This patch lets user-space request a non-coherent memory
allocation during CREATE_BUFS and REQBUFS ioctl calls.
= CREATE_BUFS
struct v4l2_create_buffers has seven 4-byte reserved areas,
so reserved[0] is renamed to ->flags. The struct, thus, now
has six reserved 4-byte regions.
= CREATE_BUFS32
struct v4l2_create_buffers32 has seven 4-byte reserved areas,
so reserved[0] is renamed to ->flags. The struct, thus, now
has six reserved 4-byte regions.
= REQBUFS
We use one byte of a 4 byte ->reserved[1] member of struct
v4l2_requestbuffers. The struct, thus, now has reserved 3 bytes.
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Preparations for future V4L2_MEMORY_FLAG_NON_COHERENT support.
Extend the vb2_core_reqbufs() parameters list to accept requests'
->flags, which will be used for memory coherency configuration.
An attempt to allocate a buffer with coherency requirements
that do not match the queue's consistency model will fail.
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
By setting or clearing the V4L2_MEMORY_FLAG_NON_COHERENT flag
user-space should be able to hint vb2 that either non-coherent
(if supported) or coherent memory should be used for the buffer
allocation.
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
It would be less error prone if the default cache hints value
(we kzalloc() structs, so it's zeroed out by default) would be
to "always sync/flush" caches. Inverse and rename cache hints
flags.
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
With the new DMA API we need an extension of the videobuf2 API.
Previously, videobuf2 core would set the non-coherent DMA bit
in the vb2_queue dma_attr field (if user-space would pass a
corresponding memory hint); the vb2 core then would pass the
vb2_queue dma_attrs to the vb2 allocators. The vb2 allocator
would use the queue's dma_attr and the DMA API would allocate
either coherent or non-coherent memory.
But we cannot do this anymore, since there is no corresponding DMA
attr flag and, hence, there is no way for the allocator to become
aware of what type of allocation user-space has requested. So we
need to pass more context from videobuf2 core to the allocators.
Fix this by changing the call_ptr_memop() macro to pass the
vb2 pointer to the corresponding op callbacks.
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
We add a new control V4L2_CID_NOTIFY_GAINS which allows the sensor to
be notified what gains will be applied to the different colour
channels by subsequent processing (such as by an ISP), even though the
sensor will not apply any of these gains itself.
For Bayer sensors this will be an array control taking 4 values which
are the 4 gains arranged in the fixed order B, Gb, Gr and R,
irrespective of the exact Bayer order of the sensor itself. The use of
an array makes it straightforward to extend this control to non-Bayer
sensors (for example, sensors with an RGBW pattern) in future.
The units are in all cases linear with the default value indicating a
gain of exactly 1.0. For example, if the default value were reported as
128 then the value 192 would represent a gain of exactly 1.5.
Signed-off-by: David Plowman <david.plowman@raspberrypi.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Add Mediatek's non-compressed 8 bit block video mode. This format is
produced by the MT8183 codec and can be converted to a non-proprietary
format by the MDP3 component.
Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@google.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Add more information about V4L2_PIX_FMT_NV12MT and
V4L2_PIX_FMT_NV12M_16X16, so it's clearer for driver authors and users.
Also, group the two pixel formats with the other tiled formats,
for clarity.
Unlike the recently introduced tiled formats (V4L2_PIX_FMT_NV12_4L4, etc)
these formats have remained Samsung-specific until now. Therefore, and
although the NV12MT and NV12MT_16X16 nomenclatures are less clear, we are
keeping them as-is.
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The V4L2_PIX_FMT_HM12 format is actually a simple NV12 tiled format,
with 16x16 linear tiles. Rename the format and move its documentation
together with the other tiled NV12 formats.
Keep V4L2_PIX_FMT_HM12 for application compatibility.
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The V4L2_PIX_FMT_SUNXI_TILED_NV12 format is actually a fairly
common NV12 tiled format, with 32x32 linear tiles. Rename the format
and move its documentation together with the other tiled NV12 formats.
Keep V4L2_PIX_FMT_SUNXI_TILED_NV12 for application compatibility.
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Rename V4L2 async notifier functions, replacing "notifier" with "nf" and
removing "_subdev" at the end of the function names adding subdevs as you
can only add subdevs to a notifier. Also wrap and otherwise clean up long
lines.
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com> (imx7)
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Pull x86 fixes from Thomas Gleixner:
"A set of fixes for X86:
- Prevent sending the wrong signal when protection keys are enabled
and the kernel handles a fault in the vsyscall emulation.
- Invoke early_reserve_memory() before invoking e820_memory_setup()
which is required to make the Xen dom0 e820 hooks work correctly.
- Use the correct data type for the SETZ operand in the EMQCMDS
instruction wrapper.
- Prevent undefined behaviour to the potential unaligned accesss in
the instruction decoder library"
* tag 'x86-urgent-2021-09-26' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/insn, tools/x86: Fix undefined behavior due to potential unaligned accesses
x86/asm: Fix SETZ size enqcmds() build failure
x86/setup: Call early_reserve_memory() earlier
x86/fault: Fix wrong signal when vsyscall fails with pkey
Pull irq fixes from Thomas Gleixner:
"A set of fixes for interrupt chip drivers:
- Work around a bad GIC integration on a Renesas platform which can't
handle byte-sized MMIO access
- Plug a potential memory leak in the GICv4 driver
- Fix a regression in the Armada 370-XP IPI code which was caused by
issuing EOI instack of ACK.
