Expose the ability to control time through the plugin API. Only one
plugin can control time so it has to request control when loaded.
There are probably more corner cases to catch here.
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
[AJB: tweaked user-mode handling, merged QEMU_PLUGIN_API fix]
Message-Id: <20240530220610.1245424-6-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240620152220.2192768-9-alex.bennee@linaro.org>
Move the key functionality of moving time forward into the clock
sub-system itself. This will allow us to plumb in time control into
plugins.
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240620152220.2192768-7-alex.bennee@linaro.org>
This generalises the qtest_clock_warp code to use the AccelOps
handlers for updating its own sense of time. This will make the next
patch which moves the warp code closer to pure code motion.
From: Alex Bennée <alex.bennee@linaro.org>
Acked-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240530220610.1245424-3-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240620152220.2192768-6-alex.bennee@linaro.org>
We are about to remove direct calls to individual accelerators for
this information and will need a central point for plugins to hook
into time changes.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240530220610.1245424-2-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240620152220.2192768-5-alex.bennee@linaro.org>
Ensure register handles are not NULL so that a plugin can assume NULL is
invalid as a register handle.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240229-null-v1-1-e716501d981e@daynix.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240620152220.2192768-4-alex.bennee@linaro.org>
This is an experiment to further reduce the amount we throw into the
exec headers. It might not be as useful as I initially thought because
just under half of the users also need gdbserver_start().
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240620152220.2192768-3-alex.bennee@linaro.org>
Last use of VMSTATE_ARRAY_TEST() was removed in commit 46baa9007f
("migration/i386: Remove old non-softfloat 64bit FP support"), we
can safely get rid of it.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Li Zhijian <lizhijian@fujitsu.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Make sure there will be an event for postcopy recovery, irrelevant of
whether the reconnect will success, or when the failure happens.
The added new case is to fail early in postcopy recovery, in which case it
didn't even reach RECOVER stage on src (and in real life it'll be the same
to dest, but the test case is just slightly more involved due to the dual
socketpair setup).
To do that, rename the postcopy_recovery_test_fail to reflect either stage
to fail, instead of a boolean.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Making sure the postcopy-recover-setup status is present in the postcopy
failure unit test. Note that it only applies to src QEMU not dest.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Enable CPU cluster support on SbsaQemu platform, so that users can
specify a 4-level CPU hierarchy sockets/clusters/cores/threads. And
this topology can be passed to the firmware through /cpus/topology
Device Tree.
Signed-off-by: Xiong Yining <xiongyining1480@phytium.com.cn>
Reviewed-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Reviewed-by: Leif Lindholm <quic_llindhol@quicinc.com>
Message-id: 20240607103825.1295328-2-xiongyining1480@phytium.com.cn
Tested-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This changes the way the ohci emulation handles a Transfer Descriptor
with "Buffer End" set to "Current Buffer Pointer" - 1, specifically
in the case of a zero-length packet.
The OHCI spec 4.3.1.2 Table 4-2 specifies td.cbp to be zero for a
zero-length packet. Peter Maydell tracked down commit 1328fe0c32
(hw: usb: hcd-ohci: check len and frame_number variables) where qemu
started checking this according to the spec.
What this patch does is loosen the qemu ohci implementation to allow a
zero-length packet if td.be (Buffer End) is set to td.cbp - 1, and with a
non-zero td.cbp value.
The spec is unclear whether this is valid or not -- it is not the
clearly documented way to send a zero length TD (which is CBP=BE=0),
but it isn't specifically forbidden. Actual hw seems to be ok with it.
Does any OS rely on this behavior? There have been no reports to
qemu-devel of this problem.
This is attempting to have qemu behave like actual hardware,
but this is just a minor change.
