All Ethernet AVB instances on R-Car V3U have registers related to UDP/IP
support, but the declared register blocks for the first two instances
are too small to cover them.
Ethernet IRQ GPIOs are marked as GPIO hogs. Thus, these GPIOs are
requested at probe time without considering if there are other
peripherals that need them. The Ethernet IRQ GPIOs are shared with
SDHI2. Selection between Ethernet and SDHI2 is done through a hardware
switch. To avoid scenarios where one wants to boot with SDHI2 support
and some SDHI pins are not propertly configured because of the GPIO
hogs, guard the Ethernet IRQ GPIO hogs with the proper build flag.
The IRQC IP block supports Bus error and ECCRAM interrupts on RZ/G2L and
alike SoC's (listed below). Update the IRQC nodes with the missing
interrupts, and additionally, include the 'interrupt-names' properties
in the IRQC nodes so that the driver can parse interrupts by name.
Because the number of channels to be configured is calculated using the %,
and it results in 0 when there's an exact division, this leads to some
channels not having their tx power configured.
Fixes: 7801da338856 ("wifi: mt76: mt7921: enable set txpower for UNII-4") Signed-off-by: Ming Yen Hsieh <mingyen.hsieh@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
In some case, the MTCL table will exist, but MTDS table will not.
So the SAR will init fail. This patch make MTCL and MTDS can exist
with no dependence.
Fixes: f965333e491e ("mt76: mt7921: introduce ACPI SAR support") Signed-off-by: Ming Yen Hsieh <mingyen.hsieh@mediatek.com> Signed-off-by: Leon Yen <leon.yen@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
From commit a304e1b82808 ("[PATCH] Debug shared irqs"), there is a test
to make sure the shared irq handler should be able to handle the unexpected
event after deregistration. For this case, let's apply MT76_REMOVED flag to
indicate the device was removed and do not run into the resource access
anymore.
Fixes: c948b5da6bbe ("wifi: mt76: mt7925: add Mediatek Wi-Fi7 driver for mt7925 chips") Signed-off-by: Deren Wu <deren.wu@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
From commit a304e1b82808 ("[PATCH] Debug shared irqs"), there is a test
to make sure the shared irq handler should be able to handle the unexpected
event after deregistration. For this case, let's apply MT76_REMOVED flag to
indicate the device was removed and do not run into the resource access
anymore.
The EHT MCS map subfield of 20 MHz-Only is not present in the EHT
capability of AP, so STA does not need to parse the subfield.
Moreover, AP should parse the subfield only if STA is 20 MHz-Only, which
can be confirmed by checking supported channel width in HE capability.
Fixes: 92aa2da9fa49 ("wifi: mt76: mt7996: enable EHT support in firmware") Co-developed-by: Shayne Chen <shayne.chen@mediatek.com> Signed-off-by: Shayne Chen <shayne.chen@mediatek.com> Signed-off-by: Benjamin Lin <benjamin-jw.lin@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
This patch fixes the following TWT issues:
- Change table_mask to u16 to support up to 16 TWT stations
- Reject TWT flows for duplicated establishment
- Fix possible unaligned pointer
- Remove unsupported TWT_CONTROL_WAKE_DUR_UNIT flag
- The minimum TWT duration supported by mt7996 chipsets is 64. Reply
with TWT_SETUP_CMD_DICTATE if the min_twt_dur is smaller than 64
Fixes: 98686cd21624 ("wifi: mt76: mt7996: add driver for MediaTek Wi-Fi 7 (802.11be) devices") Signed-off-by: Peter Chiu <chui-hao.chiu@mediatek.com> Signed-off-by: Shayne Chen <shayne.chen@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
The PPDU TXS does not include the error bit so it cannot use to report
status to mac80211. This patch fixes issue that STA wrongly detects if AP
is still alive.
Fixes: 2569ea5326e2 ("wifi: mt76: mt7996: enable PPDU-TxS to host") Signed-off-by: Peter Chiu <chui-hao.chiu@mediatek.com> Signed-off-by: Shayne Chen <shayne.chen@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
There's a race between driver and fw on some tx/rx control registers
when setting ifs, which will cause accidental hw queue pause problems.
Avoid this by setting ifs time with bss_info mcu command.
Fixes: c948b5da6bbe ("wifi: mt76: mt7925: add Mediatek Wi-Fi7 driver for mt7925 chips") Co-developed-by: Deren Wu <deren.wu@mediatek.com> Signed-off-by: Deren Wu <deren.wu@mediatek.com> Signed-off-by: Ming Yen Hsieh <mingyen.hsieh@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
A sub-process of Wifi L0.5 reset will make chip common partition
enter low power, and have chance lead to Bluetooth host-to-chip
command timeout, modify the software flow according to the chip's
design to solve the problem.
Fixes: c948b5da6bbe ("wifi: mt76: mt7925: add Mediatek Wi-Fi7 driver for mt7925 chips") Signed-off-by: Quan Zhou <quan.zhou@mediatek.com> Signed-off-by: Ming Yen Hsieh <mingyen.hsieh@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
The header translation config should set to broadcast and unicast
cases correctly, not only unicast case. And also remove the cmds
of wtbl (wlan table) series, because these MCU commands have
already been replaced by other commands in mt7925.
Fixes: c948b5da6bbe ("wifi: mt76: mt7925: add Mediatek Wi-Fi7 driver for mt7925 chips") Signed-off-by: Ming Yen Hsieh <mingyen.hsieh@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
When in suspend mode, WoW (Wake-on-WLAN) fails to wake the system remotely
due to incorrect encryption mode settings. For the new mt7925 chipset, the
old STA_REC_KEY_V2 command will send incorrect parameters to the firmware.
