The flags field is declared with default update rule 'Preserve',
this patch extends aml_field() to support UpdateRule so that we
can specify different values per field.
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Zhu Guihua <zhugh.fnst@cn.fujitsu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch adds unplug cb for memory device. It resets memory status
"is_enabled" in acpi_memory_unplug_cb(), removes the corresponding
memory region, unregisters vmstate, and unparents the object.
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Signed-off-by: Zhu Guihua <zhugh.fnst@cn.fujitsu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch adds unplug request cb for memory device, and adds the
is_removing boolean field to MemStatus. This field is used to indicate
whether the memory device in slot has been requested to be ejected.
This field is set to true in acpi_memory_unplug_request_cb().
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Signed-off-by: Zhu Guihua <zhugh.fnst@cn.fujitsu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Add a new API named acpi_memory_slot_status() to obtain a single memory
slot status. Doing this is because this procedure will be used by other
functions in the next coming patches.
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Signed-off-by: Zhu Guihua <zhugh.fnst@cn.fujitsu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Add specification about how to use memory hot unplug, and add
a flow diagram to explain memory hot unplug process.
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Zhu Guihua <zhugh.fnst@cn.fujitsu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch lets msix_init_exclusive_bar() can calculate the bar and
pba size based on the number of MSI-X vectors other than using a
hard-coded limit 4096. This is needed to allow device to have more
than 128 MSI_X vectors. To keep migration compatibility, keep using
4096 as bar size and 2048 for pba offset.
Notes: We don't care about the case that using vectors > 128 for
legacy machine type. Since we limit the queue max to 64, so vectors >=
65 is meaningless.
Virtio device will be the first user for this.
Cc: Keith Busch <keith.busch@intel.com>
Cc: Kevin Wolf <kwolf@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch tries to speed up the MSI-X masking and unmasking through
the mapping between vector and queues. With this patch it will there's
no need to go through all possible virtqueues, which may help to
reduce the time spent when doing MSI-X masking/unmasking a single
vector when more than hundreds or even thousands of virtqueues were
supported.
Tested with 80 queue pairs virito-net-pci by changing the smp affinity
in the background and doing netperf in the same time:
Before the patch:
5711.70 Gbits/sec
After the patch:
6830.98 Gbits/sec
About 19.6% improvements in throughput.
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently we will try to traverse all virtqueues to find a subset that
using a specific vector. This is sub optimal when we will support
hundreds or even thousands of virtqueues. So this patch introduces a
method which could be used by transport to get all virtqueues that
using a same vector. This is done through QLISTs and the number of
QLISTs was queried through a transport specific method. When guest
setting vectors, the virtqueue will be linked and helpers for traverse
the list was also introduced.
The first user will be virtio pci which will use this to speed up
MSI-X masking and unmasking handling.
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
It's a bad idea to need to use vector 0 for invalid virtqueue. So this patch
changes to using VIRTIO_NO_VECTOR instead.
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
CC: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
qemu_find_net_clients_except() may return a value which is greater
than the size of array we provided. So we should check this value
before using it, otherwise this may cause unexpected memory access.
This patch fixes the net related command completion when we have a
virtio-net nic with more than 255 queues.
Cc: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch replace the magic number 255, and increase it to
MAX_QUEUE_NUM which is maximum number of queues supported by a nic.
Cc: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The following patches will limit the following things to legacy
machine type:
- maximum number of virtqueues for virtio-pci were limited to 64
Cc: Alexander Graf <agraf@suse.de>
Cc: qemu-ppc@nongnu.org
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Alexander Graf <agraf@suse.de>
This patches adds machine type specific instance initialization
functions. Those functions will be used by following patches to compat
class properties for legacy machine types.
Cc: Alexander Graf <agraf@suse.de>
Cc: qemu-ppc@nongnu.org
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The following patches will limit the following things to legacy
machine type:
- maximum number of virtqueues for virtio-pci were limited to 64
- auto msix bar size for virtio-net-pci were disabled by default
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Virtqueue were indexed from zero, so don't delete virtqueue whose
index is n->max_queues * 2 + 1.
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: qemu-stable <qemu-stable@nongnu.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Ensure that the vhost-user slave knows when the vrings are valid and
when they are invalid, for example during a guest reboot.
