When two split-phy devices that support overlapping frequency ranges within
the same band(say 5 GHz low and 5 GHz high) are grouped into an ath12k
hardware (HW) setup, they share a common wiphy instance. Consequently, the
channel list (wiphy->bands[]) becomes unified across all associated
radios (ar).
When a scan is triggered with frequency list containing frequencies of
both 5 GHz low and 5 GHz high, mac80211 generates a single scan request
to driver with all the frequencies. This is because mac80211 splits the
scan request based on band.
ath12k checks the first frequency in the requested scan frequency list and
initiates scan to corresponding radio's(ar) firmware with all the
frequencies. Firmware rejects this scan since some of the frequencies in
the scan request are not supported, resulting in a scan failure.
To fix this ath12k driver should split the scan request into multiple
scans based on requested frequencies and schedule them to corresponding
underlying radio(s) in parallel.
Currently, ath12k driver assigns the scan link (link 15) in ahvif->links[]
for scan vdev creation. However, with parallel scan support being
introduced in the following patch, multiple radios (e.g., 5 GHz low and
5 GHz high) in the same HW group may attempt to use the same scan link
concurrently, causing conflicts where the vdev created by one radio could
be deleted and re-initialized by another.
To address this, reserve space for additional scan links for each radio in
a MLO group and allow subsequent radios to use different available scan
links (ahvif->link[15..MAX_SCAN_LINKS]) when scan link (15) is
pre-occupied.
While at it, rename ATH12K_DEFAULT_SCAN_LINK as ATH12K_FIRST_SCAN_LINK
as there is no longer only one scan link.
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.4.1-00199-QCAHKSWPL_SILICONZ-1
Signed-off-by: Rameshkumar Sundaram <rameshkumar.sundaram@oss.qualcomm.com>
Reviewed-by: Mahendran P <quic_mahep@quicinc.com>
Link: https://patch.msgid.link/20250507194832.2501668-2-rameshkumar.sundaram@oss.qualcomm.com
Signed-off-by: Jeff Johnson <jeff.johnson@oss.qualcomm.com>
In MLO configuration, ath12k_mac_radio_start() iterates through all
the radios and makes the ar state 'ON'. Even though some bands are
not supported in certain countries, ath12k_reg_update_chan_list()
tries to update the channel list for all the active pdevs and ends
up in the warn_on for non-supported band.
To prevent this, disable the pdev when the number of channels across
all bands supported by the pdev is zero for a particular country.
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.3.1-00173-QCAHKSWPL_SILICONZ-1
Signed-off-by: Karthik M <quic_karm@quicinc.com>
Signed-off-by: Muna Sinada <muna.sinada@oss.qualcomm.com>
Reviewed-by: Vasanthakumar Thiagarajan <vasanthakumar.thiagarajan@oss.qualcomm.com>
Link: https://patch.msgid.link/20250506214930.3561348-1-muna.sinada@oss.qualcomm.com
Signed-off-by: Jeff Johnson <jeff.johnson@oss.qualcomm.com>
During beacon miss handling, ath12k driver iterates over active virtual
interfaces (vifs) and attempts to access the radio object (ar) via
arvif->deflink->ar.
However, after commit aa80f12f3b ("wifi: ath12k: defer vdev creation for
MLO"), arvif is linked to a radio only after vdev creation, typically when
a channel is assigned or a scan is requested.
For P2P capable devices, a default P2P interface is created by
wpa_supplicant along with regular station interfaces, these serve as dummy
interfaces for P2P-capable stations, lack an associated netdev and initiate
frequent scans to discover neighbor p2p devices. When a scan is initiated
on such P2P vifs, driver selects destination radio (ar) based on scan
frequency, creates a scan vdev, and attaches arvif to the radio. Once the
scan completes or is aborted, the scan vdev is deleted, detaching arvif
from the radio and leaving arvif->ar uninitialized.
While handling beacon miss for station interfaces, P2P interface is also
encountered in the vif iteration and ath12k_mac_handle_beacon_miss_iter()
tries to dereference the uninitialized arvif->deflink->ar.
Fix this by verifying that vdev is created for the arvif before accessing
its ar during beacon miss handling and similar vif iterator callbacks.
