This patch introduces the helper "gen-features" which allows to generate
feature list definitions at compile time. Its flexibility is better and the
error-proneness is lower when compared to static programming time added
statements.

The helper includes "target-s390x/cpu_features.h" to be able to use named
facility bits instead of numbers. The generated defines will be used for
the definition of CPU models.

We generate feature lists for each HW generation and GA for EC models. BC
models are always based on a EC version and have no separate definitions.

Base features: Features we expect to be always available in sane setups.
Migration safe - will never change. Can be seen as "minimum features
required for a CPU model".

Default features: Features we expect to be stable and around in latest
setups (e.g. having KVM support) - not migration safe.

Max features: All supported features that are theoretically allowed for a
CPU model. Exceeding these features could otherwise produce problems with
IBC (instruction blocking controls) in KVM.
Acked-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMichael Mueller <mimu@linux.vnet.ibm.com>
Signed-off-by: NDavid Hildenbrand <dahi@linux.vnet.ibm.com>
[generate base, default and models. renaming and cleanup]
Message-Id: <20160905085244.99980-6-dahi@linux.vnet.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

The patch introduces s390x CPU features (most of them refered to as
facilities) along with their discription and some functions that will be
helpful when working with the features later on.

Please note that we don't introduce all known CPU features, only the
ones currently supported by KVM + QEMU. We don't want to enable later
on blindly any facilities, for which we don't know yet if we need QEMU
support to properly support them (e.g. migrate additional state when
active). We can update QEMU later on.
Acked-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMichael Mueller <mimu@linux.vnet.ibm.com>
Signed-off-by: NDavid Hildenbrand <dahi@linux.vnet.ibm.com>
[reworked to include non-stfle features, added definitions]
Message-Id: <20160905085244.99980-5-dahi@linux.vnet.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

Let's expose the description and migration safety and whether a definition
is static, as class properties, this can be helpful in the future.
Acked-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NDavid Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <20160905085244.99980-4-dahi@linux.vnet.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

This patch introduces two CPU models, "host" and "qemu".
"qemu" is used as default when running under TCG. "host" is used
as default when running under KVM. "host" cannot be used without KVM.
"host" is not migration-safe. They both inherit from the base s390x CPU,
which is turned into an abstract class.

This patch also changes CPU creation to take care of the passed CPU string
and reuses common code parse_features() function for that purpose. Unknown
CPU definitions are now reported. The "-cpu ?" and "query-cpu-definition"
commands are changed to list all CPU subclasses automatically, including
migration-safety and whether static.
Acked-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NDavid Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <20160905085244.99980-3-dahi@linux.vnet.ibm.com>
[CH: fix up self-assignments in s390_cpu_list, as spotted by clang]
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

It might be of interest for tooling whether a CPU definition can be safely
used when migrating, or if e.g. CPU features might get lost during
migration when migrationg from/to a different QEMU version or host, even if
the same compatibility machine is used.

Also, we want to know if a CPU definition is static and will never change.
Beause these definitions can then be used independantly of a compatibility
machine and will always have the same feature set, they can e.g. be used
to indicate the "host" model in libvirt later on.

Let's add two return values to query-cpu-definitions, stating for each
returned CPU definition, if it is migration-safe and if it is static.

While "migration-safe" is optional, "static" will be set to "false"
automatically by all implementing architectures. If a model really was
static all the time and will be in the future, this can simply be changed
later.
Reviewed-by: NEric Blake <eblake@redhat.com>
Signed-off-by: NDavid Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <20160905085244.99980-2-dahi@linux.vnet.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

Diag 501 (4 bytes) was used until now for software breakpoints on s390.
As instructions on s390 might be 2 bytes long, temporarily overwriting them
with 4 bytes is evil and can result in very strange guest behaviour.

We make use of invalid instruction 0x0000 as new sw breakpoint instruction.
We have to enable interception of that instruction in KVM using a
capability.

