrown away immediately, we should
just pass NULL to hash_bytesv(). There is also the same problem in
afalg-backend cipher & hmac, let's fix them together.

Reviewed-by: Eric Blake <eblake@redhat.com>
Reported-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Longpeng <longpeng2@huawei.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agotests: Run the luks tests in test-crypto-block only if encryption is available
Thomas Huth [Fri, 3 Nov 2017 11:54:52 +0000 (12:54 +0100)]
tests: Run the luks tests in test-crypto-block only if encryption is available

The test-crypto-block currently fails if encryption has not been
compiled into QEMU:

TEST: tests/test-crypto-block... (pid=22231)
  /crypto/block/qcow:                                                  OK
  /crypto/block/luks/default:
  Unexpected error in qcrypto_pbkdf2() at qemu/crypto/pbkdf-stub.c:41:
FAIL
GTester: last random seed: R02Sbbb5b6f299c6727f41bb50ba4aa6ef5c
(pid=22237)
  /crypto/block/luks/aes-256-cbc-plain64:
  Unexpected error in qcrypto_pbkdf2() at qemu/crypto/pbkdf-stub.c:41:
FAIL
GTester: last random seed: R02S3e27992a5ab4cc95e141c4ed3c7f0d2e
(pid=22239)
  /crypto/block/luks/aes-256-cbc-essiv:
  Unexpected error in qcrypto_pbkdf2() at qemu/crypto/pbkdf-stub.c:41:
FAIL
GTester: last random seed: R02S51b52bb02a66c42d8b331fd305384f53
(pid=22241)
FAIL: tests/test-crypto-block

So run the luks test only if the required encryption support is available.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agos390x/pci: let pci devices start in configured mode
Christian Borntraeger [Tue, 7 Nov 2017 17:54:55 +0000 (18:54 +0100)]
s390x/pci: let pci devices start in configured mode

Currently, to enable a pci device in the guest, the user has to issue
echo 1 > /sys/bus/pci/slots/00000000/power. This is not what people
expect. On an LPAR, the user can put a PCI device in configured or
deconfigured state via IOCDS. The "start in deconfigured state" can be
used for "sharing" a pci function across LPARs. This is not what we are
going to use in KVM, so always start configured.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Yi Min Zhao <zyimin@linux.vnet.ibm.com>
Reviewed-by: Pierre Morel <pmorel@linux.vnet.ibm.com>
Message-Id: <20171107175455.73793-2-borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
6 years agotests-aio-multithread: fix /aio/multi/schedule race condition
Stefan Hajnoczi [Mon, 6 Nov 2017 19:02:33 +0000 (19:02 +0000)]
tests-aio-multithread: fix /aio/multi/schedule race condition

test_multi_co_schedule_entry() set to_schedule[id] in the final loop
iteration before terminating the coroutine.  There is a race condition
where the main thread attempts to enter the terminating or terminated
coroutine when signalling coroutines to stop:

  atomic_mb_set(&now_stopping, true);
  for (i = 0; i < NUM_CONTEXTS; i++) {
      ctx_run(i, finish_cb, NULL);  <--- enters dead coroutine!
      to_schedule[i] = NULL;
  }

Make sure only to set to_schedule[id] if this coroutine really needs to
be scheduled!

Reported-by: "R.Nageswara Sastry" <nasastry@in.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20171106190233.1175-1-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
6 years agoe500: ppce500_init_mpic() return device instead of IRQ array
Michael Davidsaver [Fri, 20 Oct 2017 14:38:52 +0000 (09:38 -0500)]
e500: ppce500_init_mpic() return device instead of IRQ array

Actual number of interrupt pins isn't known
in ppce500_init_mpic() so a hardcoded number
was used, which causes a crash with older openpic.

Instead, return the DeviceState* and change ppce500_init()
to call qdev_get_gpio_in() to get only the irq pins
which are needed.

Signed-off-by: Michael Davidsaver <mdavidsaver@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
6 years agohw/display/sm501: Fix comment in sm501_sysbus_class_init()
Thomas Huth [Fri, 20 Oct 2017 09:53:53 +0000 (11:53 +0200)]
hw/display/sm501: Fix comment in sm501_sysbus_class_init()

The "cannot_instantiate_with_device_add_yet" flag has been renamed
to "user_creatable" a while ago.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
6 years agoppc: fix setting of compat mode
Greg Kurz [Tue, 17 Oct 2017 19:49:14 +0000 (21:49 +0200)]
ppc: fix setting of compat mode

While trying to make KVM PR usable again, commit 5dfaa532ae introduced a
regression: the current compat_pvr value is passed to KVM instead of the
new one. This means that we always pass 0 instead of the max-cpu-compat
PVR during the initial machine reset. And at CAS time, we either pass
the PVR from the command line or even don't call kvmppc_set_compat() at
all, ie, the PCR will not be set as expected.

For example if we start a big endian fedora26 guest in power7 compat
mode on a POWER8 host, we get this in the guest:

$ cat /proc/cpuinfo
processor       : 0
cpu             : POWER7 (architected), altivec supported
clock           : 4024.000000MHz
revision        : 2.0 (pvr 004d 0200)

timebase        : 512000000
platform        : pSeries
model           : IBM pSeries (emulated by qemu)
machine         : CHRP IBM pSeries (emulated by qemu)
MMU             : Hash

but the guest can still execute POWER8 instructions, and the following
program succeeds:

int main()
{
        asm("vncipher 0,0,0"); // ISA 2.07 instruction
}

Let's pass the new compat_pvr to kvmppc_set_compat() and the program fails
with SIGILL as expected.

Reported-by: Nageswara R Sastry <rnsastry@linux.vnet.ibm.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
6 years agoUpdate version for v2.11.0-rc0 release
Peter Maydell [Tue, 7 Nov 2017 16:05:28 +0000 (16:05 +0000)]
Update version for v2.11.0-rc0 release

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/berrange/tags/pull-2017-11-06-2' into staging
Peter Maydell [Tue, 7 Nov 2017 14:43:35 +0000 (14:43 +0000)]
Merge remote-tracking branch 'remotes/berrange/tags/pull-2017-11-06-2' into staging

Pull IO 2017/11/06 v2

# gpg: Signature made Tue 07 Nov 2017 14:35:25 GMT
# gpg:                using RSA key 0xBE86EBB415104FDF
# gpg: Good signature from "Daniel P. Berrange <dan@berrange.com>"
# gpg:                 aka "Daniel P. Berrange <berrange@redhat.com>"
# Primary key fingerprint: DAF3 A6FD B26B 6291 2D0E  8E3F BE86 EBB4 1510 4FDF

* remotes/berrange/tags/pull-2017-11-06-2:
  sockets: avoid leak of listen file descriptor

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agosockets: avoid leak of listen file descriptor
Daniel P. Berrange [Fri, 20 Oct 2017 09:18:39 +0000 (10:18 +0100)]
sockets: avoid leak of listen file descriptor

If we iterate over the full port range without successfully binding+listening
on the socket, we'll try the next address, whereupon we overwrite the slisten
file descriptor variable without closing it.