With a tiny OS[1] that boots and executes a test, the issue can be seen:
* OS that sends USB requests to a USB mass storage device
but sends td.cbp = td.be + 1
* qemu 4.2
* qemu HEAD (4e66a0854)
* Actual OHCI controller (hardware)
Command line:
qemu-system-x86_64 -m 20 \
-device pci-ohci,id=ohci \
-drive if=none,format=raw,id=d,file=testmbr.raw \
-device usb-storage,bus=ohci.0,drive=d \
--trace "usb_*" --trace "ohci_*" -D qemu.log
Results are:
qemu 4.2 | qemu HEAD | actual HW
-----------+------------+-----------
works fine | ohci_die() | works fine
Tip: if the flags "-serial pty -serial stdio" are added to the command line
the test will output USB requests like this:
Testing qemu HEAD:
> Free mem 2M ohci port2 conn FS
> setup { 80 6 0 1 0 0 8 0 }
> ED info=80000 { mps=8 en=0 d=0 } tail=c20920
> td0 c20880 nxt=c20960 f2000000 setup cbp=c20900 be=c20907
> td1 c20960 nxt=c20980 f3140000 in cbp=c20908 be=c2090f
> td2 c20980 nxt=c20920 f3080000 out cbp=c20910 be=c2090f ohci20 host err
> usb stopped
And in qemu.log:
usb_ohci_iso_td_bad_cc_overrun ISO_TD start_offset=0x00c20910 > next_offset=0x00c2090f
Testing qemu 4.2:
> Free mem 2M ohci port2 conn FS
> setup { 80 6 0 1 0 0 8 0 }
> ED info=80000 { mps=8 en=0 d=0 } tail=620920
> td0 620880 nxt=620960 f2000000 setup cbp=620900 be=620907 cbp=0 be=620907
> td1 620960 nxt=620980 f3140000 in cbp=620908 be=62090f cbp=0 be=62090f
> td2 620980 nxt=620920 f3080000 out cbp=620910 be=62090f cbp=0 be=62090f
> rx { 12 1 0 2 0 0 0 8 }
> setup { 0 5 1 0 0 0 0 0 } tx {}
> ED info=80000 { mps=8 en=0 d=0 } tail=620880
> td0 620920 nxt=620960 f2000000 setup cbp=620900 be=620907 cbp=0 be=620907
> td1 620960 nxt=620880 f3100000 in cbp=620908 be=620907 cbp=0 be=620907
> setup { 80 6 0 1 0 0 12 0 }
> ED info=80001 { mps=8 en=0 d=1 } tail=620960
> td0 620880 nxt=6209c0 f2000000 setup cbp=620920 be=620927 cbp=0 be=620927
> td1 6209c0 nxt=6209e0 f3140000 in cbp=620928 be=620939 cbp=0 be=620939
> td2 6209e0 nxt=620960 f3080000 out cbp=62093a be=620939 cbp=0 be=620939
> rx { 12 1 0 2 0 0 0 8 f4 46 1 0 0 0 1 2 3 1 }
> setup { 80 6 0 2 0 0 0 1 }
> ED info=80001 { mps=8 en=0 d=1 } tail=620880
> td0 620960 nxt=6209a0 f2000000 setup cbp=620a20 be=620a27 cbp=0 be=620a27
> td1 6209a0 nxt=6209c0 f3140004 in cbp=620a28 be=620b27 cbp=620a48 be=620b27
> td2 6209c0 nxt=620880 f3080000 out cbp=620b28 be=620b27 cbp=0 be=620b27
> rx { 9 2 20 0 1 1 4 c0 0 9 4 0 0 2 8 6 50 0 7 5 81 2 40 0 0 7 5 2 2 40 0 0 }
> setup { 0 9 1 0 0 0 0 0 } tx {}
> ED info=80001 { mps=8 en=0 d=1 } tail=620900
> td0 620880 nxt=620940 f2000000 setup cbp=620a00 be=620a07 cbp=0 be=620a07
> td1 620940 nxt=620900 f3100000 in cbp=620a08 be=620a07 cbp=0 be=620a07
[1] The OS disk image has been emailed to philmd@linaro.org, mjt@tls.msk.ru,
and kraxel@redhat.com:
* testCbpOffBy1.img.xz
* sha256: f87baddcb86de845de12f002c698670a426affb40946025cc32694f9daa3abed
Signed-off-by: David Hubbard <dmamfmgm@gmail.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The Exynos4210 RNG module requires 32-bit (4-byte) accesses to its registers.