==========================================================================
wlp6s0: detected beacon loss from AP (missed 7 beacons) - probing
KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017]
CPU: 5 UID: 0 PID: 0 Comm: swapper/5 Not tainted 6.16.0-rc1-wt-ath+ #2 PREEMPT(full)
RIP: 0010:ath12k_mac_handle_beacon_miss_iter+0xb5/0x1a0 [ath12k]
Call Trace:
__iterate_interfaces+0x11a/0x410 [mac80211]
ieee80211_iterate_active_interfaces_atomic+0x61/0x140 [mac80211]
ath12k_mac_handle_beacon_miss+0xa1/0xf0 [ath12k]
ath12k_roam_event+0x393/0x560 [ath12k]
ath12k_wmi_op_rx+0x1486/0x28c0 [ath12k]
ath12k_htc_process_trailer.isra.0+0x2fb/0x620 [ath12k]
ath12k_htc_rx_completion_handler+0x448/0x830 [ath12k]
ath12k_ce_recv_process_cb+0x549/0x9e0 [ath12k]
ath12k_ce_per_engine_service+0xbe/0xf0 [ath12k]
ath12k_pci_ce_workqueue+0x69/0x120 [ath12k]
process_one_work+0xe3a/0x1430
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.4.1-00199-QCAHKSWPL_SILICONZ-1
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00284.1-QCAHMTSWPL_V1.0_V2.0_SILICONZ-3
Fixes: aa80f12f3b ("wifi: ath12k: defer vdev creation for MLO")
Signed-off-by: Rameshkumar Sundaram <rameshkumar.sundaram@oss.qualcomm.com>
Reviewed-by: Vasanthakumar Thiagarajan <vasanthakumar.thiagarajan@oss.qualcomm.com>
Link: https://patch.msgid.link/20250618185635.750470-1-rameshkumar.sundaram@oss.qualcomm.com
Signed-off-by: Jeff Johnson <jeff.johnson@oss.qualcomm.com>
Some users and customers reported that their backup/copy tools started
to fail when the directory being copied contained symlink targets that
the client couldn't parse - even when those symlinks weren't followed.
Fix this by allowing lstat(2) and readlink(2) to succeed even when the
client can't resolve the symlink target, restoring old behavior.
Cc: linux-cifs@vger.kernel.org
Cc: stable@vger.kernel.org
Reported-by: Remy Monsen <monsen@monsen.cc>
Closes: https://lore.kernel.org/r/CAN+tdP7y=jqw3pBndZAGjQv0ObFq8Q=+PUDHgB36HdEz9QA6FQ@mail.gmail.com
Reported-by: Pierguido Lambri <plambri@redhat.com>
Fixes: 12b466eb52 ("cifs: Fix creating and resolving absolute NT-style symlinks")
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
When operating in split mode, it is a valid usecase to have
only one core enabled in the cluster. Remove exact core count
expecatation from the driver.
Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
Link: https://lore.kernel.org/r/20250618181933.1253033-1-tanmay.shah@amd.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Add unsafe accessors for the region for reading or writing large
blocks of data.
Reviewed-by: Andreas Hindborg <a.hindborg@kernel.org>
Signed-off-by: Abdiel Janulgue <abdiel.janulgue@gmail.com>
Reviewed-by: Alexandre Courbot <acourbot@nvidia.com>
Link: https://lore.kernel.org/r/20250602085444.1925053-4-abdiel.janulgue@gmail.com
[ Fix line length and slightly reword safety comment in doc-test of
CoherentAllocation::write(); fix formatting issue. - Danilo ]
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Sustained frequency when greater than or equal to 4Ghz on 64-bit devices
currently result in marking all frequencies as turbo. Address the turbo
frequency selection bug by fixing the truncation.
Fixes: a897575e79 ("firmware: arm_scmi: Add support for marking certain frequencies as turbo")
Signed-off-by: Sibi Sankar <quic_sibis@quicinc.com>
Message-Id: <20250514214719.203607-1-quic_sibis@quicinc.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
We could do better here by having the macros return `Result`,
so that we don't have to wrap these calls in a closure for
validation which is confusing.