Rather than having two places where we open + close socket FDs on different
iterations of nested for loops, re-arrange the code to always open+close
within the same loop iteration.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20171107' into...
Peter Maydell [Tue, 7 Nov 2017 13:54:41 +0000 (13:54 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20171107' into staging

target-arm queue:
 * arm_gicv3_its: Don't abort on table save failure
 * arm_gicv3_its: Fix the VM termination in vm_change_state_handler()
 * translate.c: Fix usermode big-endian AArch32 LDREXD and STREXD
 * hw/arm: Mark the "fsl,imx31/25/6" devices with user_creatable = false
 * arm: implement cache/shareability attribute bits for PAR registers

# gpg: Signature made Tue 07 Nov 2017 13:33:58 GMT
# gpg:                using RSA key 0x3C2525ED14360CDE
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>"
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>"
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>"
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* remotes/pmaydell/tags/pull-target-arm-20171107:
  hw/intc/arm_gicv3_its: Don't abort on table save failure
  hw/intc/arm_gicv3_its: Fix the VM termination in vm_change_state_handler()
  translate.c: Fix usermode big-endian AArch32 LDREXD and STREXD
  hw/arm: Mark the "fsl,imx31" device with user_creatable = false
  hw/arm: Mark the "fsl,imx25" device with user_creatable = false
  hw/arm: Mark the "fsl,imx6" device with user_creatable = false
  arm: implement cache/shareability attribute bits for PAR registers

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agohw/intc/arm_gicv3_its: Don't abort on table save failure
Eric Auger [Tue, 7 Nov 2017 13:03:52 +0000 (13:03 +0000)]
hw/intc/arm_gicv3_its: Don't abort on table save failure

The ITS is not fully properly reset at the moment. Caches are
not emptied.

After a reset, in case we attempt to save the state before
the bound devices have registered their MSIs and after the
1st level table has been allocated by the ITS driver
(device BASER is valid), the first level entries are still
invalid. If the device cache is not empty (devices registered
before the reset), vgic_its_save_device_tables fails with -EINVAL.
This causes a QEMU abort().

Cc: qemu-stable@nongnu.org
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reported-by: wanghaibin <wanghaibin.wang@huawei.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agohw/intc/arm_gicv3_its: Fix the VM termination in vm_change_state_handler()
Shanker Donthineni [Tue, 7 Nov 2017 13:03:51 +0000 (13:03 +0000)]
hw/intc/arm_gicv3_its: Fix the VM termination in vm_change_state_handler()

The commit cddafd8f353d ("hw/intc/arm_gicv3_its: Implement state save
/restore") breaks the backward compatibility with the older kernels
where vITS save/restore support is not available. The vmstate function
vm_change_state_handler() should not be registered if the running kernel
doesn't support ITS save/restore feature. Otherwise VM instance will be
killed whenever vmstate callback function is invoked.

Observed a virtual machine shutdown with QEMU-2.10+linux-4.11 when testing
the reboot command "virsh reboot <domain> --mode acpi" instead of reboot.

KVM Error: 'KVM_SET_DEVICE_ATTR failed: Group 4 attr 0x00000000000001'

Signed-off-by: Shanker Donthineni <shankerd@codeaurora.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Message-id: 1509712671-16299-1-git-send-email-shankerd@codeaurora.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agotranslate.c: Fix usermode big-endian AArch32 LDREXD and STREXD
Peter Maydell [Tue, 7 Nov 2017 13:03:51 +0000 (13:03 +0000)]
translate.c: Fix usermode big-endian AArch32 LDREXD and STREXD

For AArch32 LDREXD and STREXD, architecturally the 32-bit word at the
lowest address is always Rt and the one at addr+4 is Rt2, even if the
CPU is big-endian. Our implementation does these with a single
64-bit store, so if we're big-endian then we need to put the two
32-bit halves together in the opposite order to little-endian,
so that they end up in the right places. We were trying to do
this with the gen_aa32_frob64() function, but that is not correct
for the usermode emulator, because there there is a distinction
between "load a 64 bit value" (which does a BE 64-bit access
and doesn't need swapping) and "load two 32 bit values as one
64 bit access" (where we still need to do the swapping, like
system mode BE32).

Fixes: https://bugs.launchpad.net/qemu/+bug/1725267
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 1509622400-13351-1-git-send-email-peter.maydell@linaro.org

6 years agohw/arm: Mark the "fsl,imx31" device with user_creatable = false
Thomas Huth [Tue, 7 Nov 2017 13:03:51 +0000 (13:03 +0000)]
hw/arm: Mark the "fsl,imx31" device with user_creatable = false

QEMU currently crashes when the user tries to instantiate the fsl,imx31
device manually:

$ aarch64-softmmu/qemu-system-aarch64 -M kzm -device fsl,,imx31
**
ERROR:/home/thuth/devel/qemu/tcg/tcg.c:538:tcg_register_thread:
 assertion failed: (n < max_cpus)
Aborted (core dumped)

The kzm board (which is the one that uses this CPU type) only supports
one CPU, and the realize function of the "fsl,imx31" device also uses
serial_hds[] directly, so this device clearly can not be instantiated
twice and thus we should mark it with user_creatable = false.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-id: 1509519537-6964-4-git-send-email-thuth@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agohw/arm: Mark the "fsl,imx25" device with user_creatable = false
Thomas Huth [Tue, 7 Nov 2017 13:03:51 +0000 (13:03 +0000)]
hw/arm: Mark the "fsl,imx25" device with user_creatable = false

QEMU currently crashes when the user tries to instantiate the fsl,imx25
device manually:

$ aarch64-softmmu/qemu-system-aarch64 -S -M imx25-pdk -device fsl,,imx25
**
ERROR:/home/thuth/devel/qemu/tcg/tcg.c:538:tcg_register_thread:
 assertion failed: (n < max_cpus)

The imx25-pdk board (which is the one that uses this CPU type) only
supports one CPU, and the realize function of the "fsl,imx25" device
also uses serial_hds[] directly, so this device clearly can not be
instantiated twice and thus we should mark it with user_creatable = 0.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-id: 1509519537-6964-3-git-send-email-thuth@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agohw/arm: Mark the "fsl,imx6" device with user_creatable = false
Thomas Huth [Tue, 7 Nov 2017 13:03:51 +0000 (13:03 +0000)]
hw/arm: Mark the "fsl,imx6" device with user_creatable = false

