Load the virtio.c state into vring.c when we start dataplane mode and
vice versa when stopping dataplane mode. This patch makes it possible
to start and stop dataplane any time while the guest is running.
This will eventually allow us to go back to QEMU main loop for
bdrv_drain_all() and live migration. In the meantime, this patch makes
the dataplane lifecycle more robust but should make no visible
difference. It may be useful in the virtio-net dataplane effort.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit 9154b02c53)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Fix following bugs in "fallback implementation of counting semaphores
with mutex+condvar" added in c166cb72f1:
- waiting threads are not restarted properly if more than one threads
are waiting unblock signals in qemu_sem_timedwait()
- possible missing pthread_cond_signal(3) calls when waiting threads
are returned by ETIMEDOUT
- fix an uninitialized variable
The problem is analyzed by and fix is provided by Noriyuki Soda.
Also put additional cleanup suggested by Laszlo Ersek:
- make QemuSemaphore.count unsigned (it won't be negative)
- check a return value of in pthread_cond_wait() in qemu_sem_wait()
Signed-off-by: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Message-id: 1372841894-10634-1-git-send-email-tsutsui@ceres.dti.ne.jp
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
(cherry picked from commit 79761c6681)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
As of bd5c51ee6c, chardevs no longer use
bottom-halves to issue CHR_EVENT_OPENED events. To maintain past
semantics, we instead defer the CHR_EVENT_OPENED events toward the end
of chardev initialization.
For muxes, this isn't good enough, since a range of FEs must be able
to attach to the mux prior to any CHR_EVENT_OPENED being issued, else
each FE will immediately print it's initial output (prompts, banners,
etc.) just prior to us switching to the next FE as part of
initialization.
The is new and confusing behavior for users, as they'll see output for
things like the HMP monitor, even though their the current mux focus
may be a guest serial port with potentially no output.
We fix this by further deferring CHR_EVENT_OPENED events for FEs
associated with muxes until after machine init by flagging mux chardevs
with 'explicit_be_open', which suppresses emission of CHR_EVENT_OPENED
events until we explicitly set the mux as opened later.
Currently, we must defer till after machine init since we potentially
associate FEs with muxes as part of realize (for instance,
serial_isa_realizefn).
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Message-id: 1375207462-8141-1-git-send-email-mdroth@linux.vnet.ibm.com
Cc: qemu-stable@nongnu.org
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
(cherry picked from commit 7b7ab18d0b)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Fix for LP#1187529: Devices on PCI bridge stop working when
live-migrated. Update bridge mappings for all PCI bridge
devices in get_pci_config_device().
Signed-off-by: Don Koch <dkoch@verizon.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit e78e9ae4a9)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
bdrv_flush() can fail, and bdrv_flush_all() should return an error as
well if this happens for a block device. It returns the first error
return now, but still at least tries to flush the remaining devices even
in error cases.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit f0f0fdfeec)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
In two places qemu uses openpty() which is very system-dependent,
and in both places the pty is switched to raw mode as well.
Make a wrapper function which does both steps, and move all the
system-dependent complexity into a separate file, together
with static/local implementations of openpty() and cfmakeraw()
from qemu-char.c.
It is in a separate file, not part of oslib-posix.c, because
openpty() often resides in -lutil which is not linked to
every program qemu builds.
This change removes #including of <pty.h>, <termios.h>
and other rather specific system headers out of qemu-common.h,
which isn't a place for such specific headers really.
This version has been verified to build correctly on Linux,
OpenBSD, FreeBSD and OpenIndiana. On the latter it lets qemu
to be built with gtk gui which were not possible there due to
missing openpty() and cfmakeraw().
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Tested-by: Andreas Färber <andreas.faerber@web.de>
(cherry picked from commit 4efeabbbe8)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
When CHR_EVENT_OPENED was initially added, it was CHR_EVENT_RESET,
and it was issued as a bottom-half:
86e94dea5b
Which we basically used to print out a greeting/prompt for the
monitor.
AFAICT the only reason this was ever done in a BH was because in
some cases we'd modify the chr_write handler for a new chardev
backend *after* the site where we issued the reset (see:
86e94d:qemu_chr_open_stdio())
At some point this event was renamed to CHR_EVENT_OPENED, and we've
maintained the use of this BH ever since.