If no software breakpoint has been inserted at the reported position, an
operation exception has to be injected into the guest. Otherwise a
breakpoint has been hit and the pc has to be rewound.

If KVM doesn't yet support interception of instruction 0x0000 the
existing mechanism exploiting diag 501 is used. To keep overhead low,
interception of instruction 0x0000 will only be enabled if sw breakpoints
are really used.
Reviewed-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: NDavid Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

Update headers against 4.8-rc2.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

The cssid 255 is reserved but still valid from an architectural
point of view. However, feeding a bogus schid of 0xffffffff into
the virtio hypercall will lead to a crash:

Stack trace of thread 138363:
        #0  0x00000000100d168c css_find_subch (qemu-system-s390x)
        #1  0x00000000100d3290 virtio_ccw_hcall_notify
        #2  0x00000000100cbf60 s390_virtio_hypercall
        #3  0x000000001010ff7a handle_hypercall
        #4  0x0000000010079ed4 kvm_cpu_exec (qemu-system-s390x)
        #5  0x00000000100609b4 qemu_kvm_cpu_thread_fn
        #6  0x000003ff8b887bb4 start_thread (libpthread.so.0)
        #7  0x000003ff8b78df0a thread_start (libc.so.6)

This is because the css array was only allocated for 0..254
instead of 0..255.

Let's fix this by bumping MAX_CSSID to 255 and fencing off the
reserved cssid of 255 during css image allocation.
Reported-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Tested-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

As we provide format 1 chsc scpd data (and don't support any ficon
channels), we de facto already have the ficon-cascaded-switch
facility.
Reviewed-by: NPierre Morel <pmorel@linux.vnet.ibm.com>
Reviewed-by: NHalil Pasic <pasic@linux.vnet.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

Just a simple conversion to get rid of register_savevm.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

With the current code a simple sclp command takes about 13000 ns
The biggest part seems to be the resolver of the object model. By
caching the sclp device the time for an sclp command goes down to
2500ns. Talking about real life scenarios, this change doubles
the speed of the sclp console when sending single bytes outputs
to /dev/console.
Signed-off-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

If one pci device is plugged successfully, there must be a zpci device
existing. This means that during hot-unplugging a pci device, its
corresponding zpci device must be found. Therefore we use an assert to
replace current code.
Signed-off-by: NYi Min Zhao <zyimin@linux.vnet.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

In the case that zpci is automatically created, we did not return
immediately on failure, which would lead to NULL pointer dereferencing.
Let's fix it.
Signed-off-by: NYi Min Zhao <zyimin@linux.vnet.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>

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

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

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

The 9P spec at http://man.cat-v.org/plan_9/5/intro says:

All directories must support walks to the directory .. (dot-dot) meaning
parent directory, although by convention directories contain no explicit
entry for .. or . (dot).  The parent of the root directory of a server's
tree is itself.

This means that a client cannot walk further than the root directory
exported by the server. In other words, if the client wants to walk
"/.." or "/foo/../..", the server should answer like the request was
to walk "/".

This patch just does that:
- we cache the QID of the root directory at attach time
- during the walk we compare the QID of each path component with the root
  QID to detect if we're in a "/.." situation
- if so, we skip the current component and go to the next one
Signed-off-by: NGreg Kurz <groug@kaod.org>
Reviewed-by: NEric Blake <eblake@redhat.com>
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

According to the 9P spec http://man.cat-v.org/plan_9/5/open about the
create request:

The names . and .. are special; it is illegal to create files with these
names.

This patch causes the create and lcreate requests to fail with EINVAL if
the file name is either "." or "..".

Even if it isn't explicitly written in the spec, this patch extends the
checking to all requests that may cause a directory entry to be created:

    - mknod
    - rename
    - renameat
    - mkdir
    - link
    - symlink

The unlinkat request also gets patched for consistency (even if
rmdir("foo/..") is expected to fail according to POSIX.1-2001).