This device causes QEMU to abort if the user tries to instantiate it:

$ qemu-system-aarch64 -M sabrelite -smp 1,maxcpus=2 -device fsl,,imx6
Unexpected error in qemu_chr_fe_init() at chardev/char-fe.c:222:
qemu-system-aarch64: -device fsl,,imx6: Device 'serial0' is in use
Aborted (core dumped)

The device uses serial_hds[] directly in its realize function, so it
can not be instantiated again by the user.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-id: 1509519537-6964-2-git-send-email-thuth@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoarm: implement cache/shareability attribute bits for PAR registers
Andrew Baumann [Tue, 7 Nov 2017 13:03:51 +0000 (13:03 +0000)]
arm: implement cache/shareability attribute bits for PAR registers

On a successful address translation instruction, PAR is supposed to
contain cacheability and shareability attributes determined by the
translation. We previously returned 0 for these bits (in line with the
general strategy of ignoring caches and memory attributes), but some
guest OSes may depend on them.

This patch collects the attribute bits in the page-table walk, and
updates PAR with the correct attributes for all LPAE translations.
Short descriptor formats still return 0 for these bits, as in the
prior implementation.

Signed-off-by: Andrew Baumann <Andrew.Baumann@microsoft.com>
Message-id: 20171031223830.4608-1-Andrew.Baumann@microsoft.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/pmaydell/tags/pull-cocoa-20171107' into staging
Peter Maydell [Tue, 7 Nov 2017 12:19:48 +0000 (12:19 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-cocoa-20171107' into staging

cocoa queue:
 * make scrolling work in GUI monitor windows
 * change ungrab to ctrl-alt-g (matching gtk)
 * pass unused ctrl-alt combos to guest

# gpg: Signature made Tue 07 Nov 2017 10:15:00 GMT
# gpg:                using RSA key 0x3C2525ED14360CDE
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>"
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>"
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>"
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* remotes/pmaydell/tags/pull-cocoa-20171107:
  ui/cocoa.m: Send ctrl-alt key combos to guest if QEMU isn't using them
  ui/cocoa.m: move ungrab to ctrl-alt-g
  ui/cocoa.m: Make scrolling work again in GUI monitor windows

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/berrange/tags/pull-build-2017-11-07-1' into...
Peter Maydell [Tue, 7 Nov 2017 11:42:57 +0000 (11:42 +0000)]
Merge remote-tracking branch 'remotes/berrange/tags/pull-build-2017-11-07-1' into staging

Merge build 2017/11/07 v1

# gpg: Signature made Tue 07 Nov 2017 10:14:49 GMT
# gpg:                using RSA key 0xBE86EBB415104FDF
# gpg: Good signature from "Daniel P. Berrange <dan@berrange.com>"
# gpg:                 aka "Daniel P. Berrange <berrange@redhat.com>"
# Primary key fingerprint: DAF3 A6FD B26B 6291 2D0E  8E3F BE86 EBB4 1510 4FDF

* remotes/berrange/tags/pull-build-2017-11-07-1:
  build: remove use of MAKELEVEL optimization in submodule handling
  build: delay check for empty git submodule list
  build: don't fail if given a git submodule which does not exist
  build: allow automatic git submodule updates to be disabled
  build: don't create temporary files in source dir
  build: allow setting a custom GIT binary for transparent proxying

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/gkurz/tags/for-upstream' into staging
Peter Maydell [Tue, 7 Nov 2017 10:59:07 +0000 (10:59 +0000)]
Merge remote-tracking branch 'remotes/gkurz/tags/for-upstream' into staging

This fixes a bad errno returned to the guest and a trivial coding style nit.

# gpg: Signature made Mon 06 Nov 2017 18:09:24 GMT
# gpg:                using RSA key 0x71D4D5E5822F73D6
# gpg: Good signature from "Greg Kurz <groug@kaod.org>"
# gpg:                 aka "Gregory Kurz <gregory.kurz@free.fr>"
# gpg:                 aka "[jpeg image of size 3330]"
# Primary key fingerprint: B482 8BAF 9431 40CE F2A3  4910 71D4 D5E5 822F 73D6

* remotes/gkurz/tags/for-upstream:
  9pfs: fix v9fs_mark_fids_unreclaim() return value
  9pfs: drop one user of struct V9fsFidState

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoui/cocoa.m: Send ctrl-alt key combos to guest if QEMU isn't using them
Peter Maydell [Tue, 7 Nov 2017 10:14:14 +0000 (10:14 +0000)]
ui/cocoa.m: Send ctrl-alt key combos to guest if QEMU isn't using them

Send those ctrl-alt key combos that QEMU doesn't treat specially to
the guest rather than ignoring them.

All the case where we do special handling of ctrl-alt-X exit the
event handling using a "return" statement, so we can simply allow
the rest to fall through into the normal key handling by deleting
the now-spurious "else".

We take the opportunity to clean up some oddly-formatted and
now rather uninformative comments by removing them.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoui/cocoa.m: move ungrab to ctrl-alt-g
John Arbuckle [Tue, 7 Nov 2017 10:14:14 +0000 (10:14 +0000)]
ui/cocoa.m: move ungrab to ctrl-alt-g

Currently the cocoa user interface relis on the user pushing
control-alt to ungrab the mouse.  This is patch changes the key
combination to control-alt-g to be in line with the GTK user
interface.

Signed-off-by: John Arbuckle <programmingkidx@gmail.com>
Message-id: 20171102213907.11443-1-programmingkidx@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoui/cocoa.m: Make scrolling work again in GUI monitor windows
John Arbuckle [Tue, 7 Nov 2017 10:14:14 +0000 (10:14 +0000)]
ui/cocoa.m: Make scrolling work again in GUI monitor windows

Make scrolling in the monitor work, by correctly passing through
control+key combinations.

Signed-off-by: John Arbuckle <programmingkidx@gmail.com>
Message-id: 20171101154607.1582-1-programmingkidx@gmail.com
[PMM: fixed coding style nits; cleaned up commit message]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years ago9pfs: fix v9fs_mark_fids_unreclaim() return value
Greg Kurz [Mon, 6 Nov 2017 17:05:35 +0000 (18:05 +0100)]
9pfs: fix v9fs_mark_fids_unreclaim() return value

The return value of v9fs_mark_fids_unreclaim() is then propagated to
pdu_complete(). It should be a negative errno, not -1.