- A couple of small fixes here and there"
* tag 'irq-urgent-2021-09-26' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
irqchip/gic: Work around broken Renesas integration
irqchip/renesas-rza1: Use semicolons instead of commas
irqchip/gic-v3-its: Fix potential VPE leak on error
irqchip/goldfish-pic: Select GENERIC_IRQ_CHIP to fix build
irqchip/mbigen: Repair non-kernel-doc notation
irqdomain: Change the type of 'size' in __irq_domain_add() to be consistent
irqchip/armada-370-xp: Fix ack/eoi breakage
Documentation: Fix irq-domain.rst build warning
Merge misc fixes from Andrew Morton:
"16 patches.
Subsystems affected by this patch series: xtensa, sh, ocfs2, scripts,
lib, and mm (memory-failure, kasan, damon, shmem, tools, pagecache,
debug, and pagemap)"
* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
mm: fix uninitialized use in overcommit_policy_handler
mm/memory_failure: fix the missing pte_unmap() call
kasan: always respect CONFIG_KASAN_STACK
sh: pgtable-3level: fix cast to pointer from integer of different size
mm/debug: sync up latest migrate_reason to migrate_reason_names
mm/debug: sync up MR_CONTIG_RANGE and MR_LONGTERM_PIN
mm: fs: invalidate bh_lrus for only cold path
lib/zlib_inflate/inffast: check config in C to avoid unused function warning
tools/vm/page-types: remove dependency on opt_file for idle page tracking
scripts/sorttable: riscv: fix undeclared identifier 'EM_RISCV' error
ocfs2: drop acl cache for directories too
mm/shmem.c: fix judgment error in shmem_is_huge()
xtensa: increase size of gcc stack frame check
mm/damon: don't use strnlen() with known-bogus source length
kasan: fix Kconfig check of CC_HAS_WORKING_NOSANITIZE_ADDRESS
mm, hwpoison: add is_free_buddy_page() in HWPoisonHandlable()
Pull SCSI fixes from James Bottomley:
"Thirty-three fixes, I'm afraid.
Essentially the build up from the last couple of weeks while I've been
dealling with Linux Plumbers conference infrastructure issues. It's
mostly the usual assortment of spelling fixes and minor corrections.
The only core relevant changes are to the sd driver to reduce the spin
up message spew and fix a small memory leak on the freeing path"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (33 commits)
scsi: ses: Retry failed Send/Receive Diagnostic commands
scsi: target: Fix spelling mistake "CONFLIFT" -> "CONFLICT"
scsi: lpfc: Fix gcc -Wstringop-overread warning, again
scsi: lpfc: Use correct scnprintf() limit
scsi: lpfc: Fix sprintf() overflow in lpfc_display_fpin_wwpn()
scsi: core: Remove 'current_tag'
scsi: acornscsi: Remove tagged queuing vestiges
scsi: fas216: Kill scmd->tag
scsi: qla2xxx: Restore initiator in dual mode
scsi: ufs: core: Unbreak the reset handler
scsi: sd_zbc: Support disks with more than 2**32 logical blocks
scsi: ufs: core: Revert "scsi: ufs: Synchronize SCSI and UFS error handling"
scsi: bsg: Fix device unregistration
scsi: sd: Make sd_spinup_disk() less noisy
scsi: ufs: ufs-pci: Fix Intel LKF link stability
scsi: mpt3sas: Clean up some inconsistent indenting
scsi: megaraid: Clean up some inconsistent indenting
scsi: sr: Fix spelling mistake "does'nt" -> "doesn't"
scsi: Remove SCSI CDROM MAINTAINERS entry
scsi: megaraid: Fix Coccinelle warning
...
Pull xen fixes from Juergen Gross:
"Some minor cleanups and fixes of some theoretical bugs, as well as a
fix of a bug introduced in 5.15-rc1"
* tag 'for-linus-5.15b-rc3-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
xen/x86: fix PV trap handling on secondary processors
xen/balloon: fix balloon kthread freezing
swiotlb-xen: this is PV-only on x86
xen/pci-swiotlb: reduce visibility of symbols
PCI: only build xen-pcifront in PV-enabled environments
swiotlb-xen: ensure to issue well-formed XENMEM_exchange requests
Xen/gntdev: don't ignore kernel unmapping error
xen/x86: drop redundant zeroing from cpu_initialize_context()
Pull erofs fixes from Gao Xiang:
"Two bugfixes to fix the 4KiB blockmap chunk format availability and a
dangling pointer usage. There is also a trivial cleanup to clarify
compacted_2b if compacted_4b_initial > totalidx.
Summary:
- fix the dangling pointer use in erofs_lookup tracepoint
- fix unsupported chunk format check
- zero out compacted_2b if compacted_4b_initial > totalidx"
* tag 'erofs-for-5.15-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
erofs: clear compacted_2b if compacted_4b_initial > totalidx
erofs: fix misbehavior of unsupported chunk format check
erofs: fix up erofs_lookup tracepoint
Pull char/misc driver fixes from Greg KH:
"Here are some small char and misc driver fixes for 5.15-rc3.
Nothing huge in here, just fixes for a number of small issues that
have been reported. These include:
- habanalabs race conditions and other bugs fixed
- binder driver fixes
- fpga driver fixes
- coresight build warning fix
- nvmem driver fix
- comedi memory leak fix
- bcm-vk tty race fix
- other tiny driver fixes
All of these have been in linux-next for a while with no reported
issues"
* tag 'char-misc-5.15-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc: (21 commits)
comedi: Fix memory leak in compat_insnlist()
nvmem: NVMEM_NINTENDO_OTP should depend on WII
misc: bcm-vk: fix tty registration race
fpga: dfl: Avoid reads to AFU CSRs during enumeration
fpga: machxo2-spi: Fix missing error code in machxo2_write_complete()
fpga: machxo2-spi: Return an error on failure
habanalabs: expose a single cs seq in staged submissions
habanalabs: fix wait offset handling
habanalabs: rate limit multi CS completion errors
habanalabs/gaudi: fix LBW RR configuration
habanalabs: Fix spelling mistake "FEADBACK" -> "FEEDBACK"
habanalabs: fail collective wait when not supported
habanalabs/gaudi: use direct MSI in single mode
habanalabs: fix kernel OOPs related to staged cs
habanalabs: fix potential race in interrupt wait ioctl
mcb: fix error handling in mcb_alloc_bus()
misc: genwqe: Fixes DMA mask setting
coresight: syscfg: Fix compiler warning
nvmem: core: Add stubs for nvmem_cell_read_variable_le_u32/64 if !CONFIG_NVMEM
binder: make sure fd closes complete
...