The vhost-user protocol says this of VHOST_RESET_OWNER:
Issued when a new connection is about to be closed. The Master
will no longer own this connection (and will usually close it).
Send this message to tell the vhost-user slave that the vhost session
has ended and that session state (e.g. vrings) is no longer valid.
Signed-off-by: Luke Gorrie <luke@snabb.co>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Move generic acpi building helpers into dedictated file and this
can be shared with other machines.
Signed-off-by: Shannon Zhao <zhaoshenglong@huawei.com>
Signed-off-by: Shannon Zhao <shannon.zhao@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The ACPI related header file acpi-defs.h, includes definitions that
apply on other architectures as well. Move it in `include/hw/acpi/`
to sanely include it from other architectures.
Signed-off-by: Alvise Rigo <a.rigo@virtualopensystems.com>
Signed-off-by: Shannon Zhao <zhaoshenglong@huawei.com>
Signed-off-by: Shannon Zhao <shannon.zhao@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Pretending that QMP doesn't understand a command merely because
we are not in the right mode doesn't help first-time users figure
out what to do to correct things. Although the documentation for
QMP calls out capabilities negotiation, we should also make it
clear in our error messages what we were expecting. With this
patch, I now get the following transcript:
$ ./x86_64-softmmu/qemu-system-x86_64 -qmp stdio -nodefaults
{"QMP": {"version": {"qemu": {"micro": 93, "minor": 2, "major": 2}, "package": ""}, "capabilities": []}}
{"execute":"huh"}
{"error": {"class": "CommandNotFound", "desc": "The command huh has not been found"}}
{"execute":"quit"}
{"error": {"class": "CommandNotFound", "desc": "Expecting capabilities negotiation with 'qmp_capabilities' before command 'quit'"}}
{"execute":"qmp_capabilities"}
{"return": {}}
{"execute":"qmp_capabilities"}
{"error": {"class": "CommandNotFound", "desc": "Capabilities negotiation is already complete, command 'qmp_capabilities' ignored"}}
{"execute":"quit"}
{"return": {}}
{"timestamp": {"seconds": 1429110729, "microseconds": 181935}, "event": "SHUTDOWN"}
Signed-off-by: Eric Blake <eblake@redhat.com>
Tested-By: Kashyap Chamarthy <kchamart@redhat.com>
Reviewed-by: Paulo Vital <paulo.vital@profitbricks.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
As far as the QMP parser is concerned, neither the 'O' nor the 'q' format specifiers
put any constraint on the command. However, there are two differences:
1) from a documentation point of view 'O' says that this command takes
a dictionary. The dictionary will be converted to QemuOpts in the
handler to match the corresponding HMP command.
2) 'O' sets QMP_ACCEPT_UNKNOWNS, resulting in the command accepting invalid
extra arguments. For example the following is accepted:
{ "execute": "send-key",
"arguments": { "keys": [ { "type": "qcode", "data": "ctrl" },
{ "type": "qcode", "data": "alt" },
{ "type": "qcode", "data": "delete" } ], "foo": "bar" } }
Neither send-key nor migrate-set-capabilities take a QemuOpts-like
dictionary; they take an array of dictionaries. And neither command
really wants to have extra unknown arguments. Thus, the right
specifier to use in this case is 'q'; with this patch the above
command fails with
{"error": {"class": "GenericError", "desc": "Invalid parameter 'foo'"}}
as intended.
Reported-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Defaulting a parameter to True, then having all callers omit or
pass an explicit True for that parameter, is pointless. Looks
like it has been dead since introduction in commit 06d64c6, more
than 4 years ago.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
A VM supports only one balloon device, but due to several changes
in infrastructure the error message got messed up when trying
to add a second device. Fix it.
