Jeff Johnson says:
==================
ath.git patches for v7.2 (PR #4)
An assortment of cleanups and minor bug fixes across wcn36xx, ath9k,
ath10k, ath11k, and ath12k.
==================
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When growing a zoned RT section, the newly added RTGs also need to be
tagged as free in the radix tree and add to the nr_free_zones counters.
Call xfs_add_free_zone to do that, otherwise using up the newly added
space will wait for free zones forever.
Fixes: 01b71e64bb ("xfs: support growfs on zoned file systems")
Cc: stable@vger.kernel.org # v6.15
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Carlos Maiolino <cem@kernel.org>
Add a helper for adding a zone to the free pool in preparation of adding
another caller.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Carlos Maiolino <cem@kernel.org>
Add client->sva = NULL after the unbind makes it consistent with how
amdxdna_sva_fini() already clears the pointer after unbinding. The
IS_ERR_OR_NULL guard in sva_fini will then correctly skip the second
unbind.
Fixes: 3cc5d7a595 ("accel/amdxdna: Add carveout memory support for non-IOMMU systems")
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Link: https://patch.msgid.link/20260604202815.2425882-1-lizhi.hou@amd.com
When compile testing the UAPI headers with clang, there is an warning turned
error for using a C++ style ('//') comment, which is explicitly forbidden for
UAPI headers.
In file included from <built-in>:1:
./usr/include/linux/virtio_can.h:29:35: error: // comments are not allowed in this language [-Werror,-Wcomment]
29 | #define VIRTIO_CAN_MAX_DLEN 64 // this is like CANFD_MAX_DLEN
| ^
1 error generated.
Switch to a standard C style comment.
Fixes: 2b6b4bb7d96f ("can: virtio: Add virtio CAN driver")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260604-virtio_can-fix-uapi-comment-v1-1-199fa96ec5f0@kernel.org>
Recent QEMU versions added support for virtio SR-IOV emulation,
allowing virtio devices to expose SR-IOV VFs to the guest.
However, virtio_bus does not implement the num_vf callback of bus_type,
causing dev_num_vf() to return 0 for virtio devices even when
SR-IOV VFs are active.
net/core/rtnetlink.c calls dev_num_vf(dev->dev.parent) to populate
IFLA_NUM_VF in RTM_GETLINK responses. For a virtio-net device,
dev.parent points to the virtio_device, whose busis virtio_bus.
Without num_vf, SR-IOV VF information is silently
omitted from tools that rely on rtnetlink, such as 'ip link show'.
Add a num_vf callback that delegates to dev_num_vf(dev->parent),
which in turn reaches the underlying transport (pci_bus_type for
virtio-pci) where the actual VF count is tracked. Non-PCI transports
are unaffected as dev_num_vf() returns 0 when no num_vf callback is
present.
Signed-off-by: Yui Washizu <yui.washidu@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260310061454.683894-1-yui.washidu@gmail.com>
Look up the IRQ index in oct_hw->irqs instead of assuming
irq - irqs[0]. This supports non-contiguous IRQ numbers and
avoids incorrect ring indexing when irqs[0] is not the base.
Fixes: 26f8ce06af ("vdpa/octeon_ep: enable support for multiple interrupts per device")
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260224095226.1001151-5-schalla@marvell.com>
The mailbox address was computed assuming 1 ring per VF. Read the
actual rings-per-VF from OCTEP_EPF_RINFO and use it when calculating
OCTEP_PF_MBOX_DATA offsets, fixing VF initialization when rings
per VF > 1.
Fixes: 8b6c724cda ("virtio: vdpa: vDPA driver for Marvell OCTEON DPU devices")
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260224095226.1001151-2-schalla@marvell.com>
The only reason for this janitorial change is that this initialization
adds unnecessary noise to "git grep exit_signal".
args.exit_signal has no effect with CLONE_THREAD, not to mention it is
zero-initialized by the compiler anyway.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <acAIm732QPFZs15C@redhat.com>
The vhost driver has unnecessary empty module_init and
module_exit functions. Remove them. Note that if a module_init function
exists, a module_exit function must also exist; otherwise, the module
cannot be unloaded.