Pull USB driver fixes from Greg KH:
"Here are some USB driver fixes and new device ids for 5.15-rc3.
They include:
- usb-storage quirk additions
- usb-serial new device ids
- usb-serial driver fixes
- USB roothub registration bugfix to resolve a long-reported issue
- usb gadget driver fixes for a large number of small things
- dwc2 driver fixes
All of these have been in linux-next for a while with no reported
issues"
* tag 'usb-5.15-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (28 commits)
USB: serial: option: add device id for Foxconn T99W265
USB: serial: cp210x: add ID for GW Instek GDM-834x Digital Multimeter
USB: serial: cp210x: add part-number debug printk
USB: serial: cp210x: fix dropped characters with CP2102
MAINTAINERS: usb, update Peter Korsgaard's entries
usb: musb: tusb6010: uninitialized data in tusb_fifo_write_unaligned()
usb-storage: Add quirk for ScanLogic SL11R-IDE older than 2.6c
Re-enable UAS for LaCie Rugged USB3-FW with fk quirk
USB: serial: option: remove duplicate USB device ID
USB: serial: mos7840: remove duplicated 0xac24 device ID
arm64: dts: qcom: ipq8074: remove USB tx-fifo-resize property
usb: gadget: f_uac2: Populate SS descriptors' wBytesPerInterval
usb: gadget: f_uac2: Add missing companion descriptor for feedback EP
usb: dwc2: gadget: Fix ISOC transfer complete handling for DDMA
usb: core: hcd: Modularize HCD stop configuration in usb_stop_hcd()
xhci: Set HCD flag to defer primary roothub registration
usb: core: hcd: Add support for deferring roothub registration
usb: dwc2: gadget: Fix ISOC flow for BDMA and Slave
usb: dwc3: core: balance phy init and exit
Revert "USB: bcma: Add a check for devm_gpiod_get"
...
Pull ACPI fix from Rafael Wysocki:
"Revert a recent commit related to memory management that turned out to
be problematic (Jia He)"
* tag 'acpi-5.15-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
Revert "ACPI: Add memory semantics to acpi_os_map_memory()"
Pull irqchip fixes from Marc Zyngier:
- Work around a bad GIC integration on a Renesas platform, where the
interconnect cannot deal with byte-sized MMIO accesses
- Cleanup another Renesas driver abusing the comma operator
- Fix a potential GICv4 memory leak on an error path
- Make the type of 'size' consistent with the rest of the code in
__irq_domain_add()
- Fix a regression in the Armada 370-XP IPI path
- Fix the build for the obviously unloved goldfish-pic
- Some documentation fixes
Link: https://lore.kernel.org/r/20210924090933.2766857-1-maz@kernel.org
This reverts commit 437b38c511.
The memory semantics added in commit 437b38c511 causes SystemMemory
Operation region, whose address range is not described in the EFI memory
map to be mapped as NormalNC memory on arm64 platforms (through
acpi_os_map_memory() in acpi_ex_system_memory_space_handler()).
This triggers the following abort on an ARM64 Ampere eMAG machine,
because presumably the physical address range area backing the Opregion
does not support NormalNC memory attributes driven on the bus.
Internal error: synchronous external abort: 96000410 [#1] SMP
Modules linked in:
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.14.0+ #462
Hardware name: MiTAC RAPTOR EV-883832-X3-0001/RAPTOR, BIOS 0.14 02/22/2019
pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[...snip...]
Call trace:
acpi_ex_system_memory_space_handler+0x26c/0x2c8
acpi_ev_address_space_dispatch+0x228/0x2c4
acpi_ex_access_region+0x114/0x268
acpi_ex_field_datum_io+0x128/0x1b8
acpi_ex_extract_from_field+0x14c/0x2ac
acpi_ex_read_data_from_field+0x190/0x1b8
acpi_ex_resolve_node_to_value+0x1ec/0x288
acpi_ex_resolve_to_value+0x250/0x274
acpi_ds_evaluate_name_path+0xac/0x124
acpi_ds_exec_end_op+0x90/0x410
acpi_ps_parse_loop+0x4ac/0x5d8
acpi_ps_parse_aml+0xe0/0x2c8
acpi_ps_execute_method+0x19c/0x1ac
acpi_ns_evaluate+0x1f8/0x26c
acpi_ns_init_one_device+0x104/0x140
acpi_ns_walk_namespace+0x158/0x1d0
acpi_ns_initialize_devices+0x194/0x218
acpi_initialize_objects+0x48/0x50
acpi_init+0xe0/0x498
If the Opregion address range is not present in the EFI memory map there
is no way for us to determine the memory attributes to use to map it -
defaulting to NormalNC does not work (and it is not correct on a memory
region that may have read side-effects) and therefore commit
437b38c511 should be reverted, which means reverting back to the
original behavior whereby address ranges that are mapped using
acpi_os_map_memory() default to the safe devicenGnRnE attributes on
ARM64 if the mapped address range is not defined in the EFI memory map.