Before this fix
Command-line:
qemu-qmp: -device virtio-balloon-pci,id=balloon0: Another balloon device already registered
qemu-qmp: -device virtio-balloon-pci,id=balloon0: Adding balloon handler failed
qemu-qmp: -device virtio-balloon-pci,id=balloon0: Device 'virtio-balloon-pci' could not be initialized
HMP:
Another balloon device already registered
Adding balloon handler failed
Device 'virtio-balloon-pci' could not be initialized
QMP:
{ "execute": "device_add", "arguments": { "driver": "virtio-balloon-pci", "id": "balloon0" } }
{
"error": {
"class": "GenericError",
"desc": "Adding balloon handler failed"
}
}
After this fix
Command-line:
qemu-qmp: -device virtio-balloon-pci,id=balloon0: Only one balloon device is supported
qemu-qmp: -device virtio-balloon-pci,id=balloon0: Device 'virtio-balloon-pci' could not be initialized
HMP:
(qemu) device_add virtio-balloon-pci,id=balloon0
Only one balloon device is supported
Device 'virtio-balloon-pci' could not be initialized
(qemu)
QMP:
{ "execute": "device_add",
"arguments": { "driver": "virtio-balloon-pci", "id": "balloon0" } }
{
"error": {
"class": "GenericError",
"desc": "Only one balloon device is supported"
}
}
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
VHOST_SET_LOG_BASE got an incorrect address, causing
migration errors and potentially even memory corruption.
Reported-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Amos Kong <akong@redhat.com>
Message-id: 1429283565-32265-1-git-send-email-mst@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The image field in BlockDeviceInfo should never be null, however
bdrv_block_device_info() is not filling it in.
This makes the 'info block -n -v' command crash QEMU.
The proper solution is probably to move the relevant code from
bdrv_query_info() to bdrv_block_device_info(), but since we're too
close to the release for that this simpler workaround solves the
crash.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Message-id: 1429274688-8115-1-git-send-email-berto@igalia.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
After commit 5312bd8 the bonito_readl() and bonito_writel() have been
accessing incorrect addresses. Consequently QEMU is crashing when trying
to boot Linux kernel on fulong2e machine.
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
The invalidation code introduced in commit 2360b works by inverting most bits
of env->msr to ensure that hreg_store_msr() will forcibly update the CPU env
state to reflect the new msr value post-migration. Unfortunately
hreg_store_msr() is called with alter_hv set to 0 which preserves the MSR_HVB
state from the CPU env which is now the opposite value to what it should be.
Ensure that we don't invalidate the msr MSR_HVB bit during cpu_post_load so
that the correct value is restored. This fixes suspend/resume for PPC64.
Reported-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Alexander Graf <agraf@suse.de>
Message-id: 1429255009-12751-1-git-send-email-mark.cave-ayland@ilande.co.uk
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This document covers the guest-side hardware interface, as
well as the host-side programming API of QEMU's firmware
configuration (fw_cfg) device.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Signed-off-by: Gabriel Somlo <somlo@cmu.edu>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Unfortunately it turns out that libseccomp 2.2 still does not work
correctly on non-x86 architectures; return to the previous configure
setup of insisting on libseccomp 2.1 or better and i386/x86_64 and
disabling seccomp support in all other situations.
This reverts the two commits:
* "seccomp: libseccomp version varying according to arch"
(commit 896848f0d3)
* "seccomp: update libseccomp version and remove arch restriction"
(commit 8e27fc2004)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1428670681-23032-1-git-send-email-peter.maydell@linaro.org
Current QEMU crashes when specifying an illegal model with the
"-net nic,model=xxx" option, e.g.:
$ qemu-system-x86_64 -net nic,model=n/a
qemu-system-x86_64: Unsupported NIC model: n/a
Program received signal SIGSEGV, Segmentation fault.
The gdb backtrace looks like this:
0x0000555555965fe0 in error_get_pretty (err=0x0) at util/error.c:152
152 return err->msg;
(gdb) bt
0 0x0000555555965fe0 in error_get_pretty (err=0x0) at util/error.c:152
1 0x0000555555965ffd in error_report_err (err=0x0) at util/error.c:157
2 0x0000555555809c90 in pci_nic_init_nofail (nd=0x555555e49860 <nd_table>, rootbus=0x5555564409b0,
default_model=0x55555598c37b "e1000", default_devaddr=0x0) at hw/pci/pci.c:1663
3 0x0000555555691e42 in pc_nic_init (isa_bus=0x555556f71900, pci_bus=0x5555564409b0)
at hw/i386/pc.c:1506
4 0x000055555569396b in pc_init1 (machine=0x5555562abbf0, pci_enabled=1, kvmclock_enabled=1)
at hw/i386/pc_piix.c:248
5 0x0000555555693d27 in pc_init_pci (machine=0x5555562abbf0) at hw/i386/pc_piix.c:310
6 0x000055555572ddf5 in main (argc=3, argv=0x7fffffffe018, envp=0x7fffffffe038) at vl.c:4226
The problem is that pci_nic_init_nofail() does not check whether the err
parameter from pci_nic_init has been set up and thus passes a NULL pointer
to error_report_err(). Fix it by correctly checking the err parameter.
Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The type name for the SoC device, unlike those of its sub-devices,
did not follow the QOM naming conventions. While the usage is internal
only, this is exposed through QMP and HMP, so fix it before release.
Cc: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reviewed-by: Alistair Francis <alistair@alistair23.me>
Message-id: 1428676676-23056-1-git-send-email-afaerber@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Commit 0b183fc871:"memory: move mem_path handling to
memory_region_allocate_system_memory" split memory_region_init_ram and
memory_region_init_ram_from_file. Also it moved mem-path handling a step
up from memory_region_init_ram to memory_region_allocate_system_memory.
Therefore for any board that uses memory_region_init_ram directly,
-mem-path is not supported.
Fix this by replacing memory_region_init_ram with
memory_region_allocate_system_memory.
Tested-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Cc: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Signed-off-by: Dirk Mueller <dmueller@suse.com>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Commit 0b183fc871:"memory: move mem_path handling to
memory_region_allocate_system_memory" split memory_region_init_ram and
memory_region_init_ram_from_file. Also it moved mem-path handling a step
up from memory_region_init_ram to memory_region_allocate_system_memory.
Therefore for any board that uses memory_region_init_ram directly,
-mem-path is not supported.
Fix this by replacing memory_region_init_ram with
memory_region_allocate_system_memory.
Cc: Richard Henderson <rth@twiddle.net>
Signed-off-by: Dirk Mueller <dmueller@suse.com>
Acked-by: Richard Henderson <rth@twiddle.net>
Message-id: CAL5wTH64_ykF17cw2T1Axq8P3vCWm=6WbUJ3qJrLF-u+-MmzUw@mail.gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Commit 0b183fc871:"memory: move mem_path handling to
memory_region_allocate_system_memory" split memory_region_init_ram and
memory_region_init_ram_from_file. Also it moved mem-path handling a step
up from memory_region_init_ram to memory_region_allocate_system_memory.
Therefore for any board that uses memory_region_init_ram directly,
-mem-path is not supported.
Fix this by replacing memory_region_init_ram with
memory_region_allocate_system_memory.
Cc: Michael Walle <michael@walle.cc>
Signed-off-by: Dirk Mueller <dmueller@suse.com>
Acked-by: Michael Walle <michael@walle.cc>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Otherwise the guest can abuse that control to cause e.g. PCIe
Unsupported Request responses (by disabling memory and/or I/O decoding
and subsequently causing [CPU side] accesses to the respective address
ranges), which (depending on system configuration) may be fatal to the
host.
This is CVE-2015-2756 / XSA-126.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Message-id: alpine.DEB.2.02.1503311510300.7690@kaball.uk.xensource.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
libxseg has changed license to GPLv3. QEMU includes GPL "v2 only" code
which is not compatible with GPLv3. This means the resulting binaries
may not be redistributable!
Disable Archipelago (libxseg) by default to prevent accidental license
violations. Also warn if linking against libxseg is enabled to remind
the user.
Note that this commit does not constitute any advice about software
licensing. If you have doubts you should consult a lawyer.
Cc: Chrysostomos Nanakos <cnanakos@grnet.gr>
Suggested-by: Kevin Wolf <kwolf@redhat.com>
Reported-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Andreas Färber <afaerber@suse.de>
Message-id: 1428587538-8765-1-git-send-email-stefanha@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Commit 951c6300f7 out-of-lined the 32-bit-host versions of
tcg_gen_{ld,st}_i64, but in the process it inadvertently changed
an #ifdef HOST_WORDS_BIGENDIAN to #ifdef TCG_TARGET_WORDS_BIGENDIAN.
Since the latter doesn't get defined anywhere this meant we always
took the "LE host" codepath, and stored the two halves of the value
in the wrong order on BE hosts. This typically breaks any 64-bit
guest on a 32-bit BE host completely, and will have possibly more
subtle effects even for 32-bit guests.
Switch the ifdef back to HOST_WORDS_BIGENDIAN.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Tested-by: Andreas Färber <afaerber@suse.de>
Message-id: 1428523029-13620-1-git-send-email-peter.maydell@linaro.org
newer libiscsi versions may return zero events from iscsi_which_events.
