We've got two versions of the QEMU logo in the repository, one with
the whole word "QEMU" (pc-bios/qemu_logo.svg) and one that only contains
the letter "Q" (ui/icons/qemu.svg). While qemu_logo.svg contains the
proper metadata with license and author information, this is missing
from the ui/icons/qemu.svg file. Copy the meta data there so that
people have a chance to know the license of the file if they only
look at the qemu.svg file.
Closes: https://gitlab.com/qemu-project/qemu/-/issues/3139
Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20250930071419.117592-1-thuth@redhat.com>
(cherry picked from commit 9163424c50)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
If the guest feeds invalid data to the UHCI controller, we
can assert:
qemu-system-x86_64: ../../hw/usb/core.c:744: usb_ep_get: Assertion `pid == USB_TOKEN_IN || pid == USB_TOKEN_OUT' failed.
(see issue 2548 for the repro case). This happens because the guest
attempts USB_TOKEN_SETUP to an endpoint other than 0, which is not
valid. The controller code doesn't catch this guest error, so
instead we hit the assertion in the USB core code.
Catch the case of SETUP to non-zero endpoint, and treat it as a fatal
error in the TD, in the same way we do for an invalid PID value in
the TD.
This is the UHCI equivalent of the same bug in OHCI that we fixed in
commit 3c3c233677 ("hw/usb/hcd-ohci: Fix#1510, #303: pid not IN or
OUT").
This bug has been tracked as CVE-2024-8354.
Cc: qemu-stable@nongnu.org
Fixes: https://gitlab.com/qemu-project/qemu/-/issues/2548
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
(cherry picked from commit d0af3cd027)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit e13e1195db)
(Mjt: resolve context conflict in tests/tcg/multiarch/Makefile.target)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When we remove dest from orig's links, we lose the link
that we rely on later to reset links. This can lead to
failure to release from spinlock with self-modifying code.
Cc: qemu-stable@nongnu.org
Reported-by: 李威威 <liweiwei@kubuds.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Tested-by: Anton Johansson <anjo@rev.ng>
(cherry picked from commit 03fe665980)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
While running the kvm-unit-tests on Intel platforms with "split lock
disable" feature, every test triggers a kernel warning of
x86/split lock detection: #AC: qemu-system-x86_64/373232 took a split_lock trap at address: 0x1e3
Hack KVM by exiting to QEMU on split lock #AC, we get
KVM: exception 17 exit (error code 0x0)
EAX=00000001 EBX=00000000 ECX=00000014 EDX=0001fb80
ESI=00000000 EDI=000000a8 EBP=00000000 ESP=00006f10
EIP=000001e3 EFL=00010002 [-------] CPL=0 II=0 A20=1 SMM=0 HLT=0
ES =0900 00009000 0000ffff 00009300 DPL=0 DS16 [-WA]
CS =c000 000c0000 0000ffff 00009b00 DPL=0 CS16 [-RA]
SS =0000 00000000 0000ffff 00009300 DPL=0 DS16 [-WA]
DS =c000 000c0000 0000ffff 00009300 DPL=0 DS16 [-WA]
FS =0950 00009500 0000ffff 00009300 DPL=0 DS16 [-WA]
GS =06f2 00006f20 0000ffff 00009300 DPL=0 DS16 [-WA]
LDT=0000 00000000 0000ffff 00008200 DPL=0 LDT
TR =0000 00000000 0000ffff 00008b00 DPL=0 TSS32-busy
GDT= 000c02b4 00000027
IDT= 00000000 000003ff
CR0=00000011 CR2=00000000 CR3=00000000 CR4=00000000
DR0=0000000000000000 DR1=0000000000000000 DR2=0000000000000000 DR3=0000000000000000
DR6=00000000ffff0ff0 DR7=0000000000000400
EFER=0000000000000000
Code=89 16 08 00 65 66 0f 01 16 06 00 66 b8 01 00 00 00 0f 22 c0 <65> 66 ff 2e 00 00 b8 10 00 00 00 8e d0 8e d8 8e c0 8e e0 8e e8 66 b8 08 00 66 ba 10 05 66
And it matches with what disassembled from multiboo_dma.bin:
#objdump -b binary -m i386 -D pc-bios/multiboot_dma.bin
1d1: 08 00 or %al,(%eax)
1d3: 65 66 0f 01 16 lgdtw %gs:(%esi)
1d8: 06 push %es
1d9: 00 66 b8 add %ah,-0x48(%esi)
1dc: 01 00 add %eax,(%eax)
1de: 00 00 add %al,(%eax)
1e0: 0f 22 c0 mov %eax,%cr0
> 1e3: 65 66 ff 2e ljmpw *%gs:(%esi)
1e7: 00 00 add %al,(%eax)
1e9: b8 10 00 00 00 mov $0x10,%eax
1ee: 8e d0 mov %eax,%ss
1f0: 8e d8 mov %eax,%ds
1f2: 8e c0 mov %eax,%es
1f4: 8e e0 mov %eax,%fs
1f6: 8e e8 mov %eax,%gs
1f8: 66 b8 08 00 mov $0x8,%ax
1fc: 66 ba 10 05 mov $0x510,%dx
We can see that the instruction at 0x1e3 is a far jmp through the GDT.
However, the GDT is not 8 byte aligned, the base is 0xc02b4.
Intel processors follow the LOCK semantics to set the accessed flag of the
segment descriptor when loading a segment descriptor. If the the segment
descriptor crosses two cache line, it causes split lock.
Fix it by aligning the GDT on 8 bytes, so that segment descriptor cannot
span two cache lines.
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Link: https://lore.kernel.org/r/20250808035027.2194673-1-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 4c8f69b948)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When no monitor address is given, establish the QMP communication through
a socketpair() (API is also supported on Windows since Python 3.5)
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: 20230111080101.969151-4-marcandre.lureau@redhat.com
[Resolved conflicts, fixed typing error. --js]
Signed-off-by: John Snow <jsnow@redhat.com>
(cherry picked from commit bd4c0ef409)
(Mjt: this fixes quite a few qemu iotests which was probably racing somewhere)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
On macOS, private $TMPDIR's are the default. These $TMPDIR's are
generated from a user's unix UID and UUID [1], which can create a
relatively long path:
/var/folders/d7/rz20f6hd709c1ty8f6_6y_z40000gn/T/
QEMU's avocado tests create a temporary directory prefixed by
"avo_qemu_sock_", and create QMP sockets within _that_ as well.
The QMP socket is unnecessarily long, because a temporary directory
is created for every QEMUMachine object.
/avo_qemu_sock_uh3w_dgc/qemu-37331-10bacf110-monitor.sock
The path limit for unix sockets on macOS is 104: [2]
/*
* [XSI] Definitions for UNIX IPC domain.
*/
struct sockaddr_un {
unsigned char sun_len; /* sockaddr len including null */
sa_family_t sun_family; /* [XSI] AF_UNIX */
char sun_path[104]; /* [XSI] path name (gag) */
};
This results in avocado tests failing on macOS because the QMP unix
socket can't be created, because the path is too long:
ERROR| Failed to establish connection: OSError: AF_UNIX path too long
This change resolves by reducing the size of the socket directory prefix
and the suffix on the QMP and console socket names.
The result is paths like this:
pdel@pdel-mbp:/var/folders/d7/rz20f6hd709c1ty8f6_6y_z40000gn/T
$ tree qemu*
qemu_df4evjeq
qemu_jbxel3gy
qemu_ml9s_gg7
qemu_oc7h7f3u
qemu_oqb1yf97
├── 10a004050.con
└── 10a004050.qmp
[1] https://apple.stackexchange.com/questions/353832/why-is-mac-osx-temp-directory-in-weird-path
[2] /Library/Developer/CommandLineTools/SDKs/MacOSX12.3.sdk/usr/include/sys/un.h
Signed-off-by: Peter Delevoryas <peter@pjd.dev>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230110082930.42129-2-peter@pjd.dev
Signed-off-by: John Snow <jsnow@redhat.com>
(cherry picked from commit f9922937d1)
(Mjt: pick this simple fix for 7.2.x so subsequent changes applies cleanly)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
More recent fedora does not have our minimum python versions
anymore. Stick with the most common fedora version used in
7.2.x in other places, which is 37. This way, python tests
works again.
This is a 7.2-specific change, not aimed for the master branch.
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
USB NICs have a "40:" prefix hardcoded for all MAC addresses when we
return the guest the MAC address if it queries the STRING_ETHADDR USB
string property. This doesn't match what we use for the
OID_802_3_PERMANENT_ADDRESS or OID_802_3_CURRENT_ADDRESS OIDs for
NDIS, or the MAC address we actually use in the QEMU networking code
to send/receive packets for this device, or the NIC info string we
print for users. In all those other places we directly use
s->conf.macaddr.a, which is the full thing the user asks for.
This overrides user-provided configuration and leads to an inconsistent
experience.
I couldn't find any documented reason (comment or git commits) for
this behavior. It seems like everyone is just expecting the MAC
address to be fully passed through to the guest, but it isn't.
This may have been a debugging hack that accidentally made it through
to the accepted patch: it has been in the code since it was originally
added back in 2008.
This is also particularly problematic as the "40:" prefix isn't a
reserved prefix for MAC addresses (IEEE OUI). There are a number of
valid allocations out there which use this prefix, meaning that QEMU
may be causing MAC address conflicts.
Cc: qemu-stable@nongnu.org
Fixes: 6c9f886cea ("Add CDC-Ethernet usb NIC (original patch from Thomas Sailer)"
Signed-off-by: Stéphane Graber <stgraber@stgraber.org>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2951
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[PMM: beef up commit message based on mailing list discussion]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit aaf042299a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The tests are built once for all the targets, so as long as one QEMU
binary is built with CONFIG_LSI_SCSI_PCI=y, this test will
run. However some binaries might not include the device. So check this
again in runtime.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230208194700.11035-3-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit dee66bc969)
(Mjt: pick this trivial focused fix up for 7.2.x so that subsequent change(s) in this area apply cleanly)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In the scenario where vhost-user sets eventfd to -1,
qemu_chr_fe_get_msgfds retrieves fd as -1. When vhost_user_read
receives, it does not perform blocking operations on the descriptor
with fd=-1, so non-blocking operations should not be performed here
either.This is a normal use case. Calling g_unix_set_fd_nonblocking
at this point will cause the test to interrupt.
When vhost_user_write sets the call fd to -1, it sets the number of
fds to 0, so the fds obtained by qemu_chr_fe_get_msgfds will also
be 0.
Signed-off-by: Yuxue Liu <yuxue.liu@jaguarmicro.com>
Message-Id: <20240411073555.1357-1-yuxue.liu@jaguarmicro.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit f72fc16910)
(Mjt: pick this trivial focused change up for 7.2.x so that subsequent change(s) in this area apply cleanly)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
chr_read() is printing an error message and returning with s->data_mutex taken.
This can potentially cause a hang. Reported by Coverity.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20230427125423.103536-1-pbonzini@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit 3ee7f21ed2)
(Mjt: pick this trivial focused change up for 7.2.x so that subsequent fixes in this area applies cleanly)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
block/curl.c uses CURLMOPT_SOCKETFUNCTION to register a socket callback.
According to the documentation, this callback is called not just with
application-created sockets but also with internal curl sockets, - and
for such sockets, user data pointer is not set by the application, so
the result qemu crashing.
Pass BDRVCURLState directly to the callback function as user pointer,
instead of relying on CURLINFO_PRIVATE.
This problem started happening with update of libcurl from 8.9 to 8.10 --
apparently with this change curl started using private handles more.
(CURLINFO_PRIVATE is used in one more place, in curl_multi_check_completion() -
it might need a similar fix too)
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/3081
Cc: qemu-stable@qemu.org
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(cherry picked from commit 606978500c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Use extract64(entry, psn, 1) instead of (entry & (1 << psn)) to avoid
undefined behavior for shifts by 32–63 and to make bit extraction intent explicit.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Signed-off-by: Denis Rastyogin <gerben@altlinux.org>
Message-ID: <20250814104914.13101-1-gerben@altlinux.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 1f82ca7234)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Per the release 6.06 revision history:
5.03 August 21, 2013
• ABS2008 and NAN2008 fields of Table 5.7 “FCSR RegisterField
Descriptions” were optional in release 3 and could be R/W,
but as of release 5 are required, read-only, and preset by
hardware.
The P5600 core implements the release 5, and has the ABS2008
and NAN2008 bits set in CP1_fcr31. Therefore it is able to run
ELF binaries compiled with EF_MIPS_NAN2008, such the CIP United
Debian NaN2008 distribution:
http://repo.oss.cipunited.com/mipsel-nan2008/README.txt
In order to run such compiled binaries, select by default the
P5600 core when the ELF 'MIPS_NAN2008' flag is set.
Reported-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20230327162444.388-1-philmd@linaro.org>
(cherry picked from commit 450cb7ec2c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
A race condition between guest driver actions and QEMU timers can lead
to an assertion failure when the guest switches the e1000e from legacy
interrupt mode to MSI-X. If a legacy interrupt delay timer (TIDV or
RDTR) is active, but the guest enables MSI-X before the timer fires,
the pending interrupt cause can trigger an assert in
e1000e_intmgr_collect_delayed_causes().
This patch removes the assertion and executes the code that clears the
pending legacy causes. This change is safe and introduces no unintended
behavioral side effects, as it only alters a state that previously led
to termination.
- when core->delayed_causes == 0 the function was already a no-op and
remains so.
- when core->delayed_causes != 0 the function would previously
crash due to the assertion failure. The patch now defines a safe
outcome by clearing the cause and returning. Since behavior after
the assertion never existed, this simply corrects the crash.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1863
Suggested-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Message-ID: <20250807110806.409065-1-lvivier@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 8e4649cac9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
If capture-output is requested but one of the channels goes unused (eg.
we attempt to capture stderr but the command never writes to stderr), we
can leak memory.
guest_exec_output_watch() is (from what I understand) unconditionally
called for both streams if output capture is requested. The first call
will always pass the `p->size == p->length` check b/c both values are
0. Then GUEST_EXEC_IO_SIZE bytes will be allocated for the stream.
But when we reap the exited process there's a `gei->err.length > 0`
check to actually free the buffer. Which does not get run if the command
doesn't write to the stream.
Fix by making free() unconditional.
Reviewed-by: Konstantin Kostiuk <kkostiuk@redhat.com>
Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
Signed-off-by: Konstantin Kostiuk <kkostiuk@redhat.com>
(cherry picked from commit d6f67b83b8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In stm32f250_soc_initfn() we mostly use the standard pattern
for child objects of calling object_initialize_child(). However
for s->adc_irqs we call object_new() and then later qdev_realize(),
and we never unref the object on deinit. This causes a leak,
detected by ASAN on the device-introspect-test:
Indirect leak of 10 byte(s) in 1 object(s) allocated from:
#0 0x5b9fc4789de3 in malloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/qemu-system-arm+0x21f1de3) (BuildId: 267a2619a026ed91c78a07b1eb2ef15381538efe)
#1 0x740de3f28b09 in g_malloc (/lib/x86_64-linux-gnu/libglib-2.0.so.0+0x62b09) (BuildId: 1eb6131419edb83b2178b682829a6913cf682d75)
#2 0x740de3f3e4d8 in g_strdup (/lib/x86_64-linux-gnu/libglib-2.0.so.0+0x784d8) (BuildId: 1eb6131419edb83b2178b682829a6913cf682d75)
#3 0x5b9fc70159e1 in g_strdup_inline /usr/include/glib-2.0/glib/gstrfuncs.h:321:10
#4 0x5b9fc70159e1 in object_property_try_add /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../qom/object.c:1276:18
#5 0x5b9fc7015f94 in object_property_add /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../qom/object.c:1294:12
#6 0x5b9fc701b900 in object_add_link_prop /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../qom/object.c:2021:10
#7 0x5b9fc701b3fc in object_property_add_link /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../qom/object.c:2037:12
#8 0x5b9fc4c299fb in qdev_init_gpio_out_named /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../hw/core/gpio.c:90:9
#9 0x5b9fc4c29b26 in qdev_init_gpio_out /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../hw/core/gpio.c:101:5
#10 0x5b9fc4c0f77a in or_irq_init /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../hw/core/or-irq.c:70:5
#11 0x5b9fc70257e1 in object_init_with_type /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../qom/object.c:428:9
#12 0x5b9fc700cd4b in object_initialize_with_type /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../qom/object.c:570:5
#13 0x5b9fc700e66d in object_new_with_type /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../qom/object.c:774:5
#14 0x5b9fc700e750 in object_new /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../qom/object.c:789:12
#15 0x5b9fc68b2162 in stm32f205_soc_initfn /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/arm-asan/../../hw/arm/stm32f205_soc.c:69:26
Switch to using object_initialize_child() like all our
other child objects for this SoC object.
Cc: qemu-stable@nongnu.org
Fixes: b63041c8f6 ("STM32F205: Connect the ADC devices")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20250821154229.2417453-1-peter.maydell@linaro.org
(cherry picked from commit 2e27650bdd)
(Mjt: adjust for 7.2, for before qemu_or_irq rename to OrIRQState)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Newer versions of Perl (5.41.x and up) emit a warning for code in
kernel-doc:
Possible precedence problem between ! and pattern match (m//) at /scripts/kernel-doc line 1597.
This is because the code does:
if (!$param =~ /\w\.\.\.$/) {
In Perl, the ! operator has higher precedence than the =~
pattern-match binding, so the effect of this condition is to first
logically-negate the string $param into a true-or-false value and
then try to pattern match it against the regex, which in this case
will always fail. This is almost certainly not what the author
intended.
In the new Python version of kernel-doc in the Linux kernel,
the equivalent code is written:
if KernRe(r'\w\.\.\.$').search(param):
# For named variable parameters of the form `x...`,
# remove the dots
param = param[:-3]
else:
# Handles unnamed variable parameters
param = "..."
which is a more sensible way of writing the behaviour you would
get if you put in brackets to make the regex match first and
then negate the result.
Take this as the intended behaviour, and update the Perl to match.
For QEMU, this produces no change in output, presumably because we
never used the "unnamed variable parameters" syntax.
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250819115648.2125709-1-peter.maydell@linaro.org
(cherry picked from commit 5ffd387e9e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
virtio_net_pre_load_queues() inspects vdev->guest_features to tell if
VIRTIO_NET_F_RSS or VIRTIO_NET_F_MQ is enabled to infer the required
number of queues. This works for VIRTIO_NET_F_MQ but it doesn't for
VIRTIO_NET_F_RSS because only the lowest 32 bits of vdev->guest_features
is set at the point and VIRTIO_NET_F_RSS uses bit 60 while
VIRTIO_NET_F_MQ uses bit 22.
Instead of inferring the required number of queues from
vdev->guest_features, use the number loaded from the vm state. This
change also has a nice side effect to remove a duplicate peer queue
pair change by circumventing virtio_net_set_multiqueue().
Also update the comment in include/hw/virtio/virtio.h to prevent an
implementation of pre_load_queues() from refering to any fields being
loaded during migration by accident in the future.
Fixes: 8c49756825 ("virtio-net: Add only one queue pair when realizing")
Fixes: 653217a501 ("virtio-net: Add only one queue pair when realizing" in 7.2.x)
Tested-by: Lei Yang <leiyang@redhat.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit adda0ad56b)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Call virtio_net_set_multiqueue() to add queues before loading their
states. Otherwise the loaded queues will not have handlers and elements
in them will not be processed.