Co-developed-by: Andreas Hindborg <a.hindborg@kernel.org>
Signed-off-by: Andreas Hindborg <a.hindborg@kernel.org>
Link: https://lore.kernel.org/rust-for-linux/87h63qhz4q.fsf@kernel.org/
Reviewed-by: Andreas Hindborg <a.hindborg@kernel.org>
Signed-off-by: Abdiel Janulgue <abdiel.janulgue@gmail.com>
Link: https://lore.kernel.org/r/20250602085444.1925053-3-abdiel.janulgue@gmail.com
[ Fix line length in dma_read!(). - Danilo ]
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
In the kernel, `consistent` and `coherent` are used interchangeably for the
region described in this api. Stick with `coherent` nomenclature
to show that dma_alloc_coherent() is being used, in addition to improving
the clarity in the DMA mapping attributes documentation.
Reviewed-by: Benno Lossin <benno.lossin@proton.me>
Signed-off-by: Abdiel Janulgue <abdiel.janulgue@gmail.com>
Reviewed-by: Andreas Hindborg <a.hindborg@kernel.org>
Link: https://lore.kernel.org/r/20250602085444.1925053-2-abdiel.janulgue@gmail.com
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
clang gets a bit confused by the code in the qed_mfw_process_tlv_req and
ends up spilling registers to the stack hundreds of times. When sanitizers
are enabled, this can end up blowing the stack warning limit:
drivers/net/ethernet/qlogic/qed/qed_mng_tlv.c:1244:5: error: stack frame size (1824) exceeds limit (1280) in 'qed_mfw_process_tlv_req' [-Werror,-Wframe-larger-than]
Apparently the problem is the complexity of qed_mfw_update_tlvs()
after inlining, and marking the four main branches of that function
as noinline_for_stack makes this problem completely go away, the stack
usage goes down to 100 bytes.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
During the hibernate entry sequence, ACP registers will be reset to
default values and acp ip will be completely powered off including acp
SoundWire pads. During resume sequence, if acp SoundWire pad keeper enable
register is not restored along with pad pulldown control register value,
then SoundWire manager links won't be powered on correctly results in
peripheral register access failures and completely audio function is
broken.
Add code to store the acp SoundWire pad keeper enable register and acp pad
pulldown ctrl register values before entering into suspend state and
restore the register values during resume sequence based on condition check
for acp SoundWire pad keeper enable register for ACP6.3, ACP7.0 & ACP7.1
platforms.
Fixes: 4916283880 ("ASoC: amd: ps: add callback functions for acp pci driver pm ops")
Signed-off-by: Vijendar Mukunda <Vijendar.Mukunda@amd.com>
Link: https://patch.msgid.link/20250623084630.3100279-1-Vijendar.Mukunda@amd.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Reading DPCD registers has side-effects in general. In particular
accessing registers outside of the link training register range
(0x102-0x106, 0x202-0x207, 0x200c-0x200f, 0x2216) is explicitly
forbidden by the DP v2.1 Standard, see
3.6.5.1 DPTX AUX Transaction Handling Mandates
3.6.7.4 128b/132b DP Link Layer LTTPR Link Training Mandates
Based on my tests, accessing the DPCD_REV register during the link
training of an UHBR TBT DP tunnel sink leads to link training failures.
Solve the above by using the DP_LANE0_1_STATUS (0x202) register for the
DPCD register access quirk.
Cc: <stable@vger.kernel.org>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250605082850.65136-2-imre.deak@intel.com
(cherry picked from commit a40c5d727b)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Ensure that we handle the case where task creation fails and pid->attr
was never accessed at all.
Signed-off-by: Christian Brauner <brauner@kernel.org>
During a hash resize operation the new private hash is stored in
mm_struct::futex_phash_new if the current hash can not be immediately
replaced.
The new hash must not be copied during fork() into the new task. Doing
so will lead to a double-free of the memory by the two tasks.
Initialize the mm_struct::futex_phash_new during fork().
Closes: https://lore.kernel.org/all/aFBQ8CBKmRzEqIfS@mozart.vkv.me/
Fixes: bd54df5ea7 ("futex: Allow to resize the private local hash")
Reported-by: Calvin Owens <calvin@wbinvd.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Calvin Owens <calvin@wbinvd.org>
Link: https://lkml.kernel.org/r/20250623083408.jTiJiC6_@linutronix.de
Some drivers (e.g. ice) don't enable any edges by default when external
timestamping is requested by the PTP_EXTTS_REQUEST ioctl, which makes
testptp -e unusable for testing hardware supported by these drivers.