Signed-off-by: Ethan Nelson-Moore <enelsonmoore@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260131020010.45647-1-enelsonmoore@gmail.com>
The create mkey command memory embeds the MTT array as a flexible array
member. Use kvzalloc_flex() to allocate it directly instead of open-coding
the struct_size() calculation with kvcalloc().
The MTT allocation still needs to be aligned to MLX5_VDPA_MTT_ALIGN bytes.
Since each MTT entry is __be64, align the entry count directly and avoid
carrying a separate byte length variable.
Assisted-by: Codex:GPT-5.5
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260508051837.1744409-1-rosenp@gmail.com>
Driver core expects devices to be dynamically allocated and will, for
example, complain loudly when no release function has been provided.
Use root_device_register() to allocate and register the root device
instead of open coding using a static device.
Signed-off-by: Johan Hovold <johan@kernel.org>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260424104703.2619093-3-johan@kernel.org>
Driver core expects devices to be dynamically allocated and will, for
example, complain loudly when no release function has been provided.
Use root_device_register() to allocate and register the root device
instead of open coding using a static device.
Signed-off-by: Johan Hovold <johan@kernel.org>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260424104703.2619093-2-johan@kernel.org>
Add stubs for kmalloc_obj() and kmalloc_objs() to the tools/virtio
test harness, matching the new kernel allocator API. Also add the
DMA_ATTR_CPU_CACHE_CLEAN definition and include kernel.h from err.h
for the unlikely() macro.
Assisted-by: Claude:claude-opus-4-6
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <a0bd4b5bed56c49626c92a754d7aceab3325de25.1780520728.git.mst@redhat.com>
KCSAN reports data races when accessing virtio ring fields that are
concurrently written by the device (host). These are legitimate
concurrent accesses where the CPU reads fields that the device updates
via DMA-like mechanisms.
Add accessor functions that use READ_ONCE() to properly annotate these
device-writable fields and prevent compiler optimizations that could in
theory break the code. This also serves as documentation showing which
fields are shared with the device.
The affected fields are:
- Split ring: used->idx, used->ring[].id, used->ring[].len
- Packed ring: desc[].flags, desc[].id, desc[].len
This patch was partially written using the help of Kiro, an
AI coding assistant, to automate the mechanical work of generating the
inline function definition.
Signed-off-by: Alexander Graf <graf@amazon.com>
[jth: Add READ_ONCE in virtqueue_kick_prepare_split ]
Co-developed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Alexander Graf <graf@amazon.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260131102810.1254845-1-johannes.thumshirn@wdc.com>
These two ioctls are incompatible on 32-bit x86 userspace, because
the data structures are shorter than they are on 64-bit.
Add a proper .compat_ioctl handler for x86 that reads the structures
with the smaller padding before calling the internal handlers. On
all other architectures, CONFIG_COMPAT_FOR_U64_ALIGNMENT is disabled
and no special handling is required.
Fixes: ad146355bf ("vduse: Support querying information of IOVA regions")
Fixes: c8a6153b6c ("vduse: Introduce VDUSE - vDPA Device in Userspace")
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260213154051.4172275-1-arnd@kernel.org>
In parallel_test(), the return values of mmap() for both host_map and
guest_map are not checked against MAP_FAILED. If mmap() fails, the
subsequent code will dereference the invalid pointer, leading to a
segmentation fault.
Add MAP_FAILED checks after both mmap() calls, using err() to report
the error and exit, consistent with the existing error handling style
in this file (e.g., the open() call on line 149).
Fixes: 1515c5ce26 ("tools/virtio: add vring_test.")