Cc: qemu-stable@nongnu.org
Fixes: 8c49756825 ("virtio-net: Add only one queue pair when realizing")
Fixes: 653217a501 ("virtio-net: Add only one queue pair when realizing" in 7.2.x)
Reported-by: Laurent Vivier <lvivier@redhat.com>
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit 9379ea9db3)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Multiqueue usage is not negotiated yet when realizing. If more than
one queue is added and the guest never requests to enable multiqueue,
the extra queues will not be deleted when unrealizing and leak.
Fixes: f9d6dbf0bf ("virtio-net: remove virtio queues if the guest doesn't support multiqueue")
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit 8c49756825)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
For all 32-bit systems and 64-bit Windows systems, "long" is 4 bytes long.
Due to using "long" for a linear address, svm_canonicalization would
set all high bits to 1 when (assuming 48-bit linear address) the segment
base is bigger than 0x7FFF.
This fixes booting guests under TCG when the guest IDT and GDT bases are
above 0x7FFF, thereby resulting in incorrect bases. When an interrupt
arrives, it would trigger a #PF exception; the #PF would trigger again,
resulting in a #DF exception; the #PF would trigger for the third time,
resulting in triple-fault, and eventually causes a shutdown VM-Exit to
the hypervisor right after guest boot.
Cc: qemu-stable@nongnu.org
Signed-off-by: Zero Tang <zero.tangptr@gmail.com>
(cherry picked from commit c12cbaa007)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Commit 1585ab9f1b ("hw/sd/sdcard: Fill SPI response bits in card
code") exposed a bug in the SPI adapter: if no SD card is plugged,
we are returning "there is a card with an error". This is wrong,
we shouldn't return any particular packet response, but the noise
shifted on the MISO line. Return the dummy byte, otherwise we get:
qemu-system-riscv64: ../hw/sd/ssi-sd.c:160: ssi_sd_transfer: Assertion `s->arglen > 0' failed.
Reported-by: Guenter Roeck <linux@roeck-us.net>
Fixes: 775616c3ae ("Partial SD card SPI mode support")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Gustavo Romero <gustavo.romero@linaro.org>
Tested-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20250812140415.70153-2-philmd@linaro.org>
(cherry picked from commit e262646e12)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
On AST1030, reading the FMC_WDT2 register always returns 0xFFFFFFFF.
This issue is due to the aspeed_smc_read function, which checks for the
ASPEED_SMC_FEATURE_WDT_CONTROL feature. Since AST1030 was missing this
feature flag, the read operation fails and returns -1.
To resolve this, add the WDT_CONTROL feature to AST1030's feature set
so that FMC_WDT2 can be correctly accessed by firmware.
Signed-off-by: Jamin Lin <jamin_lin@aspeedtech.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Fixes: 2850df6a81 ("aspeed/smc: Add AST1030 support ")
Link: https://lore.kernel.org/qemu-devel/20250804014633.512737-1-jamin_lin@aspeedtech.com
Signed-off-by: Cédric Le Goater <clg@redhat.com>
(cherry picked from commit 13ed972b4c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The code to handle setting SVE registers via the gdbstub is broken:
* it sets each pair of elements in the zregs[].d[] array in the
wrong order for the most common (little endian) case: the least
significant 64-bit value comes first
* it makes no attempt to handle target_endian()
* it does a simple copy out of the (target endian) gdbstub buffer
into the (host endan) zregs data structure, which is wrong on
big endian hosts
Fix all these problems:
* use ldq_p() to read from the gdbstub buffer
* check target_big_endian() to see if we need to handle the
128-bit values the opposite way around
Cc: qemu-stable@nongnu.org
Signed-off-by: Vacha Bhavsar <vacha.bhavsar@oss.qualcomm.com>
Message-id: 20250722173736.2332529-3-vacha.bhavsar@oss.qualcomm.com
[PMM: adjusted commit message, fixed spacing]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 97b3d732af)
(Mjt: s/target_big_endian/target_words_bigendian/ due to missing
v10.0.0-277-gb939b8e42a "exec: Rename target_words_bigendian() -> target_big_endian()")
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In the code for allowing the gdbstub to set the value of an AArch64
FP/SIMD register, we weren't accounting for target_big_endian()
being true. This meant that for aarch64_be-linux-user we would
set the two halves of the FP register the wrong way around.
The much more common case of a little-endian guest is not affected;
nor are big-endian hosts.
Correct the handling of this case.
Cc: qemu-stable@nongnu.org
Signed-off-by: Vacha Bhavsar <vacha.bhavsar@oss.qualcomm.com>
Message-id: 20250722173736.2332529-2-vacha.bhavsar@oss.qualcomm.com
[PMM: added comment, expanded commit message, fixed missing space]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 35cca0f95f)
(Mjt: s/target_big_endian/target_words_bigendian/ due to missing
v10.0.0-277-gb939b8e42a "exec: Rename target_words_bigendian() -> target_big_endian()")
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
KVM's userspace access interface to the GICD enable and active bits
is via set/clear register pairs which implement the hardware's "write
1s to the clear register to clear the 0 bits, and write 1s to the set
register to set the 1 bits" semantics. We didn't get this right,
because we were writing 0 to the clear register.
Writing 0 to GICD_IC{ENABLE,ACTIVE}R architecturally has no effect on
interrupt status (all writes are simply ignored by KVM) and doesn't
comply with the intention of "first write to the clear-reg to clear
all bits".
Write all 1's to actually clear the enable/active status.
This didn't have any adverse effects on migration because there
we start with a clean VM state; it would be guest-visible when
doing a system reset, but since Linux always cleans up the
register state of the GIC during bootup before it enables it
most users won't have run into a problem here.
Cc: qemu-stable@nongnu.org
Fixes: 367b9f527b ("hw/intc/arm_gicv3_kvm: Implement get/put functions")
Signed-off-by: Zenghui Yu <zenghui.yu@linux.dev>
Message-id: 20250729161650.43758-3-zenghui.yu@linux.dev
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit b10bd4bd17)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
According to the 'MIPS MT Application-Specific Extension' manual:
If the VPE executing the instruction is not a Master VPE,
with the MVP bit of the VPEConf0 register set, the EVP bit
is unchanged by the instruction.
Modify the DVPE/EVPE opcodes to only update the MVPControl.EVP bit
if executed on a master VPE.
Cc: qemu-stable@nongnu.org
Reported-by: Hansni Bu
Buglink: https://bugs.launchpad.net/qemu/+bug/1926277
Fixes: f249412c74 ("mips: Add MT halting and waking of VPEs")
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Message-ID: <20210427133343.159718-1-f4bug@amsat.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit e895095c78)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Table A-5 of the Intel manual incorrectly lists the third operand of
VINSERTx128 as Wqq, but it is actually a 128-bit value. This is
visible when W is a memory operand close to the end of the page.
Fixes the recently-added poly1305_kunit test in linux-next.
(No testcase yet, but I plan to modify test-avx2 to use memory
close to the end of the page. This would work because the test
vectors correctly have the memory operand as xmm2/m128).
Reported-by: Eric Biggers <ebiggers@kernel.org>
Tested-by: Eric Biggers <ebiggers@kernel.org>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: "Jason A. Donenfeld" <Jason@zx2c4.com>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: qemu-stable@nongnu.org
Fixes: 7906847768 ("target/i386: reimplement 0x0f 0x3a, add AVX", 2022-10-18)
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit feea87cd6b)
(Mjt: adjust for 7.2.x due to lack of v8.1.0-2167-ge000687f12
"target/i386: validate VEX.W for AVX instructions")
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In case of multiple chunks, code in qxl_unpack_chunks() takes size of the
wrong (next in the chain) chunk, instead of using current chunk size.
This leads to wrong number of bytes being copied, and to crashes if next
chunk size is larger than the current one.
Based on the code by Gao Yong.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1628
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit b8882becd5)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
vnc_worker_thread_loop() copies z_stream stored in its local VncState to
the persistent VncState, and the copied one is freed with deflateEnd()
later. However, deflateEnd() refuses to operate with a copied z_stream
and returns Z_STREAM_ERROR, leaking the allocated memory.
Avoid copying the zlib state to fix the memory leak.
Fixes: bd023f953e ("vnc: threaded VNC server")
Signed-off-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250603-zlib-v3-1-20b857bd8d05@rsg.ci.i.u-tokyo.ac.jp>
(cherry picked from commit aef22331b5)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Fix an off-by-one error in the definition of AMDVI_IR_PHYS_ADDR_MASK. The
current definition masks off the most significant bit of the Interrupt Table
Root ptr i.e. it only generates a mask with bits [50:6] set. See the AMD I/O
Virtualization Technology (IOMMU) Specification for the Interrupt Table
Root Pointer[51:6] field in the Device Table Entry format.
Cc: qemu-stable@nongnu.org
Fixes: b44159fe00 ("x86_iommu/amd: Add interrupt remap support when VAPIC is not enabled")
Signed-off-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Message-Id: <20250617150427.20585-6-alejandro.j.jimenez@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 123cf4bdd3)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Address various issues with definitions of the MMIO registers e.g. for the
Device Table Address Register, the size mask currently encompasses reserved
bits [11:9], so change it to only extract the bits [8:0] encoding size.
Convert masks to use GENMASK64 for consistency, and make unrelated
definitions independent.
Cc: qemu-stable@nongnu.org
Fixes: d29a09ca68 ("hw/i386: Introduce AMD IOMMU")
Signed-off-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Message-Id: <20250617150427.20585-5-alejandro.j.jimenez@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 108e10ff69)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The DTE validation method verifies that all bits in reserved DTE fields are
unset. Update them according to the latest definition available in AMD I/O
Virtualization Technology (IOMMU) Specification - Section 2.2.2.1 Device
Table Entry Format. Remove the magic numbers and use a macro helper to
generate bitmasks covering the specified ranges for better legibility.
Note that some reserved fields specify that events are generated when they
contain non-zero bits, or checks are skipped under certain configurations.
This change only updates the reserved masks, checks for special conditions
are not yet implemented.
Cc: qemu-stable@nongnu.org
Signed-off-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Message-Id: <20250617150427.20585-4-alejandro.j.jimenez@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit ff3dcb3bf6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The documentation for the -icount option incorrectly describes the behavior
of the sleep suboption. Based on the actual implementation and system
behavior, the effects of sleep=on and sleep=off were inadvertently reversed.
This commit updates the description to reflect their intended functionality.
Cc: qemu-stable@nongnu.org
Fixes: fa647905e6 ("qemu-options.hx: Fix minor issues in icount documentation")
Signed-off-by: Ethan Chen <ethan84@andestech.com>
Message-id: 20250606095728.3672832-1-ethan84@andestech.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit e372214e66)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
With API build_mcfg(), it is not necessary with parameter structure
AcpiMcfgInfo to convert to little endian since it is directly used
with host native endian.
Here remove endian conversion before calling function build_mcfg().
With this patch, bios-tables-test passes to run on big endian host
machine S390.
Fixes: 735143f10d ("hw/loongarch: Add acpi ged support")
Cc: qemu-stable@nongnu.org
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20250604065502.1114098-2-maobibo@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>
(cherry picked from commit 9c55c03c05)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Currently, changing the 'drive' property of e.g. a scsi-hd object will
result in an assertion failure if the aio context of the block node
it's replaced with doesn't match the current aio context:
> bdrv_replace_child_noperm: Assertion `bdrv_get_aio_context(old_bs) ==
> bdrv_get_aio_context(new_bs)' failed.
The problematic scenario is already detected, but a 'return' statement
was missing.
Cc: qemu-stable@nongnu.org
Fixes: d1a58c176a ("qdev: allow setting drive property for realized device")
Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Message-ID: <20250523070211.280498-1-f.ebner@proxmox.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit eef2dd03f9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
As far as the emulated audio devices are concerned the pointer
returned by AUD_open_out() is an opaque handle. This includes
the NULL pointer. In this case, AUD_get_buffer_size_out() should
return a sensible buffer size instead of triggering a segmentation
fault. All other public AUD_*_out() and audio_*_out() functions
handle this case.
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20250515054429.7385-2-vr_qemu@t-online.de>
(cherry picked from commit 5ddd6c8dc8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The vring call fd is set even when the guest does not use MSI-X (e.g., in the
case of virtio PMD), leading to unnecessary CPU overhead for processing
interrupts.
The commit 96a3d98d2c("vhost: don't set vring call if no vector") optimized the
case where MSI-X is enabled but the queue vector is unset. However, there's an
additional case where the guest uses INTx and the INTx_DISABLED bit in the PCI
config is set, meaning that no interrupt notifier will actually be used.
In such cases, the vring call fd should also be cleared to avoid redundant
interrupt handling.
Fixes: 96a3d98d2c("vhost: don't set vring call if no vector")
Reported-by: Zhiyuan Yuan <yuanzhiyuan@chinatelecom.cn>
Signed-off-by: Jidong Xia <xiajd@chinatelecom.cn>
Signed-off-by: Huaitong Han <hanht2@chinatelecom.cn>
Message-Id: <20250522100548.212740-1-hanht2@chinatelecom.cn>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit a9403bfcd9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Currently, if the program encounters an unsupported algorithm, it does not set
the HASH_IRQ bit in the status register and send an interrupt to indicate
command completion. As a result, the FW gets stuck waiting for a completion
signal from the HACE module.
Additionally, in do_hash_operation, if an error occurs within the conditional
statement, the HASH_IRQ bit is not set in the status register. This causes the
firmware to continuously send HASH commands, as it is unaware that the HACE
model has completed processing the command.
To fix this, the HASH_IRQ bit in the status register must always be set to
ensure that the firmware receives an interrupt from the HACE module, preventing
it from getting stuck or repeatedly sending HASH commands.
Signed-off-by: Jamin Lin <jamin_lin@aspeedtech.com>
Fixes: c5475b3 ("hw: Model ASPEED's Hash and Crypto Engine")
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Link: https://lore.kernel.org/qemu-devel/20250515081008.583578-4-jamin_lin@aspeedtech.com
Signed-off-by: Cédric Le Goater <clg@redhat.com>
(cherry picked from commit fb8e59abbe)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
There're 2 QMP commands: query-sgx and query-sgx-capabilities, but
their outputs are very similar and the documentation lacks clear
differentiation.
From the codes, query-sgx is used to gather guest's SGX capabilities
(including SGX related CPUIDs and EPC sections' size, in SGXInfo), and
if guest doesn't have SGX, then QEMU will report the error message.
On the other hand, query-sgx-capabilities is used to gather host's SGX
capabilities (descripted by SGXInfo as well). And if host doesn't
support SGX, then QEMU will also report the error message.
Considering that SGXInfo is already documented and both these 2 commands
have enough error messages (for the exception case in their codes).
Therefore the QAPI documentation for these two commands only needs to
emphasize that one of them applies to the guest and the other to the
host.
Fix their documentation to reflect this difference.
Reported-by: Markus Armbruster <armbru@redhat.com>
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Link: https://lore.kernel.org/r/20250513143131.2008078-3-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 7f2131c35c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The j pseudoinstruction maps to a JAL instruction, which can only handle
a jump to somewhere with a signed 20-bit destination. In case of static
linking and LTO'ing this easily leads to "relocation truncated to fit"
error.
Switch to use tail pseudoinstruction, which is the standard way to
tail-call a function in medium code model (emits AUIPC+JALR).
Signed-off-by: Icenowy Zheng <uwu@icenowy.me>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250417072206.364008-1-uwu@icenowy.me>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Cc: qemu-stable@nongnu.org
(cherry picked from commit 22b448ccc6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
virtio-net expects set_features() will be called when the feature set
used by the guest changes to update the number of virtqueues but it is
not called during reset, which will clear all features, leaving the
queues added for VIRTIO_NET_F_MQ or VIRTIO_NET_F_RSS. Not only these
extra queues are visible to the guest, they will cause segmentation
fault during migration.
Call set_features() during reset to remove those queues for virtio-net
as we call set_status(). It will also prevent similar bugs for
virtio-net and other devices in the future.
Fixes: f9d6dbf0bf ("virtio-net: remove virtio queues if the guest doesn't support multiqueue")
Buglink: https://issues.redhat.com/browse/RHEL-73842
Cc: qemu-stable@nongnu.org
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20250421-reset-v2-1-e4c1ead88ea1@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 0caed25cd1)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This patch fixes two different bugs in v9fs_reclaim_fd():
1. Reduce latency:
This function calls v9fs_co_close() and v9fs_co_closedir() in a loop. Each
one of the calls adds two thread hops (between main thread and a fs driver
background thread). Each thread hop adds latency, which sums up in
function's loop to a significant duration.
Reduce overall latency by open coding what v9fs_co_close() and
v9fs_co_closedir() do, executing those and the loop itself altogether in
only one background thread block, hence reducing the total amount of
thread hops to only two.
2. Fix file descriptor leak:
The existing code called v9fs_co_close() and v9fs_co_closedir() to close
file descriptors. Both functions check right at the beginning if the 9p
request was cancelled:
if (v9fs_request_cancelled(pdu)) {
return -EINTR;
}
So if client sent a 'Tflush' message, v9fs_co_close() / v9fs_co_closedir()
returned without having closed the file descriptor and v9fs_reclaim_fd()
subsequently freed the FID without its file descriptor being closed, hence
leaking those file descriptors.
This 2nd bug is fixed by this patch as well by open coding v9fs_co_close()
and v9fs_co_closedir() inside of v9fs_reclaim_fd() and not performing the
v9fs_request_cancelled(pdu) check there.
Fixes: 7a46274529 ('hw/9pfs: Add file descriptor reclaim support')
Fixes: bccacf6c79 ('hw/9pfs: Implement TFLUSH operation')
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <5747469d3f039c53147e850b456943a1d4b5485c.1741339452.git.qemu_oss@crudebyte.com>
(cherry picked from commit 89f7b4da76)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Even though this function is serialized to be always called from main
thread, v9fs_reclaim_fd() is dispatching the coroutine to a worker thread
in between via its v9fs_co_*() calls, hence leading to the situation where
v9fs_reclaim_fd() is effectively executed multiple times simultaniously,
which renders its LRU algorithm useless and causes high latency.
Fix this by adding a simple boolean variable to ensure this function is
only called once at a time. No synchronization needed for this boolean
variable as this function is only entered and returned on main thread.
Fixes: 7a46274529 ('hw/9pfs: Add file descriptor reclaim support')
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <5c622067efd66dd4ee5eca740dcf263f41db20b2.1741339452.git.qemu_oss@crudebyte.com>
(cherry picked from commit 61da38db70)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
According to the i.MX 8M Plus reference manual, the status flag
I2C_I2SR[IIF] continues to be set when an interrupt condition
occurs even when I2C interrupts are disabled (I2C_I2CR[IIEN] is
clear). However, the device model only sets the flag when I2C
interrupts are enabled which causes U-Boot to loop forever. Fix
the device model by always setting the flag and let I2C_I2CR[IIEN]
guard I2C interrupts only.
Also remove the comment in the code since it merely stated the
obvious and would be outdated now.