The various error values come from the linux manual pages.
Suggested-by: NPeter Maydell <peter.maydell@linaro.org>
Signed-off-by: NGreg Kurz <groug@kaod.org>
Reviewed-by: NEric Blake <eblake@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Empty path components don't make sense for most commands and may cause
undefined behavior, depending on the backend.

Also, the walk request described in the 9P spec [1] clearly shows that
the client is supposed to send individual path components: the official
linux client never sends portions of path containing the / character for
example.

Moreover, the 9P spec [2] also states that a system can decide to restrict
the set of supported characters used in path components, with an explicit
mention "to remove slashes from name components".

This patch introduces a new name_is_illegal() helper that checks the
names sent by the client are not empty and don't contain unwanted chars.
Since 9pfs is only supported on linux hosts, only the / character is
checked at the moment. When support for other hosts (AKA. win32) is added,
other chars may need to be blacklisted as well.

If a client sends an illegal path component, the request will fail and
ENOENT is returned to the client.

[1] http://man.cat-v.org/plan_9/5/walk
[2] http://man.cat-v.org/plan_9/5/introSuggested-by: NPeter Maydell <peter.maydell@linaro.org>
Signed-off-by: NGreg Kurz <groug@kaod.org>
Reviewed-by: NEric Blake <eblake@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

* pc-bios/optionrom/Makefile fix for -O0
* revert socket_connect change

# gpg: Signature made Tue 30 Aug 2016 15:36:59 BST
# 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:
  optionrom: cope with multiple -O options
  Revert "Change net/socket.c to use socket_*() functions"
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Reproducer:

    CFLAGS="-g3 -O0" ./configure --target-list=aarch64-softmmu,arm-softmmu --enable-vhost-net --enable-virtfs

Here CFLAGS ends up with "-O2 -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 ... -g3 -O0"
and pc-bios/optionrom/Makefile forgets to add the -O2 it needs.
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Since commit 7e844959, the socket connect code is blocking, because
calling socket_connect() without callback is blocking.  This reverts the
commit.
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

tb_flush does all kind of things, which are very tcg specific. As it
is called from some places even for KVM (e.g. gdb server) it is better
to detect these cases and do an early exit.
This also fixes a crash in the gdb server that was triggered by
commit 909eaac9 ("tb hash: track translated blocks with qht").
Suggested-by: NPaolo Bonzini <pbonzini@redhat.com>
Reported-by: NRichard Henderson <rth@twiddle.net>
Reported-by: NBrent Baccala <cosine@freesoft.org>
Signed-off-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Message-id: 1472148686-39841-1-git-send-email-borntraeger@de.ibm.com>
Reviewed-by: NPeter Maydell <peter.maydell@linaro.org>
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

In previous commit

  commit c7628bff
  Author: Gerd Hoffmann <kraxel@redhat.com>
  Date:   Fri Oct 30 12:10:09 2015 +0100

    vnc: only alloc server surface with clients connected

the VNC server was changed so that the 'vd->server' pixman
image was only allocated when a client is connected.

Since then if a client disconnects and then reconnects to
the VNC server all they will see is a black screen until
they do something that triggers a refresh. On a graphical
desktop this is not often noticed since there's many things
going on which cause a refresh. On a plain text console it
is really obvious since nothing refreshes frequently.

The problem is that the VNC server didn't update the guest
dirty bitmap, so still believes its server image is in sync
with the guest contents.

To fix this we must explicitly mark the entire guest desktop
as dirty after re-creating the server surface. Move this
logic into vnc_update_server_surface() so it is guaranteed
to be call in all code paths that re-create the surface
instead of only in vnc_dpy_switch()
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>
Reviewed-by: NPeter Lieven <pl@kamp.de>
Tested-by: NPeter Lieven <pl@kamp.de>
Message-id: 1471365032-18096-1-git-send-email-berrange@redhat.com
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

virtio: fixes

some bugfixes for virtio
balloon is still broken wrt migration
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

# gpg: Signature made Tue 23 Aug 2016 17:33:11 BST
# gpg:                using RSA key 0x281F0DB8D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>"
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* remotes/mst/tags/for_upstream:
  virtio: decrement vq->inuse in virtqueue_discard()
  virtio: recalculate vq->inuse after migration
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Must include "qemu-version.h" for the QEMU_PKGVERSION definition.
Signed-off-by: NEd Maste <emaste@freebsd.org>
Message-id: 1471877833-52343-1-git-send-email-emaste@freebsd.org
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

virtqueue_discard() moves vq->last_avail_idx back so the element can be
popped again.  It's necessary to decrement vq->inuse to avoid "leaking"
the element count.