Signed-off-by: longlong yan <yanlonglong@kylinos.cn>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260605021446.1611-1-yanlonglong@kylinos.cn>
vhost-net initializes one ubuf_info per outstanding zerocopy TX
descriptor and hands it to the backend socket. The networking stack may
then clone a zerocopy skb before all skb references are released. For
example, batman-adv fragmentation reaches skb_split(), which calls
skb_zerocopy_clone() and increments the same ubuf_info refcount.
vhost_zerocopy_complete() currently treats every ubuf callback as a
completed vhost descriptor. It dereferences ubuf->ctx, writes the
descriptor completion state, and drops the vhost_net_ubuf_ref even when
the callback only releases a cloned skb reference. A backend reset can
therefore wait for and free the vhost_net_ubuf_ref while another cloned
skb still carries the same ubuf_info. A later completion then
dereferences the freed ubufs pointer.
KASAN reports the stale completion as:
BUG: KASAN: slab-use-after-free in vhost_zerocopy_complete+0x1d7/0x1f0
BUG: KASAN: slab-use-after-free in vhost_zerocopy_complete+0x101/0x1f0
vhost_zerocopy_complete
skb_copy_ubufs
__dev_forward_skb2
veth_xmit
The freed object was allocated from vhost_net_ioctl() while setting the
backend and freed through kfree_rcu()/kvfree_rcu_bulk after backend
removal, while delayed skb completion still reached
vhost_zerocopy_complete().
Honor the generic ubuf_info refcount before touching vhost state, and run
the vhost descriptor completion only for the final ubuf reference. This
matches the msg_zerocopy_complete() ownership rule for cloned zerocopy
skbs.
Fixes: bab632d69e ("vhost: vhost TX zero-copy support")
Signed-off-by: Qing Ming <a0yami@mailbox.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260601104300.197210-1-a0yami@mailbox.org>
There is one race case in vduse_dev_msg_sync and vduse_dev_read_iter:
vduse_dev_read_iter():
lock(msg_lock);
dequeue_msg(send_list);
unlock(msg_lock);
vduse_dev_msg_sync():
wait_timeout() finish
lock(msg_lock);
check msg->complete is false
list_del(msg); <- double list_del() crash!
To fix this case, we shall ensure vduse_msg is on send_list or recv_list
outside the msg_lock critical section.
Fixes: c8a6153b6c ("vduse: Introduce VDUSE - vDPA Device in Userspace")
Cc: stable@vger.kernel.org
Signed-off-by: Zhang Tianci <zhangtianci.1997@bytedance.com>
Reviewed-by: Xie Yongji <xieyongji@bytedance.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260226115550.1814-3-zhangtianci.1997@bytedance.com>
Improve MAC address handling in mlx5_vdpa_set_attr() to ensure that
old MAC entries are properly removed from the MPFS table before
adding a new one. The new MAC address is then added to both the MPFS
and VLAN tables.
This change fixes an issue where the updated MAC address would not
take effect until QEMU was rebooted.
Signed-off-by: Cindy Lu <lulu@redhat.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260126094848.9601-4-lulu@redhat.com>
Add logic in mlx5_vdpa_set_attr() to ensure the VIRTIO_NET_F_MAC
feature bit is properly set only when the device is not yet in
the DRIVER_OK (running) state.
This makes the MAC address visible in the output of:
vdpa dev config show -jp
when the device is created without an initial MAC address.
Signed-off-by: Cindy Lu <lulu@redhat.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260126094848.9601-2-lulu@redhat.com>
dev_set_name() may fail and return an error, but its return value
is currently ignored and overwritten by _vdpa_register_device().
Abort device creation if dev_set_name() fails and release the
device reference to avoid continuing with an improperly initialized
struct device.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Signed-off-by: Evgenii Burenchev <evg28bur@yandex.ru>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Zhu Lingshan <lingshan.zhu@kernel.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260226152924.38790-1-evg28bur@yandex.ru>
Previously, the size was only read upon receiving the config interrupt.