Cc: qemu-stable@nongnu.org
Fixes: 20d0f9cf6a ("i.MX: Add I2C controller emulator")
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Message-ID: <20250507124040.425773-1-shentey@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 54e54e594b)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
According to the i.MX 8M Plus reference manual, a GPIO pin is
configured as an output when the corresponding bit in the GDIR
register is set. The function imx_gpio_set_int_line() is intended to
be a no-op if the pin is configured as an output, returning early in
such cases. However, it inverts the condition. Fix this by
returning early when the bit is set.
cc: qemu-stable@nongnu.org
Fixes: f442728097 ("i.MX: Add GPIO device")
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-id: 20250501183445.2389-4-shentey@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit eba837a31b)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
If the guest code has an ISB or SB insn inside an IT block, we
generate incorrect code which trips a TCG assertion:
qemu-system-arm: ../tcg/tcg-op.c:3343: void tcg_gen_goto_tb(unsigned int): Assertion `(tcg_ctx->goto_tb_issue_mask & (1 << idx)) == 0' failed.
This is because we call gen_goto_tb(dc, 1, ...) twice:
brcond_i32 ZF,$0x0,ne,$L1
add_i32 pc,pc,$0x4
goto_tb $0x1
exit_tb $0x73d948001b81
set_label $L1
add_i32 pc,pc,$0x4
goto_tb $0x1
exit_tb $0x73d948001b81
Both calls are in arm_tr_tb_stop(), one for the
DISAS_NEXT/DISAS_TOO_MANY handling, and one for the dc->condjump
condition-failed codepath. The DISAS_NEXT handling doesn't have this
problem because arm_post_translate_insn() does the handling of "emit
the label for the condition-failed conditional execution" and so
arm_tr_tb_stop() doesn't have dc->condjump set. But for
DISAS_TOO_MANY we don't do that.
Fix the bug by making arm_post_translate_insn() handle the
DISAS_TOO_MANY case. This only affects the SB and ISB insns when
used in Thumb mode inside an IT block: only these insns specifically
set is_jmp to TOO_MANY, and their A32 encodings are unconditional.
For the major TOO_MANY case (breaking the TB because it would cross a
page boundary) we do that check and set is_jmp to TOO_MANY only after
the call to arm_post_translate_insn(); so arm_post_translate_insn()
sees is_jmp == DISAS_NEXT, and we emit the correct code for that
situation.
With this fix we generate the somewhat more sensible set of TCG ops:
brcond_i32 ZF,$0x0,ne,$L1
set_label $L1
add_i32 pc,pc,$0x4
goto_tb $0x1
exit_tb $0x7c5434001b81
(NB: the TCG optimizer doesn't optimize out the jump-to-next, but
we can't really avoid emitting it because we don't know at the
point we're emitting the handling for the condexec check whether
this insn is going to happen to be a nop for us or not.)
Cc: qemu-stable@nongnu.org
Fixes: https://gitlab.com/qemu-project/qemu/-/issues/2942
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250501125544.727038-1-peter.maydell@linaro.org
(cherry picked from commit 8ed7c0b648)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The comment about not being able to define a field with
zero bits is out of date since 94597b6146
("decodetree: Allow !function with no input bits").
This fixes the missing load of imm in the disassembler.
Cc: qemu-stable@nongnu.org
Fixes: 9d8caa67a2 ("target/avr: Add support for disassembling via option '-d in_asm'")
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit 6b661b7ed7)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When update_pagemask was split from helper_mtc0_pagemask,
we failed to actually write to the new parameter but continue
to write to env->CP0_PageMask. Thus the use within
page_table_walk_refill modifies cpu state and not the local
variable as expected.
Simplify by renaming to compute_pagemask and returning the
value directly. No need for either env or pointer return.
Fixes: 074cfcb4da ("target/mips: Implement hardware page table walker for MIPS32")
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250328175526.368121-4-richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Cc: qemu-stable@nongnu.org
(cherry picked from commit 256ba7715b)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Revert ee3863b9d4 and a08d60bc6c. The logic behind changing
the system page size because of what the Loongson kernel "prefers"
is flawed.
In the Loongson-2E manual, section 5.5, it is clear that the cpu
supports a 4k page size (along with many others). Similarly for
the Loongson-3 series CPUs, the 4k page size is mentioned in the
section 7.7 (PageMask Register). Therefore we must continue to
support a 4k page size.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250328175526.368121-2-richard.henderson@linaro.org>
[PMD: Mention Loongson-3 series CPUs]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit fca2817fdc)
(Mjt: context fixes for 7.2.x)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The maximum padding size is either 64 or 128 bytes and should always be smaller
than "req_len". If "padding_size" exceeds "req_len", then
"req_len - padding_size" underflows due to "uint32_t" data type, leading to a
large incorrect value (e.g., `0xFFXXXXXX`). This causes an out-of-bounds memory
access, potentially leading to a buffer overflow.
Added a check to ensure "padding_size" does not exceed "req_len" before
computing "pad_offset". This prevents "req_len - padding_size" from underflowing
and avoids accessing invalid memory.
Signed-off-by: Jamin Lin <jamin_lin@aspeedtech.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Fixes: 5cd7d8564a ("aspeed/hace: Support AST2600 HACE ")
Link: https://lore.kernel.org/qemu-devel/20250321092623.2097234-3-jamin_lin@aspeedtech.com
Signed-off-by: Cédric Le Goater <clg@redhat.com>
(cherry picked from commit 78877b2e06)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Add .set_vnet_le() function that always returns success, assuming that
vDPA h/w always implements LE data format. Otherwise, QEMU disables vDPA and
outputs the message:
"backend does not support LE vnet headers; falling back on userspace virtio"
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Konstantin Shkolnyy <kshk@linux.ibm.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit b027f55a99)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The check for fp_excp_el in assert_fp_access_checked is
incorrect. For SME, with StreamingMode enabled, the access
is really against the streaming mode vectors, and access
to the normal fp registers is allowed to be disabled.
C.f. sme_enabled_check.
Convert sve_access_checked to match, even though we don't
currently check the exception state.
Cc: qemu-stable@nongnu.org
Fixes: 3d74825f4d ("target/arm: Add SME enablement checks")
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250307190415.982049-2-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 298a04998f)
(Mjt: minor context fix in target/arm/tcg/translate.h,
target/arm/tcg/translate-a64.c is target/arm/translate-a64.c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
On ARM hosts with CTR_EL0.DIC and CTR_EL0.IDC set, this would only cause
an ISB to be executed during cache maintenance, which could lead to QEMU
executing TBs containing garbage instructions.
This seems to be because the ISB finishes executing instructions and
flushes the pipeline, but the ISB doesn't guarantee that writes from the
executed instructions are committed. If a small enough TB is created, it's
possible that the writes setting up the TB aren't committed by the time the
TB is executed.
This function is intended to be a port of the gcc implementation
(https://github.com/gcc-mirror/gcc/blob/85b46d0795ac76bc192cb8f88b646a647acf98c1/libgcc/config/aarch64/sync-cache.c#L67)
which makes the first DSB unconditional, so we can fix the synchronization
issue by doing that as well.
Cc: qemu-stable@nongnu.org
Fixes: 664a79735e ("util: Specialize flush_idcache_range for aarch64")
Signed-off-by: Joe Komlodi <komlodi@google.com>
Message-id: 20250310203622.1827940-2-komlodi@google.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit e6c38d2ab5)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
These warnings are breaking some build configurations since 2 months
now (https://gitlab.com/qemu-project/qemu/-/issues/2575):
ui/cocoa.m:662:14: error: 'CVDisplayLinkCreateWithCGDisplay' is deprecated: first deprecated in macOS 15.0 - use NSView.displayLink(target:selector:), NSWindow.displayLink(target:selector:), or NSScreen.displayLink(target:selector:) [-Werror,-Wdeprecated-declarations]
662 | if (!CVDisplayLinkCreateWithCGDisplay(display, &displayLink)) {
| ^
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/System/Library/Frameworks/CoreVideo.framework/Headers/CVDisplayLink.h:89:20: note: 'CVDisplayLinkCreateWithCGDisplay' has been explicitly marked deprecated here
89 | CV_EXPORT CVReturn CVDisplayLinkCreateWithCGDisplay(
| ^
ui/cocoa.m:663:29: error: 'CVDisplayLinkGetNominalOutputVideoRefreshPeriod' is deprecated: first deprecated in macOS 15.0 - use NSView.displayLink(target:selector:), NSWindow.displayLink(target:selector:), or NSScreen.displayLink(target:selector:) [-Werror,-Wdeprecated-declarations]
663 | CVTime period = CVDisplayLinkGetNominalOutputVideoRefreshPeriod(displayLink);
| ^
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/System/Library/Frameworks/CoreVideo.framework/Headers/CVDisplayLink.h:182:18: note: 'CVDisplayLinkGetNominalOutputVideoRefreshPeriod' has been explicitly marked deprecated here
182 | CV_EXPORT CVTime CVDisplayLinkGetNominalOutputVideoRefreshPeriod( CVDisplayLinkRef CV_NONNULL displayLink );
| ^
ui/cocoa.m:664:13: error: 'CVDisplayLinkRelease' is deprecated: first deprecated in macOS 15.0 - use NSView.displayLink(target:selector:), NSWindow.displayLink(target:selector:), or NSScreen.displayLink(target:selector:) [-Werror,-Wdeprecated-declarations]
664 | CVDisplayLinkRelease(displayLink);
| ^
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/System/Library/Frameworks/CoreVideo.framework/Headers/CVDisplayLink.h:249:16: note: 'CVDisplayLinkRelease' has been explicitly marked deprecated here
249 | CV_EXPORT void CVDisplayLinkRelease( CV_RELEASES_ARGUMENT CVDisplayLinkRef CV_NULLABLE displayLink );
| ^
3 errors generated.
For the next release, ignore the warnings using #pragma directives.
At least until we figure the correct new API usage.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Phil Dennis-Jordan <phil@philjordan.eu>
Tested-by: Phil Dennis-Jordan <phil@philjordan.eu>
Message-Id: <20241121131954.98949-1-philmd@linaro.org>
(cherry picked from commit 9cf6e41fe2)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Block drivers assume in their .bdrv_open() implementation that their
state in bs->opaque has been zeroed; it is initially allocated with
g_malloc0() in bdrv_open_driver().
bdrv_snapshot_goto() needs to make sure that it is zeroed again before
calling drv->bdrv_open() to avoid that block drivers use stale values.
One symptom of this bug is VMDK running into a double free when the user
tries to apply an internal snapshot like 'qemu-img snapshot -a test
test.vmdk'. This should be a graceful error because VMDK doesn't support
internal snapshots.
==25507== Invalid free() / delete / delete[] / realloc()
==25507== at 0x484B347: realloc (vg_replace_malloc.c:1801)
==25507== by 0x54B592A: g_realloc (gmem.c:171)
==25507== by 0x1B221D: vmdk_add_extent (../block/vmdk.c:570)
==25507== by 0x1B1084: vmdk_open_sparse (../block/vmdk.c:1059)
==25507== by 0x1AF3D8: vmdk_open (../block/vmdk.c:1371)
==25507== by 0x1A2AE0: bdrv_snapshot_goto (../block/snapshot.c:299)
==25507== by 0x205C77: img_snapshot (../qemu-img.c:3500)
==25507== by 0x58FA087: (below main) (libc_start_call_main.h:58)
==25507== Address 0x832f3e0 is 0 bytes inside a block of size 272 free'd
==25507== at 0x4846B83: free (vg_replace_malloc.c:989)
==25507== by 0x54AEAC4: g_free (gmem.c:208)
==25507== by 0x1AF629: vmdk_close (../block/vmdk.c:2889)
==25507== by 0x1A2A9C: bdrv_snapshot_goto (../block/snapshot.c:290)
==25507== by 0x205C77: img_snapshot (../qemu-img.c:3500)
==25507== by 0x58FA087: (below main) (libc_start_call_main.h:58)
This error was discovered by fuzzing qemu-img.
Cc: qemu-stable@nongnu.org
Closes: https://gitlab.com/qemu-project/qemu/-/issues/2853
Closes: https://gitlab.com/qemu-project/qemu/-/issues/2851
Reported-by: Denis Rastyogin <gerben@altlinux.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20250310104858.28221-1-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit b75c5f9879)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
For accesses to the 91c111 data register, the address within the
packet's data frame is determined by a combination of the pointer
register and the offset used to access the data register, so that you
can access data at effectively wider than byte width. The pointer
register's pointer field is 11 bits wide, which is exactly the size
to index a 2048-byte data frame.
We weren't quite getting the logic right for ensuring that we end up
with a pointer value to use in the s->data[][] array that isn't out
of bounds:
* we correctly mask when getting the initial pointer value
* for the "autoincrement the pointer register" case, we
correctly mask after adding 1 so that the pointer register
wraps back around at the 2048 byte mark
* but for the non-autoincrement case where we have to add the
low 2 bits of the data register offset, we don't account
for the possibility that the pointer register is 0x7ff
and the addition should wrap
Fix this bug by factoring out the "get the p value to use as an array
index" into a function, making it use FIELD macro names rather than
hard-coded constants, and having a utility function that does "add a
value and wrap it" that we can use both for the "autoincrement" and
"add the offset bits" codepaths.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2758
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250228191652.1957208-1-peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 700d3d6dd4)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When the smc91c111 transmits a packet, it must read a control byte
which is at the end of the data area and CRC. However, we don't
sanitize the length field in the packet buffer, so if the guest sets
the length field to something large we will try to read past the end
of the packet data buffer when we access the control byte.
As usual, the datasheet says nothing about the behaviour of the
hardware if the guest misprograms it in this way. It says only that
the maximum valid length is 2048 bytes. We choose to log the guest
error and silently drop the packet.
This requires us to factor out the "mark the tx packet as complete"
logic, so we can call it for this "drop packet" case as well as at
the end of the loop when we send a valid packet.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2742
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250228174802.1945417-3-peter.maydell@linaro.org>
[PMD: Update smc91c111_do_tx() as len > MAX_PACKET_SIZE]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit aad6f264ad)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The smc91c111 uses packet numbers as an index into its internal
s->data[][] array. Valid packet numbers are between 0 and 3, but
the code does not generally check this, and there are various
places where the guest can hand us an arbitrary packet number
and cause an out-of-bounds access to the data array.
Add validation of packet numbers. The datasheet is not very
helpful about how guest errors like this should be handled:
it says nothing on the subject, and none of the documented
error conditions are relevant. We choose to log the situation
with LOG_GUEST_ERROR and silently ignore the attempted operation.
In the places where we are about to access the data[][] array
using a packet number and we know the number is valid because
we got it from somewhere that has already validated, we add
an assert() to document that belief.
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250228174802.1945417-2-peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 2fa3a5b946)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The SMC91C111 includes an MMU Command register which permits
the guest to remove entries from the RX FIFO. The datasheet
does not specify what happens if the guest tries to do this
when the FIFO is already empty; there are no status registers
containing error bits which might be applicable.
Currently we don't guard at all against pop of an empty
RX FIFO, with the result that we allow the guest to drive
the rx_fifo_len index to negative values, which will cause
smc91c111_receive() to write to the rx_fifo[] array out of
bounds when we receive the next packet.
Instead ignore attempts to pop an empty RX FIFO.
Cc: qemu-stable@nongnu.org
Fixes: 80337b66a8 ("NIC emulation for qemu arm-softmmu")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2780
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250207151157.3151776-1-peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 937df81af6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The commit to fix the OCC common area sensor mappings didn't update the
register offsets to match.
Before this change, skiboot reports:
[ 0.347100086,3] OCC: Chip 0 sensor data invalid
Afterward, there is no error and the sensor_groups directory appears
under /sys/firmware/opal/.
The SLW_IMAGE_BASE address looks like a workaround to intercept firmware
memory accesses, but that does not seem to be required now (and would
have been broken by the OCC common area region mapping change anyway).
So it can be removed.
Fixes: 3a1b70b66b ("ppc/pnv: Fix OCC common area region mapping")
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
(cherry picked from commit 29c041ca7f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This error was discovered by fuzzing qemu-img.
In the QED block driver, the need_check_timer timer is freed in
bdrv_qed_detach_aio_context, but the pointer to the timer is not
set to NULL. This can lead to a use-after-free scenario
in bdrv_qed_drain_begin().
The need_check_timer pointer is set to NULL after freeing the timer.
Which helps catch this condition when checking in bdrv_qed_drain_begin().
Closes: https://gitlab.com/qemu-project/qemu/-/issues/2852
Signed-off-by: Denis Rastyogin <gerben@altlinux.org>
Message-ID: <20250304083927.37681-1-gerben@altlinux.org>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit 2ad638a3d1)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Instead of migrating the raw tick_offset, goldfish_rtc migrates a
recalculated value based on QEMU_CLOCK_VIRTUAL. As QEMU_CLOCK_VIRTUAL
stands still across a save-and-restore cycle, the guest RTC becomes out
of sync with the host RTC when the VM is restored.
As described in the bug description, it looks like this calculation was
copied from pl031 RTC, which had its tick_offset migration fixed by
Commit 032cfe6a79 ("pl031: Correctly migrate state when using -rtc
clock=host").
Migrate the tick_offset directly, adding it as a version-dependent field
to VMState. Keep the old behavior when migrating from previous versions.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2033
Signed-off-by: Rodrigo Dias Correa <r@drigo.nl>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20250114212150.228241-1-r@drigo.nl>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 3521f9cadc)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In the RISC-V privileged ISA section 3.1.15 table 15, it is determined
that a debug exception that is triggered from a load/store has a higher
priority than a possible fault that this access might trigger.
This is not the case ATM as shown in [1]. Adding a breakpoint in an
address that deliberately will fault is causing a load page fault
instead of a debug exception. The reason is that we're throwing in the
page fault as soon as the fault occurs (end of riscv_cpu_tlb_fill(),
raise_mmu_exception()), not allowing the installed watchpoints to
trigger.
Call cpu_check_watchpoint() in the page fault path to search and execute
any watchpoints that might exist for the address, never returning back
to the fault path. If no watchpoints are found cpu_check_watchpoint()
will return and we'll fall-through the regular path to
raise_mmu_exception().
[1] https://gitlab.com/qemu-project/qemu/-/issues/2627
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2627
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250121170626.1992570-3-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit c86edc5476)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The mcontrol select bit (19) is always zero, meaning our triggers will
always match virtual addresses. In this condition, if the user does not
specify a size for the trigger, the access size defaults to XLEN.
At this moment we're using def_size = 8 regardless of CPU XLEN. Use
def_size = 4 in case we're running 32 bits.
Fixes: 95799e36c1 ("target/riscv: Add initial support for the Sdtrig extension")
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250121170626.1992570-2-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 3fba76e61c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
According to the Vector Reduction Operations section in the RISC-V "V"
Vector Extension spec,
"If vl=0, no operation is performed and the destination register is not
updated."
The vd should be updated when vl is larger than 0.
Fixes: fe5c9ab1fc ("target/riscv: vector single-width integer reduction instructions")
Fixes: f714361ed7 ("target/riscv: rvv-1.0: implement vstart CSR")
Signed-off-by: Max Chou <max.chou@sifive.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20250124101452.2519171-1-max.chou@sifive.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit ffd455963f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The function `vhost_dev_init()` expects the `struct vhost_dev`
(passed as a parameter) to be fully initialized. This is important
because some parts of the code check whether `vhost_dev->config_ops`
is NULL to determine if it has been set (e.g. later via
`vhost_dev_set_config_notifier`).
To ensure this initialization, it’s better to allocate the entire
`CryptoDevBackendVhost` structure (which includes `vhost_dev`) using
`g_mem0()`, following the same approach used for other vhost devices,
such as in `vhost_net_init()`.