Therefore, STA_REC_KEY_V3 has been introduced as a replacement for it.
Fixes: c948b5da6bbe ("wifi: mt76: mt7925: add Mediatek Wi-Fi7 driver for mt7925 chips") Signed-off-by: Ming Yen Hsieh <mingyen.hsieh@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
Suggested-by: "Sujuan Chen" <sujuan.chen@mediatek.com> Fixes: 4920a3a1285f ("wifi: mt76: mt7996: set DMA mask to 36 bits for boards with more than 4GB of RAM") Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
Some of the names for the Tegra234 DLA clients are not unique and do not
align with the name of the client ID definitions. Therefore, it is not
possible to determine the exact DLA client from messages that print the
client name. Fix this by correcting the DLA memory client names for
Tegra234 to align with the name of the corresponding memory client ID.
Note that although the client names are also used by the interconnect
framework, interconnect support for the DLA clients has not been added
and so this issue does not impact the interconnect support.
Fixes: 5cd24ca0985f ("memory: tegra: Add DLA clients for Tegra234") Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Link: https://lore.kernel.org/r/20240220124430.19072-1-jonathanh@nvidia.com Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When the development chip ROM was added, the "direct-mapped" compatible
value was already obsolete. In addition, the device node lacked the
accompanying "probe-type" property, causing the old physmap_of_core
driver to fall back to trying all available probe types.
Unfortunately this fallback was lost when the DT and pdata cases were
merged.
Fix this by using the modern "mtd-rom" compatible value instead.
When using a wilc1000 chip over a spi bus, users can optionally define a
reset gpio and a chip enable gpio. The reset line of wilc1000 is active
low, so to hold the chip in reset, a low (physical) value must be applied.
The corresponding device tree binding documentation was introduced by
commit f31ee3c0a555 ("wilc1000: Document enable-gpios and reset-gpios
properties") and correctly indicates that the reset line is an active-low
signal. The corresponding driver part, brought by commit ec031ac4792c
("wilc1000: Add reset/enable GPIO support to SPI driver") was applying the
correct logic. But commit fcf690b0b474 ("wifi: wilc1000: use correct
sequence of RESET for chip Power-UP/Down") eventually flipped this logic
and started misusing the gpiod APIs, applying an inverted logic when
powering up/down the chip (for example, setting the reset line to a logic
"1" during power up, which in fact asserts the reset line when device tree
describes the reset line as GPIO_ACTIVE_LOW). As a consequence, any
platform currently using the driver in SPI mode must use a faulty reset
line description in device tree, or else chip will be maintained in reset
and will not even allow to bring up the chip.
Fix reset line usage by inverting back the gpiod APIs usage, setting the
reset line to the logic value "0" when powering the chip, and the logic
value "1" when powering off the chip.
Fixes: fcf690b0b474 ("wifi: wilc1000: use correct sequence of RESET for chip Power-UP/Down") Signed-off-by: Alexis Lothoré <alexis.lothore@bootlin.com> Acked-by: Conor Dooley <conor.dooley@microchip.com> Acked-by: Ajay Singh <ajay.kathat@microchip.com> Signed-off-by: Kalle Valo <kvalo@kernel.org> Link: https://msgid.link/20240217-wilc_1000_reset_line-v2-1-b216f433d7d5@bootlin.com Signed-off-by: Sasha Levin <sashal@kernel.org>
The INTR module for DMASS1 (CSI specific DMASS) is outside the currently
available ranges, as it starts at 0x4e400000. So fix the ranges property
to enable programming the interrupts correctly.
Fixes: 29075cc09f43 ("arm64: dts: ti: Introduce AM62P5 family of SoCs") Reviewed-by: Vaishnav Achath <vaishnav.a@ti.com> Signed-off-by: Jai Luthra <j-luthra@ti.com> Link: https://lore.kernel.org/r/20240220-am62p_csi-v2-1-3e71d9945571@ti.com Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Change offset in mux-reg-masks property for hbmc_mux node
since reg-mux property is used in compatible.
While here, update the reg region to include 4 bytes as this
is a 32bit register.
Fixes: 2765149273f4 ("mux: mmio: use reg property when parent device is not a syscon") Suggested-by: Peter Rosin <peda@axentia.se> Signed-off-by: Andrew Davis <afd@ti.com> Link: https://lore.kernel.org/r/20240215141957.13775-1-afd@ti.com Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The ML element generation code to create a BSS entry from a per-STA
profile was not overwriting the BSS parameter change count. This meant
that the incorrect parameter change count would be reported within the
multi-link element.
Fix this by returning the BSS parameter change count from the function
and placing it into the ML element. The returned tbtt info was never
used, so just drop that to simplify the code.
Fixes: 5f478adf1f99 ("wifi: cfg80211: generate an ML element for per-STA profiles") Signed-off-by: Benjamin Berg <benjamin.berg@intel.com> Reviewed-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com> Link: https://msgid.link/20240216135047.f2a507634692.I06b122c7a319a38b4e970f5e0bd3d3ef9cac4cbe@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When the generic SCMI code tears down a channel, it calls the chan_free
callback function, defined by each transport. Since multiple protocols
might share the same transport_info member, chan_free() might want to
clean up the same member multiple times within the given SCMI transport
implementation. In this case, it is SMC transport. This will lead to a NULL
pointer dereference at the second time:
This adds common1 register space for AM62x SoC which is using TI's Keystone
display hardware and supporting it as described in
Documentation/devicetree/bindings/display/ti/ti,am65x-dss.yaml
This adds common1 register space for AM65x SoC which is using TI's Keystone
display hardware and supporting it as described in
Documentation/devicetree/bindings/display/ti/ti,am65x-dss.yaml
The external output reset signal was originally disabled and sent from
firmware. However, an unfixed bug in the firmware on tomato prevents
the signal from being sent, causing the device to fail to boot. To fix
this, enable external output reset signal to allow the device to reboot
normally.