This interrupt is sent when the size changes. However, we also need to
read the initial size.
Also make sure to only read the size from config if F_SIZE is enabled.
Fixes: 9778829cff ("virtio: console: Store each console's size in the console structure")
Signed-off-by: Filip Hejsek <filip.hejsek@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260223-virtio-console-fix-v1-1-0cf08303b428@gmail.com>
virtio_device_restore() resets the device and restores the negotiated
features before calling ->restore(). viortc_freeze() intentionally
leaves the existing virtqueues in place so the alarm queue can still
wake the system, but viortc_restore() immediately calls
viortc_init_vqs() without first deleting those old queues.
If virtqueue reinitialization fails on virtio-pci, the transport error
path can run vp_del_vqs() against a newly allocated vp_dev->vqs array
while vdev->vqs still contains the old virtqueues. vp_del_vqs() then
looks up queue state through the new array and can dereference a NULL
info pointer in vp_del_vq(), crashing the guest kernel during restore.
This can also happen during a non-faulty reinitialization, when one of
the vp_find_vqs_msix() attempts is unsuccessful before a later attempt
would succeed.
Delete the stale virtqueues before rebuilding them. If restore fails
before virtio_device_ready(), reuse the remove path to stop the device.
Once the device is ready, return errors directly instead of deleting the
virtqueues again.
Fixes: 0623c75927 ("virtio_rtc: Add module and driver core")
Signed-off-by: Jia Jia <physicalmtea@gmail.com>
Reviewed-by: Peter Hilber <peter.hilber@oss.qualcomm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260507120801.3677552-1-physicalmtea@gmail.com>
Driver core expects devices to be allocated dynamically and complains
loudly when a device that lacks a release function is freed.
Use __root_device_register() to allocate and register the root device
instead of open coding using a static device.
Note that root_device_register(), which also creates a link to the
module, cannot be used as the device is registered when parsing the
module parameters which happens before the module kobject has been set
up.
Fixes: 81a054ce0b ("virtio-mmio: Devices parameter parsing")
Cc: stable@vger.kernel.org # 3.5
Cc: Pawel Moll <pawel.moll@arm.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260427143710.14702-1-johan@kernel.org>
vhost_vdpa_mmap() and vhost_vdpa_fault() use vma->vm_pgoff as a
virtqueue index for get_vq_notification(), but they do not validate
that the index is smaller than v->nvqs.
The ioctl path already performs both a bounds check and
array_index_nospec(), but the mmap/fault path only checks that the
index fits in u16. This allows an out-of-range queue index to reach
driver-specific get_vq_notification() callbacks.
Fix this by extracting a unified vhost_vdpa_get_vq_notification()
helper that validates the queue index against v->nvqs and applies
array_index_nospec() before calling the driver callback. Both the
mmap and fault paths use this helper, and the bounds checking is
consolidated into a single location.
From source inspection, the most defensible impact is out-of-bounds
access in the callback path, potentially leading to invalid PFN
remaps and crash/DoS.
Fixes: ddd89d0a05 ("vhost_vdpa: support doorbell mapping via mmap")
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Qihang Tang <q.h.hack.winter@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260508075821.92656-1-q.h.hack.winter@gmail.com>
vduse_dev_open() looks up struct vduse_dev through the IDR and then
acquires dev->lock only after vduse_lock has been dropped.
This leaves a window where a concurrent VDUSE_DESTROY_DEV can remove the
same object from the IDR and free it before the open path locks the
device, leading to a use-after-free.
Close this race by keeping vduse_lock held until dev->lock has been
acquired in the open path, matching the lock ordering already used by
the destroy path.
Fixes: c8a6153b6c ("vduse: Introduce VDUSE - vDPA Device in Userspace")
Signed-off-by: Qihang Tang <q.h.hack.winter@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260508094659.94647-1-q.h.hack.winter@gmail.com>
When the host initiates an AF_VSOCK connect() to a guest that has not
yet loaded the virtio-vsock transport (i.e. still booting), the caller
blocks for VSOCK_DEFAULT_CONNECT_TIMEOUT.