However, due to 9f939df955, we schedule
the BH via g_idle_add(), which is causing events to sometimes be
delivered after we've already begun processing data from backends,
leading to:
known bugs:
QMP:
session negotation resets with OPENED event, in some cases this
is causing new sessions to get sporadically reset
potential bugs:
hw/usb/redirect.c:
can_read handler checks for dev->parser != NULL, which may be
true if CLOSED BH has not been executed yet. In the past, OPENED
quiesced outstanding CLOSED events prior to us reading client
data. If it's delayed, our check may allow reads to occur even
though we haven't processed the OPENED event yet, and when we
do finally get the OPENED event, our state may get reset.
qtest.c:
can begin session before OPENED event is processed, leading to
a spurious reset of the system and irq_levels
gdbstub.c:
may start a gdb session prior to the machine being paused
To fix these, let's just drop the BH.
Since the initial reasoning for using it still applies to an extent,
work around that by deferring the delivery of CHR_EVENT_OPENED until
after the chardevs have been fully initialized, toward the end of
qmp_chardev_add() (or some cases, qemu_chr_new_from_opts()). This
defers delivery long enough that we can be assured a CharDriverState
is fully initialized before CHR_EVENT_OPENED is sent.
Also, rather than requiring each chardev to do an explicit open, do it
automatically, and allow the small few who don't desire such behavior to
suppress the OPENED-on-init behavior by setting a 'explicit_be_open'
flag.
We additionally add missing OPENED events for stdio backends on w32,
which were previously not being issued, causing us to not recieve the
banner and initial prompts for qmp/hmp.
Reported-by: Stefan Priebe <s.priebe@profihost.ag>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Message-id: 1370636393-21044-1-git-send-email-mdroth@linux.vnet.ibm.com
Cc: qemu-stable@nongnu.org
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
(cherry picked from commit bd5c51ee6c)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
We are currently setting the PCI hole to start at HVM_BELOW_4G_RAM_END,
that is 0xf0000000.
Start the PCI hole at 0xe0000000 instead, that is the same value used by
pc_init1 and qemu-xen-traditional.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
CC: qemu-stable@nongnu.org
(cherry picked from commit 9f24a8030a)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
We switched from qemu_memalign to mmap() but then we don't modify
qemu_vfree() to do a munmap() over free(). Which we cannot do
because qemu_vfree() frees memory allocated by qemu_{mem,block}align.
Introduce a new function that does the munmap(), luckily the size is
available in the RAMBlock.
Reported-by: Amos Kong <akong@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Amos Kong <akong@redhat.com>
Message-id: 1368454796-14989-3-git-send-email-pbonzini@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
This patch adds a small typename cache to ObjectClass. This allows
caching positive casts within each ObjectClass. Benchmarking a
PPC workload provided by Aurelien, this patch eliminates every
single g_hash_table_lookup() happening during the benchmark (which
was about 2 million per-second).
With this patch applied, I get exactly the same performance (within
the margin of error) as with --disable-qom-cast-debug.
N.B. it's safe to cache typenames only from the _assert() macros
because they are always called with string literals.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
This reverts commit 9953f8822c.
While Markus's analysis is entirely correct, there are 1.6 patches
that fix the bug for real and without requiring machine type hacks.
Let's think of the children who will have to read this code, and
avoid a complicated mess of semantics that differ between <1.5,
1.5, and >1.5.
Conflicts:
hw/i386/pc_piix.c
hw/i386/pc_q35.c
include/hw/i386/pc.h
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Anthony Liguori <aliguori@us.ibm.com>
Message-id: 1368189483-7915-1-git-send-email-pbonzini@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Cast debugging can have a substantial cost (20% or more). Instead of adding
special-cased "fast casts" in the hot paths, we can just disable it in
releases. The tracing facilities we just added make it easier to analyze
those problems that cast debugging would reveal.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1368188203-3407-7-git-send-email-pbonzini@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
sys/types.h is taken out from "ifdef __OpenBSD__" guard. It should be
safe for other systems, according to following survey:
http://hacks.owlfolio.org/header-survey/
This fixes build for CONFIG_IOVEC-less systems (mingw).
Signed-off-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This changes the model number of 486 to 8 (DX4) which matches the
feature set presented, and actually has the CPUID instruction.
This adds a compatibility property, to keep model=0 on pc-*-1.4 and older.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
[AF: Add compat_props entry]
Tested-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reuse it in qdev_prop_set_globals().