Fixes: 5eb2e303ec6b ("arm64: dts: mediatek: Introduce MT8195 Cherry platform's Tomato") Signed-off-by: Hsin-Te Yuan <yuanhsinte@chromium.org> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20240124-send-upstream-v3-1-5097c9862a73@chromium.org Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Only Tx and Rx Signal lines for wkup_uart0 are brought out on
the J784S4 EVM from SoC, but CTS and RTS signal lines are not
brought on the EVM. Thus, remove pinmux for CTS and RTS signal
lines for wkup_uart0 in J784S4.
Only Tx and Rx Signal lines for wkup_uart0 are brought out on
the Common Proc Board through SoM, but CTS and RTS signal lines
are not brought on the board. Thus, remove pinmux for CTS and RTS
signal lines for wkup_uart0 in J721S2.
Clock-frequency property is already present in mcu_uart0 node of the
k3-j7200-mcu-wakeup.dtsi file. Thus, remove redundant clock-frequency
property from mcu_uart0 node.
Avoid the following warnings by removing the ena_select_queue() function
and rely on the net core to do the queue selection, The issue happen
when an skb received from an interface with more queues than ena is
forwarded to the ena interface.
[ 1176.159959] eth0 selects TX queue 11, but real number of TX queues is 8
[ 1176.863976] eth0 selects TX queue 14, but real number of TX queues is 8
[ 1180.767877] eth0 selects TX queue 14, but real number of TX queues is 8
[ 1188.703742] eth0 selects TX queue 14, but real number of TX queues is 8
Fixes: 1738cd3ed342 ("net: ena: Add a driver for Amazon Elastic Network Adapters (ENA)") Signed-off-by: Kamal Heib <kheib@redhat.com> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The caller of the function freq_qos_add_request() checks again a non
zero value but freq_qos_add_request() can return '1' if the request
already exists. Therefore, the setup function fails while the QoS
request actually did not failed.
Fix that by changing the check against a negative value like all the
other callers of the function.
Fixes: 0e8f68d7f0485 ("Add CPU energy model based support") Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
On many systems that have an AMD IOMMU the following sequence of
warnings is observed during bootup.
```
pci 0000:00:00.2 can't derive routing for PCI INT A
pci 0000:00:00.2: PCI INT A: not connected
```
This series of events happens because of the IOMMU initialization
sequence order and the lack of _PRT entries for the IOMMU.
During initialization the IOMMU driver first enables the PCI device
using pci_enable_device(). This will call acpi_pci_irq_enable()
which will check if the interrupt is declared in a PCI routing table
(_PRT) entry. According to the PCI spec [1] these routing entries
are only required under PCI root bridges:
The _PRT object is required under all PCI root bridges
The IOMMU is directly connected to the root complex, so there is no
parent bridge to look for a _PRT entry. The first warning is emitted
since no entry could be found in the hierarchy. The second warning is
then emitted because the interrupt hasn't yet been configured to any
value. The pin was configured in pci_read_irq() but the byte in
PCI_INTERRUPT_LINE return 0xff which means "Unknown".
After that sequence of events pci_enable_msi() is called and this
will allocate an interrupt.
That is both of these warnings are totally harmless because the IOMMU
uses MSI for interrupts. To avoid even trying to probe for a _PRT
entry mark the IOMMU as IRQ managed. This avoids both warnings.
Update the architecture dependency to be the generic Tegra
because the driver works on the four latest Tegra generations
not just Tegra210, if you build a kernel with a specific
ARCH_TEGRA_xxx_SOC option that excludes Tegra210 you don't get
this driver.
Fixes: 46a88534afb59 ("bus: Add support for Tegra ACONNECT") Signed-off-by: Peter Robinson <pbrobinson@gmail.com> Cc: Jon Hunter <jonathanh@nvidia.com> Cc: Thierry Reding <treding@nvidia.com> Signed-off-by: Thierry Reding <treding@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fix this by freeing the CPU idle device after unregistering it.
Fixes: 3d339dcbb56d ("cpuidle / ACPI : move cpuidle_device field out of the acpi_processor_power structure") Signed-off-by: Armin Wolf <W_Armin@gmx.de> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
AM62 USB works with some devices, while failing to operate with others.
[ 560.189822] xhci-hcd xhci-hcd.4.auto: xHCI Host Controller
[ 560.195631] xhci-hcd xhci-hcd.4.auto: new USB bus registered, assigned bus number 2
[ 574.388509] xhci-hcd xhci-hcd.4.auto: can't setup: -110
[ 574.393814] xhci-hcd xhci-hcd.4.auto: USB bus 2 deregistered
[ 574.399544] xhci-hcd: probe of xhci-hcd.4.auto failed with error -110
This seems to be related to LPM (Link Power Management), and disabling it
turns USB into reliable working state.
As per AM62 reference manual:
> 4.8.2.1 USB2SS Unsupported Features
>
> The following features are not supported on this family of devices:
> ...
> - USB 2.0 ECN: Link Power Management (LPM)
> ...