According to the User Manual Section 25.3[1], the registers for RNG operations
are 32-bit. This change ensures that the memory region operations for the
RNG module enforce the correct access sizes, preventing invalid memory accesses.
[1] http://www.mediafire.com/view/8ly2fqls3c9c31c/Exynos_4412_SCP_Users_Manual_Ver.0.10.00_Preliminary0.pdf
Reproducer:
cat << EOF | qemu-system-aarch64 -display none \
-machine accel=qtest, -m 512M -machine smdkc210 -qtest stdio
readb 0x10830454
EOF
Suggested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Message-id: 20240618163701.3204975-1-zheyuma97@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Multiple warning messages and corresponding backtraces are observed when Linux
guest is booted on the host with Fujitsu CPUs. One of them is shown as below.
[ 0.032443] ------------[ cut here ]------------
[ 0.032446] uart-pl011 9000000.pl011: ARCH_DMA_MINALIGN smaller than
CTR_EL0.CWG (128 < 256)
[ 0.032454] WARNING: CPU: 0 PID: 1 at arch/arm64/mm/dma-mapping.c:54
arch_setup_dma_ops+0xbc/0xcc
[ 0.032470] Modules linked in:
[ 0.032475] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.14.0-452.el9.aarch64
[ 0.032481] Hardware name: linux,dummy-virt (DT)
[ 0.032484] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[ 0.032490] pc : arch_setup_dma_ops+0xbc/0xcc
[ 0.032496] lr : arch_setup_dma_ops+0xbc/0xcc
[ 0.032501] sp : ffff80008003b860
[ 0.032503] x29: ffff80008003b860 x28: 0000000000000000 x27: ffffaae4b949049c
[ 0.032510] x26: 0000000000000000 x25: 0000000000000000 x24: 0000000000000000
[ 0.032517] x23: 0000000000000100 x22: 0000000000000000 x21: 0000000000000000
[ 0.032523] x20: 0000000100000000 x19: ffff2f06c02ea400 x18: ffffffffffffffff
[ 0.032529] x17: 00000000208a5f76 x16: 000000006589dbcb x15: ffffaae4ba071c89
[ 0.032535] x14: 0000000000000000 x13: ffffaae4ba071c84 x12: 455f525443206e61
[ 0.032541] x11: 68742072656c6c61 x10: 0000000000000029 x9 : ffffaae4b7d21da4
[ 0.032547] x8 : 0000000000000029 x7 : 4c414e494d5f414d x6 : 0000000000000029
[ 0.032553] x5 : 000000000000000f x4 : ffffaae4b9617a00 x3 : 0000000000000001
[ 0.032558] x2 : 0000000000000000 x1 : 0000000000000000 x0 : ffff2f06c029be40
[ 0.032564] Call trace:
[ 0.032566] arch_setup_dma_ops+0xbc/0xcc
[ 0.032572] of_dma_configure_id+0x138/0x300
[ 0.032591] amba_dma_configure+0x34/0xc0
[ 0.032600] really_probe+0x78/0x3dc
[ 0.032614] __driver_probe_device+0x108/0x160
[ 0.032619] driver_probe_device+0x44/0x114
[ 0.032624] __device_attach_driver+0xb8/0x14c
[ 0.032629] bus_for_each_drv+0x88/0xe4
[ 0.032634] __device_attach+0xb0/0x1e0
[ 0.032638] device_initial_probe+0x18/0x20
[ 0.032643] bus_probe_device+0xa8/0xb0
[ 0.032648] device_add+0x4b4/0x6c0
[ 0.032652] amba_device_try_add.part.0+0x48/0x360
[ 0.032657] amba_device_add+0x104/0x144
[ 0.032662] of_amba_device_create.isra.0+0x100/0x1c4
[ 0.032666] of_platform_bus_create+0x294/0x35c
[ 0.032669] of_platform_populate+0x5c/0x150
[ 0.032672] of_platform_default_populate_init+0xd0/0xec
[ 0.032697] do_one_initcall+0x4c/0x2e0
[ 0.032701] do_initcalls+0x100/0x13c
[ 0.032707] kernel_init_freeable+0x1c8/0x21c
[ 0.032712] kernel_init+0x28/0x140
[ 0.032731] ret_from_fork+0x10/0x20
[ 0.032735] ---[ end trace 0000000000000000 ]---
In Linux, a check is applied to every device which is exposed through
device-tree node. The warning message is raised when the device isn't
DMA coherent and the cache line size is larger than ARCH_DMA_MINALIGN
(128 bytes). The cache line is sorted from CTR_EL0[CWG], which corresponds
to 256 bytes on the guest CPUs. The DMA coherent capability is claimed
through 'dma-coherent' in their device-tree nodes or parent nodes.