Fixes: 437b38c511 ("ACPI: Add memory semantics to acpi_os_map_memory()")
Signed-off-by: Jia He <justin.he@arm.com>
Acked-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Pull rseq fixes from Paolo Bonzini:
"A fix for a bug with restartable sequences and KVM.
KVM's handling of TIF_NOTIFY_RESUME, e.g. for task migration, clears
the flag without informing rseq and leads to stale data in userspace's
rseq struct"
* tag 'for-linus-rseq' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: selftests: Remove __NR_userfaultfd syscall fallback
KVM: selftests: Add a test for KVM_RUN+rseq to detect task migration bugs
tools: Move x86 syscall number fallbacks to .../uapi/
entry: rseq: Call rseq_handle_notify_resume() in tracehook_notify_resume()
KVM: rseq: Update rseq when processing NOTIFY_RESUME on xfer to KVM guest
Pull networking fixes from Jakub Kicinski:
"Current release - regressions:
- dsa: bcm_sf2: fix array overrun in bcm_sf2_num_active_ports()
Previous releases - regressions:
- introduce a shutdown method to mdio device drivers, and make DSA
switch drivers compatible with masters disappearing on shutdown;
preventing infinite reference wait
- fix issues in mdiobus users related to ->shutdown vs ->remove
- virtio-net: fix pages leaking when building skb in big mode
- xen-netback: correct success/error reporting for the
SKB-with-fraglist
- dsa: tear down devlink port regions when tearing down the devlink
port on error
- nexthop: fix division by zero while replacing a resilient group
- hns3: check queue, vf, vlan ids range before using
Previous releases - always broken:
- napi: fix race against netpoll causing NAPI getting stuck
- mlx4_en: ensure link operstate is updated even if link comes up
before netdev registration
- bnxt_en: fix TX timeout when TX ring size is set to the smallest
- enetc: fix illegal access when reading affinity_hint; prevent oops
on sysfs access
- mtk_eth_soc: avoid creating duplicate offload entries
Misc:
- core: correct the sock::sk_lock.owned lockdep annotations"
* tag 'net-5.15-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (51 commits)
atlantic: Fix issue in the pm resume flow.
net/mlx4_en: Don't allow aRFS for encapsulated packets
net: mscc: ocelot: fix forwarding from BLOCKING ports remaining enabled
net: ethernet: mtk_eth_soc: avoid creating duplicate offload entries
nfc: st-nci: Add SPI ID matching DT compatible
MAINTAINERS: remove Guvenc Gulce as net/smc maintainer
nexthop: Fix memory leaks in nexthop notification chain listeners
mptcp: ensure tx skbs always have the MPTCP ext
qed: rdma - don't wait for resources under hw error recovery flow
s390/qeth: fix deadlock during failing recovery
s390/qeth: Fix deadlock in remove_discipline
s390/qeth: fix NULL deref in qeth_clear_working_pool_list()
net: dsa: realtek: register the MDIO bus under devres
net: dsa: don't allocate the slave_mii_bus using devres
Doc: networking: Fox a typo in ice.rst
net: dsa: fix dsa_tree_setup error path
net/smc: fix 'workqueue leaked lock' in smc_conn_abort_work
net/smc: add missing error check in smc_clc_prfx_set()
net: hns3: fix a return value error in hclge_get_reset_status()
net: hns3: check vlan id before using it
...
Invoke rseq_handle_notify_resume() from tracehook_notify_resume() now
that the two function are always called back-to-back by architectures
that have rseq. The rseq helper is stubbed out for architectures that
don't support rseq, i.e. this is a nop across the board.
Note, tracehook_notify_resume() is horribly named and arguably does not
belong in tracehook.h as literally every line of code in it has nothing
to do with tracing. But, that's been true since commit a42c6ded82
("move key_repace_session_keyring() into tracehook_notify_resume()")
first usurped tracehook_notify_resume() back in 2012. Punt cleaning that
mess up to future patches.
No functional change intended.
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210901203030.1292304-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Pull AFS fixes from David Howells:
"Fixes for AFS problems that can cause data corruption due to
interaction with another client modifying data cached locally:
- When d_revalidating a dentry, don't look at the inode to which it
points. Only check the directory to which the dentry belongs. This
was confusing things and causing the silly-rename cleanup code to
remove the file now at the dentry of a file that got deleted.
- Fix mmap data coherency. When a callback break is received that
relates to a file that we have cached, the data content may have
been changed (there are other reasons, such as the user's rights
having been changed). However, we're checking it lazily, only on
entry to the kernel, which doesn't happen if we have a writeable
shared mapped page on that file.
We make the kernel keep track of mmapped files and clear all PTEs
mapping to that file as soon as the callback comes in by calling
unmap_mapping_pages() (we don't necessarily want to zap the
pagecache). This causes the kernel to be reentered when userspace
tries to access the mmapped address range again - and at that point
we can query the server and, if we need to, zap the page cache.
Ideally, I would check each file at the point of notification, but
that involves poking the server[*] - which is holding an exclusive
lock on the vnode it is changing, waiting for all the clients it
notified to reply. This could then deadlock against the server.
Further, invalidating the pagecache might call ->launder_page(),
which would try to write to the file, which would definitely
deadlock. (AFS doesn't lease file access).
[*] Checking to see if the file content has changed is a matter of
comparing the current data version number, but we have to ask
the server for that. We also need to get a new callback promise
and we need to poke the server for that too.
- Add some more points at which the inode is validated, since we're
doing it lazily, notably in ->read_iter() and ->page_mkwrite(), but
also when performing some directory operations.
Ideally, checking in ->read_iter() would be done in some derivation
of filemap_read(). If we're going to call the server to read the
file, then we get the file status fetch as part of that.