Add -E option to specify if the rising, falling, or both edges should
be enabled by the ioctl.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use sysfs_emit() instead of snprintf() in version_show() function to
follow the preferred kernel API.
Signed-off-by: Ankit Chauhan <ankitchauhan2065@gmail.com>
Link: https://lore.kernel.org/20250619031536.19352-1-ankitchauhan2065@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
Add a compatible for the MediaTek MT8196 Chromebook SoC's
regulators over DVFSRC.
This SoC has only one regulator "dvfsrc-vcore" and different
values for vsel compared to the others.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://patch.msgid.link/20250623120144.109359-4-angelogioacchino.delregno@collabora.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The MediaTek Dimensity 1200 (MT6893) features the same DVFSRC
regulators as the other currently supported SoCs, but with a
different select value: add an array describing the possible
voltages for the VCORE and VSCP regulators, and assign it to
a new compatible for this SoC.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://patch.msgid.link/20250623120144.109359-3-angelogioacchino.delregno@collabora.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Add a compatible for the MediaTek Dimensity 1200 (MT6893) SoC's
regulators over DVFSRC.
This SoC uses different values for the vsel, hence it is not
compatible with the currently supported ones.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Acked-by: "Rob Herring (Arm)" <robh@kernel.org>
Link: https://patch.msgid.link/20250623120144.109359-2-angelogioacchino.delregno@collabora.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Add SoundWire machine with RT722 multi functional codec support for ACP7.0
and ACP7.1 platforms at sdw link0 for sof stack.
Signed-off-by: Vijendar Mukunda <Vijendar.Mukunda@amd.com>
Link: https://patch.msgid.link/20250620120942.1168827-2-Vijendar.Mukunda@amd.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The config file related to the memory windows start the numbering of
the MW from 1. The other NTB function does the same, yet the enumeration
defining the BARs of the vNTB function starts numbering the MW from 0.
Both numbering should be fine, but mixing the two is a bit confusing. The
configfs file being the interface with userspace, keep that stable and
consistently start the numbering of the MW from 1.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
[mani: commit message rewording]
Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Link: https://patch.msgid.link/20250603-pci-vntb-bar-mapping-v2-2-fc685a22ad28@baylibre.com
According the function documentation of epf_ntb_init_epc_bar(), the
function should return an error code on error. However, it returns -1 when
no BAR is available i.e., when pci_epc_get_next_free_bar() fails.
Return -ENOENT instead.
Fixes: e35f56bb03 ("PCI: endpoint: Support NTB transfer between RC and EP")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
[mani: changed err code to -ENOENT]
Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Link: https://patch.msgid.link/20250603-pci-vntb-bar-mapping-v2-1-fc685a22ad28@baylibre.com
DIV_ROUND_CLOSEST_ULL uses do_div(), which expects a 32-bit divisor.
When passing a 64-bit constant like CURVE2_MULTIPLIER, the value is
silently truncated to u32, potentially leading to incorrect results
on large divisors.
Replace DIV_ROUND_CLOSEST_ULL with DIV64_U64_ROUND_CLOSEST which correctly
handles full 64-bit division.
v2: Use DIV64_U64_ROUND_CLOSEST instead of div64_u64 macro. (Jani)
Fixes: 5947642004 ("drm/i915/display: Add support for SNPS PHY HDMI PLL algorithm for DG2")
Reported-by: Vas Novikov <vasya.novikov@gmail.com>
Closes: https://lore.kernel.org/all/8d7c7958-9558-4c8a-a81a-e9310f2d8852@gmail.com/
Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Cc: Suraj Kandpal <suraj.kandpal@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Vas Novikov <vasya.novikov@gmail.com>
Cc: stable@vger.kernel.org # v6.15+
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Link: https://lore.kernel.org/r/20250618130951.1596587-2-ankit.k.nautiyal@intel.com
(cherry picked from commit b300a175a1)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
By default, the driver relies on the default hardware defined value for the
Max Link Width (MLW) capability. But if the "num-lanes" DT property is
present, assume that the chip's default capability information is incorrect
or undesired, and use the specified value instead.