This happens even when the device doesn't implement or use DMA at all,
for legacy reasons.
Fix the issue by adding 'dma-coherent' property to the device-tree root
node, meaning all devices are capable of DMA coherent by default.
This both suppresses the spurious kernel warnings and also guards
against possible future QEMU bugs where we add a DMA-capable device
and forget to mark it as dma-coherent.
Signed-off-by: Zhenyu Zhang <zhenyzha@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Donald Dutile <ddutile@redhat.com
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Message-id: 20240612020506.307793-1-zhenyzha@redhat.com
[PMM: tweaked commit message]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
For some use-cases, it is helpful to have more than one UART
available to the guest. If the second UART slot is not already used
for a TrustZone Secure-World-only UART, create it as a NonSecure UART
only when the user provides a serial backend (e.g. via a second
-serial command line option).
This avoids problems where existing guest software only expects a
single UART, and gets confused by the second UART in the DTB. The
major example of this is older EDK2 firmware, which will send the
GRUB bootloader output to UART1 and the guest serial output to UART0.
Users who want to use both UARTs with a guest setup including EDK2
are advised to update to EDK2 release edk2-stable202311 or newer.
(The prebuilt EDK2 blobs QEMU upstream provides are new enough.)
The relevant EDK2 changes are the ones described here:
https://bugzilla.tianocore.org/show_bug.cgi?id=4577
Inspired-by: Axel Heider <axel.heider@hensoldt.net>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Tested-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240610162343.2131524-4-peter.maydell@linaro.org
We're going to make the second UART not always a secure-only device.
Rename the constants VIRT_UART and VIRT_SECURE_UART to VIRT_UART0
and VIRT_UART1 accordingly.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240610162343.2131524-3-peter.maydell@linaro.org
If there is more than one UART in the DTB, then there is no guarantee
on which order a guest is supposed to initialise them. The standard
solution to this is "serialN" entries in the "/aliases" node of the
dtb which give the nodename of the UARTs.
At the moment we only have two UARTs in the DTB when one is for
the Secure world and one for the Non-Secure world, so this isn't
really a problem. However if we want to add a second NS UART we'll
need the aliases to ensure guests pick the right one.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240610162343.2131524-2-peter.maydell@linaro.org
This commit modifies the dwc2_hsotg_read() and dwc2_hsotg_write() functions
to handle invalid address access gracefully. Instead of using
g_assert_not_reached(), which causes the program to abort, the functions
now log an error message and return a default value for reads or do
nothing for writes.
This change prevents the program from aborting and provides clear log
messages indicating when an invalid memory address is accessed.
Reproducer:
cat << EOF | qemu-system-aarch64 -display none \
-machine accel=qtest, -m 512M -machine raspi2b -m 1G -nodefaults \
-usb -drive file=null-co://,if=none,format=raw,id=disk0 -device \
usb-storage,port=1,drive=disk0 -qtest stdio
readl 0x3f980dfb
EOF
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Reviewed-by: Paul Zimmerman <pauldzim@gmail.com>
Message-id: 20240618135610.3109175-1-zheyuma97@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This commit updates the a9_gtimer_get_current_cpu() function to handle
cases where QTest is enabled. When QTest is used, it returns 0 instead
of dereferencing the current_cpu, which can be NULL. This prevents the
program from crashing during QTest runs.