Fixes: 2240f96cf3cd ("arm64: dts: ti: k3-am62-main: Add support for USB") Signed-off-by: Andrejs Cainikovs <andrejs.cainikovs@toradex.com> Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com> Reviewed-by: Roger Quadros <rogerq@ti.com> Link: https://lore.kernel.org/r/20240209130213.38908-1-andrejs.cainikovs@gmail.com Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
wilc_netdev_cleanup currently triggers a KASAN warning, which can be
observed on interface registration error path, or simply by
removing the module/unbinding device from driver:
==================================================================
BUG: KASAN: slab-use-after-free in wilc_netdev_cleanup+0x508/0x5cc
Read of size 4 at addr c54d1ce8 by task sh/86
CPU: 0 PID: 86 Comm: sh Not tainted 6.8.0-rc1+ #117
Hardware name: Atmel SAMA5
unwind_backtrace from show_stack+0x18/0x1c
show_stack from dump_stack_lvl+0x34/0x58
dump_stack_lvl from print_report+0x154/0x500
print_report from kasan_report+0xac/0xd8
kasan_report from wilc_netdev_cleanup+0x508/0x5cc
wilc_netdev_cleanup from wilc_bus_remove+0xc8/0xec
wilc_bus_remove from spi_remove+0x8c/0xac
spi_remove from device_release_driver_internal+0x434/0x5f8
device_release_driver_internal from unbind_store+0xbc/0x108
unbind_store from kernfs_fop_write_iter+0x398/0x584
kernfs_fop_write_iter from vfs_write+0x728/0xf88
vfs_write from ksys_write+0x110/0x1e4
ksys_write from ret_fast_syscall+0x0/0x1c
David Mosberger-Tan initial investigation [1] showed that this
use-after-free is due to netdevice unregistration during vif list
traversal. When unregistering a net device, since the needs_free_netdev has
been set to true during registration, the netdevice object is also freed,
and as a consequence, the corresponding vif object too, since it is
attached to it as private netdevice data. The next occurrence of the loop
then tries to access freed vif pointer to the list to move forward in the
list.
Fix this use-after-free thanks to two mechanisms:
- navigate in the list with list_for_each_entry_safe, which allows to
safely modify the list as we go through each element. For each element,
remove it from the list with list_del_rcu
- make sure to wait for RCU grace period end after each vif removal to make
sure it is safe to free the corresponding vif too (through
unregister_netdev)
Since we are in a RCU "modifier" path (not a "reader" path), and because
such path is expected not to be concurrent to any other modifier (we are
using the vif_mutex lock), we do not need to use RCU list API, that's why
we can benefit from list_for_each_entry_safe.
It is still possible to compile-test a kernel without CONFIG_COMMON_CLK
for some ancient ARM boards or other architectures, but this causes a
link failure in the qcom-cpufreq-hw driver:
Add a Kconfig dependency here to make sure this always work. Apparently
this bug has been in the kernel for a while without me running into it
on randconfig builds as COMMON_CLK is almost always enabled.
I have cross-checked by building an allmodconfig kernel with COMMON_CLK
disabled, which showed no other driver having this problem.
The mtu3 usb controllers don't list the xhci clock, though they require
it, and thus rely on the bootloader leaving it on in order to work.
When booting with the upstream arm64 defconfig, the usb controllers will
defer probe until modules have loaded since they have an indirect
dependency on CONFIG_MTK_CMDQ, which is configured as a module. However
at the point where modules are loaded, unused clocks are also disabled,
causing the usb controllers to probe without the xhci clock enabled and
fail to probe:
mtu3 11201000.usb: clks of sts1 are not stable!
mtu3 11201000.usb: device enable failed -110
mtu3 11201000.usb: mtu3 hw init failed:-110
mtu3 11201000.usb: failed to initialize gadget
mtu3: probe of 11201000.usb failed with error -110
The ssusb power domains currently don't list any clocks, despite
depending on some, and thus rely on the bootloader leaving the required
clocks on in order to work.
When booting with the upstream arm64 defconfig, the power domain
controller will defer probe until modules have loaded since it has an
indirect dependency on CONFIG_MTK_CMDQ, which is configured as a module.
However at the point where modules are loaded, unused clocks are also
disabled, causing the ssusb domains to fail to be enabled and
consequently the controller to fail probe:
mtk-power-controller 10006000.syscon:power-controller: /soc/syscon@10006000/power-controller/power-domain@4: failed to power on domain: -110
mtk-power-controller: probe of 10006000.syscon:power-controller failed with error -110
Add the missing clocks for the ssusb power domains so that they can
successfully probe without relying on the bootloader state.
The qfprom actually is bigger than 0x1000, so adjust the reg.
Note that the non-ECC-corrected qfprom can be found at 0xfc4b8000
(-0x4000). The current reg points to the ECC-corrected qfprom block
which should have equivalent values at all offsets compared to the
non-corrected version.
[luca@z3ntu.xyz: extract to standalone patch and adjust for review
comments]
Commit c72ca343f911 ("soc: qcom: llcc: Add v4.1 HW version support")
introduced a new 4.1 if statement in llcc_update_act_ctrl() without
considering that ret might be overwritten. So, add return value check
after Broadcast_OR register read in llcc_update_act_ctrl().
Currently tracing is supposed not to allow for bpf_spin_{lock,unlock}()
helper calls. This is to prevent deadlock for the following cases:
- there is a prog (prog-A) calling bpf_spin_{lock,unlock}().
- there is a tracing program (prog-B), e.g., fentry, attached
to bpf_spin_lock() and/or bpf_spin_unlock().
- prog-B calls bpf_spin_{lock,unlock}().
For such a case, when prog-A calls bpf_spin_{lock,unlock}(),
a deadlock will happen.
The related source codes are below in kernel/bpf/helpers.c:
notrace BPF_CALL_1(bpf_spin_lock, struct bpf_spin_lock *, lock)
notrace BPF_CALL_1(bpf_spin_unlock, struct bpf_spin_lock *, lock)
notrace is supposed to prevent fentry prog from attaching to
bpf_spin_{lock,unlock}().