Fixes: 042cea274c ("cryptodev: add vhost-user as a new cryptodev backend")
Cc: qemu-stable@nongnu.org
Reported-by: myluo24@m.fudan.edu.cn
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-Id: <20250211135523.101203-1-sgarzare@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 83cb18ac45)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
AMD IOMMU provides the base address of control registers through
IVRS table and PCI capability. Since this base address is of 64 bit,
use 32 bits mask (instead of 16 bits) to set BAR low and high.
Fixes: d29a09ca68 ("hw/i386: Introduce AMD IOMMU")
Signed-off-by: Sairaj Kodilkar <sarunkod@amd.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Message-Id: <20250207045354.27329-3-sarunkod@amd.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 3684717b74)
(Mjt: adjust for 7.2.x)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
By accessing MemoryRegion internals, amdvi_init() gives the false
idea that the PCI BAR can be modified. However this isn't true
(at least the model isn't ready for that): the device is explicitly
maps at the BAR at the fixed AMDVI_BASE_ADDR address in
amdvi_sysbus_realize(). Since the SysBus API isn't designed to
remap regions, directly use the fixed address in amdvi_init().
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230313153031.86107-3-philmd@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 6291a28645)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Interrupt passthrough is determine by the bits 191,190,187-184.
These bits are part of the 3rd quad word (i.e. index 2) in DTE. Hence
replace dte[3] by dte[2].
Fixes: b44159fe0 ("x86_iommu/amd: Add interrupt remap support when VAPIC is not enabled")
Signed-off-by: Sairaj Kodilkar <sarunkod@amd.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Message-Id: <20250207045354.27329-2-sarunkod@amd.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 63dc0b8647)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
TYPE_CHIPIDEA models an IP block which is also used in TYPE_ZYNQ_MACHINE which
itself is not an IMX device. CONFIG_ZYNQ selects CONFIG_USB_EHCI_SYSBUS while
TYPE_CHIPIDEA is a separate compilation unit, so only works by accident if
CONFIG_IMX is given. Fix that by extracting CONFIG_USB_CHIPIDEA from CONFIG_IMX.
cc: qemu-stable@nongnu.org
Fixes: 616ec12d0f "hw/arm/xilinx_zynq: Fix USB port instantiation"
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-id: 20250209103604.29545-1-shentey@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 464ce71a96)
(Mjt: context fixup due to missing
v8.0.0-1939-gde6cd7599b51 "meson: Replace softmmu_ss -> system_ss",
v9.1.0-609-ge02491903d50 "hw/usb: Remove tusb6010 USB controller",
v9.2.0-1303-g1b326f278d05 "hw/pci-host/designware: Expose MSI IRQ")
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In the gicv3_{irq,fiq,irqfiq}_access() functions, there is a check
which downgrades a CP_ACCESS_TRAP_EL3 to CP_ACCESS_TRAP if EL3 is not
AArch64. This has been there since the GIC was first implemented,
but it isn't right: if we are trapping because of SCR.IRQ or SCR.FIQ
then we definitely want to be going to EL3 (doing
AArch32.TakeMonitorTrapException() in pseudocode terms). We might
want to not take a trap at all, but we don't ever want to go to the
default target EL, because that would mean, for instance, taking a
trap to Hyp mode if the trapped access was made from Hyp mode.
(This might have been an attempt to work around our failure to
properly implement Monitor Traps.)
Remove the bogus check.
Cc: qemu-stable@nongnu.org
Fixes: 359fbe65e0 ("hw/intc/arm_gicv3: Implement GICv3 CPU interface registers")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250130182309.717346-7-peter.maydell@linaro.org
(cherry picked from commit d04c6c3c00)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In system register access pseudocode the common pattern for
AArch32 registers with access traps to EL3 is:
at EL1 and EL2:
if HaveEL(EL3) && !ELUsingAArch32(EL3) && (SCR_EL3.TERR == 1) then
AArch64.AArch32SystemAccessTrap(EL3, 0x03);
elsif HaveEL(EL3) && ELUsingAArch32(EL3) && (SCR.TERR == 1) then
AArch32.TakeMonitorTrapException();
at EL3:
if (PSTATE.M != M32_Monitor) && (SCR.TERR == 1) then
AArch32.TakeMonitorTrapException();
(taking as an example the ERRIDR access pseudocode).
This implements the behaviour of (in this case) SCR.TERR that
"Accesses to the specified registers from modes other than Monitor
mode generate a Monitor Trap exception" and of SCR_EL3.TERR that
"Accesses of the specified Error Record registers at EL2 and EL1
are trapped to EL3, unless the instruction generates a higher
priority exception".
In QEMU we don't implement this pattern correctly in two ways:
* in access_check_cp_reg() we turn the CP_ACCESS_TRAP_EL3 into
an UNDEF, not a trap to Monitor mode
* in the access functions, we check trap bits like SCR.TERR
only when arm_current_el(env) < 3 -- this is correct for
AArch64 EL3, but misses the "trap non-Monitor-mode execution
at EL3 into Monitor mode" case for AArch32 EL3
In this commit we fix the first of these two issues, by
making access_check_cp_reg() handle CP_ACCESS_TRAP_EL3
as a Monitor trap. This is a kind of exception that we haven't
yet implemented(!), so we need a new EXCP_MON_TRAP for it.
This diverges from the pseudocode approach, where every access check
function explicitly checks for "if EL3 is AArch32" and takes a
monitor trap; if we wanted to be closer to the pseudocode we could
add a new CP_ACCESS_TRAP_MONITOR and make all the accessfns use it
when appropriate. But because there are no non-standard cases in the
pseudocode (i.e. where either it raises a Monitor trap that doesn't
correspond to an AArch64 SystemAccessTrap or where it raises a
SystemAccessTrap that doesn't correspond to a Monitor trap), handling
this all in one place seems less likely to result in future bugs
where we forgot again about this special case when writing an
accessor.
(The cc of stable here is because "hw/intc/arm_gicv3_cpuif: Don't
downgrade monitor traps for AArch32 EL3" which is also cc:stable
will implicitly use the new EXCP_MON_TRAP code path.)
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250130182309.717346-6-peter.maydell@linaro.org
(cherry picked from commit 4cf4948651)
(Mjt: context fix due to missing
v9.0.0-151-gb36a32ead159 "target/arm: Add support for Non-maskable Interrupt",
v8.0.0-2011-g11b76fda0adc "target/arm: Implement GPC exceptions")
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The pseudocode for AT S1E2R and AT S1E2W says that they should be
UNDEFINED if executed at EL3 when EL2 is not enabled. We were
incorrectly using CP_ACCESS_TRAP and reporting the wrong exception
syndrome as a result. Use CP_ACCESS_TRAP_UNCATEGORIZED.
Cc: qemu-stable@nongnu.org
Fixes: 2a47df9532 ("target-arm: Wire up AArch64 EL2 and EL3 address translation ops")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250130182309.717346-4-peter.maydell@linaro.org
(cherry picked from commit ccda792945)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The access pseudocode for the CNTPS_TVAL_EL1, CNTPS_CTL_EL1 and
CNTPS_CVAL_EL1 secure timer registers says that they are UNDEFINED
from EL2 or NS EL1. We incorrectly return CP_ACCESS_TRAP from the
access function in these cases, which means that we report the wrong
syndrome value to the target EL.
Use CP_ACCESS_TRAP_UNCATEGORIZED, which reports the correct syndrome
value for an UNDEFINED instruction.
Cc: qemu-stable@nongnu.org
Fixes: b4d3978c2f ("target-arm: Add the AArch64 view of the Secure physical timer")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250130182309.717346-2-peter.maydell@linaro.org
(cherry picked from commit b819fd6994)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The gdbstub implementation for the Sparc architecture would
incorrectly calculate the the floating point register offset.
This resulted in, for example, registers f32 and f34 to point to
the same value.
The issue was caused by the confusion between even register numbers
and even register indexes. For example, the register index of f32 is 64
and f34 is 65.
Cc: qemu-stable@nongnu.org
Fixes: 30038fd818 ("target-sparc: Change fpr representation to doubles.")
Signed-off-by: Mikael Szreder <git@miszr.win>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250214070343.11501-1-git@miszr.win>
(cherry picked from commit 7a74e46808)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Windows only:
The libSDL2 Windows message loop needs the libSDL2 Windows low
level keyboard hook procedure to grab the left and right Windows
keys correctly. Reenable the SDL2 Windows keyboard hook procedure.
Since SDL2 2.30.4 the SDL2 keyboard hook procedure also filters
out the special left Control key event for every Alt Gr key event
on keyboards with an international layout. This means the QEMU low
level keyboard hook procedure is no longer needed. Remove the QEMU
Windows keyboard hook procedure.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2139
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2323
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Link: https://lore.kernel.org/r/20241231115950.6732-1-vr_qemu@t-online.de
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 4dafba778a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fix in ui/sdl2.c and adaptation in ui/meson.build)
When USBPacket in OUT direction has larger payload
than the ep_out_buffer (of size 512), a buffer overflow
would occur.
It could be fixed by limiting the size of usb_packet_copy
to be at most buffer size. Further optimization gets rid
of the ep_out_buffer and directly uses ep_out as the target
buffer.
This is reported by a security researcher who artificially
constructed an OUT packet of size 2047. The report has gone
through the QEMU security process, and as this device is for
testing purpose and no deployment of it in virtualization
environment is observed, it is triaged not to be a security bug.
Cc: qemu-stable@nongnu.org
Fixes: d7d3491855 ("hw/usb: Add CanoKey Implementation")
Reported-by: Juan Jose Lopez Jaimez <thatjiaozi@gmail.com>
Signed-off-by: Hongren Zheng <i@zenithal.me>
Message-id: Z4TfMOrZz6IQYl_h@Sun
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 664280abdd)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The pseudocode ResetSVEState() does:
FPSR = ZeroExtend(0x0800009f<31:0>, 64);
but QEMU's arm_reset_sve_state() called vfp_set_fpcr() by accident.
Before the advent of FEAT_AFP, this was only setting a collection of
RES0 bits, which vfp_set_fpsr() would then ignore, so the only effect
was that we didn't actually set the FPSR the way we are supposed to
do. Once FEAT_AFP is implemented, setting the bottom bits of FPSR
will change the floating point behaviour.
Call vfp_set_fpsr(), as we ought to.
(Note for stable backports: commit 7f2a01e736 moved this function
from sme_helper.c to helper.c, but it had the same bug before the
move too.)
Cc: qemu-stable@nongnu.org
Fixes: f84734b874 ("target/arm: Implement SMSTART, SMSTOP")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250124162836.2332150-4-peter.maydell@linaro.org
(cherry picked from commit 1edc3d43f2)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(it is in sme_helper.c in 7.2, not in helper.c)
Current versions of Windows call _DSM(func=7) regardless
of whether it is supported or not. It leads to NICs having bogus
'PCI Label Id = 0', where none should be set at all.
Also presence of 'PCI Label Id' triggers another Windows bug
on localized versions that leads to hangs. The later bug is fixed
in latest updates for 'Windows Server' but not in consumer
versions of Windows (and there is no plans to fix it
as far as I'm aware).
Given it's easy, implement Microsoft suggested workaround
(return invalid Package) so that affected Windows versions
could boot on QEMU.
This would effectvely remove bogus 'PCI Label Id's on NICs,
but MS teem confirmed that flipping 'PCI Label Id' should not
change 'Network Connection' ennumeration, so it should be safe
for QEMU to change _DSM without any compat code.
Smoke tested with WinXP and WS2022
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/774
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20250115125342.3883374-3-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 0b05339198)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The end vector calculation has a bug that results in polling fewer
than required vectors when reading at a non-zero offset in PBA memory.
Fixes: bbef882cc1 ("msi: add API to get notified about pending bit poll")
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20241212120402.1475053-1-npiggin@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 42e2a7a0ab)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
PCI hotplug for downstream endpoints on arm fails because Linux'
PCIe hotplug driver doesn't like the QEMU provided LNKSTA:
pcieport 0000:08:01.0: pciehp: Slot(2): Card present
pcieport 0000:08:01.0: pciehp: Slot(2): Link Up
pcieport 0000:08:01.0: pciehp: Slot(2): Cannot train link: status 0x2000
There's 2 cases where LNKSTA isn't setup properly:
* the downstream device has no express capability
* max link width of the bridge is 0
Move the sanity checks added via 88c869198a
("pci: Sanity test minimum downstream LNKSTA") outside of the
branch to make sure downstream ports always have a valid LNKSTA.
Signed-off-by: Sebastian Ott <sebott@redhat.com>
Tested-by: Zhenyu Zhang <zhenyzha@redhat.com>
Message-Id: <20241203121928.14861-1-sebott@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 694632fd44)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
QEMU would crash with a failed assertion if the XHCI controller
attempted to raise the interrupt on an interrupter corresponding
to a MSI vector with a higher index than the highest configured
for the device by the guest driver.
This behaviour is correct on the MSI/PCI side: per PCI 3.0 spec,
devices must ensure they do not send MSI notifications for
vectors beyond the range of those allocated by the system/driver
software. Unlike MSI-X, there is no generic way for handling
aliasing in the case of fewer allocated vectors than requested,
so the specifics are up to device implementors. (Section
6.8.3.4. "Sending Messages")
It turns out the XHCI spec (Implementation Note in section 4.17,
"Interrupters") requires that the host controller signal the MSI
vector with the number computed by taking the interrupter number
modulo the number of enabled MSI vectors.
This change introduces that modulo calculation, fixing the
failed assertion. This makes the device work correctly in MSI mode
with macOS's XHCI driver, which only allocates a single vector.
Signed-off-by: Phil Dennis-Jordan <phil@philjordan.eu>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250112210056.16658-2-phil@philjordan.eu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit bb5b7fced6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
TCG trace-events were deprecated before the v6.2 release,
and removed for v7.0.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(cherry picked from commit b4859e8f33)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
KVM is not happy when starting a VM with weird RAM sizes:
# qemu-system-s390x --enable-kvm --nographic -m 1234K
qemu-system-s390x: kvm_set_user_memory_region: KVM_SET_USER_MEMORY_REGION
failed, slot=0, start=0x0, size=0x244000: Invalid argument
kvm_set_phys_mem: error registering slot: Invalid argument
Aborted (core dumped)
Let's handle that in a better way by rejecting such weird RAM sizes
right from the start:
# qemu-system-s390x --enable-kvm --nographic -m 1234K
qemu-system-s390x: ram size must be multiples of 1 MiB
Message-ID: <20241219144115.2820241-2-david@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Acked-by: Janosch Frank <frankja@linux.ibm.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
(cherry picked from commit 14e568ab48)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The macOS builds in our CI (and possibly other very recent distros)
are currently broken since the update to libnfs version 6 there.
That version apparently comes with a big API breakage. v5.0.3 was
the final release of the old API (see the libnfs commit here:
https://github.com/sahlberg/libnfs/commit/4379837 ).
Disallow version 6.x for now to get the broken CI job working
again. Once somebody had enough time to adapt our code in
block/nfs.c, we can revert this change again.
Message-ID: <20241218065157.209020-1-thuth@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit e2d98f2571)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fixup)
In the GICv3 ITS model, we have a common coding pattern which has a
local C struct like "DTEntry dte", which is a C representation of an
in-guest-memory data structure, and we call a function such as
get_dte() to read guest memory and fill in the C struct. These
functions to read in the struct sometimes have cases where they will
leave early and not fill in the whole struct (for instance get_dte()
will set "dte->valid = false" and nothing else for the case where it
is passed an entry_addr implying that there is no L2 table entry for
the DTE). This then causes potential use of uninitialized memory
later, for instance when we call a trace event which prints all the
fields of the struct. Sufficiently advanced compilers may produce
-Wmaybe-uninitialized warnings about this, especially if LTO is
enabled.
Rather than trying to carefully separate out these trace events into
"only the 'valid' field is initialized" and "all fields can be
printed", zero-init all the structs when we define them. None of
these structs are large (the biggest is 24 bytes) and having
consistent behaviour is less likely to be buggy.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2718
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20241213182337.3343068-1-peter.maydell@linaro.org
(cherry picked from commit 9678b9c505)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
If the binary loaded via -kernel is *not* a linux kernel (in which
case protocol == 0), do not patch the linux kernel header fields.
It's (a) pointless and (b) might break binaries by random patching
and (c) changes the binary hash which in turn breaks secure boot
verification.
Background: OVMF happily loads and runs not only linux kernels but
any efi binary via direct kernel boot.
Note: Breaking the secure boot verification is a problem for linux
kernels too, but fixed that is left for another day ...
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-ID: <20240905141211.1253307-3-kraxel@redhat.com>
(cherry picked from commit 57e2cc9abf)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: it is in hw/i386/x86.c not hw/i386/x86-common.c in 7.2.x)
We used to only have a single UART on the platform and it was located at
address 0x90000000. When the number of UARTs was increased to 4, the
first UART remained at it's location, but instead of being the first one
to be registered, it became the last.
This caused QEMU to pick 0x90000300 as the default UART, which broke
software that hardcoded the address of 0x90000000 and expected it's
output to be visible when the user configured only a single console.
This caused regressions[1] in the barebox test suite when updating to a
newer QEMU. As there seems to be no good reason to register the UARTs in
inverse order, let's register them by ascending address, so existing
software can remain oblivious to the additional UART ports.
Changing the order of uart registration alone breaks Linux which
was choosing the UART at 0x90000300 as the default for ttyS0. To fix
Linux we fix three things in the device tree:
1. Define stdout-path only one time for the first registered UART
instead of incorrectly defining for each UART.
2. Change the UART alias name from 'uart0' to 'serial0' as almost all
Linux tty drivers look for an alias starting with "serial".
3. Add the UART nodes so they appear in the final DTB in the
order starting with the lowest address and working upwards.
In summary these changes mean that the QEMU default UART (serial_hd(0))
is now setup where:
* serial_hd(0) is the lowest-address UART
* serial_hd(0) is listed first in the DTB
* serial_hd(0) is the /chosen/stdout-path one
* the /aliases/serial0 alias points at serial_hd(0)
[1]: https://lore.barebox.org/barebox/707e7c50-aad1-4459-8796-0cc54bab32e2@pengutronix.de/T/#m5da26e8a799033301489a938b5d5667b81cef6ad
Fixes: 777784bda4 ("hw/openrisc: support 4 serial ports in or1ksim")
Cc: qemu-stable@nongnu.org
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
[stafford: Change to serial0 alias and update change message, reverse
uart registration order]
Signed-off-by: Stafford Horne <shorne@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 26dcf2be7e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The ppc (pnv and spapr) NMI injection code does not go through the
asynchronous interrupt path and set a bit in env->pending_interrupts
and raise an interrupt request that the cpu_exec() loop can see.
Instead it injects the exception directly into registers.
This can lead to cpu_exec() missing that the thread has work to do,
if a NMI is injected while it was idle.
Fix this by clearing halted when injecting the interrupt. Probably
NMI injection should be reworked to use the interrupt request interface,
but this seems to work as a minimal fix.
Fixes: 3431648272 ("spapr: Add support for new NMI interface")
Reviewed-by: Glenn Miles <milesg@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
(cherry picked from commit fa416ae615)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
With a valid file ID (FID) of an open file, it should be possible to send
a 'Tgettattr' 9p request and successfully receive a 'Rgetattr' response,
even if the file has been removed in the meantime. Currently this would
fail with ENOENT.