Reproducer:
cat << EOF | qemu-system-aarch64 -display \
none -machine accel=qtest, -m 512M -machine npcm750-evb -qtest stdio
writel 0xf03fe20c 0x26d7468c
EOF
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240618144009.3137806-1-zheyuma97@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Add libqmp to the testlibs component.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240604145934.1230583-6-peter.maydell@linaro.org
Update the 'monitor' component:
* qapi/ and monitor/ are now subdirectories
* add job-qmp.c
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240604145934.1230583-5-peter.maydell@linaro.org
host/include/*/host/crypto/ are relatively new headers; add them
to the crypto component.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240604145934.1230583-4-peter.maydell@linaro.org
The 'char' component:
* includes the no-longer-present qemu-char.c, which has been
long since split into the chardev/ backend code
* also includes the hw/char devices
Split it into two components:
* char is the hw/char devices
* chardev is the chardev backends
with regexes matching our current sources.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240604145934.1230583-3-peter.maydell@linaro.org
Since commit 83aa1baa06 we have been running the build for Coverity
Scan as a Gitlab CI job, rather than the old setup where it was run
on a local developer's machine. This is working well, but the
absolute paths of files are different for the Gitlab CI job, which
means that the regexes we use to identify Coverity components no
longer work. With Gitlab CI builds the file paths are of the form
/builds/qemu-project/qemu/accel/kvm/kvm-all.c
rather than the old
/qemu/accel/kvm/kvm-all.c
and our regexes all don't match.
Update all the regexes to start with .*/qemu/ . This will hopefully
avoid the need to change them again in future if the build path
changes again.
This change was made with a search-and-replace of (/qemu)?
to .*/qemu .
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240604145934.1230583-2-peter.maydell@linaro.org
Fix the system bus interrupt line to CPU core assignment.
Fixes: ddcf58e044 ("hw/arm/xilinx_zynq: Support up to two CPU cores")
Signed-off-by: Sebastian Huber <sebastian.huber@embedded-brains.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240610052906.4432-1-sebastian.huber@embedded-brains.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Julien reported that he has seen strange behaviour when running
Xen on QEMU using GICv2. When Xen migrates a guest's vCPU from
one pCPU to another while the vCPU is handling an interrupt, the
guest is unable to properly deactivate interrupts.
Looking at it a little closer, our GICv2 model treats
deactivation of SPI lines as if they were PPI's, i.e banked per
CPU core. The state for active interrupts should only be banked
for PPI lines, not for SPI lines.
Make deactivation of SPI lines unbanked, similar to how we
handle writes to GICD_ICACTIVER.
Reported-by: Julien Grall <julien@xen.org>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Message-id: 20240605143044.2029444-2-edgar.iglesias@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Updated firmware for QEMU CI is already in merge queue so we can move
platform to be future proof.
All supported cpus work fine with 1GHz timer frequency when firmware is
fresh enough.
Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Reviewed-by: Leif Lindholm <quic_llindhol@quicinc.com>
Message-id: 20240531093729.220758-2-marcin.juszkiewicz@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Returning an uint32_t casted to a gint from g_cmp_ids causes the tx queue to
become wrongly sorted when executing g_slist_sort. Fix this by always
returning -1 or 1 from g_cmp_ids based on the ID comparison instead.
Also, if two message IDs are the same, sort them by using their index and
transmit the message at the lowest index first.
Signed-off-by: Shiva sagar Myana <Shivasagar.Myana@amd.com>
Reviewed-by: Francisco Iglesias <francisco.iglesias@amd.com>
Message-id: 20240603051732.3334571-1-Shivasagar.Myana@amd.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Introduce a small helper to wait for a migration event, generalized from
the incoming migration path. Make the helper easier to use by allowing it
to keep waiting until the expected event is received.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Libvirt should always enable it, so it'll be nice qtest also cover that for
all tests on both sides. migrate_incoming_qmp() used to enable it only on
dst, now we enable them on both, as we'll start to sanity check events even
on the src QEMU.