But actually this is not the case and fentry prog can successfully
attached to bpf_spin_lock(). Siddharth Chintamaneni reported
the issue in [1]. The following is the macro definition for
above BPF_CALL_1:
#define BPF_CALL_x(x, name, ...) \
static __always_inline \
u64 ____##name(__BPF_MAP(x, __BPF_DECL_ARGS, __BPF_V, __VA_ARGS__)); \
typedef u64 (*btf_##name)(__BPF_MAP(x, __BPF_DECL_ARGS, __BPF_V, __VA_ARGS__)); \
u64 name(__BPF_REG(x, __BPF_DECL_REGS, __BPF_N, __VA_ARGS__)); \
u64 name(__BPF_REG(x, __BPF_DECL_REGS, __BPF_N, __VA_ARGS__)) \
{ \
return ((btf_##name)____##name)(__BPF_MAP(x,__BPF_CAST,__BPF_N,__VA_ARGS__));\
} \
static __always_inline \
u64 ____##name(__BPF_MAP(x, __BPF_DECL_ARGS, __BPF_V, __VA_ARGS__))
The notrace attribute is actually applied to the static always_inline function
____bpf_spin_{lock,unlock}(). The actual callback function
bpf_spin_{lock,unlock}() is not marked with notrace, hence
allowing fentry prog to attach to two helpers, and this
may cause the above mentioned deadlock. Siddharth Chintamaneni
actually has a reproducer in [2].
To fix the issue, a new macro NOTRACE_BPF_CALL_1 is introduced which
will add notrace attribute to the original function instead of
the hidden always_inline function and this fixed the problem.
Fixes: 9be162a7b670 ("wifi: iwlwifi: mvm: add support for the new MAC CTXT command") Signed-off-by: Ilan Peer <ilan.peer@intel.com> Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com> Link: https://msgid.link/20240208185302.a338c30ec4e9.Ic2813cdeba4443c692d462fc4859392f069d7e33@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Interrupts are enabled/disabled in more places than just m_can_start()
and m_can_stop(). Couple the polling timer with enabling/disabling of
all interrupts to achieve equivalent behavior.
Cc: Judith Mendez <jm@ti.com> Fixes: b382380c0d2d ("can: m_can: Add hrtimer to generate software interrupt") Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/all/20240207093220.2681425-2-msp@baylibre.com Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
This fixes:
arch/arm64/boot/dts/mediatek/mt7622-rfb1.dtb: /: memory@40000000: 'device_type' is a required property
from schema $id: http://devicetree.org/schemas/memory.yaml#
arch/arm64/boot/dts/mediatek/mt7622-bananapi-bpi-r64.dtb: /: memory@40000000: 'device_type' is a required property
from schema $id: http://devicetree.org/schemas/memory.yaml#
Signed-off-by: Rafał Miłecki <rafal@milecki.pl> Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20240122132357.31264-1-zajec5@gmail.com Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The larb clock is in fact a subsys clock, so it must be prefixed by
'subsys-' to be correctly identified in the driver.
Fixes: d9e43c1e7a38 ("arm64: dts: mt8186: Add power domains controller") Signed-off-by: Eugen Hristev <eugen.hristev@collabora.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20231228113245.174706-6-eugen.hristev@collabora.com Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Clock name should be `venc_sel` as per binding.
Fix the warning message :
arch/arm64/boot/dts/mediatek/mt8192-asurada-hayato-r1.dtb: vcodec@17020000: clock-names:0: 'venc_sel' was expected
from schema $id: http://devicetree.org/schemas/media/mediatek,vcodec-encoder.yaml#
Fixes: aa8f3711fc87 ("arm64: dts: mt8192: Add H264 venc device node") Signed-off-by: Eugen Hristev <eugen.hristev@collabora.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20231228113245.174706-4-eugen.hristev@collabora.com Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The commit adding the ChromeOS EC to the Asurada Devicetree mistakenly
added a base detection node. While tablet mode detection is supported by
CrosEC and used by Hayato, it is done through the cros-ec-keyb driver.
The base detection node, which is handled by the hid-google-hammer
driver, also provides tablet mode detection but by checking base
attachment status on the CrosEC, which is not supported for Asurada.
Hence, remove the unused CrosEC base detection node for Asurada.
MT7986's Infrastructure System Configuration Controller includes reset
controller. It can reset blocks as specified in the
include/dt-bindings/reset/mt7986-resets.h . Add #reset-cells so it can
be referenced properly.
This fixes:
arch/arm64/boot/dts/mediatek/mt7986a-bananapi-bpi-r3.dtb: infracfg@10001000: '#reset-cells' is a required property
from schema $id: http://devicetree.org/schemas/arm/mediatek/mediatek,infracfg.yaml#
Fixes: 1f9986b258c2 ("arm64: dts: mediatek: add clock support for mt7986a") Cc: Sam Shih <sam.shih@mediatek.com> Signed-off-by: Rafał Miłecki <rafal@milecki.pl> Link: https://lore.kernel.org/r/20240101182040.28538-2-zajec5@gmail.com Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com> Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
PWM is not a clock provider and its binding doesn't specify
"#clock-cells" property.