- The above is now causing us to make a lot more calls to
afs_validate() to check the inode - and afs_validate() takes the
RCU read lock each time to make a quick check (ie.
afs_check_validity()). This is entirely for the purpose of checking
cb_s_break to see if the server we're using reinitialised its list
of callbacks - however this isn't a very common event, so most of
the time we're taking this needlessly.
Add a new cell-wide counter to count the number of
reinitialisations done by any server and check that - and only if
that changes, take the RCU read lock and check the server list (the
server list may change, but the cell a file is part of won't).
- Don't update vnode->cb_s_break and ->cb_v_break inside the validity
checking loop. The cb_lock is done with read_seqretry, so we might
go round the loop a second time after resetting those values - and
that could cause someone else checking validity to miss something
(I think).
Also included are patches for fixes for some bugs encountered whilst
debugging this:
- Fix a leak of afs_read objects and fix a leak of keys hidden by
that.
- Fix a leak of pages that couldn't be added to extend a writeback.
- Fix the maintenance of i_blocks when i_size is changed by a local
write or a local dir edit"
Link: https://bugzilla.kernel.org/show_bug.cgi?id=214217 [1]
Link: https://lore.kernel.org/r/163111665183.283156.17200205573146438918.stgit@warthog.procyon.org.uk/ # v1
Link: https://lore.kernel.org/r/163113612442.352844.11162345591911691150.stgit@warthog.procyon.org.uk/ # i_blocks patch
* tag 'afs-fixes-20210913' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs:
afs: Fix updating of i_blocks on file/dir extension
afs: Fix corruption in reads at fpos 2G-4G from an OpenAFS server
afs: Try to avoid taking RCU read lock when checking vnode validity
afs: Fix mmap coherency vs 3rd-party changes
afs: Fix incorrect triggering of sillyrename on 3rd-party invalidation
afs: Add missing vnode validation checks
afs: Fix page leak
afs: Fix missing put on afs_read objects and missing get on the key therein
Pull cifs client fixes from Steve French:
- two deferred close fixes (for bugs found with xfstests 478 and 461)
- a deferred close improvement in rename
- two trivial fixes for incorrect Linux comment formatting of multiple
cifs files (pointed out by automated kernel test robot and
checkpatch)
* tag '5.15-rc1-smb3' of git://git.samba.org/sfrench/cifs-2.6:
cifs: Not to defer close on file when lock is set
cifs: Fix soft lockup during fsstress
cifs: Deferred close performance improvements
cifs: fix incorrect kernel doc comments
cifs: remove pathname for file from SPDX header
The function __bad_area_nosemaphore() calls kernelmode_fixup_or_oops()
with the parameter @signal being actually @pkey, which will send a
signal numbered with the argument in @pkey.
This bug can be triggered when the kernel fails to access user-given
memory pages that are protected by a pkey, so it can go down the
do_user_addr_fault() path and pass the !user_mode() check in
__bad_area_nosemaphore().
Most cases will simply run the kernel fixup code to make an -EFAULT. But
when another condition current->thread.sig_on_uaccess_err is met, which
is only used to emulate vsyscall, the kernel will generate the wrong
signal.
Add a new parameter @pkey to kernelmode_fixup_or_oops() to fix this.
[ bp: Massage commit message, fix build error as reported by the 0day
bot: https://lkml.kernel.org/r/202109202245.APvuT8BX-lkp@intel.com ]
Fixes: 5042d40a26 ("x86/fault: Bypass no_context() for implicit kernel faults from usermode")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Jiashuo Liang <liangjs@pku.edu.cn>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lkml.kernel.org/r/20210730030152.249106-1-liangjs@pku.edu.cn
The code is unreachable for HVM or PVH, and it also makes little sense
in auto-translated environments. On Arm, with
xen_{create,destroy}_contiguous_region() both being stubs, I have a hard
time seeing what good the Xen specific variant does - the generic one
ought to be fine for all purposes there. Still Arm code explicitly
references symbols here, so the code will continue to be included there.
Instead of making PCI_XEN's "select" conditional, simply drop it -
SWIOTLB_XEN will be available unconditionally in the PV case anyway, and
is - as explained above - dead code in non-PV environments.
This in turn allows dropping the stubs for
xen_{create,destroy}_contiguous_region(), the former of which was broken
anyway - it failed to set the DMA handle output.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Link: https://lore.kernel.org/r/5947b8ae-fdc7-225c-4838-84712265fc1e@suse.com
Signed-off-by: Juergen Gross <jgross@suse.com>
Nathan Chancellor reports that the recent change to pci_iounmap in
commit 9caea00076 ("parisc: Declare pci_iounmap() parisc version only
when CONFIG_PCI enabled") causes build errors on arm64.
It took me about two hours to convince myself that I think I know what
the logic of that mess of #ifdef's in the <asm-generic/io.h> header file
really aim to do, and rewrite it to be easier to follow.
Famous last words.
Anyway, the code has now been lifted from that grotty header file into
lib/pci_iomap.c, and has fairly extensive comments about what the logic
is. It also avoids indirecting through another confusing (and badly
named) helper function that has other preprocessor config conditionals.
Let's see what odd architecture did something else strange in this area
to break things. But my arm64 cross build is clean.
Fixes: 9caea00076 ("parisc: Declare pci_iounmap() parisc version only when CONFIG_PCI enabled")
Reported-by: Nathan Chancellor <nathan@kernel.org>
Cc: Helge Deller <deller@gmx.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: Ulrich Teichert <krypton@ulrich-teichert.org>
Cc: James Bottomley <James.Bottomley@hansenpartnership.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull x86 fixes from Borislav Petkov:
- Prevent a infinite loop in the MCE recovery on return to user space,
which was caused by a second MCE queueing work for the same page and
thereby creating a circular work list.
- Make kern_addr_valid() handle existing PMD entries, which are marked
not present in the higher level page table, correctly instead of
blindly dereferencing them.