We'll need to leave the one in migrate_incoming_qmp(), because
virtio-net-failover test uses that one only, and it relies on the events to
work.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Most of them are not needed, we can stick with one ifdef inside
postcopy_recover_fail() so as to cover the scm right tricks only.
The tests won't run on windows anyway due to has_uffd always false.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Firstly, the "Paused" state was added in the wrong place before. The state
machine section was describing PostcopyState, rather than MigrationStatus.
Drop the Paused state descriptions.
Then in the postcopy recover session, add more information on the state
machine for MigrationStatus in the lines. Add the new RECOVER_SETUP phase.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
[fix typo s/reconnects/reconnect]
Signed-off-by: Fabiano Rosas <farosas@suse.de>
This patch adds a migration state on src called "postcopy-recover-setup".
The new state will describe the intermediate step starting from when the
src QEMU received a postcopy recovery request, until the migration channels
are properly established, but before the recovery process take place.
The request came from Libvirt where Libvirt currently rely on the migration
state events to detect migration state changes. That works for most of the
migration process but except postcopy recovery failures at the beginning.
Currently postcopy recovery only has two major states:
- postcopy-paused: this is the state that both sides of QEMU will be in
for a long time as long as the migration channel was interrupted.
- postcopy-recover: this is the state where both sides of QEMU handshake
with each other, preparing for a continuation of postcopy which used to
be interrupted.
The issue here is when the recovery port is invalid, the src QEMU will take
the URI/channels, noticing the ports are not valid, and it'll silently keep
in the postcopy-paused state, with no event sent to Libvirt. In this case,
the only thing Libvirt can do is to poll the migration status with a proper
interval, however that's less optimal.
Considering that this is the only case where Libvirt won't get a
notification from QEMU on such events, let's add postcopy-recover-setup
state to mimic what we have with the "setup" state of a newly initialized
migration, describing the phase of connection establishment.
With that, postcopy recovery will have two paths to go now, and either path
will guarantee an event generated. Now the events will look like this
during a recovery process on src QEMU:
- Initially when the recovery is initiated on src, QEMU will go from
"postcopy-paused" -> "postcopy-recover-setup". Old QEMUs don't have
this event.
- Depending on whether the channel re-establishment is succeeded:
- In succeeded case, src QEMU will move from "postcopy-recover-setup"
to "postcopy-recover". Old QEMUs also have this event.
- In failure case, src QEMU will move from "postcopy-recover-setup" to
"postcopy-paused" again. Old QEMUs don't have this event.
This guarantees that Libvirt will always receive a notification for
recovery process properly.
One thing to mention is, such new status is only needed on src QEMU not
both. On dest QEMU, the state machine doesn't change. Hence the events
don't change either. It's done like so because dest QEMU may not have an
explicit point of setup start. E.g., it can happen that when dest QEMUs
doesn't use migrate-recover command to use a new URI/channel, but the old
URI/channels can be reused in recovery, in which case the old ports simply
can work again after the network routes are fixed up.
Add a new helper postcopy_is_paused() detecting whether postcopy is still
paused, taking RECOVER_SETUP into account too. When using it on both
src/dst, a slight change is done altogether to always wait for the
semaphore before checking the status, because for both sides a sem_post()
will be required for a recovery.
Cc: Jiri Denemark <jdenemar@redhat.com>
Cc: Prasad Pandit <ppandit@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Buglink: https://issues.redhat.com/browse/RHEL-38485
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Destination QEMU can setup incoming ports for two purposes: either a fresh
new incoming migration, in which QEMU will switch to SETUP for channel
establishment, or a paused postcopy migration, in which QEMU will stay in
POSTCOPY_PAUSED until kicking off the RECOVER phase.