This fixes:
arch/arm64/boot/dts/mediatek/mt7986a-bananapi-bpi-r3.dtb: pwm@10048000: '#clock-cells' does not match any of the regexes: 'pinctrl-[0-9]+'
from schema $id: http://devicetree.org/schemas/pwm/mediatek,mt2712-pwm.yaml#
Fixes: eabb04df46c6 ("arm64: dts: mt7986: add PWM") Cc: Daniel Golle <daniel@makrotopia.org> Signed-off-by: Rafał Miłecki <rafal@milecki.pl> Link: https://lore.kernel.org/r/20240101182040.28538-1-zajec5@gmail.com Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com> Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
This fixes following validation errors:
arch/arm64/boot/dts/mediatek/mt7986a-rfb.dtb: spi_nand@0: $nodename:0: 'spi_nand@0' does not match '^(flash|.*sram|nand)(@.*)?$'
from schema $id: http://devicetree.org/schemas/mtd/spi-nand.yaml#
arch/arm64/boot/dts/mediatek/mt7986b-rfb.dtb: spi_nand@0: $nodename:0: 'spi_nand@0' does not match '^(flash|.*sram|nand)(@.*)?$'
from schema $id: http://devicetree.org/schemas/mtd/spi-nand.yaml#
Fixes: 885e153ed7c1 ("arm64: dts: mt7986: add spi related device nodes") Signed-off-by: Rafał Miłecki <rafal@milecki.pl> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20231116130952.5099-2-zajec5@gmail.com Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com> Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
This fixes SPI setup and resolves following validation errors:
arch/arm64/boot/dts/mediatek/mt7986a-rfb.dtb: spi_nand@0: Unevaluated properties are not allowed ('spi-rx-buswidth', 'spi-tx-buswidth' were unexpected)
from schema $id: http://devicetree.org/schemas/mtd/spi-nand.yaml#
arch/arm64/boot/dts/mediatek/mt7986b-rfb.dtb: spi_nand@0: Unevaluated properties are not allowed ('spi-rx-buswidth', 'spi-tx-buswidth' were unexpected)
from schema $id: http://devicetree.org/schemas/mtd/spi-nand.yaml#
Fixes: 885e153ed7c1 ("arm64: dts: mt7986: add spi related device nodes") Signed-off-by: Rafał Miłecki <rafal@milecki.pl> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20231116130952.5099-1-zajec5@gmail.com Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com> Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
According to the "inside-secure,safexcel-eip97" binding "clock-names" is
required only if there are two clocks specified. If present the first
name must by "core".
Name "infra_eip97_ck" is invalid and was probably just a typo. Drop it.
Fixes: ecc5287cfe53 ("arm64: dts: mt7986: add crypto related device nodes") Cc: Sam Shih <sam.shih@mediatek.com> Signed-off-by: Rafał Miłecki <rafal@milecki.pl> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20231116132411.7665-1-zajec5@gmail.com Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com> Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
This fixes typo and resolves following validation error:
arch/arm64/boot/dts/mediatek/mt7986a-bananapi-bpi-r3.dtb: pwm-fan: pwms: [[54, 0, 10000], [0]] is too long
from schema $id: http://devicetree.org/schemas/hwmon/pwm-fan.yaml#
Fixes: c26f779a2295 ("arm64: dts: mt7986: add pwm-fan and cooling-maps to BPI-R3 dts") Cc: Daniel Golle <daniel@makrotopia.org> Signed-off-by: Rafał Miłecki <rafal@milecki.pl> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20231116130816.4932-1-zajec5@gmail.com Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com> Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The cbas node is used to describe base detection functionality in the
ChromeOS EC, which is used for units that have a detachable keyboard and
thus rely on this functionality to switch between tablet and laptop
mode.
Despite the original commit having added the cbas node to the
mt8183-kukui.dtsi, not all machines that include it are detachables. In
fact all machines that include from mt8183-kukui-jacuzzi.dtsi are either
clamshells (ie normal laptops) or convertibles, meaning the keyboard can
be flipped but not detached. The detection for the keyboard getting
flipped is handled by the driver bound to the keyboard-controller node
in the EC.
Move the base detection node from the base kukui dtsi to the dtsis where
all machines are detachables, and thus actually make use of the node.
Fixes: 4fa8492d1e5b ("arm64: dts: mt8183: add cbas node under cros_ec") Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20240116-mt8183-kukui-cbas-remove-v3-1-055e21406e86@collabora.com Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com> Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
As discussed in the past (commit 2d3916f31891 ("ipv6: fix skb drops
in igmp6_event_query() and igmp6_event_report()")) I think the
synchronize_net() call in ipv6_mc_down() is not needed.
Under load, synchronize_net() can last between 200 usec and 5 ms.
KASAN seems to agree as well.
Fixes: f185de28d9ae ("mld: add new workqueues for process mld events") Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Taehee Yoo <ap420073@gmail.com> Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: David Ahern <dsahern@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Stop selecting UTMI clock as the USB3 PIPE clock. This setting is
incompatible with the USB host working in USB3 (SuperSpeed) mode.
While we are at it, also drop the default setting for the port speed.
The two tests that make use of multicast routig (router.sh and
router_multicast.sh) are currently failing in the netdev CI because the
kernel is missing multicast routing support.
The config file contains a partial kernel configuration to be used by
`virtme-configkernel --custom'. The presumption is that the config file
contains all Kconfig options needed by the selftests from the directory.
In net/forwarding/config, many are missing, which manifests as spurious
failures when running the selftests, with messages about unknown device
types, qdisc kinds or classifier actions. Add the missing configurations.
Tested the resulting configuration using virtme-ng as follows:
# vng -b -f tools/testing/selftests/net/forwarding/config
# vng --user root
(within the VM:)
# make -C tools/testing/selftests TARGETS=net/forwarding run_tests
'-fPIC' as an option to the linker does not do what it seems like it
should. With ld.bfd, it is treated as '-f PIC', which does not make
sense based on the meaning of '-f':
-f SHLIB, --auxiliary SHLIB Auxiliary filter for shared object symbol table
When building with ld.lld (currently under review in a GitHub pull
request), it just errors out because '-f' means nothing and neither does
'-fPIC':
ld.lld: error: unknown argument '-fPIC'
'-fPIC' was blindly copied from CFLAGS when the vDSO stopped being
linked with '$(CC)', it should not be needed. Remove it to clear up the
build failure with ld.lld.