Cc: qemu-stable@nongnu.org
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

The vq->inuse field is not migrated.  Many devices don't hold
VirtQueueElements across migration so it doesn't matter that vq->inuse
starts at 0 on the destination QEMU.

At least virtio-serial, virtio-blk, and virtio-balloon migrate while
holding VirtQueueElements.  For these devices we need to recalculate
vq->inuse upon load so the value is correct.

Cc: qemu-stable@nongnu.org
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

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

# gpg: Signature made Mon 22 Aug 2016 09:06:32 BST
# gpg:                using RSA key 0xEF04965B398D6211
# 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:
  e1000e: remove internal interrupt flag
  slirp: fix segv when init failed
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Commit 66bf7d58 removed internal msi state flag E1000E_USE_MSI, E1000E_USE_MSIX
is not necessary too, remove it now. And interrupt flag field intr_state also
can be removed now.

CC: Dmitry Fleytman <dmitry@daynix.com>
CC: Jason Wang <jasowang@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
CC: Marcel Apfelbaum <marcel@redhat.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: NCao jin <caoj.fnst@cn.fujitsu.com>
Reviewed-by: NMarkus Armbruster <armbru@redhat.com>
Acked-by: NDmitry Fleytman <dmitry@daynix.com>
Reviewed-by: NPaolo Bonzini <pbonzini@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Since commit f6c2e66a, slirp uses an exit notifier to call
slirp_smb_cleanup. However, if init() failed, the notifier isn't added,
and removing it will fail:

==18447== Invalid write of size 8
==18447==    at 0x7EF2B5: notifier_remove (notify.c:32)
==18447==    by 0x48E80C: qemu_remove_exit_notifier (vl.c:2661)
==18447==    by 0x6A2187: net_slirp_cleanup (slirp.c:134)
==18447==    by 0x69419D: qemu_cleanup_net_client (net.c:338)
==18447==    by 0x69445B: qemu_del_net_client (net.c:401)
==18447==    by 0x6A2B81: net_slirp_init (slirp.c:366)
==18447==    by 0x6A4241: net_init_slirp (slirp.c:865)
==18447==    by 0x695C6D: net_client_init1 (net.c:1051)
==18447==    by 0x695F6E: net_client_init (net.c:1108)
==18447==    by 0x696DBA: net_init_netdev (net.c:1498)
==18447==    by 0x7F1F99: qemu_opts_foreach (qemu-option.c:1116)
==18447==    by 0x696E60: net_init_clients (net.c:1516)
==18447==  Address 0x0 is not stack'd, malloc'd or (recently) free'd
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Since f6880b7f [qemu-log: support simple pid substitution for logs],
test-logging creates files with hard-coded names in /tmp. In the best
case, this prevents multiple developers from running "make check" on
the same machine. In the worst case, it allows for symlink attacks,
enabling an attacker to overwrite files that are writable to the
developer running "make check".

Instead of hard-coding the paths, create a temporary directory using
g_dir_make_tmp() and clean it up afterwards.

Fixes: f6880b7f ("qemu-log: support simple pid substitution for logs")
Signed-off-by: NSascha Silbe <silbe@linux.vnet.ibm.com>
Message-id: 1471545963-11720-3-git-send-email-silbe@linux.vnet.ibm.com
Reviewed-by: NPeter Maydell <peter.maydell@linaro.org>
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

We're going to make use of g_dir_make_tmp() in test-logging. Provide a
compatibility implementation of it for glib < 2.30.