- Pass a valid address to sanitize_phys(). This was caused by the
mixture of inclusive and exclusive ranges. memtype_reserve() expect
'end' being exclusive, but sanitize_phys() wants it inclusive. This
worked so far, but with end being the end of the physical address
space the fail is exposed.
- Increase the maximum supported GPIO numbers for 64bit. Newer SoCs
exceed the previous maximum.
* tag 'x86_urgent_for_v5.15_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/mce: Avoid infinite loop for copy from user recovery
x86/mm: Fix kern_addr_valid() to cope with existing but not present entries
x86/platform: Increase maximum GPIO number for X86_64
x86/pat: Pass valid address to sanitize_phys()
Commit 86f8b1c01a ("net: dsa: Do not make user port errors fatal")
decided it was fine to ignore errors on certain ports that fail to
probe, and go on with the ports that do probe fine.
Commit fb6ec87f72 ("net: dsa: Fix type was not set for devlink port")
noticed that devlink_port_type_eth_set(dlp, dp->slave); does not get
called, and devlink notices after a timeout of 3600 seconds and prints a
WARN_ON. So it went ahead to unregister the devlink port. And because
there exists an UNUSED port flavour, we actually re-register the devlink
port as UNUSED.
Commit 08156ba430 ("net: dsa: Add devlink port regions support to
DSA") added devlink port regions, which are set up by the driver and not
by DSA.
When we trigger the devlink port deregistration and reregistration as
unused, devlink now prints another WARN_ON, from here:
devlink_port_unregister:
WARN_ON(!list_empty(&devlink_port->region_list));
So the port still has regions, which makes sense, because they were set
up by the driver, and the driver doesn't know we're unregistering the
devlink port.
Somebody needs to tear them down, and optionally (actually it would be
nice, to be consistent) set them up again for the new devlink port.
But DSA's layering stays in our way quite badly here.
The options I've considered are:
1. Introduce a function in devlink to just change a port's type and
flavour. No dice, devlink keeps a lot of state, it really wants the
port to not be registered when you set its parameters, so changing
anything can only be done by destroying what we currently have and
recreating it.
2. Make DSA cache the parameters passed to dsa_devlink_port_region_create,
and the region returned, keep those in a list, then when the devlink
port unregister needs to take place, the existing devlink regions are
destroyed by DSA, and we replay the creation of new regions using the
cached parameters. Problem: mv88e6xxx keeps the region pointers in
chip->ports[port].region, and these will remain stale after DSA frees
them. There are many things DSA can do, but updating mv88e6xxx's
private pointers is not one of them.
3. Just let the driver do it (i.e. introduce a very specific method
called ds->ops->port_reinit_as_unused, which unregisters its devlink
port devlink regions, then the old devlink port, then registers the
new one, then the devlink port regions for it). While it does work,
as opposed to the others, it's pretty horrible from an API
perspective and we can do better.
4. Introduce a new pair of methods, ->port_setup and ->port_teardown,
which in the case of mv88e6xxx must register and unregister the
devlink port regions. Call these 2 methods when the port must be
reinitialized as unused.
Naturally, I went for the 4th approach.
Fixes: 08156ba430 ("net: dsa: Add devlink port regions support to DSA")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
lock_sock_fast() and lock_sock_nested() contain lockdep annotations for the
sock::sk_lock.owned 'mutex'. sock::sk_lock.owned is not a regular mutex. It
is just lockdep wise equivalent. In fact it's an open coded trivial mutex
implementation with some interesting features.
sock::sk_lock.slock is a regular spinlock protecting the 'mutex'
representation sock::sk_lock.owned which is a plain boolean. If 'owned' is
true, then some other task holds the 'mutex', otherwise it is uncontended.
As this locking construct is obviously endangered by lock ordering issues as
any other locking primitive it got lockdep annotated via a dedicated
dependency map sock::sk_lock.dep_map which has to be updated at the lock
and unlock sites.
lock_sock_nested() is a straight forward 'mutex' lock operation:
might_sleep();
spin_lock_bh(sock::sk_lock.slock)
while (!try_lock(sock::sk_lock.owned)) {
spin_unlock_bh(sock::sk_lock.slock);
wait_for_release();
spin_lock_bh(sock::sk_lock.slock);
}
The lockdep annotation for sock::sk_lock.owned is for unknown reasons
_after_ the lock has been acquired, i.e. after the code block above and
after releasing sock::sk_lock.slock, but inside the bottom halves disabled
region:
spin_unlock(sock::sk_lock.slock);
mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_);
local_bh_enable();
The placement after the unlock is obvious because otherwise the
mutex_acquire() would nest into the spin lock held region.
But that's from the lockdep perspective still the wrong place:
1) The mutex_acquire() is issued _after_ the successful acquisition which
is pointless because in a dead lock scenario this point is never
reached which means that if the deadlock is the first instance of
exposing the wrong lock order lockdep does not have a chance to detect
it.
2) It only works because lockdep is rather lax on the context from which
the mutex_acquire() is issued. Acquiring a mutex inside a bottom halves
and therefore non-preemptible region is obviously invalid, except for a
trylock which is clearly not the case here.
This 'works' stops working on RT enabled kernels where the bottom halves
serialization is done via a local lock, which exposes this misplacement
because the 'mutex' and the local lock nest the wrong way around and
lockdep complains rightfully about a lock inversion.
The placement is wrong since the initial commit a5b5bb9a05 ("[PATCH]
lockdep: annotate sk_locks") which introduced this.