I.e. this fixes the following misbehaviour with a 9p Linux client:
open("/home/tst/filename", O_RDWR|O_CREAT|O_EXCL, 0600) = 3
unlink("/home/tst/filename") = 0
fstat(3, 0x23aa1a8) = -1 ENOENT (No such file or directory)
Expected results:
open("/home/tst/filename", O_RDWR|O_CREAT|O_EXCL, 0600) = 3
unlink("/home/tst/filename") = 0
fstat(3, {st_mode=S_IFREG|0600, st_size=0, ...}) = 0
This is because 9p server is always using a path name based lstat() call
which fails as soon as the file got removed. So to fix this, use fstat()
whenever we have an open file descriptor already.
Fixes: 00ede4c252 ("virtio-9p: getattr server implementation...")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/103
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <4c41ad47f449a5cc8bfa9285743e029080d5f324.1732465720.git.qemu_oss@crudebyte.com>
(cherry picked from commit c81e7219e0)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
After removing a file from the file system, we should still be able to
work with the file if we already had it open before removal.
As a first step we verify that it is possible to write to an unlinked
file, as this is what already works. This test is extended later on
after having fixed other use cases after unlink that are not working
yet.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <3d6449d4df25bcdd3e807eff169f46f1385e5257.1732465720.git.qemu_oss@crudebyte.com>
(cherry picked from commit 462db8fb1d)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fix, pick it to stable so the next patch in this place applies)
The libssh does not handle non-blocking mode in SFTP correctly. The
driver code already changes the mode to blocking for the SFTP
initialization, but for some reason changes to non-blocking mode.
This used to work accidentally until libssh in 0.11 branch merged
the patch to avoid infinite looping in case of network errors:
https://gitlab.com/libssh/libssh-mirror/-/merge_requests/498
Since then, the ssh driver in qemu fails to read files over SFTP
as the first SFTP messages exchanged after switching the session
to non-blocking mode return SSH_AGAIN, but that message is lost
int the SFTP internals and interpretted as SSH_ERROR, which is
returned to the caller:
https://gitlab.com/libssh/libssh-mirror/-/issues/280
This is indeed an issue in libssh that we should address in the
long term, but it will require more work on the internals. For
now, the SFTP is not supported in non-blocking mode.
Fixes: https://gitlab.com/libssh/libssh-mirror/-/issues/280
Signed-off-by: Jakub Jelen <jjelen@redhat.com>
Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
Message-ID: <20241113125526.2495731-1-rjones@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit fbdea3d6c1)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
pci_devfn properties accept either a string or an integer as input. To
implement this, set_pci_devfn() first tries to visit the option as a
string, and if that fails, it visits it as an integer instead. While the
QemuOpts visitor happens to accept this, it is invalid according to the
visitor interface. QObject input visitors run into an assertion failure
when this is done.
QObject input visitors are used with the JSON syntax version of -device
on the command line:
$ ./qemu-system-x86_64 -enable-kvm -M q35 -device pcie-pci-bridge,id=pci.1,bus=pcie.0 -blockdev null-co,node-name=disk -device '{ "driver": "virtio-blk-pci", "drive": "disk", "id": "virtio-disk0", "bus": "pci.1", "addr": 1 }'
qemu-system-x86_64: ../qapi/qobject-input-visitor.c:143: QObject *qobject_input_try_get_object(QObjectInputVisitor *, const char *, _Bool): Assertion `removed' failed.
The proper way to accept both strings and integers is using the
alternate mechanism, which tells us the type of the input before it's
visited. With this information, we can directly visit it as the right
type.
This fixes set_pci_devfn() by using the alternate mechanism.
Cc: qemu-stable@nongnu.org
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20241119120353.57812-1-kwolf@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 5102f9df4a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
work_around_broken_dhclient() accesses IP and UDP headers to detect
relevant packets and to calculate checksums, but it didn't check if
the packet has size sufficient to accommodate them, causing out-of-bound
access hazards. Fix this by correcting the size requirement.
Fixes: 1d41b0c1ec ("Work around dhclient brokenness")
Cc: qemu-stable@nongnu.org
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit a8575f7fb2)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
- Various developers are reluctant to git Cirrus-CI the permissions
requested to access their GitHub account.
- When we use the cirrus-run script to trigger Cirrus-CI job from
GitLab-CI, the GitLab-CI job is restricted to a 1h timeout
(often not enough).
- Although Cirrus-CI VMs are more powerful than GitLab-CI ones,
its free plan is limited in 2 concurrent jobs.
- The GitLab-CI MSYS2 jobs are a 1:1 mapping with the Cirrus-CI ones
(modulo the environment caching).
Reduce the maintenance burden by removing the Cirrus-CI config file,
keeping the GitLab-CI jobs.
Update Yonggang Luo's maintenance file list to the new file, which
use the same environment shell.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230322135721.61138-3-philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit da80f11efe)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: ignore previous changes in .cirrus.yml since it is being removed anyway)
In extioi_setirq() we try to operate on a bit array stored as an
array of uint32_t using the set_bit() and clear_bit() functions
by casting the pointer to 'unsigned long *'.
This has two problems:
* the alignment of 'uint32_t' is less than that of 'unsigned long'
so we pass an insufficiently aligned pointer, which is
undefined behaviour
* on big-endian hosts the 64-bit 'unsigned long' will have
its two halves the wrong way around, and we will produce
incorrect results
The undefined behaviour is shown by the clang undefined-behaviour
sanitizer when running the loongarch64-virt functional test:
/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/include/qemu/bitops.h:41:5: runtime error: store to misaligned address 0x555559745d9c for type 'unsigned long', which requires 8 byte alignment
0x555559745d9c: note: pointer points here
ff ff ff ff 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
^
#0 0x555556fb81c4 in set_bit /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/include/qemu/bitops.h:41:9
#1 0x555556fb81c4 in extioi_setirq /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/clang/../../hw/intc/loongarch_extioi.c:65:9
#2 0x555556fb6e90 in pch_pic_irq_handler /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/clang/../../hw/intc/loongarch_pch_pic.c:75:5
#3 0x555556710265 in serial_ioport_write /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/clang/../../hw/char/serial.c
Fix these problems by using set_bit32() and clear_bit32(),
which work with bit arrays stored as an array of uint32_t.
Cc: qemu-stable@nongnu.org
Fixes: cbff2db1e9 ("hw/intc: Add LoongArch extioi interrupt controller(EIOINTC)")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Message-id: 20241108135514.4006953-4-peter.maydell@linaro.org
(cherry picked from commit 335be5bc44)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: drop hunk in hw/intc/loongarch_extioi.c:extioi_update_sw_coremap()
due to missing v8.2.0-548-g428a6ef4396a "hw/intc/loongarch_extioi: Add vmstate post_load support")
Currently bitops.h defines a set of operations that work on
arbitrary-length bit arrays. However (largely because they
originally came from the Linux kernel) the bit array storage is an
array of 'unsigned long'. This is OK for the kernel and even for
parts of QEMU where we don't really care about the underlying storage
format, but it is not good for devices, where we often want to expose
the storage to the guest and so need a type that is not
variably-sized between host OSes.
We already have a workaround for this in the GICv3 model:
arm_gicv3_common.h defines equivalents of the bit operations that
work on uint32_t. It turns out that we should also be using
something similar in hw/intc/loongarch_extioi.c, which currently
casts a pointer to a uint32_t array to 'unsigned long *' in
extio_setirq(), which is both undefined behaviour and not correct on
a big-endian host.
Define equivalents of the set_bit() function family which work
with a uint32_t array.
(Cc stable because we're about to provide a bugfix to
loongarch_extioi which will depend on this commit.)
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20241108135514.4006953-2-peter.maydell@linaro.org
(cherry picked from commit 3d7680fb18)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The clang sanitizer complains about the code in the EOI handling
of openpic_cpu_write_internal():
UBSAN_OPTIONS=halt_on_error=1:abort_on_error=1 ./build/clang/qemu-system-ppc -M mac99,graphics=off -display none -kernel day15/invaders.elf
../../hw/intc/openpic.c:1034:16: runtime error: index -1 out of bounds for type 'IRQSource[264]' (aka 'struct IRQSource[264]')
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior ../../hw/intc/openpic.c:1034:16 in
This is because we do
src = &opp->src[n_IRQ];
when n_IRQ may be -1. This is in practice harmless because if n_IRQ
is -1 then we don't do anything with the src pointer, but it is
undefined behaviour. (This has been present since this device
was first added to QEMU.)
Rearrange the code so we only do the array index when n_IRQ is not -1.
Cc: qemu-stable@nongnu.org
Fixes: e9df014c0b ("Implement embedded IRQ controller for PowerPC 6xx/740 & 75")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Message-id: 20241105180205.3074071-1-peter.maydell@linaro.org
(cherry picked from commit 3bf7dcd47a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The ClearPortFeature control message fails for PORT_POWER because there
is no break; at the end of the case statement, causing it to fall through
to the failure handler. Add the missing break; to solve the problem.
Fixes: 1cc403eb21 ("usb-hub: emulate per port power switching")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20241112170152.217664-11-linux@roeck-us.net>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit b2cc699979)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When SET_STREAM_FORMAT is called, the st->buft timer is overwritten, thus
causing a memory leak. This was originally fixed in commit 816139ae6a5
("hw/audio/hda: fix memory leak on audio setup", 2024-11-14) but that
caused the audio to break in SPICE.
Fortunately, a simpler fix is possible. The timer only needs to be
reset, because the callback is always the same (st->output is set at
realize time in hda_audio_init); call to timer_new_ns overkill. Replace
it with timer_del and only initialize the timer once; for simplicity,
do it even if use_timer is false.
An even simpler fix would be to free the old time in hda_audio_setup().
However, it seems better to place the initialization of the timer close
to that of st->ouput.
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Message-ID: <20241114125318.1707590-3-pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 626b39006d)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When compiling QEMU with --enable-cfi, the "q800" m68k machine
currently crashes very early, when the q800_machine_init() function
tries to wire the interrupts of the "via1" device.
This happens because TYPE_MOS6522_Q800_VIA1 is supposed to be a
proper SysBus device, but its parent (TYPE_MOS6522) has a mistake
in its class definition where it is only derived from DeviceClass,
and not from SysBusDeviceClass, so we end up in funny memory access
issues here. Using the right class hierarchy for the MOS6522 device
fixes the problem.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2675
Signed-off-by: Thomas Huth <thuth@redhat.com>
Fixes: 51f233ec92 ("misc: introduce new mos6522 VIA device")
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Message-ID: <20241114104653.963812-1-thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit c3d7c18b0d)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In simd_desc() we create a SIMD descriptor from various pieces
including an arbitrary data value from the caller. We try to
sanitize these to make sure everything will fit: the 'data' value
needs to fit in the SIMD_DATA_BITS (== 22) sized field. However we
do that sanitizing with:
tcg_debug_assert(data == sextract32(data, 0, SIMD_DATA_BITS));
This works for the case where the data is supposed to be considered
as a signed integer (which can then be returned via simd_data()).
However, some callers want to treat the data value as unsigned.
Specifically, for the Arm SVE operations, make_svemte_desc()
assembles a data value as a collection of fields, and it needs to use
all 22 bits. Currently if MTE is enabled then its MTEDESC SIZEM1
field may have the most significant bit set, and then it will trip
this assertion.
Loosen the assertion so that we only check that the data value will
fit into the field in some way, either as a signed or as an unsigned
value. This means we will fail to detect some kinds of bug in the
callers, but we won't spuriously assert for intentional use of the
data field as unsigned.
Cc: qemu-stable@nongnu.org
Fixes: db432672dc ("tcg: Add generic vector expanders")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2601
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-ID: <20241115172515.1229393-1-peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit 8377e3fb85)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The commit fd6f7798ac ("linux-user: Use direct syscalls for setuid(),
etc") added direct syscall wrappers for setuid(), setgid(), etc since the
system calls have different semantics than the libc functions.
Add and use the corresponding wrappers for setreuid and setregid which
were missed in that commit.
This fixes the build of the debian package of the uid_wrapper library
(https://cwrap.org/uid_wrapper.html) when running linux-user.
Cc: qemu-stable@nongnu.org
Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-ID: <Zyo2jMKqq8hG8Pkz@p100>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit 8491026a08)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Commit b56617bbcb ("target/i386: Walk NPT in guest real mode") added
logic to run the page table walker even in real mode if we are in NPT
mode. That function then determined whether real mode or paging is
active based on whether the pg_mode variable was 0.
Unfortunately pg_mode is 0 in two situations:
1) Paging is disabled (real mode)
2) Paging is in 2-level paging mode (32bit without PAE)
That means the walker now assumed that 2-level paging mode was real
mode, breaking NetBSD as well as Windows XP.
To fix that, this patch adds a new PG flag to pg_mode which indicates
whether paging is active at all and uses that to determine whether we
are in real mode or not.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2654
Fixes: b56617bbcb ("target/i386: Walk NPT in guest real mode")
Fixes: 253c0a06e0 (commit b56617bbcb in stable-7.2.x series)
Signed-off-by: Alexander Graf <graf@amazon.com>
Reported-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Link: https://lore.kernel.org/r/20241106154329.67218-1-graf@amazon.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 8fa11a4df3)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
A bad (broken or malicious) 9p client (guest) could cause QEMU host to
crash by sending a 9p 'Treaddir' request with a numeric file ID (FID) that
was previously opened for a file instead of an expected directory:
#0 0x0000762aff8f4919 in __GI___rewinddir (dirp=0xf) at
../sysdeps/unix/sysv/linux/rewinddir.c:29
#1 0x0000557b7625fb40 in do_readdir_many (pdu=0x557bb67d2eb0,
fidp=0x557bb67955b0, entries=0x762afe9fff58, offset=0, maxsize=131072,
dostat=<optimized out>) at ../hw/9pfs/codir.c:101
#2 v9fs_co_readdir_many (pdu=pdu@entry=0x557bb67d2eb0,
fidp=fidp@entry=0x557bb67955b0, entries=entries@entry=0x762afe9fff58,
offset=0, maxsize=131072, dostat=false) at ../hw/9pfs/codir.c:226
#3 0x0000557b7625c1f9 in v9fs_do_readdir (pdu=0x557bb67d2eb0,
fidp=0x557bb67955b0, offset=<optimized out>,
max_count=<optimized out>) at ../hw/9pfs/9p.c:2488
#4 v9fs_readdir (opaque=0x557bb67d2eb0) at ../hw/9pfs/9p.c:2602
That's because V9fsFidOpenState was declared as union type. So the
same memory region is used for either an open POSIX file handle (int),
or a POSIX DIR* pointer, etc., so 9p server incorrectly used the
previously opened (valid) POSIX file handle (0xf) as DIR* pointer,
eventually causing a crash in glibc's rewinddir() function.
Root cause was therefore a missing check in 9p server's 'Treaddir'
request handler, which must ensure that the client supplied FID was
really opened as directory stream before trying to access the
aforementioned union and its DIR* member.
Cc: qemu-stable@nongnu.org
Fixes: d62dbb51f7 ("virtio-9p: Add fidtype so that we can do type ...")
Reported-by: Akihiro Suda <suda.kyoto@gmail.com>
Tested-by: Akihiro Suda <suda.kyoto@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <E1t8GnN-002RS8-E2@kylie.crudebyte.com>
(cherry picked from commit 042b4ebfd2)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
If a host chooses to use the SQHD "hint" in the CQE to know if there is
room in the submission queue for additional commands, it may result in a
situation where there are not enough internal resources (struct
NvmeRequest) available to process the command. For a lack of a better
term, the host may "over-commit" the device (i.e., it may have more
inflight commands than the queue size).
For example, assume a queue with N entries. The host submits N commands
and all are picked up for processing, advancing the head and emptying
the queue. Regardless of which of these N commands complete first, the
SQHD field of that CQE will indicate to the host that the queue is
empty, which allows the host to issue N commands again. However, if the
device has not posted CQEs for all the previous commands yet, the device
will have less than N resources available to process the commands, so
queue processing is suspended.
And here lies an 11 year latent bug. In the absense of any additional
tail updates on the submission queue, we never schedule the processing
bottom-half again unless we observe a head update on an associated full
completion queue. This has been sufficient to handle N-to-1 SQ/CQ setups
(in the absense of over-commit of course). Incidentially, that "kick all
associated SQs" mechanism can now be killed since we now just schedule
queue processing when we return a processing resource to a non-empty
submission queue, which happens to cover both edge cases. However, we
must retain kicking the CQ if it was previously full.
So, apparently, no previous driver tested with hw/nvme has ever used
SQHD (e.g., neither the Linux NVMe driver or SPDK uses it). But then OSv
shows up with the driver that actually does. I salute you.
Fixes: f3c507adcd ("NVMe: Initial commit for new storage interface")
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2388
Reported-by: Waldemar Kozaczuk <jwkozaczuk@gmail.com>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
(cherry picked from commit 9529aa6bb4)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Our implementation of the indexed version of SVE SDOT/UDOT/USDOT got
the calculation of the inner loop terminator wrong. Although we
correctly account for the element size when we calculate the
terminator for the first iteration:
intptr_t segend = MIN(16 / sizeof(TYPED), opr_sz_n);
we don't do that when we move it forward after the first inner loop
completes. The intention is that we process the vector in 128-bit
segments, which for a 64-bit element size should mean (1, 2), (3, 4),
(5, 6), etc. This bug meant that we would iterate (1, 2), (3, 4, 5,
6), (7, 8, 9, 10) etc and apply the wrong indexed element to some of
the operations, and also index off the end of the vector.
You don't see this bug if the vector length is small enough that we
don't need to iterate the outer loop, i.e. if it is only 128 bits,
or if it is the 64-bit special case from AA32/AA64 AdvSIMD. If the
vector length is 256 bits then we calculate the right results for the
elements in the vector but do index off the end of the vector. Vector
lengths greater than 256 bits see wrong answers. The instructions
that produce 32-bit results behave correctly.
Fix the recalculation of 'segend' for subsequent iterations, and
restore a version of the comment that was lost in the refactor of
commit 7020ffd656 that explains why we only need to clamp segend to
opr_sz_n for the first iteration, not the later ones.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2595
Fixes: 7020ffd656 ("target/arm: Macroize helper_gvec_{s,u}dot_idx_{b,h}")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20241101185544.2130972-1-peter.maydell@linaro.org
(cherry picked from commit e6b2fa1b81)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
divdu (without a dot) sometimes updates cr0, even though it shouldn't.
The reason is that gen_op_arith_divd() checks Rc(ctx->opcode), which is
not initialized. This field is initialized only for instructions that
go through decode_legacy(), and not decodetree.
There already was a similar issue fixed in commit 86e6202a57
("target/ppc: Make divw[u] handler method decodetree compatible.").
It's not immediately clear what else may access the uninitialized
ctx->opcode, so instead of playing whack-a-mole and changing the check
to compute_rc0, simply initialize ctx->opcode.
Cc: qemu-stable@nongnu.org
Fixes: 99082815f1 ("target/ppc: Add infrastructure for prefixed insns")
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
(cherry picked from commit c9b8a13a88)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The section 4.5.2 of the RISC-V AIA specification says that any write
to a sourcecfg register of an APLIC might (or might not) cause the
corresponding interrupt-pending bit to be set to one if the rectified
input value is high (= 1) under the new source mode.