When the device drivers are initialized, a sysfs directory
is created. This contains many attributes which are allocated with
kzalloc(). Should it fail, the memory for the attributes already
created is freed in attr_event_free(). Its second parameter is number
of attribute elements to delete. This parameter is off by one.
When i. e. the 10th attribute fails to get created, attributes
numbered 0 to 9 should be deleted. Currently only attributes
numbered 0 to 8 are deleted.
Fixes: 39d62336f5c1 ("s390/pai: add support for cryptography counters") Reported-by: Sumanth Korikkar <sumanthk@linux.ibm.com> Signed-off-by: Thomas Richter <tmricht@linux.ibm.com> Acked-by: Sumanth Korikkar <sumanthk@linux.ibm.com> Signed-off-by: Heiko Carstens <hca@linux.ibm.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The firmware (later) actually uses the values even for keys
that are invalid as far as the host is concerned, later in
rekeying, and then only sets the low 48 bits since the PNs
are only 48 bits over the air. It does, however, compare the
full 64 bits later, obviously causing problems.
Remove the memset and use kzalloc instead to avoid any old
heap data leaking to the firmware. We already init all the
other fields in the struct anyway. This leaves the data set
to zero for any unused fields, so the firmware can look at
them safely even if they're not used right now.
Fixes: 79e561f0f05a ("iwlwifi: mvm: d3: implement RSC command version 5") Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com> Link: https://msgid.link/20240206175739.462101146fef.I10f3855b99417af4247cff04af78dcbc6cb75c9c@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The firmware doesn't need the MFP flag for the GTK, it can even make the
firmware crash. in case the AP is configured with: group cipher TKIP and
MFPC. We would send the GTK with cipher = TKIP and MFP which is of course
not possible.
Fixes: 5c75a208c244 ("wifi: iwlwifi: mvm: support new key API") Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com> Link: https://msgid.link/20240206175739.2f2c602ab3c6.If13b2e2fa532381d985c07df130bee1478046c89@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When retrieving the queue index ("SCD SSN") from the TX response,
it's currently masked with 0xFFF. However, now that we have queues
longer than 4k, that became wrong, so make the mask depend on the
hardware family.
This fixes an issue where if we get a single frame reclaim while
in the top half of an 8k long queue, we'd reclaim-wrap the queue
twice (once on this and then again on the next non-single reclaim)
which at least triggers the WARN_ON_ONCE() in iwl_txq_reclaim(),
but could have other negative side effects (such as unmapping a
frame that wasn't transmitted yet, and then taking an IOMMU fault)
as well.
Fixes: 7b3e42ea2ead ("iwlwifi: support multiple tfd queue max sizes for different devices") Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com> Link: https://msgid.link/20240205211151.4148a6ef54e0.I733a70f679c25f9f99097a8dcb3a1f8165da6997@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Before sending SESSION PROTECTION cmd the driver verifies that the
link for which the cmd is going to be sent is active.
The existing code is checking it only for MLD vifs,
but also the deflink (in non-MLD vifs) needs to be active in order
the have a session protection for it.
Fix this by checking if the link is active also for non-MLD vifs
Fixes: 135065837310 ("wifi: iwlwifi: support link_id in SESSION_PROTECTION cmd") Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com> Reviewed-by: Johannes Berg <johannes.berg@intel.com> Link: https://msgid.link/20240205211151.c61820f14ca6.Ibbe0f848f3e71f64313d21642650b6e4bfbe4b39@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The .BTF_ids section is pre-filled with zeroed BTF ID entries during the
build and afterwards patched by resolve_btfids with correct values.
Since resolve_btfids always writes in host-native endianness, it relies
on libelf to do the translation when the target ELF is cross-compiled to
a different endianness (this was introduced in commit 61e8aeda9398
("bpf: Fix libelf endian handling in resolv_btfids")).
Unfortunately, the translation will corrupt the flags fields of SET8
entries because these were written during vmlinux compilation and are in
the correct endianness already. This will lead to numerous selftests
failures such as:
Since it's not possible to instruct libelf to translate just certain
values, let's manually bswap the flags (both global and entry flags) in
resolve_btfids when needed, so that libelf then translates everything
correctly.
Instead of using magic offsets to access BTF ID set data, leverage types
from btf_ids.h (btf_id_set and btf_id_set8) which define the actual
layout of the data. Thanks to this change, set sorting should also
continue working if the layout changes.
This requires to sync the definition of 'struct btf_id_set8' from
include/linux/btf_ids.h to tools/include/linux/btf_ids.h. We don't sync
the rest of the file at the moment, b/c that would require to also sync
multiple dependent headers and we don't need any other defs from
btf_ids.h.
The driver only used the number of pwm channels to set the pwm_chip's
npwm member. The result is that if there are more capture channels than
PWM channels specified in the device tree, only a part of the capture
channel is usable. Fix that by passing the bigger channel count to the
pwm framework. This makes it possible that the .apply() callback is
called with .hwpwm >= pwm_num_devs, catch that case and return an error
code.
The commit d51507098ff91 ("printk: disable optimistic spin
during panic") added checks to avoid becoming a console waiter
if a panic is in progress.
However, the transition to panic can occur while there is
already a waiter. The current owner should not pass the lock to
the waiter because it might get stopped or blocked anytime.
Also the panic context might pass the console lock owner to an
already stopped waiter by mistake. It might happen when
console_flush_on_panic() ignores the current lock owner, for
example:
CPU0 CPU1
---- ----
console_lock_spinning_enable()
console_trylock_spinning()
[CPU1 now console waiter]
NMI: panic()
panic_other_cpus_shutdown()
[stopped as console waiter]
console_flush_on_panic()
console_lock_spinning_enable()
[print 1 record]
console_lock_spinning_disable_and_check()
[handover to stopped CPU1]
This results in panic() not flushing the panic messages.