Fix it by moving the mutex_acquire() in front of the actual lock
acquisition, which is what the regular mutex_lock() operation does as well.
lock_sock_fast() is not that straight forward. It looks at the first glance
like a convoluted trylock operation:
spin_lock_bh(sock::sk_lock.slock)
if (!sock::sk_lock.owned)
return false;
while (!try_lock(sock::sk_lock.owned)) {
spin_unlock_bh(sock::sk_lock.slock);
wait_for_release();
spin_lock_bh(sock::sk_lock.slock);
}
spin_unlock(sock::sk_lock.slock);
mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_);
local_bh_enable();
return true;
But that's not the case: lock_sock_fast() is an interesting optimization
for short critical sections which can run with bottom halves disabled and
sock::sk_lock.slock held. This allows to shortcut the 'mutex' operation in
the non contended case by preventing other lockers to acquire
sock::sk_lock.owned because they are blocked on sock::sk_lock.slock, which
in turn avoids the overhead of doing the heavy processing in release_sock()
including waking up wait queue waiters.
In the contended case, i.e. when sock::sk_lock.owned == true the behavior
is the same as lock_sock_nested().
Semantically this shortcut means, that the task acquired the 'mutex' even
if it does not touch the sock::sk_lock.owned field in the non-contended
case. Not telling lockdep about this shortcut acquisition is hiding
potential lock ordering violations in the fast path.
As a consequence the same reasoning as for the above lock_sock_nested()
case vs. the placement of the lockdep annotation applies.
The current placement of the lockdep annotation was just copied from
the original lock_sock(), now renamed to lock_sock_nested(),
implementation.
Fix this by moving the mutex_acquire() in front of the actual lock
acquisition and adding the corresponding mutex_release() into
unlock_sock_fast(). Also document the fast path return case with a comment.
Reported-by: Sebastian Siewior <bigeasy@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: netdev@vger.kernel.org
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Lino reports that on his system with bcmgenet as DSA master and KSZ9897
as a switch, rebooting or shutting down never works properly.
What does the bcmgenet driver have special to trigger this, that other
DSA masters do not? It has an implementation of ->shutdown which simply
calls its ->remove implementation. Otherwise said, it unregisters its
network interface on shutdown.
This message can be seen in a loop, and it hangs the reboot process there:
unregister_netdevice: waiting for eth0 to become free. Usage count = 3
So why 3?
A usage count of 1 is normal for a registered network interface, and any
virtual interface which links itself as an upper of that will increment
it via dev_hold. In the case of DSA, this is the call path:
dsa_slave_create
-> netdev_upper_dev_link
-> __netdev_upper_dev_link
-> __netdev_adjacent_dev_insert
-> dev_hold
So a DSA switch with 3 interfaces will result in a usage count elevated
by two, and netdev_wait_allrefs will wait until they have gone away.
Other stacked interfaces, like VLAN, watch NETDEV_UNREGISTER events and
delete themselves, but DSA cannot just vanish and go poof, at most it
can unbind itself from the switch devices, but that must happen strictly
earlier compared to when the DSA master unregisters its net_device, so
reacting on the NETDEV_UNREGISTER event is way too late.
It seems that it is a pretty established pattern to have a driver's
->shutdown hook redirect to its ->remove hook, so the same code is
executed regardless of whether the driver is unbound from the device, or
the system is just shutting down. As Florian puts it, it is quite a big
hammer for bcmgenet to unregister its net_device during shutdown, but
having a common code path with the driver unbind helps ensure it is well
tested.
So DSA, for better or for worse, has to live with that and engage in an
arms race of implementing the ->shutdown hook too, from all individual
drivers, and do something sane when paired with masters that unregister
their net_device there. The only sane thing to do, of course, is to
unlink from the master.
However, complications arise really quickly.
The pattern of redirecting ->shutdown to ->remove is not unique to
bcmgenet or even to net_device drivers. In fact, SPI controllers do it
too (see dspi_shutdown -> dspi_remove), and presumably, I2C controllers
and MDIO controllers do it too (this is something I have not researched
too deeply, but even if this is not the case today, it is certainly
plausible to happen in the future, and must be taken into consideration).
Since DSA switches might be SPI devices, I2C devices, MDIO devices, the
insane implication is that for the exact same DSA switch device, we
might have both ->shutdown and ->remove getting called.
So we need to do something with that insane environment. The pattern
I've come up with is "if this, then not that", so if either ->shutdown
or ->remove gets called, we set the device's drvdata to NULL, and in the
other hook, we check whether the drvdata is NULL and just do nothing.
This is probably not necessary for platform devices, just for devices on
buses, but I would really insist for consistency among drivers, because
when code is copy-pasted, it is not always copy-pasted from the best
sources.
So depending on whether the DSA switch's ->remove or ->shutdown will get
called first, we cannot really guarantee even for the same driver if
rebooting will result in the same code path on all platforms. But
nonetheless, we need to do something minimally reasonable on ->shutdown
too to fix the bug. Of course, the ->remove will do more (a full
teardown of the tree, with all data structures freed, and this is why
the bug was not caught for so long). The new ->shutdown method is kept
separate from dsa_unregister_switch not because we couldn't have
unregistered the switch, but simply in the interest of doing something
quick and to the point.
The big question is: does the DSA switch's ->shutdown get called earlier
than the DSA master's ->shutdown? If not, there is still a risk that we
might still trigger the WARN_ON in unregister_netdevice that says we are
attempting to unregister a net_device which has uppers. That's no good.
Although the reference to the master net_device won't physically go away
even if DSA's ->shutdown comes afterwards, remember we have a dev_hold
on it.
The answer to that question lies in this comment above device_link_add:
* A side effect of the link creation is re-ordering of dpm_list and the
* devices_kset list by moving the consumer device and all devices depending
* on it to the ends of these lists (that does not happen to devices that have
* not been registered when this function is called).
so the fact that DSA uses device_link_add towards its master is not
exactly for nothing. device_shutdown() walks devices_kset from the back,
so this is our guarantee that DSA's shutdown happens before the master's
shutdown.