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
6 years ago9pfs: drop one user of struct V9fsFidState
Greg Kurz [Mon, 6 Nov 2017 17:05:35 +0000 (18:05 +0100)]
9pfs: drop one user of struct V9fsFidState

To comply with QEMU coding style.

Signed-off-by: Greg Kurz <groug@kaod.org>
6 years agoMerge remote-tracking branch 'remotes/yongbok/tags/mips-20171106' into staging
Peter Maydell [Mon, 6 Nov 2017 16:13:09 +0000 (16:13 +0000)]
Merge remote-tracking branch 'remotes/yongbok/tags/mips-20171106' into staging

MIPS patches 2017-11-06

Changes:
Update email addresses of Yongbok Kim, James Hogan and Paul Burton.

# gpg: Signature made Mon 06 Nov 2017 15:38:58 GMT
# gpg:                using RSA key 0x2238EB86D5F797C2
# gpg: Good signature from "Yongbok Kim <yongbok.kim@mips.com>"
# gpg:                 aka "Yongbok Kim <yongbok.kim@imgtec.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: 8600 4CF5 3415 A5D9 4CFA  2B5C 2238 EB86 D5F7 97C2

* remotes/yongbok/tags/mips-20171106:
  MAINTAINERS: Update Paul Burton's email address
  MAINTAINERS: Update James Hogan's email address
  MAINTAINERS: Update Yongbok Kim's email address

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMAINTAINERS: Update Paul Burton's email address
Paul Burton [Fri, 20 Oct 2017 23:21:14 +0000 (16:21 -0700)]
MAINTAINERS: Update Paul Burton's email address

Update my email address from paul.burton@imgtec.com to
paul.burton@mips.com, since MIPS will soon no longer be a part of
Imagination Technologies & as such the @imgtec.com address will soon
cease to function.

A mapping is added in .mailmap such that git reports the new @mips.com
address, and get_maintainer.pl in turn reports it when examining git
history. Whilst here add a mapping for my also-defunct
paul@archlinuxmips.org email address too.

Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: Laurent Vivier <laurent@vivier.eu>
Cc: Michael Tokarev <mjt@tls.msk.ru>
Cc: Yongbok Kim <yongbok.kim@imgtec.com>
Cc: Yongbok Kim <yongbok.kim@mips.com>
Cc: qemu-devel@nongnu.org
Cc: qemu-trivial@nongnu.org
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
[Yongbok Kim:
  Updated message subject]
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
6 years agoMAINTAINERS: Update James Hogan's email address
James Hogan [Tue, 17 Oct 2017 14:13:11 +0000 (15:13 +0100)]
MAINTAINERS: Update James Hogan's email address

Update my imgtec.com email address to my kernel.org one in MAINTAINERS
as MIPS will soon no longer be part of Imagination Technologies, and add
a mapping in .mailcap so get_maintainer.pl reports the right address.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Michael Tokarev <mjt@tls.msk.ru>
Cc: Laurent Vivier <laurent@vivier.eu>
Cc: qemu-devel@nongnu.org
Cc: qemu-trivial@nongnu.org
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
[Yongbok Kim:
  Updated message subject]
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
6 years agoMAINTAINERS: Update Yongbok Kim's email address
Yongbok Kim [Tue, 17 Oct 2017 15:07:19 +0000 (16:07 +0100)]
MAINTAINERS: Update Yongbok Kim's email address

Update my email address to mips.com

Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
6 years agobuild: remove use of MAKELEVEL optimization in submodule handling
Daniel P. Berrange [Mon, 6 Nov 2017 11:08:14 +0000 (11:08 +0000)]
build: remove use of MAKELEVEL optimization in submodule handling

The Makefile attempts to optimize the handling of submodules by using MAKELEVEL
to only check the submodule status when running from the top level make
invokation. This causes problems for people who are using a makefile of their
own to in turn invoke QEMU's makefile, as MAKELEVEL is already set to 1 (or
more) when QEMU's makefile runs.

This optimization should not really be needed, since the git-submodule.sh
script is already used to detect if a submodule update is required. This by
removing the MAKELEVEL check, we at most add an extra 'git-submodule.sh status'
call to each make level, the overhead of which is lost in noise of building
QEMU.

Reviewed-by: Greg Kurz <groug@kaod.org>
Tested-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging
Peter Maydell [Mon, 6 Nov 2017 11:24:14 +0000 (11:24 +0000)]
Merge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging

# gpg: Signature made Mon 06 Nov 2017 11:20:18 GMT
# gpg:                using RSA key 0x9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>"
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>"
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35  775A 9CA4 ABB3 81AB 73C8

* remotes/stefanha/tags/block-pull-request:
  aio-posix: drop QEMU_AIO_POLL_MAX_NS env var

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agobuild: delay check for empty git submodule list
Daniel P. Berrange [Mon, 30 Oct 2017 08:29:29 +0000 (09:29 +0100)]
build: delay check for empty git submodule list

We short circuit the git submodule update when passed an empty module list.
This accidentally causes the 'status' command to write to the status file. The
test needs to be delayed into the individual commands to avoid this premature
writing of the status file.

Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agobuild: don't fail if given a git submodule which does not exist
Daniel P. Berrange [Fri, 27 Oct 2017 09:49:58 +0000 (10:49 +0100)]
build: don't fail if given a git submodule which does not exist

If going back in time in git history, across a commit that introduces a new
submodule, the 'git-submodule.sh' script will fail, causing rebuild to fail.

This is because config-host.mak contains a GIT_SUBMODULES variable that lists
a submodule that only exists in the later commit. config-host.mak won't get
repopulated until config.status is invoked, but make won't get this far due to
the submodule error.

This change makes 'git-submodule.sh' check whether each module is known to git
and drops any which are not present. A warning message will be printed when any
submodule is dropped in this manner.

Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agoaio-posix: drop QEMU_AIO_POLL_MAX_NS env var
Stefan Hajnoczi [Fri, 3 Nov 2017 15:40:41 +0000 (15:40 +0000)]
aio-posix: drop QEMU_AIO_POLL_MAX_NS env var

This hunk should not have been merged but I forgot to remove it.  Let's
remove it before it slips into a QEMU release.

¯\_(ツ)_/¯

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20171103154041.12617-1-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
6 years agobuild: allow automatic git submodule updates to be disabled
Daniel P. Berrange [Thu, 26 Oct 2017 12:52:26 +0000 (13:52 +0100)]
build: allow automatic git submodule updates to be disabled

Some people building QEMU use VPATH builds where the source directory is on a
read-only volume. In such a case 'scripts/git-submodules.sh update' will always
fail and users are required to run it manually themselves on their original
writable source directory.