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
[AF: Renamed from qdev_prop_set_custom_globals()]
Signed-off-by: Andreas Färber <afaerber@suse.de>
The Linux nbd driver recently increased the maximum supported request
size up to 32 MB:
commit 078be02b80359a541928c899c2631f39628f56df
Author: Michal Belczyk <belczyk@bsd.krakow.pl>
Date: Tue Apr 30 15:28:28 2013 -0700
nbd: increase default and max request sizes
Raise the default max request size for nbd to 128KB (from 127KB) to get it
4KB aligned. This patch also allows the max request size to be increased
(via /sys/block/nbd<x>/queue/max_sectors_kb) to 32MB.
QEMU's 1 MB buffers are too small to handle these requests.
This patch allocates data buffers dynamically and allows up to 32 MB per
request.
Reported-by: Nick Thomas <nick@bytemark.co.uk>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This adds the Castagnoli CRC32C algorithm, using the 0x11EDC6F41
polynomial.
This is extracted from the linux kernel cryptographic crc32.c module.
The algorithm is based on:
Castagnoli93: Guy Castagnoli and Stefan Braeuer and Martin Herrman
"Optimization of Cyclic Redundancy-Check Codes with 24
and 32 Parity Bits", IEEE Transactions on Communication,
Volume 41, Number 6, June 1993
Signed-off-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
# By Igor Mammedov (21) and others
# Via Andreas Färber
* afaerber/qom-cpu: (29 commits)
Drop redundant resume_all_vcpus() from main()
cpus: Fix pausing TCG CPUs while in vCPU thread
target-i386: Replace cpuid_*features fields with a feature word array
target-i386: Break CPUID feature definition lines
target-i386/kvm.c: Code formatting changes
target-i386: Group together level, xlevel, xlevel2 fields
pc: Implement QEMUMachine::hot_add_cpu hook
QMP: Add cpu-add command
Add hot_add_cpu hook to QEMUMachine
target-i386: Move APIC to ICC bus
target-i386: Attach ICC bus to CPU on its creation
target-i386: Introduce ICC bus/device/bridge
cpu: Move cpu_write_elfXX_note() functions to CPUState
kvmvapic: Make dependency on sysbus.h explicit
target-i386: Replace MSI_SPACE_SIZE with APIC_SPACE_SIZE
target-i386: Do not allow to set apic-id once CPU is realized
target-i386: Introduce apic-id CPU property
target-i386: Introduce feat2prop() for CPU properties
acpi_piix4: Add infrastructure to send CPU hot-plug GPE to guest
cpu: Add helper cpu_exists(), to check if CPU with specified id exists
...
similiar -> similar
recieve -> receive
transfered -> transferred
preperation -> preparation
Most changes are in comments, one modifies a parameter name in a function
prototype.
The spelling fixes were made using codespell.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Acked-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Hook should be set by machines that implement CPU hot-add
via cpu-add QMP command.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
It allows APIC to be hotplugged.
* map APIC's mmio at board level if it is present
* do not register mmio region for each APIC, since
only one is used/mapped
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
X86CPU should have parent bus so it could provide bus for child APIC.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Provides a hotpluggable bus for APIC and CPU.
* icc-bridge will serve as a parent for icc-bus and provide
mmio mapping services to child icc-devices.
* icc-device will replace SysBusDevice as a parent of APIC
and IOAPIC devices.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Put APIC_SPACE_SIZE in a public header so that it can be
reused elsewhere later.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Wrapper to avoid open-coded loops and to make CPUState iteration
independent of CPUArchState.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
get_arch_id() adds possibility for generic code to get a guest-visible
CPU ID without accessing CPUArchState.
If derived classes don't override it, it will return cpu_index.
Override it on target-i386 in X86CPU to return the APIC ID.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: liguang <lig.fnst@cn.fujitsu.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
It provides updated currently available CPUs count to BIOS on reboot.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Hot-add CPU event will be distributed to acpi_piix4 and rtc_cmos.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Also add a stub for it, to make possible to use it in qom/cpu.c,
which is shared with user emulators.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Stefan Hajnoczi
Izumi Tsutsui
Michael Roth
Don Koch
Kevin Wolf
Markus Armbruster
Michael Tokarev
Brad Smith
Stefano Stabellini
Ed Maste
Paolo Bonzini
KONRAD Frederic
Anthony Liguori
Igor Mitsyanko
Andreas Färber
Jeff Cody
Stefan Weil
Igor Mammedov
Jens Freimann
Michael S. Tsirkin