Fixes: 2f1e8ea726 ("net: dsa: link interfaces with the DSA master to get rid of lockdep warnings")
Link: https://lore.kernel.org/netdev/20210909095324.12978-1-LinoSanfilippo@gmx.de/
Reported-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
MDIO-attached devices might have interrupts and other things that might
need quiesced when we kexec into a new kernel. Things are even more
creepy when those interrupt lines are shared, and in that case it is
absolutely mandatory to disable all interrupt sources.
Moreover, MDIO devices might be DSA switches, and DSA needs its own
shutdown method to unlink from the DSA master, which is a new
requirement that appeared after commit 2f1e8ea726 ("net: dsa: link
interfaces with the DSA master to get rid of lockdep warnings").
So introduce a ->shutdown method in the MDIO device driver structure.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull io_uring iov_iter retry fixes from Jens Axboe:
"This adds a helper to save/restore iov_iter state, and modifies
io_uring to use it.
After that is done, we can now kill the iter->truncated addition that
we added for this release. The io_uring change is being overly
cautious with the save/restore/advance, but better safe than sorry and
we can always improve that and reduce the overhead if it proves to be
of concern. The only case to be worried about in this regard is huge
IO, where iteration can take a while to iterate segments.
I spent some time writing test cases, and expanded the coverage quite
a bit from the last posting of this. liburing carries this regression
test case now:
https://git.kernel.dk/cgit/liburing/tree/test/file-verify.c
which exercises all of this. It now also supports provided buffers,
and explicitly tests for end-of-file/device truncation as well.
On top of that, Pavel sanitized the IOPOLL retry path to follow the
exact same pattern as normal IO"
* tag 'iov_iter.3-5.15-2021-09-17' of git://git.kernel.dk/linux-block:
io_uring: move iopoll reissue into regular IO path
Revert "iov_iter: track truncated size"
io_uring: use iov_iter state save/restore helpers
iov_iter: add helper to save iov_iter state
Pull io_uring fixes from Jens Axboe:
"Mostly fixes for regressions in this cycle, but also a few fixes that
predate this release.
The odd one out is a tweak to the direct files added in this release,
where attempting to reuse a slot is allowed instead of needing an
explicit removal of that slot first. It's a considerable improvement
in usability to that API, hence I'm sending it for -rc2.
- io-wq race fix and cleanup (Hao)
- loop_rw_iter() type fix
- SQPOLL max worker race fix
- Allow poll arm for O_NONBLOCK files, fixing a case where it's
impossible to properly use io_uring if you cannot modify the file
flags
- Allow direct open to simply reuse a slot, instead of needing it
explicitly removed first (Pavel)
- Fix a case where we missed signal mask restoring in cqring_wait, if
we hit -EFAULT (Xiaoguang)"
* tag 'io_uring-5.15-2021-09-17' of git://git.kernel.dk/linux-block:
io_uring: allow retry for O_NONBLOCK if async is supported
io_uring: auto-removal for direct open/accept
io_uring: fix missing sigmask restore in io_cqring_wait()
io_uring: pin SQPOLL data before unlocking ring lock
io-wq: provide IO_WQ_* constants for IORING_REGISTER_IOWQ_MAX_WORKERS arg items
io-wq: fix potential race of acct->nr_workers
io-wq: code clean of io_wqe_create_worker()
io_uring: ensure symmetry in handling iter types in loop_rw_iter()
NXP Legal insists that the following are not fine:
- Saying "NXP Semiconductors" instead of "NXP", since the company's
registered name is "NXP"
- Putting a "(c)" sign in the copyright string
- Putting a comma in the copyright string
The only accepted copyright string format is "Copyright <year-range> NXP".
This patch changes the copyright headers in the networking files that
were sent by me, or derived from code sent by me.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Sometimes when unbinding the mv88e6xxx driver on Turris MOX, these error
messages appear:
mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete be:79:b4:9e:9e:96 vid 1 from fdb: -2
mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete be:79:b4:9e:9e:96 vid 0 from fdb: -2
mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete d8:58:d7:00:ca:6d vid 100 from fdb: -2
mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete d8:58:d7:00:ca:6d vid 1 from fdb: -2
mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete d8:58:d7:00:ca:6d vid 0 from fdb: -2
(and similarly for other ports)
What happens is that DSA has a policy "even if there are bugs, let's at
least not leak memory" and dsa_port_teardown() clears the dp->fdbs and
dp->mdbs lists, which are supposed to be empty.
But deleting that cleanup code, the warnings go away.
=> the FDB and MDB lists (used for refcounting on shared ports, aka CPU
and DSA ports) will eventually be empty, but are not empty by the time
we tear down those ports. Aka we are deleting them too soon.
The addresses that DSA complains about are host-trapped addresses: the
local addresses of the ports, and the MAC address of the bridge device.
The problem is that offloading those entries happens from a deferred
work item scheduled by the SWITCHDEV_FDB_DEL_TO_DEVICE handler, and this
races with the teardown of the CPU and DSA ports where the refcounting
is kept.
In fact, not only it races, but fundamentally speaking, if we iterate
through the port list linearly, we might end up tearing down the shared
ports even before we delete a DSA user port which has a bridge upper.
So as it turns out, we need to first tear down the user ports (and the
unused ones, for no better place of doing that), then the shared ports
(the CPU and DSA ports). In between, we need to ensure that all work
items scheduled by our switchdev handlers (which only run for user
ports, hence the reason why we tear them down first) have finished.
Fixes: 161ca59d39 ("net: dsa: reference count the MDB entries at the cross-chip notifier level")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20210914134726.2305133-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