Now the state machine worked on dest node for the latter, only because
migrate_set_state() implicitly will become a noop if the current state
check failed. It wasn't clear at all.
Clean it up by providing a helper migration_incoming_state_setup() doing
proper checks over current status. Postcopy-paused will be explicitly
checked now, and then we can bail out for unknown states.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
QEMU uses "int" in most cases even if it stores MigrationStatus. I don't
know why, so let's try to do that right and see what blows up..
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
The postcopy thread names on dest QEMU are slightly confusing, partly I'll
need to blame myself on 36f62f11e4 ("migration: Postcopy preemption
preparation on channel creation"). E.g., "fault-fast" reads like a fast
version of "fault-default", but it's actually the fast version of
"postcopy/listen".
Taking this chance, rename all the migration threads with proper rules.
Considering we only have 15 chars usable, prefix all threads with "mig/",
meanwhile identify src/dst threads properly this time. So now most thread
names will look like "mig/DIR/xxx", where DIR will be "src"/"dst", except
the bg-snapshot thread which doesn't have a direction.
For multifd threads, making them "mig/{src|dst}/{send|recv}_%d".
We used to have "live_migration" thread for a very long time, now it's
called "mig/src/main". We may hope to have "mig/dst/main" soon but not
yet.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Zhijian Li (Fujitsu) <lizhijian@fujitsu.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
We always do the flush when finishing one round of scan, and during
complete() phase we should scan one more round making sure no dirty page
existed. In that case we shouldn't need one explicit FLUSH at the end of
complete(), as when reaching there all pages should have been flushed.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Tested-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Add a multifd test for mapped-ram with passing of fds into QEMU. This
is how libvirt will consume the feature.
There are a couple of details to the fdset mechanism:
- multifd needs two distinct file descriptors (not duplicated with
dup()) so it can enable O_DIRECT only on the channels that do
aligned IO. The dup() system call creates file descriptors that
share status flags, of which O_DIRECT is one.
- the open() access mode flags used for the fds passed into QEMU need
to match the flags QEMU uses to open the file. Currently O_WRONLY
for src and O_RDONLY for dst.
Note that fdset code goes under _WIN32 because fd passing is not
supported on Windows.
Reviewed-by: Peter Xu <peterx@redhat.com>
[brought back the qmp_remove_fd() call at the end of the tests]
Signed-off-by: Fabiano Rosas <farosas@suse.de>
With the last few changes to the fdset infrastructure, we now allow
multifd to use an fdset when migrating to a file. This is useful for
the scenario where the management layer wants to have control over the
migration file.
By receiving the file descriptors directly, QEMU can delegate some
high level operating system operations to the management layer (such
as mandatory access control). The management layer might also want to
add its own headers before the migration stream.
Document the "file:/dev/fdset/#" syntax for the multifd migration with
mapped-ram. The requirements for the fdset mechanism are:
- the fdset must contain two fds that are not duplicates between
themselves;
- if direct-io is to be used, exactly one of the fds must have the
O_DIRECT flag set;
- the file must be opened with WRONLY on the migration source side;
- the file must be opened with RDONLY on the migration destination
side.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
We're about to enable the use of O_DIRECT in the migration code and
due to the alignment restrictions imposed by filesystems we need to
make sure the flag is only used when doing aligned IO.
The migration will do parallel IO to different regions of a file, so
we need to use more than one file descriptor. Those cannot be obtained
by duplicating (dup()) since duplicated file descriptors share the
file status flags, including O_DIRECT. If one migration channel does
unaligned IO while another sets O_DIRECT to do aligned IO, the
filesystem would fail the unaligned operation.
The add-fd QMP command along with the fdset code are specifically
designed to allow the user to pass a set of file descriptors with
different access flags into QEMU to be later fetched by code that
needs to alternate between those flags when doing IO.
Extend the fdset matching to behave the same with the O_DIRECT flag.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
The tests are only allowed to run in systems that know about the
O_DIRECT flag and in filesystems which support it.