While this is already supported, it is nice to give users a command line flag
to configure to permanently disable automatic submodule updates, as it means
they won't get hard to diagnose failures from git-submodules.sh at an arbitrary
later date.

This patch thus introduces a flag '--disable-git-update' which will prevent
'make' from ever running 'scripts/git-submodules.sh update'. It will still run
the 'status' command to determine if a submodule update is needed, but when it
does this it'll simply stop and print a message instructing the developer what
todo. eg

$ ./configure  --target-list=x86_64-softmmu --disable-git-update
...snip...

$ make
  GEN     config-host.h
  GEN     trace/generated-tcg-tracers.h
  GEN     trace/generated-helpers-wrappers.h
  GEN     trace/generated-helpers.h
  GEN     trace/generated-helpers.c
  GEN     module_block.h

GIT submodule checkout is out of date. Please run
  scripts/git-submodule.sh update ui/keycodemapdb
from the source directory checkout /home/berrange/src/virt/qemu

make: *** [Makefile:31: git-submodule-update] Error 1

Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agobuild: don't create temporary files in source dir
Daniel P. Berrange [Thu, 26 Oct 2017 12:45:38 +0000 (13:45 +0100)]
build: don't create temporary files in source dir

There are cases where users do VPATH builds with the source directory being on
a read-only volume. In such a case they have to manually run the command
'git-submodule.sh ...modules...' ahead of time. When checking for status we
should not then write into the source dir.

Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agobuild: allow setting a custom GIT binary for transparent proxying
Daniel P. Berrange [Fri, 20 Oct 2017 14:02:43 +0000 (15:02 +0100)]
build: allow setting a custom GIT binary for transparent proxying

Some users can't run a bare 'git' command, due to need for a transparent
proxying solution such as 'tsocks'. This adds an argument to configure to
let users specify such a thing:

  ./configure --with-git="tsocks git"

The submodule script is also updated to give the user a hint about using this
flag, if we fail to checkout modules.

Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
Peter Maydell [Mon, 6 Nov 2017 10:04:16 +0000 (10:04 +0000)]
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging

* Remove legacy KVM PCI device assignment
* Fix qemu-pr-helper.c compilation on older compilers

# gpg: Signature made Sun 05 Nov 2017 13:52:36 GMT
# gpg:                using RSA key 0xBFFBD25F78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>"
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* remotes/bonzini/tags/for-upstream:
  build: disable -Wmissing-braces on older compilers
  pci-assign: Remove

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agobuild: disable -Wmissing-braces on older compilers
Paolo Bonzini [Fri, 20 Oct 2017 10:11:32 +0000 (12:11 +0200)]
build: disable -Wmissing-braces on older compilers

GCC 4.9 and newer stopped warning for missing braces around the
"universal" C zero initializer {0}.  One such initializer sneaked
into scsi/qemu-pr-helper.c and is breaking the build with such
older GCC versions.

Detect the lack of support for the idiom, and disable the warning
in that case.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
6 years agopci-assign: Remove
Paolo Bonzini [Fri, 20 Oct 2017 08:17:50 +0000 (10:17 +0200)]
pci-assign: Remove

Legacy PCI device assignment has been removed from Linux in 4.12,
and had been deprecated 2 years ago there.  We can remove it from
QEMU as well.

The ROM loading code was shared with Xen PCI passthrough, so move
it to hw/xen.

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/stsquad/tags/pull-ci-updates-for-softfreeze...
Peter Maydell [Fri, 3 Nov 2017 10:08:34 +0000 (10:08 +0000)]
Merge remote-tracking branch 'remotes/stsquad/tags/pull-ci-updates-for-softfreeze-021117-2' into staging

One fix for win32/64 cross compiles.

# gpg: Signature made Thu 02 Nov 2017 11:55:46 GMT
# gpg:                using RSA key 0xFBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <alex.bennee@linaro.org>"
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8  DF35 FBD0 DB09 5A9E 2A44

* remotes/stsquad/tags/pull-ci-updates-for-softfreeze-021117-2:
  docker: add python stdlib dependency (required by keycodemapdb)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/rth/tags/pull-tcg-20171103' into staging
Peter Maydell [Fri, 3 Nov 2017 09:31:33 +0000 (09:31 +0000)]
Merge remote-tracking branch 'remotes/rth/tags/pull-tcg-20171103' into staging

Queued tcg patches

# gpg: Signature made Fri 03 Nov 2017 08:37:58 GMT
# gpg:                using RSA key 0x64DF38E8AF7E215F
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>"
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* remotes/rth/tags/pull-tcg-20171103:
  cpu-exec: Exit exclusive region on longjmp from step_atomic
  tcg/s390x: Use constant pool for prologue
  tcg: Allow constant pool entries in the prologue

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agocpu-exec: Exit exclusive region on longjmp from step_atomic
Peter Maydell [Thu, 2 Nov 2017 16:35:36 +0000 (16:35 +0000)]
cpu-exec: Exit exclusive region on longjmp from step_atomic

Commit ac03ee5331612e44be narrowed the scope of the exclusive
region so it only covers when we're executing the TB, not when
we're generating it. However it missed that there is more than
one execution path out of cpu_tb_exec -- if the atomic insn
causes an exception then the code will longjmp out, skipping
the code to end the exclusive region. This causes QEMU to hang
the next time the CPU calls start_exclusive(), waiting for
itself to exit the region.

Move the "end the region" code out to the end of the
function so that it is run for both normal exit and also
for exit-via-longjmp. We have to use a volatile bool flag
to decide whether we need to end the region, because we
can longjump out of the codegen as well as the execution.

(For some reason this only reproduces for me with a clang
optimized build, not a gcc debug build.)

Reviewed-by: Emilio G. Cota <cota@braap.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Fixes: ac03ee5331612e44beb393df2b578c951d27dc0d
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <1509640536-32160-1-git-send-email-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6 years agotcg/s390x: Use constant pool for prologue
Richard Henderson [Wed, 25 Oct 2017 16:03:27 +0000 (18:03 +0200)]
tcg/s390x: Use constant pool for prologue

Rather than have separate code only used for guest_base,
rely on a recent change to handle constant pool entries.

Cc: qemu-s390x@nongnu.org
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6 years agotcg: Allow constant pool entries in the prologue
Richard Henderson [Wed, 25 Oct 2017 14:14:20 +0000 (07:14 -0700)]
tcg: Allow constant pool entries in the prologue

Both ARMv6 and AArch64 currently may drop complex guest_base values
into the constant pool.  But generic code wasn't expecting that, and
the pool is not emitted.  Correct that.