A caller that wants to know if the guest is up yet instead of waiting
could theoretically tune SO_VM_SOCKETS_CONNECT_TIMEOUT, but it's tricky
to find the right timeout, if not impossible: there's no way to
distinguish "guest won't reply because it's not up yet" vs "guest is up
and tried to reply, but was too slow".
Furthermore, this delay is pointless:
- If the guest doesn't initialize within this timeout, connect()
returns ETIMEDOUT.
- If the guest **does** initialize, it'll reply with RST immediately,
because there won't be a listener on the port yet; connect() returns
ECONNRESET.
That's also inconsistent with the behavior at other initialization
stages: if a connection is attempted when the guest driver is already
loaded, but nothing is listening yet, we return ECONNRESET immediately
without waiting.
Fix this by checking the RX virtqueue backend in
vhost_transport_send_pkt() before queuing. If it's NULL, return
-EHOSTUNREACH immediately.
Callers that used to get ETIMEDOUT will now usually get EHOSTUNREACH.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Co-developed-by: Polina Vishneva <polina.vishneva@virtuozzo.com>
Signed-off-by: Polina Vishneva <polina.vishneva@virtuozzo.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260513145842.809404-1-polina.vishneva@virtuozzo.com>
Commit bee8c7c24b ("virtio: introduce map ops in virtio core") and
commit b16060c5c7 ("virtio: introduce virtio_map container union")
added 'map' and 'vmap' members to struct virtio_device but did not
update the kernel-doc comment block. This caused 'make htmldocs' to
emit warnings:
./include/linux/virtio.h:188 struct member 'map' not described in 'virtio_device'
./include/linux/virtio.h:188 struct member 'vmap' not described in 'virtio_device'
Add the missing entries in struct-declaration order to match the
existing convention in the file. After this patch, 'make htmldocs'
no longer emits these warnings.
Fixes: bee8c7c24b ("virtio: introduce map ops in virtio core")
Fixes: b16060c5c7 ("virtio: introduce virtio_map container union")
Reported-by: Luis Felipe Hernandez <luis.hernandez093@gmail.com>
Signed-off-by: Christian Fontanez <christfontanez@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260519013321.32511-1-christfontanez@gmail.com>
The Kconfig logic for selecting the scheduler clocksource on
NXP Vybrid (VF610) uses a `choice` block restricted to 32-bit ARM. This
prevents 64-bit architectures, such as the NXP S32 family, from enabling
the NXP Periodic Interrupt Timer (PIT) driver (CONFIG_NXP_PIT_TIMER).
Relocate the NXP clocksource selection from arch/arm/mach-imx/Kconfig to
drivers/clocksource/Kconfig. This allows the configuration to be shared
across different architectures.
Update the selection to include support for ARCH_S32 and add a "None"
option restricted to ARCH_S32, since Vybrid lacks the ARM Architected
Timer. The Vybrid Global Timer option is restricted to ARCH_MULTI_V7
SOC_VF610 platforms to prevent it from being visible on Cortex-M4 builds,
which lack the ARM Global Timer hardware.
Fixes: bee33f22d7 ("clocksource/drivers/nxp-pit: Add NXP Automotive s32g2 / s32g3 support")
Signed-off-by: Enric Balletbo i Serra <eballetb@redhat.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@kernel.org>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Link: https://patch.msgid.link/20260514-fix-nxp-timer-v3-1-a3e68fdb505e@redhat.com
Tegra SoCs supports multiple watchdog timers. If the kernel crashes or
hangs before userspace enables a watchdog, the system cannot recover and
may remain bricked, e.g. after a failed OTA update. The driver currently
leaves all watchdogs disabled until userspace configures them.