Signed-off-by: Jim Quinlan <james.quinlan@broadcom.com>
[mani: reworded the description and comments]
Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Link: https://patch.msgid.link/20250530224035.41886-3-james.quinlan@broadcom.com
The TBU clock belongs to the Translation Buffer Unit, part of the SMMU.
The ref clock is already being driven upstream through some of the
branches.
Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://patch.msgid.link/20250521-topic-8150_pcie_drop_clocks-v1-2-3d42e84f6453@oss.qualcomm.com
The TBU clock belongs to the Translation Buffer Unit, part of the SMMU.
The ref clock is already being driven upstream through some of the
branches.
Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://patch.msgid.link/20250521-topic-8150_pcie_drop_clocks-v1-1-3d42e84f6453@oss.qualcomm.com
Song Liu <song@kernel.org> says:
Introduce a new kfunc bpf_cgroup_read_xattr, which can read xattr from
cgroupfs nodes. The primary users are LSMs, cgroup programs, and sched_ext.
* patches from https://lore.kernel.org/20250623063854.1896364-1-song@kernel.org:
selftests/bpf: Add tests for bpf_cgroup_read_xattr
bpf: Mark cgroup_subsys_state->cgroup RCU safe
bpf: Introduce bpf_cgroup_read_xattr to read xattr of cgroup's node
kernfs: remove iattr_mutex
Link: https://lore.kernel.org/20250623063854.1896364-1-song@kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Add tests for different scenarios with bpf_cgroup_read_xattr:
1. Read cgroup xattr from bpf_cgroup_from_id;
2. Read cgroup xattr from bpf_cgroup_ancestor;
3. Read cgroup xattr from css_iter;
4. Use bpf_cgroup_read_xattr in LSM hook security_socket_connect.
5. Use bpf_cgroup_read_xattr in cgroup program.
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/20250623063854.1896364-5-song@kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Mark struct cgroup_subsys_state->cgroup as safe under RCU read lock. This
will enable accessing css->cgroup from a bpf css iterator.
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/20250623063854.1896364-4-song@kernel.org
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
BPF programs, such as LSM and sched_ext, would benefit from tags on
cgroups. One common practice to apply such tags is to set xattrs on
cgroupfs folders.
Introduce kfunc bpf_cgroup_read_xattr, which allows reading cgroup's
xattr.
Note that, we already have bpf_get_[file|dentry]_xattr. However, these
two APIs are not ideal for reading cgroupfs xattrs, because:
1) These two APIs only works in sleepable contexts;
2) There is no kfunc that matches current cgroup to cgroupfs dentry.
bpf_cgroup_read_xattr is generic and can be useful for many program
types. It is also safe, because it requires trusted or rcu protected
argument (KF_RCU). Therefore, we make it available to all program types.
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/20250623063854.1896364-3-song@kernel.org
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
All allocations of struct kernfs_iattrs are serialized through a global
mutex. Simply do a racy allocation and let the first one win. I bet most
callers are under inode->i_rwsem anyway and it wouldn't be needed but
let's not require that.
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/20250623063854.1896364-2-song@kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
The current KERN_WARNING level message for detecting absent memory
controllers is overly dramatic. The BIOS likely had valid reasons to
disable the memory controller (e.g. it isn't connected to any DIMM
slots on the motherboard for this system). So there's nothing actually
wrong that needs to be fixed.
Reduce the log level to KERN_DEBUG to eliminate the false warning.
Suggested-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Qiuxu Zhuo <qiuxu.zhuo@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20250618162307.1523736-2-qiuxu.zhuo@intel.com
Device ID '0xa740' is shared by i7-14700, i7-14700K, and i7-14700T.
Device ID '0xa704' is shared by i5-14600, i5-14600K, and i5-14600T.
Tested locally on my i7-14700K.
Signed-off-by: George Gaidarov <gdgaidarov+lkml@gmail.com>
Reviewed-by: Qiuxu Zhuo <qiuxu.zhuo@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20250529162933.1228735-1-gdgaidarov+lkml@gmail.com
Bartlett Lake-S is a derivative of Raptor Lake-S and is optimized for
IoT/Edge applications. It shares the same memory controller registers
as Raptor Lake-S. Add compute die IDs of Bartlett Lake-S and reuse the
configuration data of Raptor Lake-S for Bartlett Lake-S EDAC support.