Note: this also brings back migrate_set_parameter_bool() which went
away when we removed the compression tests. I copied it verbatim.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
When multifd is used along with mapped-ram, we can take benefit of a
filesystem that supports the O_DIRECT flag and perform direct I/O in
the multifd threads. This brings a significant performance improvement
because direct-io writes bypass the page cache which would otherwise
be thrashed by the multifd data which is unlikely to be needed again
in a short period of time.
To be able to use a multifd channel opened with O_DIRECT, we must
ensure that a certain aligment is used. Filesystems usually require a
block-size alignment for direct I/O. The way to achieve this is by
enabling the mapped-ram feature, which already aligns its I/O properly
(see MAPPED_RAM_FILE_OFFSET_ALIGNMENT at ram.c).
By setting O_DIRECT on the multifd channels, all writes to the same
file descriptor need to be aligned as well, even the ones that come
from outside multifd, such as the QEMUFile I/O from the main migration
code. This makes it impossible to use the same file descriptor for the
QEMUFile and for the multifd channels. The various flags and metadata
written by the main migration code will always be unaligned by virtue
of their small size. To workaround this issue, we'll require a second
file descriptor to be used exclusively for direct I/O.
The second file descriptor can be obtained by QEMU by re-opening the
migration file (already possible), or by being provided by the user or
management application (support to be added in future patches).
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Add the direct-io migration parameter that tells the migration code to
use O_DIRECT when opening the migration stream file whenever possible.
This is currently only used with the mapped-ram migration that has a
clear window guaranteed to perform aligned writes.
Acked-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
We want to make use of the Error object to report fdset errors from
qemu_open_internal() and passing the error pointer to qemu_open_old()
would require changing all callers. Move the file channel to the new
API instead.
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
I'm keeping the EACCES because callers expect to be able to look at
errno.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Remove fds right away instead of setting the ->removed flag. We don't
need the extra complexity of having a cleanup function reap the
removed entries at a later time.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
monitor_fdsets_cleanup() currently has three responsibilities:
1- Remove the fds that have been marked for removal(->removed=true) by
qmp_remove_fd(). This is overly complicated, but ok.
2- Remove any file descriptors that have been passed into QEMU and
never duplicated[1,2]. A file descriptor without duplicates
indicates that no part of QEMU has made use of it. This is
problematic because the current implementation does it only if the
guest is not running and the monitor is closed.
3- Remove/free fdsets that have become empty due to the above
removals. This is ok.
The scenario described in (2) is starting to show some cracks now that
we're trying to consume fds from the migration code:
- Doing cleanup every time the last monitor connection closes works to
reap unused fds, but also has the side effect of forcing the
management layer to pass the file descriptors again in case of a
disconnect/re-connect, if that happened to be the only monitor
connection.
Another side effect is that removing an fd with qmp_remove_fd() is
effectively delayed until the last monitor connection closes.
The usage of mon_refcount is also problematic because it's racy.
- Checking runstate_is_running() skips the cleanup unless the VM is
running and avoids premature cleanup of the fds, but also has the
side effect of blocking the legitimate removal of an fd via
qmp_remove_fd() if the VM happens to be in another state.
This affects qmp_remove_fd() and qmp_query_fdsets() in particular
because requesting a removal at a bad time (guest stopped) might
cause an fd to never be removed, or to be removed at a much later
point in time, causing the query command to continue showing the
supposedly removed fd/fdset.
Note that file descriptors that *have* been duplicated are owned by
the code that uses them and will be removed after qemu_close() is
called. Therefore we've decided that the best course of action to
avoid the undesired side-effects is to stop managing non-duplicated
file descriptors.
1- efb87c1697 ("monitor: Clean up fd sets on monitor disconnect")
2- ebe52b592d ("monitor: Prevent removing fd from set during init")
Reviewed-by: Peter Xu <peterx@redhat.com>
[fix logic mistake: s/fdset_free/fdset_free_if_empty]
Signed-off-by: Fabiano Rosas <farosas@suse.de>