If an interrupt is asserted before the driver configs its interrupt
type to APLIC, it's pending bit will not be set except a relevant
write to a setip or setipnum register. When we write the interrupt
type to sourcecfg register, if the APLIC device doesn't check
rectified input value and update the pending bit, this interrupt
might never becomes pending.
For APLIC.m, we can manully set pending by setip or setipnum
registers in driver. But for APLIC.w, the pending status totally
depends on the rectified input value, we can't control the pending
status via mmio registers. In this case, hw should check and update
pending status for us when writing sourcecfg registers.
Update QEMU emulation to handle "pre-existing" interrupts.
Signed-off-by: Yong-Xuan Wang <yongxuan.wang@sifive.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20241004104649.13129-1-yongxuan.wang@sifive.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 2ae6cca1d3)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fixup in hw/intc/riscv_aplic.c)
The reads to in_clrip[x] registers return rectified input values of the
interrupt sources.
A rectified input value of an interrupt source is defined by the section
"4.5.2 Source configurations (sourcecfg[1]–sourcecfg[1023])" of the RISC-V
AIA specification as:
"rectified input value = (incoming wire value) XOR (source is inverted)"
Update the riscv_aplic_read_input_word() implementation to match the above.
Fixes: e8f79343cf ("hw/intc: Add RISC-V AIA APLIC device emulation")
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20240306095722.463296-3-apatel@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 0678e9f29c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The RISC-V unprivileged specification "31.3.11. State of Vector
Extension at Reset" has a note that recommends vtype.vill be set on
reset as part of ensuring that the vector extension have a consistent
state at reset.
This change now makes QEMU consistent with Spike which sets vtype.vill
on reset.
Signed-off-by: Rob Bradford <rbradford@rivosinc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20240930165258.72258-1-rbradford@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit f8c1f36a2e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The register VXSAT should be RW only to the first bit.
The remaining bits should be 0.
The RISC-V Instruction Set Manual Volume I: Unprivileged Architecture
The vxsat CSR has a single read-write least-significant bit (vxsat[0])
that indicates if a fixed-point instruction has had to saturate an output
value to fit into a destination format. Bits vxsat[XLEN-1:1]
should be written as zeros.
Signed-off-by: Evgenii Prokopiev <evgenii.prokopiev@syntacore.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20241002084436.89347-1-evgenii.prokopiev@syntacore.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 5a60026cad)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In regime_is_user() we assert if we're passed an ARMMMUIdx_E10_*
mmuidx value. This used to make sense because we only used this
function in ptw.c and would never use it on this kind of stage 1+2
mmuidx, only for an individual stage 1 or stage 2 mmuidx.
However, when we implemented FEAT_E0PD we added a callsite in
aa64_va_parameters(), which means this can now be called for
stage 1+2 mmuidx values if the guest sets the TCG_ELX.{E0PD0,E0PD1}
bits to enable use of the feature. This will then result in
an assertion failure later, for instance on a TLBI operation:
#6 0x00007ffff6d0e70f in g_assertion_message_expr
(domain=0x0, file=0x55555676eeba "../../target/arm/internals.h", line=978, func=0x555556771d48 <__func__.5> "regime_is_user", expr=<optimised out>)
at ../../../glib/gtestutils.c:3279
#7 0x0000555555f286d2 in regime_is_user (env=0x555557f2fe00, mmu_idx=ARMMMUIdx_E10_0) at ../../target/arm/internals.h:978
#8 0x0000555555f3e31c in aa64_va_parameters (env=0x555557f2fe00, va=18446744073709551615, mmu_idx=ARMMMUIdx_E10_0, data=true, el1_is_aa32=false)
at ../../target/arm/helper.c:12048
#9 0x0000555555f3163b in tlbi_aa64_get_range (env=0x555557f2fe00, mmuidx=ARMMMUIdx_E10_0, value=106721347371041) at ../../target/arm/helper.c:5214
#10 0x0000555555f317e8 in do_rvae_write (env=0x555557f2fe00, value=106721347371041, idxmap=21, synced=true) at ../../target/arm/helper.c:5260
#11 0x0000555555f31925 in tlbi_aa64_rvae1is_write (env=0x555557f2fe00, ri=0x555557fbeae0, value=106721347371041) at ../../target/arm/helper.c:5302
#12 0x0000555556036f8f in helper_set_cp_reg64 (env=0x555557f2fe00, rip=0x555557fbeae0, value=106721347371041) at ../../target/arm/tcg/op_helper.c:965
Since we do know whether these mmuidx values are for usermode
or not, we can easily make regime_is_user() handle them:
ARMMMUIdx_E10_0 is user, and the other two are not.
Cc: qemu-stable@nongnu.org
Fixes: e4c93e44ab ("target/arm: Implement FEAT_E0PD")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Tested-by: Alex Bennée <alex.bennee@linaro.org>
Message-id: 20241017172331.822587-1-peter.maydell@linaro.org
(cherry picked from commit 1505b651fd)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The patch fixes the following errors generated by GCC 14.2:
../src/net/tap-win32.c:343:19: error: '%s' directive output may be truncated writing up to 255 bytes into a region of size 176 [-Werror=format-truncation=]
343 | "%s\\%s\\Connection",
| ^~
344 | NETWORK_CONNECTIONS_KEY, enum_name);
| ~~~~~~~~~
../src/net/tap-win32.c:341:9: note: 'snprintf' output between 92 and 347 bytes into a destination of size 256
341 | snprintf(connection_string,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~
342 | sizeof(connection_string),
| ~~~~~~~~~~~~~~~~~~~~~~~~~~
343 | "%s\\%s\\Connection",
| ~~~~~~~~~~~~~~~~~~~~~
344 | NETWORK_CONNECTIONS_KEY, enum_name);
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../src/net/tap-win32.c:242:58: error: '%s' directive output may be truncated writing up to 255 bytes into a region of size 178 [-Werror=format-truncation=]
242 | snprintf (unit_string, sizeof(unit_string), "%s\\%s",
| ^~
243 | ADAPTER_KEY, enum_name);
| ~~~~~~~~~
../src/net/tap-win32.c:242:9: note: 'snprintf' output between 79 and 334 bytes into a destination of size 256
242 | snprintf (unit_string, sizeof(unit_string), "%s\\%s",
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
243 | ADAPTER_KEY, enum_name);
| ~~~~~~~~~~~~~~~~~~~~~~~
../src/net/tap-win32.c:620:52: error: '%s' directive output may be truncated writing up to 255 bytes into a region of size 245 [-Werror=format-truncation=]
620 | snprintf (device_path, sizeof(device_path), "%s%s%s",
| ^~
621 | USERMODEDEVICEDIR,
622 | device_guid,
| ~~~~~~~~~~~
../src/net/tap-win32.c:620:5: note: 'snprintf' output between 16 and 271 bytes into a destination of size 256
620 | snprintf (device_path, sizeof(device_path), "%s%s%s",
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
621 | USERMODEDEVICEDIR,
| ~~~~~~~~~~~~~~~~~~
622 | device_guid,
| ~~~~~~~~~~~~
623 | TAPSUFFIX);
| ~~~~~~~~~~
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2607
Cc: qemu-stable@nongnu.org
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit 75fe36b4e8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When translating virtual to physical address with a guest CPU that
supports nested paging (NPT), we need to perform every page table walk
access indirectly through the NPT, which we correctly do.
However, we treat real mode (no page table walk) special: In that case,
we currently just skip any walks and translate VA -> PA. With NPT
enabled, we also need to then perform NPT walk to do GVA -> GPA -> HPA
which we fail to do so far.
The net result of that is that TCG VMs with NPT enabled that execute
real mode code (like SeaBIOS) end up with GPA==HPA mappings which means
the guest accesses host code and data. This typically shows as failure
to boot guests.
This patch changes the page walk logic for NPT enabled guests so that we
always perform a GVA -> GPA translation and then skip any logic that
requires an actual PTE.
That way, all remaining logic to walk the NPT stays and we successfully
walk the NPT in real mode.
Cc: qemu-stable@nongnu.org
Fixes: fe441054bb ("target-i386: Add NPT support")
Signed-off-by: Alexander Graf <graf@amazon.com>
Reported-by: Eduard Vlad <evlad@amazon.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20240921085712.28902-1-graf@amazon.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit b56617bbcb)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Coverity complains (CID 1507880) that the declaration "int error_code;"
in mmu_translate() is unreachable code. Since this is only a declaration,
this isn't actually a bug, but:
* it's a bear-trap for future changes, because if it was changed to
include an initialization 'int error_code = foo;' then the
initialization wouldn't actually happen (being dead code)
* it's against our coding style, which wants declarations to be
at the start of blocks
* it means that anybody reading the code has to go and look up
exactly what the C rules are for skipping over variable declarations
using a goto
Move the declaration to the top of the function.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20230406155946.3362077-1-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 987b63f24a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: cherry-pick this for stable-7.2 so that the next patch applies cleanly)
s->offset and s->size are only set at the end of the function and still
contain the old values when formatting the error message. Print the
parameters with the new values that we actually checked instead.
Fixes: 500e243420 ('raw-format: Split raw_read_options()')
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20240829185527.47152-1-kwolf@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 04bbc3ee52)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Rather than waiting for the completion of migration on the source side,
wait for it on the destination QEMU side to avoid accessing the TPM TIS
memory mapped registers before QEMU could restore their state. This
error condition could be triggered on busy systems where the destination
QEMU did not have enough time to restore the TIS state while the test case
was already reading its registers. The test case was for example reading
the STS register and received an unexpected value (0xffffffff), which
lead to a segmentation fault later on due to trying to read 0xffff bytes
from the TIS into a buffer.
Cc: <qemu-stable@nongnu.org>
Reported-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
(cherry picked from commit d9280ea317)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Zhiyi reported an infinite loop issue in VFIO use case. The cause of that
was a separate discussion, however during that I found a regression of
dirty sync slowness when profiling.
Each KVMMemoryListerner maintains an array of kvm memslots. Currently it's
statically allocated to be the max supported by the kernel. However after
Linux commit 4fc096a99e ("KVM: Raise the maximum number of user memslots"),
the max supported memslots reported now grows to some number large enough
so that it may not be wise to always statically allocate with the max
reported.
What's worse, QEMU kvm code still walks all the allocated memslots entries
to do any form of lookups. It can drastically slow down all memslot
operations because each of such loop can run over 32K times on the new
kernels.
Fix this issue by making the memslots to be allocated dynamically.
Here the initial size was set to 16 because it should cover the basic VM
usages, so that the hope is the majority VM use case may not even need to
grow at all (e.g. if one starts a VM with ./qemu-system-x86_64 by default
it'll consume 9 memslots), however not too large to waste memory.
There can also be even better way to address this, but so far this is the
simplest and should be already better even than before we grow the max
supported memslots. For example, in the case of above issue when VFIO was
attached on a 32GB system, there are only ~10 memslots used. So it could
be good enough as of now.
In the above VFIO context, measurement shows that the precopy dirty sync
shrinked from ~86ms to ~3ms after this patch applied. It should also apply
to any KVM enabled VM even without VFIO.
NOTE: we don't have a FIXES tag for this patch because there's no real
commit that regressed this in QEMU. Such behavior existed for a long time,
but only start to be a problem when the kernel reports very large
nr_slots_max value. However that's pretty common now (the kernel change
was merged in 2021) so we attached cc:stable because we'll want this change
to be backported to stable branches.
Cc: qemu-stable <qemu-stable@nongnu.org>
Reported-by: Zhiyi Guo <zhguo@redhat.com>
Tested-by: Zhiyi Guo <zhguo@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240917163835.194664-2-peterx@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 5504a81261)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fixup in accel/kvm/kvm-all.c and accel/kvm/trace-events;
also remove now-unused local variable `KVMState *s` in kvm-all.c:kvm_log_sync_global() )
The result of 1 << regbit with regbit==31 has a 1 in the 32nd bit.
When cast to uint64_t (for further bitwise OR), the 32 most
significant bits will be filled with 1s. However, the documentation
states that the upper 32 bits of ICH_AP[0/1]R<n>_EL2 are reserved.
Add an explicit cast to match the documentation.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Cc: qemu-stable@nongnu.org
Fixes: c3f21b065a ("hw/intc/arm_gicv3_cpuif: Support vLPIs")
Signed-off-by: Alexandra Diupina <adiupina@astralinux.ru>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 3db74afec3)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Commit 1880ad4f4e ("virtio-scsi: Batched prepare for cmd reqs") split
calls to scsi_req_new() and scsi_req_enqueue() in the virtio-scsi device.
No ill effects were observed until commit 8cc5583abe ("virtio-scsi: Send
"REPORTED LUNS CHANGED" sense data upon disk hotplug events") added a
unit attention that was easy to trigger with device hotplug and
hot-unplug.
Because the two calls were separated, all requests in the batch were
prepared calling scsi_req_new() to report a sense. The first one
submitted would report the right sense and reset it to NO_SENSE, while
the others reported CHECK_CONDITION with no sense data. This caused
SCSI errors in Linux.
To solve this issue, let's fetch the unit attention as early as possible
when we prepare the request, so that only the first request in the batch
will use the unit attention SCSIReqOps and the others will not report
CHECK CONDITION.
Fixes: 1880ad4f4e ("virtio-scsi: Batched prepare for cmd reqs")
Fixes: 8cc5583abe ("virtio-scsi: Send "REPORTED LUNS CHANGED" sense data upon disk hotplug events")
Reported-by: Thomas Huth <thuth@redhat.com>
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2176702
Co-developed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-ID: <20230712134352.118655-2-sgarzare@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 9472083e64)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Comparing a string of 4 bytes only works in little-endian.
Adjust bulk bswap to only apply to the note payload.
Perform swapping of the note header manually; the magic
is defined so that it does not need a runtime swap.
Fixes: 83f990eb5a ("linux-user/elfload: Parse NT_GNU_PROPERTY_TYPE_0 notes")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2596
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
(cherry picked from commit 2884596f5f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When we are building for OSS-Fuzz, we want to ensure that the fuzzer
targets are actually created, regardless of leaks. Leaks will be
detected by the subsequent tests of the individual fuzz-targets.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240527150001.325565-1-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit 3e964275d6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Allow overlapping request by removing the assert that made it
impossible. There are only two callers:
1. block_copy_task_create()
It already asserts the very same condition before calling
reqlist_init_req().
2. cbw_snapshot_read_lock()
There is no need to have read requests be non-overlapping in
copy-before-write when used for snapshot-access. In fact, there was no
protection against two callers of cbw_snapshot_read_lock() calling
reqlist_init_req() with overlapping ranges and this could lead to an
assertion failure [1].
In particular, with the reproducer script below [0], two
cbw_co_snapshot_block_status() callers could race, with the second
calling reqlist_init_req() before the first one finishes and removes
its conflicting request.
[0]:
> #!/bin/bash -e
> dd if=/dev/urandom of=/tmp/disk.raw bs=1M count=1024
> ./qemu-img create /tmp/fleecing.raw -f raw 1G
> (
> ./qemu-system-x86_64 --qmp stdio \
> --blockdev raw,node-name=node0,file.driver=file,file.filename=/tmp/disk.raw \
> --blockdev raw,node-name=node1,file.driver=file,file.filename=/tmp/fleecing.raw \
> <<EOF
> {"execute": "qmp_capabilities"}
> {"execute": "blockdev-add", "arguments": { "driver": "copy-before-write", "file": "node0", "target": "node1", "node-name": "node3" } }
> {"execute": "blockdev-add", "arguments": { "driver": "snapshot-access", "file": "node3", "node-name": "snap0" } }
> {"execute": "nbd-server-start", "arguments": {"addr": { "type": "unix", "data": { "path": "/tmp/nbd.socket" } } } }
> {"execute": "block-export-add", "arguments": {"id": "exp0", "node-name": "snap0", "type": "nbd", "name": "exp0"}}
> EOF
> ) &
> sleep 5
> while true; do
> ./qemu-nbd -d /dev/nbd0
> ./qemu-nbd -c /dev/nbd0 nbd:unix:/tmp/nbd.socket:exportname=exp0 -f raw -r
> nbdinfo --map 'nbd+unix:///exp0?socket=/tmp/nbd.socket'
> done
[1]:
> #5 0x000071e5f0088eb2 in __GI___assert_fail (...) at ./assert/assert.c:101
> #6 0x0000615285438017 in reqlist_init_req (...) at ../block/reqlist.c:23
> #7 0x00006152853e2d98 in cbw_snapshot_read_lock (...) at ../block/copy-before-write.c:237
> #8 0x00006152853e3068 in cbw_co_snapshot_block_status (...) at ../block/copy-before-write.c:304
> #9 0x00006152853f4d22 in bdrv_co_snapshot_block_status (...) at ../block/io.c:3726
> #10 0x000061528543a63e in snapshot_access_co_block_status (...) at ../block/snapshot-access.c:48
> #11 0x00006152853f1a0a in bdrv_co_do_block_status (...) at ../block/io.c:2474
> #12 0x00006152853f2016 in bdrv_co_common_block_status_above (...) at ../block/io.c:2652
> #13 0x00006152853f22cf in bdrv_co_block_status_above (...) at ../block/io.c:2732
> #14 0x00006152853d9a86 in blk_co_block_status_above (...) at ../block/block-backend.c:1473
> #15 0x000061528538da6c in blockstatus_to_extents (...) at ../nbd/server.c:2374
> #16 0x000061528538deb1 in nbd_co_send_block_status (...) at ../nbd/server.c:2481
> #17 0x000061528538f424 in nbd_handle_request (...) at ../nbd/server.c:2978
> #18 0x000061528538f906 in nbd_trip (...) at ../nbd/server.c:3121
> #19 0x00006152855a7caf in coroutine_trampoline (...) at ../util/coroutine-ucontext.c:175
Cc: qemu-stable@nongnu.org
Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Message-Id: <20240712140716.517911-1-f.ebner@proxmox.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
(cherry picked from commit 6475155d51)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The XT check for the lxvx/stxvx instructions is currently
inverted. This was introduced during the move to decodetree.
>From the ISA:
Chapter 7. Vector-Scalar Extension Facility
Load VSX Vector Indexed X-form
lxvx XT,RA,RB
if TX=0 & MSR.VSX=0 then VSX_Unavailable()
if TX=1 & MSR.VEC=0 then Vector_Unavailable()
...
Let XT be the value 32×TX + T.
The code currently does the opposite:
if (paired || a->rt >= 32) {
REQUIRE_VSX(ctx);
} else {
REQUIRE_VECTOR(ctx);
}
This was already fixed for lxv/stxv at commit "2cc0e449d1 (target/ppc:
Fix lxv/stxv MSR facility check)", but the indexed forms were missed.
Cc: qemu-stable@nongnu.org
Fixes: 70426b5bb7 ("target/ppc: moved stxvx and lxvx from legacy to decodtree")
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Claudio Fontana <cfontana@suse.de>
Acked-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Message-ID: <20240911141651.6914-1-farosas@suse.de>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit 8bded2e73e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The follow-up transactions may use the data in the attribution, so keep
the value of attribution from the function parameter just as
flatview_translate() above.