Signed-off-by: Qiuxu Zhuo <qiuxu.zhuo@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20250502013900.343430-1-qiuxu.zhuo@intel.com
Zhang Yi <yi.zhang@huawei.com> says:
Currently, we can use the fallocate command to quickly create a
pre-allocated file. However, on most filesystems, such as ext4 and XFS,
fallocate create pre-allocation blocks in an unwritten state, and the
FALLOC_FL_ZERO_RANGE flag also behaves similarly. The extent state must
be converted to a written state when the user writes data into this
range later, which can trigger numerous metadata changes and consequent
journal I/O. This may leads to significant write amplification and
performance degradation in synchronous write mode. Therefore, we need a
method to create a pre-allocated file with written extents that can be
used for pure overwriting. At the monent, the only method available is
to create an empty file and write zero data into it (for example, using
'dd' with a large block size). However, this method is slow and consumes
a considerable amount of disk bandwidth, we must pre-allocate files in
advance but cannot add pre-allocated files while user business services
are running.
Fortunately, with the development and more and more widely used of
flash-based storage devices, we can efficiently write zeros to SSDs
using the unmap write zeroes command if the devices do not write
physical zeroes to the media. For example, if SCSI SSDs support the
UMMAP bit or NVMe SSDs support the DEAC bit[1], the write zeroes command
does not write actual data to the device, instead, NVMe converts the
zeroed range to a deallocated state, which works fast and consumes
almost no disk write bandwidth. Consequently, this feature can provide
us with a faster method for creating pre-allocated files with written
extents and zeroed data. However, please note that this may be a
best-effort optimization rather than a mandatory requirement, some
devices may partially fall back to writing physical zeroes due to
factors such as receiving unaligned commands.
This series aims to implement this by:
1. Introduce a new feature BLK_FEAT_WRITE_ZEROES_UNMAP to the block
device queue limit features, which indicates whether the storage is
device explicitly supports the unmapped write zeroes command. This
flag should be set to 1 by the driver if the attached disk supports
this command.
2. Introduce a queue limit flag, BLK_FLAG_WRITE_ZEROES_UNMAP_DISABLED,
along with a corresponding sysfs entry. Users can query the support
status of the unmap write zeroes operation and disable this operation
if the write zeroes operation is very slow.
/sys/block/<disk>/queue/write_zeroes_unmap
3. Introduce a new flag, FALLOC_FL_WRITE_ZEROES, into the fallocate.
Filesystems that support this operation should allocate written
extents and issue zeroes to the specified range of the device. For
local block device filesystems, this operation should depend on the
write_zeroes_unmap operaion of the underlying block device. It should
return -EOPNOTSUPP if the device doesn't enable unmap write zeroes
operaion.
This series implements the BLK_FEAT_WRITE_ZEROES_UNMAP feature and
BLK_FLAG_WRITE_ZEROES_UNMAP_DISABLED flag for SCSI, NVMe and
device-mapper drivers, and add the FALLOC_FL_WRITE_ZEROES and
STATX_ATTR_WRITE_ZEROES_UNMAP support for ext4 and raw bdev devices.
Any comments are welcome.
I've tested performance with this series on ext4 filesystem on my
machine with an Intel Xeon Gold 6248R CPU, a 7TB KCD61LUL7T68 NVMe SSD
which supports unmap write zeroes command with the Deallocated state
and the DEAC bit. Feel free to give it a try.
0. Ensure the NVMe device supports WRITE_ZERO command.
$ cat /sys/block/nvme5n1/queue/write_zeroes_max_bytes
8388608
$ nvme id-ns -H /dev/nvme5n1 | grep -i -A 3 "dlfeat"
dlfeat : 25
[4:4] : 0x1 Guard Field of Deallocated Logical Blocks is set to CRC
of The Value Read
[3:3] : 0x1 Deallocate Bit in the Write Zeroes Command is Supported
[2:0] : 0x1 Bytes Read From a Deallocated Logical Block and its
Metadata are 0x00
1. Compare 'dd' and fallocate with unmap write zeroes, the later one is
significantly faster than 'dd'.