In this case iscsi_service will return immediately without any progress.
To avoid busy waiting for iscsi_which_events to change we deregister all
read and write handlers in this case and schedule a timer to periodically
check iscsi_which_events for changed events.
Next libiscsi version will introduce async reconnects and zero events
are returned while libiscsi is waiting for a reconnect retry.
Signed-off-by: Peter Lieven <pl@kamp.de>
Message-id: 1428437295-29577-1-git-send-email-pl@kamp.de
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
There are two problems with memory barriers in async.c. The fix is
to use atomic_xchg in order to achieve sequential consistency between
the scheduling of a bottom half and the corresponding execution.
First, if bh->scheduled is already 1 in qemu_bh_schedule, QEMU does
not execute a memory barrier to order any writes needed by the callback
before the read of bh->scheduled. If the other side sees req->state as
THREAD_ACTIVE, the callback is not invoked and you get deadlock.
Second, the memory barrier in aio_bh_poll is too weak. Without this
patch, it is possible that bh->scheduled = 0 is not "published" until
after the callback has returned. Another thread wants to schedule the
bottom half, but it sees bh->scheduled = 1 and does nothing. This causes
a lost wakeup. The memory barrier should have been changed to smp_mb()
in commit 924fe12 (aio: fix qemu_bh_schedule() bh->ctx race condition,
2014-06-03) together with qemu_bh_schedule()'s. Guess who reviewed
that patch?
Both of these involve a store and a load, so they are reproducible on
x86_64 as well. It is however much easier on aarch64, where the
libguestfs test suite triggers the bug fairly easily. Even there the
failure can go away or appear depending on compiler optimization level,
tracing options, or even kernel debugging options.
Paul Leveille however reported how to trigger the problem within 15
minutes on x86_64 as well. His (untested) recipe, reproduced here
for reference, is the following:
1) Qcow2 (or 3) is critical – raw files alone seem to avoid the problem.
2) Use “cache=directsync” rather than the default of
“cache=none” to make it happen easier.
3) Use a server with a write-back RAID controller to allow for rapid
IO rates.
4) Run a random-access load that (mostly) writes chunks to various
files on the virtual block device.
a. I use ‘diskload.exe c:25’, a Microsoft HCT load
generator, on Windows VMs.
b. Iometer can probably be configured to generate a similar load.
5) Run multiple VMs in parallel, against the same storage device,
to shake the failure out sooner.
6) IvyBridge and Haswell processors for certain; not sure about others.
A similar patch survived over 12 hours of testing, where an unpatched
QEMU would fail within 15 minutes.
This bug is, most likely, also the cause of failures in the libguestfs
testsuite on AArch64.
Thanks to Laszlo Ersek for initially reporting this bug, to Stefan
Hajnoczi for suggesting closer examination of qemu_bh_schedule, and to
Paul for providing test input and a prototype patch.
Reported-by: Laszlo Ersek <lersek@redhat.com>
Reported-by: Paul Leveille <Paul.Leveille@stratus.com>
Reported-by: John Snow <jsnow@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1428419779-26062-1-git-send-email-pbonzini@redhat.com
Suggested-by: Paul Leveille <Paul.Leveille@stratus.com>
Suggested-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Commit 0b183fc871:"memory: move mem_path handling to
memory_region_allocate_system_memory" split memory_region_init_ram and
memory_region_init_ram_from_file. Also it moved mem-path handling a step
up from memory_region_init_ram to memory_region_allocate_system_memory.
Therefore for any board that uses memory_region_init_ram directly,
-mem-path is not supported.
Fix this by replacing memory_region_init_ram with
memory_region_allocate_system_memory.
Signed-off-by: Dirk Mueller <dmueller@suse.com>
Message-id: CAL5wTH4UHYKpJF=dLJfFzxpufjY189chnCow47-ySuLf8GLbug@mail.gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
After qemu_iovec_destroy, the QEMUIOVector's size is zeroed and
the zero size ultimately is used to compute virtqueue_push's len
argument. Therefore, reads from virtio-blk devices did not
migrate their results correctly. (Writes were okay).
Save the size in virtio_blk_handle_request, and use it when the request
is completed.
Based on a patch by Wen Congyang.
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Tested-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Message-id: 1427997044-392-1-git-send-email-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