Reserve first available watchdog as a kernel-only watchdog for Tegra186
and Tegra234. Arm it during probe (120s timeout) and keep it alive in
the driver IRQ handler. Do not register it to userspace. Other available
watchdogs remain exposed to userspace. This guarantees the system can
reset itself in case of a hang or crash even when userspace never starts.
Signed-off-by: Kartik Rajput <kkartik@nvidia.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@kernel.org>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260507154557.2082697-5-kkartik@nvidia.com
Tegra SoCs support multiple watchdogs; currently only one (WDT0) is
used. When multiple watchdogs are registered, tegra186_wdt_enable()
overwrites the TKEIE(x) register, discarding any existing watchdog
interrupt enable bits. As a result, enabling one watchdog inadvertently
disables interrupts for the others.
Fix this by preserving the existing TKEIE(x) value and updating it
using a read-modify-write sequence.
Fixes: 42cee19a9f ("clocksource: Add Tegra186 timers support")
Cc: stable@vger.kernel.org
Signed-off-by: Kartik Rajput <kkartik@nvidia.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@kernel.org>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260507154557.2082697-2-kkartik@nvidia.com
Add focused map_kptr coverage for BPF-side map updates that touch values
containing referenced kptrs.
The new syscall programs stash the testmod refcounted object in an array
map, a preallocated hash map, and a no-prealloc hash map, then update the
same map from BPF. The refcount must remain elevated after the update,
while the userspace runner destroys the skeleton and reuses the existing
refcount wait to confirm map teardown releases the kptr.
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20260609202548.3571690-5-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add task_kfunc failure cases for bpf_obj_drop() on local objects with
referenced kptr fields from tracing and NMI tracing programs. These programs
must be rejected because dropping the object would run full special-field
destruction synchronously in an unsafe context.
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20260609202548.3571690-4-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Map update and delete paths currently call bpf_obj_free_fields() when a
value is being replaced or recycled. That makes field destruction depend
on the context of the update/delete operation. For tracing programs this
can include NMI context, where referenced kptr destructors, uptr
unpinning, and graph root destruction are not generally safe.
Introduce bpf_obj_cancel_fields() for the reusable-value path. It only
performs NMI-safe cleanup for timer, workqueue, and task_work fields.
Fields that need full destruction are left attached to the recycled value
and are destroyed by the final cleanup path instead.
Switch array and hashtab update/delete/recycle paths to this cancel
helper. Keep bpf_obj_free_fields() for final map destruction and for
bpf_mem_alloc destructors. Preallocated hashtabs do not have allocator
destructors, so teardown continues to walk the normal and extra elements
and fully destroy their fields.
This deliberately relaxes the eager-free semantics of map update/delete
for special fields. Programs that relied on a recycled map slot becoming
empty immediately after update/delete were relying on behavior that
cannot be implemented safely from every BPF execution context without
offloading arbitrary destructors.
There is a chance this change breaks programs making assumptions
regarding the eager freeing of fields. If so, we can relax semantics to
cancellation only when irqs_disabled() is true in the future. However,
theoretically, map values that get reused eagerly already have weaker
guarantees as parallel users can recreate freed fields before the new
element becomes visible again.
Fixes: 14a324f6a6 ("bpf: Wire up freeing of referenced kptr")
Signed-off-by: Justin Suess <utilityemal77@gmail.com>
Co-developed-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20260609202548.3571690-3-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
bpf_obj_drop() runs bpf_obj_free_fields() synchronously for
program-allocated objects. When such an object contains NMI unsafe
fields, tracing programs that can run from arbitrary instrumented
context can reach that destruction from unsafe contexts, including NMI.
NMI is likely one instance of this problem, and other instances would
include possible unsafe reentrancy. Deferring bpf_obj_drop() is not
appealing either: it would add delayed-free machinery to a release
operation that otherwise has straightforward synchronous ownership
semantics.