Signed-off-by: Fea.Wang <fea.wang@sifive.com>
Cc: qemu-stable@nongnu.org
Fixes: f26404fbee ("Make address_space_map() take a MemTxAttrs argument")
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Link: https://lore.kernel.org/r/20240912070404.2993976-2-fea.wang@sifive.com
Signed-off-by: Peter Xu <peterx@redhat.com>
(cherry picked from commit d8d5ca4004)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fix due to lack of
v9.1.0-134-g637b0aa13956 "softmmu: Support concurrent bounce buffers"
v9.0.0-564-g69e78f1b3484 "system/physmem: Per-AddressSpace bounce buffering")
On GICv2 and later, level triggered interrupts are pending when either
the interrupt line is asserted or the interrupt was made pending by a
GICD_ISPENDRn write. Making a level triggered interrupt pending by
software persists until either the interrupt is acknowledged or cleared
by writing GICD_ICPENDRn. As long as the interrupt line is asserted,
the interrupt is pending in any case.
This logic is transparently implemented in gic_test_pending() for
GICv1 and GICv2. The function combines the "pending" irq_state flag
(used for edge triggered interrupts and software requests) and the
line status (tracked in the "level" field). However, we also
incorrectly set the pending flag on a guest write to GICD_ISENABLERn
if the line of a level triggered interrupt was asserted. This keeps
the interrupt pending even if the line is de-asserted after some
time.
This incorrect logic is a leftover of the initial 11MPCore GIC
implementation. That handles things slightly differently to the
architected GICv1 and GICv2. The 11MPCore TRM does not give a lot of
detail on the corner cases of its GIC's behaviour, and historically
we have not wanted to investigate exactly what it does in reality, so
QEMU's GIC model takes the approach of "retain our existing behaviour
for 11MPCore, and implement the architectural standard for later GIC
revisions".
On that basis, commit 8d999995e4 in 2013 is where we added the
"level-triggered interrupt with the line asserted" handling to
gic_test_pending(), and we deliberately kept the old behaviour of
gic_test_pending() for REV_11MPCORE. That commit should have added
the "only if 11MPCore" condition to the setting of the pending bit on
writes to GICD_ISENABLERn, but forgot it.
Add the missing "if REV_11MPCORE" condition, so that our behaviour
on GICv1 and GICv2 matches the GIC architecture requirements.
Cc: qemu-stable@nongnu.org
Fixes: 8d999995e4 ("arm_gic: Fix GIC pending behavior")
Signed-off-by: Jan Klötzke <jan.kloetzke@kernkonzept.com>
Message-id: 20240911114826.3558302-1-jan.kloetzke@kernkonzept.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: expanded comment a little and converted to coding-style form;
expanded commit message with the historical backstory]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 110684c9a6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
As debian-11 transitions to LTS we are starting to have problems
building the image. While we could update to a later Debian building a
32 bit QEMU without modern floating point is niche host amongst the
few remaining 32 bit hosts we regularly build for. For now we still
have armhf-debian-cross-container which is currently built from the
more recent debian-12.
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-2-alex.bennee@linaro.org>
(cherry picked from commit d0068b746a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: force-delete tests/docker/dockerfiles/debian-armel-cross.docker,
fixups in .gitlab-ci.d/crossbuilds.yml,
context fixup in tests/lcitool/refresh)
Both gnutls and gcrypt can be configured to exclude support for certain
algorithms via a runtime check against system crypto policies. Thus it
is not sufficient to have a compile time test for hash support in their
pbkdf implementations.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
(cherry picked from commit e6c09ea4f9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
As reported by Peter, we might be leaking memory when removing the
highest RAMBlock (in the weird ram_addr_t space), and adding a new one.
We will fail to realize that we already allocated bitmaps for more
dirty memory blocks, and effectively discard the pointers to them.
Fix it by getting rid of last_ram_page() and by remembering the number
of dirty memory blocks that have been allocated already.
While at it, let's use "unsigned int" for the number of blocks, which
should be sufficient until we reach ~32 exabytes.
Looks like this leak was introduced as we switched from using a single
bitmap_zero_extend() to allocating multiple bitmaps:
bitmap_zero_extend() relies on g_renew() which should have taken care of
this.
Resolves: https://lkml.kernel.org/r/CAFEAcA-k7a+VObGAfCFNygQNfCKL=AfX6A4kScq=VSSK0peqPg@mail.gmail.com
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Fixes: 5b82b703b6 ("memory: RCU ram_list.dirty_memory[] for safe RAM hotplug")
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Tested-by: Peter Maydell <peter.maydell@linaro.org>
Cc: qemu-stable@nongnu.org
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20240828090743.128647-1-david@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
(cherry picked from commit b84f06c2be)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fix due to lack of
v9.0.0-rc4-49-g15f7a80c49cb "RAMBlock: Add support of KVM private guest memfd")
When the creds->username property is set we allocate memory
for it in qcrypto_tls_creds_psk_prop_set_username(), but
we never free this when the QCryptoTLSCredsPSK is destroyed.
Free the memory in finalize.
This fixes a LeakSanitizer complaint in migration-test:
$ (cd build/asan; ASAN_OPTIONS="fast_unwind_on_malloc=0" QTEST_QEMU_BINARY=./qemu-system-x86_64 ./tests/qtest/migration-test --tap -k -p /x86_64/migration/precopy/unix/tls/psk)
=================================================================
==3867512==ERROR: LeakSanitizer: detected memory leaks
Direct leak of 5 byte(s) in 1 object(s) allocated from:
#0 0x5624e5c99dee in malloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-x86_64+0x218edee) (BuildId: a9e623fa1009a9435c0142c037cd7b8c1ad04ce3)
#1 0x7fb199ae9738 in g_malloc debian/build/deb/../../../glib/gmem.c:128:13
#2 0x7fb199afe583 in g_strdup debian/build/deb/../../../glib/gstrfuncs.c:361:17
#3 0x5624e82ea919 in qcrypto_tls_creds_psk_prop_set_username /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../crypto/tlscredspsk.c:255:23
#4 0x5624e812c6b5 in property_set_str /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object.c:2277:5
#5 0x5624e8125ce5 in object_property_set /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object.c:1463:5
#6 0x5624e8136e7c in object_set_properties_from_qdict /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object_interfaces.c:55:14
#7 0x5624e81372d2 in user_creatable_add_type /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object_interfaces.c:112:5
#8 0x5624e8137964 in user_creatable_add_qapi /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object_interfaces.c:157:11
#9 0x5624e891ba3c in qmp_object_add /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/qom-qmp-cmds.c:227:5
#10 0x5624e8af9118 in qmp_marshal_object_add /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qapi/qapi-commands-qom.c:337:5
#11 0x5624e8bd1d49 in do_qmp_dispatch_bh /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qapi/qmp-dispatch.c:128:5
#12 0x5624e8cb2531 in aio_bh_call /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/async.c:171:5
#13 0x5624e8cb340c in aio_bh_poll /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/async.c:218:13
#14 0x5624e8c0be98 in aio_dispatch /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/aio-posix.c:423:5
#15 0x5624e8cba3ce in aio_ctx_dispatch /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/async.c:360:5
#16 0x7fb199ae0d3a in g_main_dispatch debian/build/deb/../../../glib/gmain.c:3419:28
#17 0x7fb199ae0d3a in g_main_context_dispatch debian/build/deb/../../../glib/gmain.c:4137:7
#18 0x5624e8cbe1d9 in glib_pollfds_poll /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/main-loop.c:287:9
#19 0x5624e8cbcb13 in os_host_main_loop_wait /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/main-loop.c:310:5
#20 0x5624e8cbc6dc in main_loop_wait /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/main-loop.c:589:11
#21 0x5624e6f3f917 in qemu_main_loop /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../system/runstate.c:801:9
#22 0x5624e893379c in qemu_default_main /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../system/main.c:37:14
#23 0x5624e89337e7 in main /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../system/main.c:48:12
#24 0x7fb197972d8f in __libc_start_call_main csu/../sysdeps/nptl/libc_start_call_main.h:58:16
#25 0x7fb197972e3f in __libc_start_main csu/../csu/libc-start.c:392:3
#26 0x5624e5c16fa4 in _start (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-x86_64+0x210bfa4) (BuildId: a9e623fa1009a9435c0142c037cd7b8c1ad04ce3)
SUMMARY: AddressSanitizer: 5 byte(s) leaked in 1 allocation(s).
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240819145021.38524-1-peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 87e012f29f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Commit 3e7ef738 plugged the use-after-free of the global nbd_server
object, but overlooked a use-after-free of nbd_server->listener.
Although this race is harder to hit, notice that our shutdown path
first drops the reference count of nbd_server->listener, then triggers
actions that can result in a pending client reaching the
nbd_blockdev_client_closed() callback, which in turn calls
qio_net_listener_set_client_func on a potentially stale object.
If we know we don't want any more clients to connect, and have already
told the listener socket to shut down, then we should not be trying to
update the listener socket's associated function.
Reproducer:
> #!/usr/bin/python3
>
> import os
> from threading import Thread
>
> def start_stop():
> while 1:
> os.system('virsh qemu-monitor-command VM \'{"execute": "nbd-server-start",
+"arguments":{"addr":{"type":"unix","data":{"path":"/tmp/nbd-sock"}}}}\'')
> os.system('virsh qemu-monitor-command VM \'{"execute": "nbd-server-stop"}\'')
>
> def nbd_list():
> while 1:
> os.system('/path/to/build/qemu-nbd -L -k /tmp/nbd-sock')
>
> def test():
> sst = Thread(target=start_stop)
> sst.start()
> nlt = Thread(target=nbd_list)
> nlt.start()
>
> sst.join()
> nlt.join()
>
> test()
Fixes: CVE-2024-7409
Fixes: 3e7ef738c8 ("nbd/server: CVE-2024-7409: Close stray clients at server-stop")
CC: qemu-stable@nongnu.org
Reported-by: Andrey Drobyshev <andrey.drobyshev@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240822143617.800419-2-eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit 3874f5f73c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
A malicious client can attempt to connect to an NBD server, and then
intentionally delay progress in the handshake, including if it does
not know the TLS secrets. Although the previous two patches reduce
this behavior by capping the default max-connections parameter and
killing slow clients, they did not eliminate the possibility of a
client waiting to close the socket until after the QMP nbd-server-stop
command is executed, at which point qemu would SEGV when trying to
dereference the NULL nbd_server global which is no longer present.
This amounts to a denial of service attack. Worse, if another NBD
server is started before the malicious client disconnects, I cannot
rule out additional adverse effects when the old client interferes
with the connection count of the new server (although the most likely
is a crash due to an assertion failure when checking
nbd_server->connections > 0).
For environments without this patch, the CVE can be mitigated by
ensuring (such as via a firewall) that only trusted clients can
connect to an NBD server. Note that using frameworks like libvirt
that ensure that TLS is used and that nbd-server-stop is not executed
while any trusted clients are still connected will only help if there
is also no possibility for an untrusted client to open a connection
but then stall on the NBD handshake.
Given the previous patches, it would be possible to guarantee that no
clients remain connected by having nbd-server-stop sleep for longer
than the default handshake deadline before finally freeing the global
nbd_server object, but that could make QMP non-responsive for a long
time. So intead, this patch fixes the problem by tracking all client
sockets opened while the server is running, and forcefully closing any
such sockets remaining without a completed handshake at the time of
nbd-server-stop, then waiting until the coroutines servicing those
sockets notice the state change. nbd-server-stop now has a second
AIO_WAIT_WHILE_UNLOCKED (the first is indirectly through the
blk_exp_close_all_type() that disconnects all clients that completed
handshakes), but forced socket shutdown is enough to progress the
coroutines and quickly tear down all clients before the server is
freed, thus finally fixing the CVE.
This patch relies heavily on the fact that nbd/server.c guarantees
that it only calls nbd_blockdev_client_closed() from the main loop
(see the assertion in nbd_client_put() and the hoops used in
nbd_client_put_nonzero() to achieve that); if we did not have that
guarantee, we would also need a mutex protecting our accesses of the
list of connections to survive re-entrancy from independent iothreads.
Although I did not actually try to test old builds, it looks like this
problem has existed since at least commit 862172f45c (v2.12.0, 2017) -
even back when that patch started using a QIONetListener to handle
listening on multiple sockets, nbd_server_free() was already unaware
that the nbd_blockdev_client_closed callback can be reached later by a
client thread that has not completed handshakes (and therefore the
client's socket never got added to the list closed in
nbd_export_close_all), despite that patch intentionally tearing down
the QIONetListener to prevent new clients.
Reported-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>
Fixes: CVE-2024-7409
CC: qemu-stable@nongnu.org
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-14-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
(cherry picked from commit 3e7ef738c8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
A client that opens a socket but does not negotiate is merely hogging
qemu's resources (an open fd and a small amount of memory); and a
malicious client that can access the port where NBD is listening can
attempt a denial of service attack by intentionally opening and
abandoning lots of unfinished connections. The previous patch put a
default bound on the number of such ongoing connections, but once that
limit is hit, no more clients can connect (including legitimate ones).
The solution is to insist that clients complete handshake within a
reasonable time limit, defaulting to 10 seconds. A client that has
not successfully completed NBD_OPT_GO by then (including the case of
where the client didn't know TLS credentials to even reach the point
of NBD_OPT_GO) is wasting our time and does not deserve to stay
connected. Later patches will allow fine-tuning the limit away from
the default value (including disabling it for doing integration
testing of the handshake process itself).
Note that this patch in isolation actually makes it more likely to see
qemu SEGV after nbd-server-stop, as any client socket still connected
when the server shuts down will now be closed after 10 seconds rather
than at the client's whims. That will be addressed in the next patch.
For a demo of this patch in action:
$ qemu-nbd -f raw -r -t -e 10 file &
$ nbdsh --opt-mode -c '
H = list()
for i in range(20):
print(i)
H.insert(i, nbd.NBD())
H[i].set_opt_mode(True)
H[i].connect_uri("nbd://localhost")
'
$ kill $!
where later connections get to start progressing once earlier ones are
forcefully dropped for taking too long, rather than hanging.
Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-13-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[eblake: rebase to changes earlier in series, reduce scope of timer]
Signed-off-by: Eric Blake <eblake@redhat.com>
(cherry picked from commit b9b72cb3ce)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fixup in nbd/server.c: lack of WITH_QEMU_LOCK_GUARD in 7.2)
Allowing an unlimited number of clients to any web service is a recipe
for a rudimentary denial of service attack: the client merely needs to
open lots of sockets without closing them, until qemu no longer has
any more fds available to allocate.
For qemu-nbd, we default to allowing only 1 connection unless more are
explicitly asked for (-e or --shared); this was historically picked as
a nice default (without an explicit -t, a non-persistent qemu-nbd goes
away after a client disconnects, without needing any additional
follow-up commands), and we are not going to change that interface now
(besides, someday we want to point people towards qemu-storage-daemon
instead of qemu-nbd).
But for qemu proper, and the newer qemu-storage-daemon, the QMP
nbd-server-start command has historically had a default of unlimited
number of connections, in part because unlike qemu-nbd it is
inherently persistent until nbd-server-stop. Allowing multiple client
sockets is particularly useful for clients that can take advantage of
MULTI_CONN (creating parallel sockets to increase throughput),
although known clients that do so (such as libnbd's nbdcopy) typically
use only 8 or 16 connections (the benefits of scaling diminish once
more sockets are competing for kernel attention). Picking a number
large enough for typical use cases, but not unlimited, makes it
slightly harder for a malicious client to perform a denial of service
merely by opening lots of connections withot progressing through the
handshake.
This change does not eliminate CVE-2024-7409 on its own, but reduces
the chance for fd exhaustion or unlimited memory usage as an attack
surface. On the other hand, by itself, it makes it more obvious that
with a finite limit, we have the problem of an unauthenticated client
holding 100 fds opened as a way to block out a legitimate client from
being able to connect; thus, later patches will further add timeouts
to reject clients that are not making progress.
This is an INTENTIONAL change in behavior, and will break any client
of nbd-server-start that was not passing an explicit max-connections
parameter, yet expects more than 100 simultaneous connections. We are
not aware of any such client (as stated above, most clients aware of
MULTI_CONN get by just fine on 8 or 16 connections, and probably cope
with later connections failing by relying on the earlier connections;
libvirt has not yet been passing max-connections, but generally
creates NBD servers with the intent for a single client for the sake
of live storage migration; meanwhile, the KubeSAN project anticipates
a large cluster sharing multiple clients [up to 8 per node, and up to
100 nodes in a cluster], but it currently uses qemu-nbd with an
explicit --shared=0 rather than qemu-storage-daemon with
nbd-server-start).
We considered using a deprecation period (declare that omitting
max-parameters is deprecated, and make it mandatory in 3 releases -
then we don't need to pick an arbitrary default); that has zero risk
of breaking any apps that accidentally depended on more than 100
connections, and where such breakage might not be noticed under unit
testing but only under the larger loads of production usage. But it
does not close the denial-of-service hole until far into the future,
and requires all apps to change to add the parameter even if 100 was
good enough. It also has a drawback that any app (like libvirt) that
is accidentally relying on an unlimited default should seriously
consider their own CVE now, at which point they are going to change to
pass explicit max-connections sooner than waiting for 3 qemu releases.
Finally, if our changed default breaks an app, that app can always
pass in an explicit max-parameters with a larger value.
It is also intentional that the HMP interface to nbd-server-start is
not changed to expose max-connections (any client needing to fine-tune
things should be using QMP).
Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-12-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[ericb: Expand commit message to summarize Dan's argument for why we
break corner-case back-compat behavior without a deprecation period]
Signed-off-by: Eric Blake <eblake@redhat.com>
(cherry picked from commit c8a76dbd90)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: minor fixups in qapi/block-export.json)
Upcoming patches to fix a CVE need to track an opaque pointer passed
in by the owner of a client object, as well as request for a time
limit on how fast negotiation must complete. Prepare for that by
changing the signature of nbd_client_new() and adding an accessor to
get at the opaque pointer, although for now the two servers
(qemu-nbd.c and blockdev-nbd.c) do not change behavior even though
they pass in a new default timeout value.
Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-11-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[eblake: s/LIMIT/MAX_SECS/ as suggested by Dan]
Signed-off-by: Eric Blake <eblake@redhat.com>
(cherry picked from commit fb1c2aaa98)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When reading with `read_cluster` we get the `mapping` with
`find_mapping_for_cluster` and then we call `open_file` for this
mapping.
The issue appear when its the same file, but a second cluster that is
not immediately after it, imagine clusters `500 -> 503`, this will give
us 2 mappings one has the range `500..501` and another `503..504`, both
point to the same file, but different offsets.
When we don't open the file since the path is the same, we won't assign
`s->current_mapping` and thus accessing way out of bound of the file.
From our example above, after `open_file` (that didn't open anything) we
will get the offset into the file with
`s->cluster_size*(cluster_num-s->current_mapping->begin)`, which will
give us `0x2000 * (504-500)`, which is out of bound for this mapping and
will produce some issues.
Signed-off-by: Amjad Alsharafi <amjadsharafi10@gmail.com>
Message-ID: <1f3ea115779abab62ba32c788073cdc99f9ad5dd.1721470238.git.amjadsharafi10@gmail.com>
[kwolf: Simplified the patch based on Amjad's analysis and input]
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 5eed3db336)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
How this `abort` was intended to check for was:
- if the `mapping->first_mapping_index` is not the same as
`first_mapping_index`, which **should** happen only in one case,
when we are handling the first mapping, in that case
`mapping->first_mapping_index == -1`, in all other cases, the other
mappings after the first should have the condition `true`.