Create a 1GB and 10GB zeroed file.
$dd if=/dev/zero of=foo bs=2M count=$count oflag=direct
$time fallocate -w -l $size bar
#1G
dd: 0.5s
FALLOC_FL_WRITE_ZEROES: 0.17s
#10G
dd: 5.0s
FALLOC_FL_WRITE_ZEROES: 1.7s
2. Run fio overwrite and fallocate with unmap write zeroes
simultaneously, fallocate has little impact on write bandwidth and
only slightly affects write latency.
a) Test bandwidth costs.
$ fio -directory=/test -direct=1 -iodepth=10 -fsync=0 -rw=write \
-numjobs=10 -bs=2M -ioengine=libaio -size=20G -runtime=20 \
-fallocate=none -overwrite=1 -group_reportin -name=bw_test
Without background zero range:
bw (MiB/s): min= 2068, max= 2280, per=100.00%, avg=2186.40
With background zero range:
bw (MiB/s): min= 2056, max= 2308, per=100.00%, avg=2186.20
b) Test write latency costs.
$ fio -filename=/test/foo -direct=1 -iodepth=1 -fsync=0 -rw=write \
-numjobs=1 -bs=4k -ioengine=psync -size=5G -runtime=20 \
-fallocate=none -overwrite=1 -group_reportin -name=lat_test
Without background zero range:
lat (nsec): min=9269, max=71635, avg=9840.65
With a background zero range:
lat (usec): min=9, max=982, avg=11.03
3. Compare overwriting in a pre-allocated unwritten file and a written
file in O_DSYNC mode. Write to a file with written extents is much
faster.
# First mkfs and create a test file according to below three cases,
# and then run fio.
$ fio -filename=/test/foo -direct=1 -iodepth=1 -fdatasync=1 \
-rw=write -numjobs=1 -bs=4k -ioengine=psync -size=5G \
-runtime=20 -fallocate=none -group_reportin -name=test
unwritten file: IOPS=20.1k, BW=78.7MiB/s
unwritten file + fast_commit: IOPS=42.9k, BW=167MiB/s
written file: IOPS=98.8k, BW=386MiB/s
* patches from https://lore.kernel.org/20250619111806.3546162-1-yi.zhang@huaweicloud.com:
ext4: add FALLOC_FL_WRITE_ZEROES support
block: add FALLOC_FL_WRITE_ZEROES support
block: factor out common part in blkdev_fallocate()
fs: introduce FALLOC_FL_WRITE_ZEROES to fallocate
dm: clear unmap write zeroes limits when disabling write zeroes
scsi: sd: set max_hw_wzeroes_unmap_sectors if device supports SD_ZERO_*_UNMAP
nvmet: set WZDS and DRB if device enables unmap write zeroes operation
nvme: set max_hw_wzeroes_unmap_sectors if device supports DEAC bit
block: introduce max_{hw|user}_wzeroes_unmap_sectors to queue limits
Link: https://lore.kernel.org/20250619111806.3546162-1-yi.zhang@huaweicloud.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
Add support for FALLOC_FL_WRITE_ZEROES if the underlying device enable
the unmap write zeroes operation. This first allocates blocks as
unwritten, then issues a zero command outside of the running journal
handle, and finally converts them to a written state.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-10-yi.zhang@huaweicloud.com
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Add support for FALLOC_FL_WRITE_ZEROES, if the block device enables the
unmap write zeroes operation, it will issue a write zeroes command.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-9-yi.zhang@huaweicloud.com
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Only the flags passed to blkdev_issue_zeroout() differ among the two
zeroing branches in blkdev_fallocate(). Therefore, do cleanup by
factoring them out.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-8-yi.zhang@huaweicloud.com
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
With the development of flash-based storage devices, we can quickly
write zeros to SSDs using the WRITE_ZERO command if the devices do not
actually write physical zeroes to the media. Therefore, we can use this
command to quickly preallocate a real all-zero file with written
extents. This approach should be beneficial for subsequent pure
overwriting within this file, as it can save on block allocation and,
consequently, significant metadata changes, which should greatly improve
overwrite performance on certain filesystems.