Reject bpf_obj_drop() and bpf_percpu_obj_drop() from tracing programs
that may run from unsafe contexts unless every field in the object's BTF
record is explicitly NMI safe. Do not reject sleepable
BPF_PROG_TYPE_TRACING programs, since they are not the arbitrary/NMI
contexts that motivate the restriction.
Note that while bpf_rb_root and bpf_list_head would be NMI safe on their
own to free, the objects recursively held by them may not be; be
conservative and just mark them as not NMI safe for now.
Use a whitelist for the NMI-safe field set instead of listing only known
NMI unsafe fields. Locks, async fields, unreferenced kptrs, and
refcounts are known to be NMI safe because their destruction is either a
no-op, simple state reset, or async cancellation. Referenced kptrs,
percpu referenced kptrs, uptrs, graph roots, graph nodes, and any future
field type are rejected until audited for arbitrary tracing and NMI
contexts. This is less susceptible to future changes in fields that were
previously safe by exclusion, and to new fields being added without
updating this check.
Convert the existing recursive local-object drop success case to a
syscall program in the same commit, since this verifier change makes the
old tracing program form invalid. The test still exercises
bpf_obj_drop() releasing a referenced task kptr from a safe program
type.
Fixes: ac9f06050a ("bpf: Introduce bpf_obj_drop")
Signed-off-by: Justin Suess <utilityemal77@gmail.com>
Co-developed-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20260609202548.3571690-2-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Yonghong Song says:
====================
selftests/bpf: Fix tests for llvm23 true signature
LLVM23 ([1]) records the 'true' function signature in BTF, i.e. the
signature inferred after optimization rather than the one written in C.
This caused two kinds of selftest failures (see below).
Case 1: keep int return type for tailcall subprogs
The verifier requires any subprog that issues a bpf_tail_call to return
an 'int' (see check_btf_func() in kernel/bpf/check_btf.c, which rejects
it with "tail_call is only allowed in functions that return 'int'").
Several tailcall subprogs do 'return 0' (or another constant) whose
result no caller uses. With llvm23 the compiler folds the constant and,
since the return value is dead, optimizes the subprog to effectively
return 'void' and records 'void' in BTF, so the program fails to load.
Use barrier_var() and __sink() to prevent returned value from being
optimized.
Case 2: adjust tracing prog ctx layout for the true signature
test_pkt_access_subprog2() has an unused argument that llvm optimizes
away. Before llvm23 the BTF signature did not match the optimized
assembly, so the verifier fell back to MAX_BPF_FUNC_REG_ARGS (5) u64
arguments and the fexit return value sat after args[5]. With llvm23 the
true signature has a single argument, so the return value moves to the
slot after args[1]. Select the matching ctx struct based on __clang_major__
so the test works with both old and new llvm.
[1] https://github.com/llvm/llvm-project/pull/198426
Changelogs:
v1 -> v2:
- v1: https://lore.kernel.org/bpf/20260609163947.1717694-1-yonghong.song@linux.dev/
- Do not use bpf array map or bpf global var. Use __sink() instead.
====================
Link: https://patch.msgid.link/20260609233402.2711071-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Johannes Berg
Christoph Hellwig
Lizhi Hou
Nathan Chancellor
Matias Ezequiel Vara Larsen
Yui Washizu
Huacai Chen
Srujana Challa
Vamsi Attunuru
Oleg Nesterov
Ethan Nelson-Moore
Rosen Penev
Johan Hovold
Maurice Hieronymus
Michael S. Tsirkin
Alexander Graf
Arnd Bergmann
longlong yan
Qing Ming
Jason Wang
Zhang Tianci
Cindy Lu
Evgenii Burenchev
Filip Hejsek
Ethan Carter Edwards
Jia Jia
Qihang Tang
Denis V. Lunev
Christian Fontanez
Enric Balletbo i Serra
Kartik Rajput
Alexei Starovoitov
Kumar Kartikeya Dwivedi
Justin Suess