Tested-by: Emilio G. Cota <cota@braap.org>
Tested-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Reported-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6 years agooslib-posix: Use sysctl(2) call to resolve exec_dir on NetBSD
Kamil Rytarowski [Sat, 28 Oct 2017 19:48:33 +0000 (21:48 +0200)]
oslib-posix: Use sysctl(2) call to resolve exec_dir on NetBSD

NetBSD 8.0(beta) ships with KERN_PROC_PATHNAME in sysctl(2).
Older NetBSD versions can use argv[0] parsing fallback.

This code section is partly shared with FreeBSD.

Signed-off-by: Kamil Rytarowski <n54@gmx.com>
Message-id: 20171028194833.23858-1-n54@gmx.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoui/cocoa.m: Fix console selection keys
John Arbuckle [Thu, 5 Oct 2017 19:04:49 +0000 (15:04 -0400)]
ui/cocoa.m: Fix console selection keys

Fix console selection keys so that the right console is selected.

Signed-off-by: John Arbuckle <programmingkidx@gmail.com>
Message-id: 20171005190449.15591-1-programmingkidx@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agodocker: add python stdlib dependency (required by keycodemapdb)
Philippe Mathieu-Daudé [Tue, 31 Oct 2017 12:28:24 +0000 (09:28 -0300)]
docker: add python stdlib dependency (required by keycodemapdb)

Since 927128222b0a QEMU depends of keycodemapdb, which uses the python 'csv'
module from stdlib to parse keymaps.csv.

Without this package the build fails:

    GEN     ui/input-keymap-linux-to-qcode.c
  Traceback (most recent call last):
    File "ui/keycodemapdb/tools/keymap-gen", line 15, in <module>
      import csv
  ImportError: No module named csv
    GEN     ui/input-keymap-qcode-to-qnum.c
  Traceback (most recent call last):
    File "ui/keycodemapdb/tools/keymap-gen", line 15, in <module>
      import csv
  ImportError: No module named csv
  [...]
    CC      ui/input-keymap.o
  ui/input-keymap.c:8:44: fatal error: ui/input-keymap-linux-to-qcode.c: No such file or directory
  make: *** [ui/input-keymap.o] Error 1

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/jnsnow/tags/ide-pull-request' into staging
Peter Maydell [Thu, 2 Nov 2017 10:11:37 +0000 (10:11 +0000)]
Merge remote-tracking branch 'remotes/jnsnow/tags/ide-pull-request' into staging

# gpg: Signature made Tue 31 Oct 2017 23:01:18 GMT
# gpg:                using RSA key 0x7DEF8106AAFC390E
# gpg: Good signature from "John Snow (John Huston) <jsnow@redhat.com>"
# Primary key fingerprint: FAEB 9711 A12C F475 812F  18F2 88A9 064D 1835 61EB
#      Subkey fingerprint: F9B7 ABDB BCAC DF95 BE76  CBD0 7DEF 8106 AAFC 390E

* remotes/jnsnow/tags/ide-pull-request:
  ide: avoid referencing NULL dev in rotational rate setting
  hw/ide/ahci: Move allwinner code into a separate file

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoide: avoid referencing NULL dev in rotational rate setting
Daniel P. Berrange [Fri, 20 Oct 2017 09:14:03 +0000 (10:14 +0100)]
ide: avoid referencing NULL dev in rotational rate setting

The 'dev' variable can be NULL when the guest OS calls identify on an IDE
unit that does not have a drive attached to it.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20171020091403.1479-1-berrange@redhat.com
Signed-off-by: John Snow <jsnow@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/mcayland/tags/qemu-sparc-signed' into staging
Peter Maydell [Tue, 31 Oct 2017 18:09:11 +0000 (18:09 +0000)]
Merge remote-tracking branch 'remotes/mcayland/tags/qemu-sparc-signed' into staging

qemu-sparc update

# gpg: Signature made Tue 31 Oct 2017 17:43:11 GMT
# gpg:                using RSA key 0x5BC2C56FAE0F321F
# gpg: Good signature from "Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>"
# Primary key fingerprint: CC62 1AB9 8E82 200D 915C  C9C4 5BC2 C56F AE0F 321F

* remotes/mcayland/tags/qemu-sparc-signed:
  sun4m: change TYPE_SUN4M_IOMMU macro from "iommu" to "sun4m-iommu"
  sun4m_iommu: remove legacy sparc_iommu_memory_rw() function
  sparc32_dma: switch over to using IOMMU memory region and DMA API
  sun4m: implement IOMMU translation using IOMMU memory region
  sparc32_dma: add len to esp/le DMA memory tracing
  sparc32_dma: remove is_ledma hack and replace with memory region alias
  sparc32_dma: introduce new SPARC32_DMA type container object
  sparc32_dma: make lance device child of ledma device
  lance: move TYPE_LANCE and SysBusPCNetState from lance.c to lance.h
  sparc32_dma: make esp device child of espdma device
  esp: move TYPE_ESP and SysBusESPState from esp.c to esp.h
  sparc32_dma: use object link instead of qdev property to pass IOMMU reference
  sun4m_iommu: move TYPE_SUN4M_IOMMU declaration to sun4m.h
  sun4m: move DMA device wiring from sparc32_dma_init() to sun4m_hw_init()
  sparc32_dma: move type declarations from sparc32_dma.c to sparc32_dma.h
  sparc32_dma: split esp and le into separate DMA devices
  sparc32_dma: rename SPARC32_DMA type to SPARC32_DMA_DEVICE

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agohw/ide/ahci: Move allwinner code into a separate file
Thomas Huth [Tue, 31 Oct 2017 17:48:59 +0000 (13:48 -0400)]
hw/ide/ahci: Move allwinner code into a separate file

The allwinner code is only needed for the allwinner board (for which
we also have a separate CONFIG_ALLWINNER_A10 config switch), so it
does not make sense that we compile this for all the other boards
that need AHCI, too. Let's move it to a separate file that is only
compiled when CONFIG_ALLWINNER_A10 is set.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 1508784509-29377-1-git-send-email-thuth@redhat.com
Signed-off-by: John Snow <jsnow@redhat.com>
6 years agosun4m: change TYPE_SUN4M_IOMMU macro from "iommu" to "sun4m-iommu"
Mark Cave-Ayland [Fri, 27 Oct 2017 12:09:03 +0000 (13:09 +0100)]
sun4m: change TYPE_SUN4M_IOMMU macro from "iommu" to "sun4m-iommu"

This is a legacy artifact from when the sun4m IOMMU implementation was
the only IOMMU available within QEMU.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
6 years agosun4m_iommu: remove legacy sparc_iommu_memory_rw() function
Mark Cave-Ayland [Fri, 27 Oct 2017 12:09:03 +0000 (13:09 +0100)]
sun4m_iommu: remove legacy sparc_iommu_memory_rw() function