May behave differently in some edge cases (e.g. pattern only at the
end of the template, the file name is not part of the error message),
but good enough in practice.
Signed-off-by: NSascha Silbe <silbe@linux.vnet.ibm.com>
Message-id: 1471545963-11720-2-git-send-email-silbe@linux.vnet.ibm.com
[PMM: removed variable "template" which caused compilation failures
 when C++ files include glib-compat.h]
Reviewed-by: NPeter Maydell <peter.maydell@linaro.org>
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

In 9c371467 I've tried to fix a broken build with older
linux-headers. However, I didn't do it properly. The solution
implemented here is to grab the enums that caused the problem
initially, and rename their values so that they are "QEMU_"
prefixed. In order to guarantee matching values with actual
enums from linux-headers, the enums are seeded with starting
values from the original enums.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NLaurent Vivier <laurent@vivier.eu>
Message-id: 75c14d6e8a97c4ff3931d69c13eab7376968d8b4.1471593869.git.mprivozn@redhat.com
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

The fix I've made there was wrong. I mean, basically what I did
there was equivalent to:

  #if 0
  some code;
  #endif

This reverts commit 9c371467.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NLaurent Vivier <laurent@vivier.eu>
Message-id: 40d61349e445c1ad5fef795da704bf7ed6e19c86.1471593869.git.mprivozn@redhat.com
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

# gpg: Signature made Thu 18 Aug 2016 14:39:31 BST
# 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:
  block: fix possible reorder of flush operations
  block: fix deadlock in bdrv_co_flush
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

This patch reduce CPU usage of flush operations a bit. When we have one
flush completed we should kick only next operation. We should not start
all pending operations in the hope that they will go back to wait on
wait_queue.

Also there is a technical possibility that requests will get reordered
with the previous approach. After wakeup all requests are removed from
the wait queue. They become active and they are processed one-by-one
adding to the wait queue in the same order. Though new flush can arrive
while all requests are not put into the queue.
Signed-off-by: NDenis V. Lunev <den@openvz.org>
Tested-by: NEvgeny Yakovlev <eyakovlev@virtuozzo.com>
Signed-off-by: NEvgeny Yakovlev <eyakovlev@virtuozzo.com>
Message-id: 1471457214-3994-3-git-send-email-den@openvz.org
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Fam Zheng <famz@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>

The following commit
    commit 3ff2f67a
    Author: Evgeny Yakovlev <eyakovlev@virtuozzo.com>
    Date:   Mon Jul 18 22:39:52 2016 +0300
    block: ignore flush requests when storage is clean
has introduced a regression.

There is a problem that it is still possible for 2 requests to execute
in non sequential fashion and sometimes this results in a deadlock
when bdrv_drain_one/all are called for BDS with such stalled requests.

1. Current flushed_gen and flush_started_gen is 1.
2. Request 1 enters bdrv_co_flush to with write_gen 1 (i.e. the same
   as flushed_gen). It gets past flushed_gen != flush_started_gen and
   sets flush_started_gen to 1 (again, the same it was before).
3. Request 1 yields somewhere before exiting bdrv_co_flush
4. Request 2 enters bdrv_co_flush with write_gen 2. It gets past
   flushed_gen != flush_started_gen and sets flush_started_gen to 2.
5. Request 2 runs to completion and sets flushed_gen to 2
6. Request 1 is resumed, runs to completion and sets flushed_gen to 1.
   However flush_started_gen is now 2.

From here on out flushed_gen is always != to flush_started_gen and all
further requests will wait on flush_queue. This change replaces
flush_started_gen with an explicitly tracked active flush request.
Signed-off-by: NEvgeny Yakovlev <eyakovlev@virtuozzo.com>
Signed-off-by: NDenis V. Lunev <den@openvz.org>
Message-id: 1471457214-3994-2-git-send-email-den@openvz.org
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Fam Zheng <famz@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>