- From above, we know that this is the first mapping, so if the offset
is not `0`, then abort, since this is an invalid state.
The issue was that `first_mapping_index` is not set if we are
checking from the middle, the variable `first_mapping_index` is
only set if we passed through the check `cluster_was_modified` with the
first mapping, and in the same function call we checked the other
mappings.
One approach is to go into the loop even if `cluster_was_modified`
is not true so that we will be able to set `first_mapping_index` for the
first mapping, but since `first_mapping_index` is only used here,
another approach is to just check manually for the
`mapping->first_mapping_index != -1` since we know that this is the
value for the only entry where `offset == 0` (i.e. first mapping).
Signed-off-by: Amjad Alsharafi <amjadsharafi10@gmail.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <b0fbca3ee208c565885838f6a7deeaeb23f4f9c2.1721470238.git.amjadsharafi10@gmail.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit f60a6f7e17)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Before this commit, the behavior when calling `commit_one_file` for
example with `offset=0x2000` (second cluster), what will happen is that
we won't fetch the next cluster from the fat, and instead use the first
cluster for the read operation.
This is due to off-by-one error here, where `i=0x2000 !< offset=0x2000`,
thus not fetching the next cluster.
Signed-off-by: Amjad Alsharafi <amjadsharafi10@gmail.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Tested-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <b97c1e1f1bc2f776061ae914f95d799d124fcd73.1721470238.git.amjadsharafi10@gmail.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit b881cf00c9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Patch 06b1297017 ("virtio-net: fix network stall under load")
added double-check to test whether the available buffer size
can satisfy the request or not, in case the guest has added
some buffers to the avail ring simultaneously after the first
check. It will be lucky if the available buffer size becomes
okay after the double-check, then the host can send the packet
to the guest. If the buffer size still can't satisfy the request,
even if the guest has added some buffers, viritio-net would
stall at the host side forever.
The patch enables notification and checks whether the guest has
added some buffers since last check of available buffers when
the available buffers are insufficient. If no buffer is added,
return false, else recheck the available buffers in the loop.
If the available buffers are sufficient, disable notification
and return true.
Changes:
1. Change the return type of virtqueue_get_avail_bytes() from void
to int, it returns an opaque that represents the shadow_avail_idx
of the virtqueue on success, else -1 on error.
2. Add a new API: virtio_queue_enable_notification_and_check(),
it takes an opaque as input arg which is returned from
virtqueue_get_avail_bytes(). It enables notification firstly,
then checks whether the guest has added some buffers since
last check of available buffers or not by virtio_queue_poll(),
return ture if yes.
The patch also reverts patch "06b12970174".
The case below can reproduce the stall.
Guest 0
+--------+
| iperf |
---------------> | server |
Host | +--------+
+--------+ | ...
| iperf |----
| client |---- Guest n
+--------+ | +--------+
| | iperf |
---------------> | server |
+--------+
Boot many guests from qemu with virtio network:
qemu ... -netdev tap,id=net_x \
-device virtio-net-pci-non-transitional,\
iommu_platform=on,mac=xx:xx:xx:xx:xx:xx,netdev=net_x
Each guest acts as iperf server with commands below:
iperf3 -s -D -i 10 -p 8001
iperf3 -s -D -i 10 -p 8002
The host as iperf client:
iperf3 -c guest_IP -p 8001 -i 30 -w 256k -P 20 -t 40000
iperf3 -c guest_IP -p 8002 -i 30 -w 256k -P 20 -t 40000
After some time, the host loses connection to the guest,
the guest can send packet to the host, but can't receive
packet from the host.
It's more likely to happen if SWIOTLB is enabled in the guest,
allocating and freeing bounce buffer takes some CPU ticks,
copying from/to bounce buffer takes more CPU ticks, compared
with that there is no bounce buffer in the guest.
Once the rate of producing packets from the host approximates
the rate of receiveing packets in the guest, the guest would
loop in NAPI.
receive packets ---
| |
v |
free buf virtnet_poll
| |
v |
add buf to avail ring ---
|
| need kick the host?
| NAPI continues
v
receive packets ---
| |
v |
free buf virtnet_poll
| |
v |
add buf to avail ring ---
|
v
... ...
On the other hand, the host fetches free buf from avail
ring, if the buf in the avail ring is not enough, the
host notifies the guest the event by writing the avail
idx read from avail ring to the event idx of used ring,
then the host goes to sleep, waiting for the kick signal
from the guest.
Once the guest finds the host is waiting for kick singal
(in virtqueue_kick_prepare_split()), it kicks the host.
The host may stall forever at the sequences below:
Host Guest
------------ -----------
fetch buf, send packet receive packet ---
... ... |
fetch buf, send packet add buf |
... add buf virtnet_poll
buf not enough avail idx-> add buf |
read avail idx add buf |
add buf ---
receive packet ---
write event idx ... |
wait for kick add buf virtnet_poll
... |
---
no more packet, exit NAPI
In the first loop of NAPI above, indicated in the range of
virtnet_poll above, the host is sending packets while the
guest is receiving packets and adding buffers.
step 1: The buf is not enough, for example, a big packet
needs 5 buf, but the available buf count is 3.
The host read current avail idx.
step 2: The guest adds some buf, then checks whether the
host is waiting for kick signal, not at this time.
The used ring is not empty, the guest continues
the second loop of NAPI.
step 3: The host writes the avail idx read from avail
ring to used ring as event idx via
virtio_queue_set_notification(q->rx_vq, 1).
step 4: At the end of the second loop of NAPI, recheck
whether kick is needed, as the event idx in the
used ring written by the host is beyound the
range of kick condition, the guest will not
send kick signal to the host.
Fixes: 06b1297017 ("virtio-net: fix network stall under load")
Cc: qemu-stable@nongnu.org
Signed-off-by: Wencheng Yang <east.moutain.yang@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit f937309fbd)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fixup in include/hw/virtio/virtio.h)
Ensure the queue index points to a valid queue when software RSS
enabled. The new calculation matches with the behavior of Linux's TAP
device with the RSS eBPF program.
Fixes: 4474e37a5b ("virtio-net: implement RX RSS processing")
Reported-by: Zhibin Hu <huzhibin5@huawei.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit f1595ceb9a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Fixes: CVE-2024-6505
The FMOPA (widening) SME instruction takes pairs of half-precision
floating point values, widens them to single-precision, does a
two-way dot product and accumulates the results into a
single-precision destination. We don't quite correctly handle the
FPCR bits FZ and FZ16 which control flushing of denormal inputs and
outputs. This is because at the moment we pass a single float_status
value to the helper function, which then uses that configuration for
all the fp operations it does. However, because the inputs to this
operation are float16 and the outputs are float32 we need to use the
fp_status_f16 for the float16 input widening but the normal fp_status
for everything else. Otherwise we will apply the flushing control
FPCR.FZ16 to the 32-bit output rather than the FPCR.FZ control, and
incorrectly flush a denormal output to zero when we should not (or
vice-versa).
(In commit 207d30b5fd we tried to fix the FZ handling but
didn't get it right, switching from "use FPCR.FZ for everything" to
"use FPCR.FZ16 for everything".)
(Mjt: it is commit d5373d7bdb in stable-7.2)
Pass the CPU env to the sme_fmopa_h helper instead of an fp_status
pointer, and have the helper pass an extra fp_status into the
f16_dotadd() function so that we can use the right status for the
right parts of this operation.
Cc: qemu-stable@nongnu.org
Fixes: 207d30b5fd ("target/arm: Use FPST_F16 for SME FMOPA (widening)")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2373
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit 55f9f4ee01)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: s/tcg_env/cpu_env/ due to missingv
8.1.0-1189-gad75a51e84af "tcg: Rename cpu_env to tcg_env")
In newer versions of Sphinx the env.doc2path() API is going to change
to return a Path object rather than a str. This was originally visible
in Sphinx 8.0.0rc1, but has been rolled back for the final 8.0.0
release. However it will probably emit a deprecation warning and is
likely to change for good in 9.0:
https://github.com/sphinx-doc/sphinx/issues/12686
Our use in depfile.py assumes a str, and if it is passed a Path
it will fall over:
Handler <function write_depfile at 0x77a1775ff560> for event 'build-finished' threw an exception (exception: unsupported operand type(s) for +: 'PosixPath' and 'str')
Wrapping the env.doc2path() call in str() will coerce a Path object
to the str we expect, and have no effect in older Sphinx versions
that do return a str.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2458
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240729120533.2486427-1-peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 48e5b5f994)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When determining the current vector length, the SMCR_EL2.LEN and
SVCR_EL2.LEN settings should only be considered if EL2 is enabled
(compare the pseudocode CurrentSVL and CurrentNSVL which call
EL2Enabled()).
We were checking against ARM_FEATURE_EL2 rather than calling
arm_is_el2_enabled(), which meant that we would look at
SMCR_EL2/SVCR_EL2 when in Secure EL1 or Secure EL0 even if Secure EL2
was not enabled.
Use the correct check in sve_vqm1_for_el_sm().
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240722172957.1041231-5-peter.maydell@linaro.org
(cherry picked from commit f573ac059e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The function tszimm_esz() returns a shift amount, or possibly -1 in
certain cases that correspond to unallocated encodings in the
instruction set. We catch these later in the trans_ functions
(generally with an "a-esz < 0" check), but before we do the
decodetree-generated code will also call tszimm_shr() or tszimm_sl(),
which will use the tszimm_esz() return value as a shift count without
checking that it is not negative, which is undefined behaviour.
Avoid the UB by checking the return value in tszimm_shr() and
tszimm_shl().
Cc: qemu-stable@nongnu.org
Resolves: Coverity CID 1547617, 1547694
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240722172957.1041231-4-peter.maydell@linaro.org
(cherry picked from commit 76916dfa89)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The UMOPA/UMOPS instructions are supposed to multiply unsigned 8 or
16 bit elements and accumulate the products into a 64-bit element.
In the Arm ARM pseudocode, this is done with the usual
infinite-precision signed arithmetic. However our implementation
doesn't quite get it right, because in the DEF_IMOP_64() macro we do:
sum += (NTYPE)(n >> 0) * (MTYPE)(m >> 0);
where NTYPE and MTYPE are uint16_t or int16_t. In the uint16_t case,
the C usual arithmetic conversions mean the values are converted to
"int" type and the multiply is done as a 32-bit multiply. This means
that if the inputs are, for example, 0xffff and 0xffff then the
result is 0xFFFE0001 as an int, which is then promoted to uint64_t
for the accumulation into sum; this promotion incorrectly sign
extends the multiply.
Avoid the incorrect sign extension by casting to int64_t before
the multiply, so we do the multiply as 64-bit signed arithmetic,
which is a type large enough that the multiply can never
overflow into the sign bit.
(The equivalent 8-bit operations in DEF_IMOP_32() are fine, because
the 8-bit multiplies can never overflow into the sign bit of a
32-bit integer.)
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2372
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240722172957.1041231-3-peter.maydell@linaro.org
(cherry picked from commit ea3f5a90f0)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
For an instruction which accesses a 128-bit element tile when
the SVL is also 128 (for example MOV z0.Q, p0/M, ZA0H.Q[w0,0]),
we will assert in get_tile_rowcol():
qemu-system-aarch64: ../../tcg/tcg-op.c:926: tcg_gen_deposit_z_i32: Assertion `len > 0' failed.
This happens because we calculate
len = ctz32(streaming_vec_reg_size(s)) - esz;$
but if the SVL and the element size are the same len is 0, and
the deposit operation asserts.
In this case the ZA storage contains exactly one 128 bit
element ZA tile, and the horizontal or vertical slice is just
that tile. This means that regardless of the index value in
the Ws register, we always access that tile. (In pseudocode terms,
we calculate (index + offset) MOD 1, which is 0.)
Special case the len == 0 case to avoid hitting the assertion
in tcg_gen_deposit_z_i32().
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240722172957.1041231-2-peter.maydell@linaro.org
(cherry picked from commit 56f1c0db92)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The documentation of the "Set palette" mailbox property at
https://github.com/raspberrypi/firmware/wiki/Mailbox-property-interface#set-palette
says it has the form:
Length: 24..1032
Value:
u32: offset: first palette index to set (0-255)
u32: length: number of palette entries to set (1-256)
u32...: RGBA palette values (offset to offset+length-1)
We get this wrong in a couple of ways:
* we aren't checking the offset and length are in range, so the guest
can make us spin for a long time by providing a large length
* the bounds check on our loop is wrong: we should iterate through
'length' palette entries, not 'length - offset' entries
Fix the loop to implement the bounds checks and get the loop
condition right. In the process, make the variables local to
this switch case, rather than function-global, so it's clearer
what type they are when reading the code.
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240723131029.1159908-2-peter.maydell@linaro.org
(cherry picked from commit 0892fffc2a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fix due to lack of
v9.0.0-1812-g5d5f1b60916a "hw/misc: Implement mailbox properties for customer OTP and device specific private keys"
v8.0.0-1924-g251918266666 "hw/misc/bcm2835_property: Use 'raspberrypi-fw-defs.h' definitions"
also remove now-unused local `n' variable which gets removed in the next change in this file,
v9.0.0-2720-g32f1c201eedf "hw/misc/bcm2835_property: Avoid overflow in OTP access properties")
When a bare-metal application on the raspi3 board reads the
AUX_MU_STAT_REG MMIO register while the device's buffer is
at full receive FIFO capacity
(i.e. `s->read_count == BCM2835_AUX_RX_FIFO_LEN`) the
assertion `assert(s->read_count < BCM2835_AUX_RX_FIFO_LEN)`
fails.
Reported-by: Cryptjar <cryptjar@junk.studio>
Suggested-by: Cryptjar <cryptjar@junk.studio>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/459
Signed-off-by: Frederik van Hövell <frederik@fvhovell.nl>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
[PMM: commit message tweaks]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 546d574b11)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The unrealize functions of the various vhost-user devices are
calling the corresponding vhost_*_set_status() functions with a
status of 0 to shut down the device correctly.
Now these vhost_*_set_status() functions all follow this scheme:
bool should_start = virtio_device_should_start(vdev, status);
if (vhost_dev_is_started(&vvc->vhost_dev) == should_start) {
return;
}
if (should_start) {
/* ... do the initialization stuff ... */
} else {
/* ... do the cleanup stuff ... */
}
The problem here is virtio_device_should_start(vdev, 0) currently
always returns "true" since it internally only looks at vdev->started
instead of looking at the "status" parameter. Thus once the device
got started once, virtio_device_should_start() always returns true
and thus the vhost_*_set_status() functions return early, without
ever doing any clean-up when being called with status == 0. This
causes e.g. problems when trying to hot-plug and hot-unplug a vhost
user devices multiple times since the de-initialization step is
completely skipped during the unplug operation.
This bug has been introduced in commit 9f6bcfd99f ("hw/virtio: move
vm_running check to virtio_device_started") which replaced
should_start = status & VIRTIO_CONFIG_S_DRIVER_OK;
with
should_start = virtio_device_started(vdev, status);
which later got replaced by virtio_device_should_start(). This blocked
the possibility to set should_start to false in case the status flag
VIRTIO_CONFIG_S_DRIVER_OK was not set.
Fix it by adjusting the virtio_device_should_start() function to
only consider the status flag instead of vdev->started. Since this
function is only used in the various vhost_*_set_status() functions
for exactly the same purpose, it should be fine to fix it in this
central place there without any risk to change the behavior of other
code.
Fixes: 9f6bcfd99f ("hw/virtio: move vm_running check to virtio_device_started")
Buglink: https://issues.redhat.com/browse/RHEL-40708
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20240618121958.88673-1-thuth@redhat.com>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit d72479b117)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
aio_context_set_thread_pool_params() takes two int64_t arguments to
set the minimum and maximum number of threads in the pool. We do
some bounds checking on these, but we don't catch the case where the
inputs are negative. This means that later in the function when we
assign these inputs to the AioContext::thread_pool_min and
::thread_pool_max fields, which are of type int, the values might
overflow the smaller type.
A negative number of threads is meaningless, so make
aio_context_set_thread_pool_params() return an error if either min or
max are negative.
Resolves: Coverity CID 1547605
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240723150927.1396456-1-peter.maydell@linaro.org
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit 851495571d)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Loongson IPI is only available in little-endian,
so use that to access the guest memory (in case
we run on a big-endian host).
Cc: qemu-stable@nongnu.org
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Fixes: f6783e3438 ("hw/loongarch: Add LoongArch ipi interrupt support")
[PMD: Extracted from bigger commit, added commit description]
Co-Developed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Tested-by: Bibo Mao <maobibo@loongson.cn>
Acked-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Tested-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Message-Id: <20240718133312.10324-3-philmd@linaro.org>
(cherry picked from commit 2465c89fb9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: fixups for 7.2, for lack of:
v9.0.0-583-g91d0b151de4c "hw/intc/loongson_ipi: Implement IOCSR address space for MIPS"
v9.0.0-582-gb4a12dfc2132 "hw/intc/loongarch_ipi: Rename as loongson_ipi"
v8.2.0-545-gfdd6ee0b7653 "hw/intc/loongarch_ipi: Use MemTxAttrs interface for ipi ops")
If I use `-serial stdio` on Windows, after QEMU exits, the terminal
could not handle arrow keys and tab any more. Because stdio backend
on Windows sets console mode to virtual terminal input when starts,
but does not restore the old mode when finalize.
This small patch saves the old console mode and set it back.
Signed-off-by: Ziming Song <s.ziming@hotmail.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <ME3P282MB25488BE7C39BF0C35CD0DA5D8CA82@ME3P282MB2548.AUSP282.PROD.OUTLOOK.COM>
(cherry picked from commit 903cc9e117)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
sgx_epc_get_section assumes a PC platform is in use:
bool sgx_epc_get_section(int section_nr, uint64_t *addr, uint64_t *size)
{
PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
However, sgx_epc_get_section is called by CPUID regardless of whether
SGX state has been initialized or which platform is in use. Check
whether the machine has the right QOM class and if not behave as if
there are no EPC sections.
Fixes: 1dec2e1f19 ("i386: Update SGX CPUID info according to hardware/KVM/user input", 2021-09-30)
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2142
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 13be929aff)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The allocated memory to hold LBA ranges leaks in the nvme_dsm function. This
happens because the allocated memory for iocb->range is not freed in all
error handling paths.
Fix this by adding a free to ensure that the allocated memory is properly freed.
ASAN log:
==3075137==ERROR: LeakSanitizer: detected memory leaks
Direct leak of 480 byte(s) in 6 object(s) allocated from:
#0 0x55f1f8a0eddd in malloc llvm/compiler-rt/lib/asan/asan_malloc_linux.cpp:129:3
#1 0x7f531e0f6738 in g_malloc (/lib/x86_64-linux-gnu/libglib-2.0.so.0+0x5e738)
#2 0x55f1faf1f091 in blk_aio_get block/block-backend.c:2583:12
#3 0x55f1f945c74b in nvme_dsm hw/nvme/ctrl.c:2609:30
#4 0x55f1f945831b in nvme_io_cmd hw/nvme/ctrl.c:4470:16
#5 0x55f1f94561b7 in nvme_process_sq hw/nvme/ctrl.c:7039:29
Cc: qemu-stable@nongnu.org
Fixes: d7d1474fd8 ("hw/nvme: reimplement dsm to allow cancellation")
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
(cherry picked from commit c510fe78f1)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Michael Tokarev
Icenowy