With the switch to the IOMMU memory region and DMA API, this is no longer
required.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
6 years agosparc32_dma: switch over to using IOMMU memory region and DMA API
Mark Cave-Ayland [Fri, 27 Oct 2017 12:09:03 +0000 (13:09 +0100)]
sparc32_dma: switch over to using IOMMU memory region and DMA API

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
6 years agosun4m: implement IOMMU translation using IOMMU memory region
Mark Cave-Ayland [Fri, 27 Oct 2017 12:09:03 +0000 (13:09 +0100)]
sun4m: implement IOMMU translation using IOMMU memory region

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
6 years agosparc32_dma: add len to esp/le DMA memory tracing
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:22 +0000 (13:22 +0100)]
sparc32_dma: add len to esp/le DMA memory tracing

This is surprisingly useful when trying to debug DMA issues.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosparc32_dma: remove is_ledma hack and replace with memory region alias
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:22 +0000 (13:22 +0100)]
sparc32_dma: remove is_ledma hack and replace with memory region alias

This hack originated from before the memory region API was introduced, and
increased the size of the ledma DMA device to capture incorrect accesses
beyond the end of the ledma device. A full analysis can be found on Artyom's
blog at http://tyom.blogspot.co.uk/2010/10/bug-in-all-solaris-versions-after-57.html.

With the memory API we can now simply alias the incorrect access onto its
intended destination allowing us to remove the hack.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosparc32_dma: introduce new SPARC32_DMA type container object
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:22 +0000 (13:22 +0100)]
sparc32_dma: introduce new SPARC32_DMA type container object

Create a new SPARC32_DMA container object (including an appropriate container
memory region) and add instances of the SPARC32_ESPDMA_DEVICE and
SPARC32_LEDMA_DEVICE as child objects. The benefit is that most of the gpio
wiring complexity between esp/espdma and lance/ledma is now hidden within the
SPARC32_DMA realize function.

Since the sun4m IOMMU is already QOMified we can find a reference to
it using object_resolve_path_type() allowing us to completely remove all external
references to the iommu pointer.

Finally we rework sun4m's sparc32_dma_init() to invoke the new SPARC32_DMA object
and wire up the remaining board memory regions/IRQs.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Acked-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosparc32_dma: make lance device child of ledma device
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:22 +0000 (13:22 +0100)]
sparc32_dma: make lance device child of ledma device

This makes it possible to reference the lance device from the ledma device as
required.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Acked-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agolance: move TYPE_LANCE and SysBusPCNetState from lance.c to lance.h
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:22 +0000 (13:22 +0100)]
lance: move TYPE_LANCE and SysBusPCNetState from lance.c to lance.h

This enables them to be used outside of lance.c. We also update the comment to
refer to the SPARC32 lance device rather than the AMD PCNet-II device (of which
lance is a register-compatible subset).

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
CC: Jason Wang <jasowang@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosparc32_dma: make esp device child of espdma device
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:22 +0000 (13:22 +0100)]
sparc32_dma: make esp device child of espdma device

This makes it possible to reference the esp device from the espdma device as
required, and by wiring up the device ourselves in sun4m.c we can drop use
of the esp_init() function.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Acked-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agoesp: move TYPE_ESP and SysBusESPState from esp.c to esp.h
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:22 +0000 (13:22 +0100)]
esp: move TYPE_ESP and SysBusESPState from esp.c to esp.h

This enables them to be used outside of esp.c.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
CC: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosparc32_dma: use object link instead of qdev property to pass IOMMU reference
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:21 +0000 (13:22 +0100)]
sparc32_dma: use object link instead of qdev property to pass IOMMU reference

This enables us to remove the last remaining (opaque) qdev property. Whilst we
are here, also update iommu_init() to use TYPE_SUN4M_IOMMU instead of a
hardcoded string.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosun4m_iommu: move TYPE_SUN4M_IOMMU declaration to sun4m.h
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:21 +0000 (13:22 +0100)]
sun4m_iommu: move TYPE_SUN4M_IOMMU declaration to sun4m.h

This is in preparation to allow the type to be used elsewhere.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosun4m: move DMA device wiring from sparc32_dma_init() to sun4m_hw_init()
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:21 +0000 (13:22 +0100)]
sun4m: move DMA device wiring from sparc32_dma_init() to sun4m_hw_init()

By using the sysbus interface it is possible to wire up the esp/le devices
to the sun4m DMA controller directly during sun4m_hw_init() instead of
passing qemu_irqs into the sparc32_dma_init() function.

This is an intermediate step to allow further reorganisation as more logic
is moved into the relevant SPARC32 DMA devices; there will be a final
refactoring of sparc32_dma_init() once this work is complete.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosparc32_dma: move type declarations from sparc32_dma.c to sparc32_dma.h
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:21 +0000 (13:22 +0100)]
sparc32_dma: move type declarations from sparc32_dma.c to sparc32_dma.h

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Acked-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosparc32_dma: split esp and le into separate DMA devices
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:21 +0000 (13:22 +0100)]
sparc32_dma: split esp and le into separate DMA devices

Due to slight differences in behaviour accessing the registers for the
esp and le devices, create two separate SPARC32_DMA_DEVICE types and
update the sun4m machine to use.

Note that by using different device types we already know the size of
the register block and the value of is_ledma at init time, allowing us to
drop the SPARC32_DMA_DEVICE realize function and the is_ledma device
property.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Acked-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agosparc32_dma: rename SPARC32_DMA type to SPARC32_DMA_DEVICE
Mark Cave-Ayland [Sat, 14 Oct 2017 12:22:21 +0000 (13:22 +0100)]
sparc32_dma: rename SPARC32_DMA type to SPARC32_DMA_DEVICE

Also update the function names to match as appropriate. While we're
here rename the type from sparc32_dma to sparc32-dma in order to
match the current QOM convention.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Artyom Tarasenko <atar4qemu@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6 years agoMerge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20171031' into...
Peter Maydell [Tue, 31 Oct 2017 14:28:25 +0000 (14:28 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20171031' into staging

target-arm queue:
 * fix instruction-length bit in syndrome for WFI/WFE traps
 * xlnx-zcu102: Specify the max number of CPUs
 * msf2: Remove dead code reported by Coverity
 * msf2: Wire up SYSRESETREQ in SoC for system reset
 * hw/pci-host/gpex: Improve INTX to gsi routing error checking