Therefore, introduce a new operation FALLOC_FL_WRITE_ZEROES to
fallocate. This flag is used to convert a specified range of a file to
zeros by issuing a zeroing operation. Blocks should be allocated for the
regions that span holes in the file, and the entire range is converted
to written extents. If the underlying device supports the actual offload
write zeroes command, the process of zeroing out operation can be
accelerated. If it does not, we currently don't prevent the file system
from writing actual zeros to the device. This provides users with a new
method to quickly generate a zeroed file, users no longer need to write
zero data to create a file with written extents.
Users can determine whether a disk supports the unmap write zeroes
feature through querying this sysfs interface:
/sys/block/<disk>/queue/write_zeroes_unmap_max_hw_bytes
Users can also enable or disable the unmap write zeroes operation
through this sysfs interface:
/sys/block/<disk>/queue/write_zeroes_unmap_max_bytes
Finally, this flag cannot be specified in conjunction with the
FALLOC_FL_KEEP_SIZE since allocating written extents beyond file EOF is
not permitted. In addition, filesystems that always require out-of-place
writes should not support this flag since they still need to allocated
new blocks during subsequent overwrites.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-7-yi.zhang@huaweicloud.com
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
The unmap write zeroes limits have been set to the stacking queue limits
by default in blk_set_stacking_limits() and blk_stack_limits(), but it
should be cleared if any underlying device does not support it.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-6-yi.zhang@huaweicloud.com
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
When the device supports the Write Zeroes command and the zeroing mode
is set to SD_ZERO_WS16_UNMAP or SD_ZERO_WS10_UNMAP, this means that the
device supports unmap Write Zeroes, so set the corresponding
max_hw_write_zeroes_unmap_sectors to max_write_zeroes_sectors on the
device's queue limit.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-5-yi.zhang@huaweicloud.com
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Set the WZDS and DRB bits to the namespace dlfeat if the underlying
block device enables the unmap write zeroes operation, make the nvme
target device supports the unmap write zeroes command.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-4-yi.zhang@huaweicloud.com
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
When the device supports the Write Zeroes command and the DEAC bit, it
indicates that the deallocate bit in the Write Zeroes command is
supported, and the bytes read from a deallocated logical block are
zeroes. This means the device supports unmap Write Zeroes operation, so
set the max_hw_wzeroes_unmap_sectors to max_write_zeroes_sectors on the
device's queue limit.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-3-yi.zhang@huaweicloud.com
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Currently, disks primarily implement the write zeroes command (aka
REQ_OP_WRITE_ZEROES) through two mechanisms: the first involves
physically writing zeros to the disk media (e.g., HDDs), while the
second performs an unmap operation on the logical blocks, effectively
putting them into a deallocated state (e.g., SSDs). The first method is
generally slow, while the second method is typically very fast.
For example, on certain NVMe SSDs that support NVME_NS_DEAC, submitting
REQ_OP_WRITE_ZEROES requests with the NVME_WZ_DEAC bit can accelerate
the write zeros operation by placing disk blocks into a deallocated
state, which opportunistically avoids writing zeroes to media while
still guaranteeing that subsequent reads from the specified block range
will return zeroed data. This is a best-effort optimization, not a
mandatory requirement, some devices may partially fall back to writing
physical zeroes due to factors such as misalignment or being asked to
clear a block range smaller than the device's internal allocation unit.
Therefore, the speed of this operation is not guaranteed.
It is difficult to determine whether the storage device supports unmap
write zeroes operation. We cannot determine this by only querying
bdev_limits(bdev)->max_write_zeroes_sectors. Therefore, first, add a new
hardware queue limit parameters, max_hw_wzeroes_unmap_sectors, to
indicate whether a device supports this unmap write zeroes operation.
Then, add two new counterpart software queue limits,
max_wzeroes_unmap_sectors and max_user_wzeroes_unmap_sectors, which
allow users to disable this operation if the speed is very slow on some
sepcial devices.
Finally, for the stacked devices cases, initialize these two parameters
to UINT_MAX. This operation should be enabled by both the stacking
driver and all underlying devices.
Thanks to Martin K. Petersen for optimizing the documentation of the
write_zeroes_unmap sysfs interface.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Link: https://lore.kernel.org/20250619111806.3546162-2-yi.zhang@huaweicloud.com
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