Fix these problems by disabling all spinning operations
completely during panic().
Another advantage is that it prevents possible deadlocks caused
by "console_owner_lock". The panic() context does not need to
take it any longer. The lockless checks are safe because the
functions become NOPs when they see the panic in progress. All
operations manipulating the state are still synchronized by the
lock even when non-panic CPUs would notice the panic
synchronously.
The current owner might stay spinning. But non-panic() CPUs
would get stopped anyway and the panic context will never start
spinning.
Fixes: dbdda842fe96 ("printk: Add console owner and waiter logic to load balance console writes") Signed-off-by: John Ogness <john.ogness@linutronix.de> Link: https://lore.kernel.org/r/20240207134103.1357162-12-john.ogness@linutronix.de Signed-off-by: Petr Mladek <pmladek@suse.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Normally a reader will stop once reaching a non-finalized
record. However, when a panic happens, writers from other CPUs
(or an interrupted context on the panic CPU) may have been
writing a record and were unable to finalize it. The panic CPU
will reserve/commit/finalize its panic records, but these will
be located after the non-finalized records. This results in
panic() not flushing the panic messages.
Extend _prb_read_valid() to skip over non-finalized records if
on the panic CPU.
Fixes: 896fbe20b4e2 ("printk: use the lockless ringbuffer") Signed-off-by: John Ogness <john.ogness@linutronix.de> Reviewed-by: Petr Mladek <pmladek@suse.com> Link: https://lore.kernel.org/r/20240207134103.1357162-11-john.ogness@linutronix.de Signed-off-by: Petr Mladek <pmladek@suse.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
With the lockless ringbuffer, it is allowed that multiple
CPUs/contexts write simultaneously into the buffer. This creates
an ambiguity as some writers will finalize sooner.
The documentation for the prb_read functions is not clear as it
refers to "not yet written" and "no data available". Clarify the
return values and language to be in terms of the reader: records
available for reading.
Signed-off-by: John Ogness <john.ogness@linutronix.de> Reviewed-by: Petr Mladek <pmladek@suse.com> Link: https://lore.kernel.org/r/20240207134103.1357162-9-john.ogness@linutronix.de Signed-off-by: Petr Mladek <pmladek@suse.com>
Stable-dep-of: b1c4c67a5e90 ("printk: ringbuffer: Skip non-finalized records in panic") Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently pr_flush() will only wait for records that were
available to readers at the time of the call (using
prb_next_seq()). But there may be more records (non-finalized)
that have following finalized records. pr_flush() should wait
for these to print as well. Particularly because any trailing
finalized records may be the messages that the calling context
wants to ensure are printed.
Add a new ringbuffer function prb_next_reserve_seq() to return
the sequence number following the most recently reserved record.
This guarantees that pr_flush() will wait until all current
printk() messages (completed or in progress) have been printed.
Fixes: 3b604ca81202 ("printk: add pr_flush()") Signed-off-by: John Ogness <john.ogness@linutronix.de> Reviewed-by: Petr Mladek <pmladek@suse.com> Link: https://lore.kernel.org/r/20240207134103.1357162-10-john.ogness@linutronix.de Signed-off-by: Petr Mladek <pmladek@suse.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit f244b4dc53e5 ("printk: ringbuffer: Improve
prb_next_seq() performance") introduced an optimization for
prb_next_seq() by using best-effort to track recently finalized
records. However, the order of finalization does not
necessarily match the order of the records. The optimization
changed prb_next_seq() to return inconsistent results, possibly
yielding sequence numbers that are not available to readers
because they are preceded by non-finalized records or they are
not yet visible to the reader CPU.
Rather than simply best-effort tracking recently finalized
records, force the committing writer to read records and
increment the last "contiguous block" of finalized records. In
order to do this, the sequence number instead of ID must be
stored because ID's cannot be directly compared.
A new memory barrier pair is introduced to guarantee that a
reader can always read the records up until the sequence number
returned by prb_next_seq() (unless the records have since
been overwritten in the ringbuffer).
This restores the original functionality of prb_next_seq()
while also keeping the optimization.
For 32bit systems, only the lower 32 bits of the sequence
number are stored. When reading the value, it is expanded to
the full 64bit sequence number using the 32bit seq macros,
which fold in the value returned by prb_first_seq().
Fixes: f244b4dc53e5 ("printk: ringbuffer: Improve prb_next_seq() performance") Signed-off-by: John Ogness <john.ogness@linutronix.de> Reviewed-by: Petr Mladek <pmladek@suse.com> Link: https://lore.kernel.org/r/20240207134103.1357162-5-john.ogness@linutronix.de Signed-off-by: Petr Mladek <pmladek@suse.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The macros __seq_to_nbcon_seq() and __nbcon_seq_to_seq() are
used to provide support for atomic handling of sequence numbers
on 32bit systems. Until now this was only used by nbcon.c,
which is why they were located in nbcon.c and include nbcon in
the name.
In a follow-up commit this functionality is also needed by
printk_ringbuffer. Rather than duplicating the functionality,
relocate the macros to printk_ringbuffer.h.
Also, since the macros will be no longer nbcon-specific, rename
them to __u64seq_to_ulseq() and __ulseq_to_u64seq().
This does not result in any functional change.
Signed-off-by: John Ogness <john.ogness@linutronix.de> Reviewed-by: Petr Mladek <pmladek@suse.com> Link: https://lore.kernel.org/r/20240207134103.1357162-2-john.ogness@linutronix.de Signed-off-by: Petr Mladek <pmladek@suse.com>
Stable-dep-of: 5f72e52ba959 ("printk: ringbuffer: Do not skip non-finalized records with prb_next_seq()") Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / thirdparty / kernel / stable.git / log