# gpg: Signature made Tue 31 Oct 2017 13:10:02 GMT
# gpg:                using RSA key 0x3C2525ED14360CDE
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>"
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>"
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>"
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* remotes/pmaydell/tags/pull-target-arm-20171031:
  hw/pci-host/gpex: Improve INTX to gsi routing error checking
  msf2: Wire up SYSRESETREQ in SoC for system reset
  msf2: Remove dead code reported by Coverity
  xlnx-zcu102: Specify the max number of CPUs
  fix WFI/WFE length in syndrome register

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2017-10-30' into staging
Peter Maydell [Tue, 31 Oct 2017 13:12:21 +0000 (13:12 +0000)]
Merge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2017-10-30' into staging

nbd patches for 2017-10-30

Vladimir Sementsov-Ogievskiy (some patches co-authored by Eric Blake):
00/12 nbd minimal structured read

# gpg: Signature made Mon 30 Oct 2017 20:54:54 GMT
# gpg:                using RSA key 0xA7A16B4A2527436A
# gpg: Good signature from "Eric Blake <eblake@redhat.com>"
# gpg:                 aka "Eric Blake (Free Software Programmer) <ebb9@byu.net>"
# gpg:                 aka "[jpeg image of size 6874]"
# Primary key fingerprint: 71C2 CC22 B1C4 6029 27D2  F3AA A7A1 6B4A 2527 436A

* remotes/ericb/tags/pull-nbd-2017-10-30:
  nbd: Minimal structured read for client
  nbd: Move nbd_read() to common header
  nbd/client: prepare nbd_receive_reply for structured reply
  nbd/client: refactor nbd_receive_starttls
  nbd/server: Include human-readable message in structured errors
  nbd: Minimal structured read for server
  nbd/server: Refactor zero-length option check
  nbd/server: Simplify nbd_negotiate_options loop
  nbd/server: Report error for write to read-only export
  nbd: Expose constants and structs for structured read
  nbd: Move nbd_errno_to_system_errno() to public header
  nbd: Include error names in trace messages

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agohw/pci-host/gpex: Improve INTX to gsi routing error checking
Eric Auger [Tue, 31 Oct 2017 11:50:52 +0000 (11:50 +0000)]
hw/pci-host/gpex: Improve INTX to gsi routing error checking

We exposed gpex_set_irq_num() for machines to set the INTx to
GSI routing. However if the machine forgets to call that
function we currently do not check the association was properly
done. Let's initialize gsi values to -1 and if this value is
found in gpex_route_intx_pin_to_irq, set the routing mode as
disabled.

Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-id: 1508776211-22175-1-git-send-email-eric.auger@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agomsf2: Wire up SYSRESETREQ in SoC for system reset
Subbaraya Sundeep [Tue, 31 Oct 2017 11:50:52 +0000 (11:50 +0000)]
msf2: Wire up SYSRESETREQ in SoC for system reset

Implemented system reset by creating SYSRESETREQ gpio
out from nvic.

Signed-off-by: Subbaraya Sundeep <sundeep.lkml@gmail.com>
Message-id: 1509253165-7434-1-git-send-email-sundeep.lkml@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agomsf2: Remove dead code reported by Coverity
Subbaraya Sundeep [Tue, 31 Oct 2017 11:50:51 +0000 (11:50 +0000)]
msf2: Remove dead code reported by Coverity

Fixed incorrect frame size mask, validated maximum frame
size in spi_write and removed dead code.

Signed-off-by: Subbaraya Sundeep <sundeep.lkml@gmail.com>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1508898544-10307-1-git-send-email-sundeep.lkml@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoxlnx-zcu102: Specify the max number of CPUs
Alistair Francis [Tue, 31 Oct 2017 11:50:51 +0000 (11:50 +0000)]
xlnx-zcu102: Specify the max number of CPUs

Specify the number of CPUs that can run on ZynqMP.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agofix WFI/WFE length in syndrome register
Stefano Stabellini [Tue, 31 Oct 2017 11:50:50 +0000 (11:50 +0000)]
fix WFI/WFE length in syndrome register

WFI/E are often, but not always, 4 bytes long. When they are, we need to
set ARM_EL_IL_SHIFT in the syndrome register.

Pass the instruction length to HELPER(wfi), use it to decrement pc
appropriately and to pass an is_16bit flag to syn_wfx, which sets
ARM_EL_IL_SHIFT if needed.

Set dc->insn in both arm_tr_translate_insn and thumb_tr_translate_insn.

Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>
Message-id: alpine.DEB.2.10.1710241055160.574@sstabellini-ThinkPad-X260
[PMM: move setting of dc->insn for Thumb so it is correct for 32 bit insns]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/dgilbert/tags/pull-hmp-20171030' into staging
Peter Maydell [Tue, 31 Oct 2017 11:18:00 +0000 (11:18 +0000)]
Merge remote-tracking branch 'remotes/dgilbert/tags/pull-hmp-20171030' into staging

hmp pull 2017-10-30

# gpg: Signature made Mon 30 Oct 2017 20:17:57 GMT
# gpg:                using RSA key 0x0516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <dgilbert@redhat.com>"
# Primary key fingerprint: 45F5 C71B 4A0C B7FB 977A  9FA9 0516 331E BC5B FDE7

* remotes/dgilbert/tags/pull-hmp-20171030:
  monitor: fix dangling CPU pointer
  hmp: Replace error_report_err

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agonbd: Minimal structured read for client
Vladimir Sementsov-Ogievskiy [Fri, 27 Oct 2017 10:40:37 +0000 (12:40 +0200)]
nbd: Minimal structured read for client

Minimal implementation: for structured error only error_report error
message.

Note that test 83 is now more verbose, because the implementation
prints more warnings about unexpected communication errors; perhaps
future patches should tone things down by using trace messages
instead of traces, but the common case of successful communication
is no noisier than before.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171027104037.8319-13-eblake@redhat.com>

6 years agonbd: Move nbd_read() to common header
Eric Blake [Fri, 27 Oct 2017 10:40:36 +0000 (12:40 +0200)]
nbd: Move nbd_read() to common header

An upcoming change to block/nbd-client.c will want to read the
tail of a structured reply chunk directly from the wire.  Move
this function to make it easier.

Based on a patch from Vladimir Sementsov-Ogievskiy.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20171027104037.8319-12-eblake@redhat.com>

6 years agonbd/client: prepare nbd_receive_reply for structured reply
Vladimir Sementsov-Ogievskiy [Fri, 27 Oct 2017 10:40:35 +0000 (12:40 +0200)]
nbd/client: prepare nbd_receive_reply for structured reply

In following patch nbd_receive_reply will be used both for simple
and structured reply header receiving.
NBDReply is altered into union of simple reply header and structured
reply chunk header, simple error translation moved to block/nbd-client
to be consistent with further structured reply error translation.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20171027104037.8319-11-eblake@redhat.com>