ieee80211_tx_control_port() can be called from nl80211_tx_control_port()
in normal process context via netlink sendmsg(), where BH/preemption
is not disabled. dev_sw_netstats_tx_add() uses this_cpu_ptr() internally,
which requires preemption to be disabled, triggering a
"BUG: using smp_processor_id() in preemptible code" warning.
Fix this by moving the local_bh_disable()/local_bh_enable() section to
also cover dev_sw_netstats_tx_add() and ieee80211_tpt_led_trig_tx().
Fixes: d5a014204a ("wifi: mac80211: tx: simplify control port frame transmission")
Reported-by: syzbot+d979bd35c8a76fd1b6f5@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=d979bd35c8a76fd1b6f5
Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
Link: https://patch.msgid.link/20260910041047.24437-1-kartikey406@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
hwsim_mcast_config_msg() falls back to genlmsg_multicast() when it has no
genl_info, and that only reaches init_net. A radio in another namespace
therefore has its DEL_RADIO announced in init_net, while its own namespace
hears nothing.
Multicast to wiphy_net() instead, the net that GET and DUMP already filter
on. Tested with a config-group listener in each namespace, on a base and a
patched kernel.
Assisted-by: Claude:claude-opus-5 codeql
Signed-off-by: Maoyi Xie <maoyixie.tju@gmail.com>
Link: https://patch.msgid.link/20260904140300.3402053-1-maoyixie.tju@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When bringing a monitor interface up and down again, the
refcount is leaked for active monitor interfaces, so the
hardware filter is never reset.
Move the decrement out so it mirrors the increment.
Signed-off-by: Devin Wittmayer <lucid_duck@justthetip.ca>
Link: https://patch.msgid.link/20260904202817.12564-1-lucid_duck@justthetip.ca
[rewrite commit message better]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Drivers without FAST_XMIT and hardware scan support currently drop TX
frames during software scanning while being off-channel. This happens
in ieee80211_tx_h_check_assoc() before frames reach internal TX queue.
Move the off-channel check to invoke_tx_handlers_late(), after
ieee80211_queue_skb(). Frames can then remain queued and be transmitted
when the interface returns on-channel. This how TX works currently when
code goes via ieee80211_xmit_fast() path.
Below is rt2x00 driver ping output example while scanning, before and
after the fix. After the fix, packets are no longer lost. Some have
increased latency, what is expected behavior.
BEFORE:
64 bytes from 192.168.0.1: icmp_seq=29 ttl=64 time=3.61 ms
64 bytes from 192.168.0.1: icmp_seq=31 ttl=64 time=3.33 ms # Missed 30
64 bytes from 192.168.0.1: icmp_seq=32 ttl=64 time=4.03 ms
64 bytes from 192.168.0.1: icmp_seq=34 ttl=64 time=3.10 ms # Missed 33
64 bytes from 192.168.0.1: icmp_seq=36 ttl=64 time=3.04 ms # Missed 35
64 bytes from 192.168.0.1: icmp_seq=38 ttl=64 time=3.05 ms # Missed 37
64 bytes from 192.168.0.1: icmp_seq=40 ttl=64 time=4.14 ms # Missed 39
64 bytes from 192.168.0.1: icmp_seq=41 ttl=64 time=4.50 ms
AFTER:
64 bytes from 192.168.0.1: icmp_seq=29 ttl=64 time=3.40 ms
64 bytes from 192.168.0.1: icmp_seq=30 ttl=64 time=6.10 ms
64 bytes from 192.168.0.1: icmp_seq=31 ttl=64 time=4.69 ms
64 bytes from 192.168.0.1: icmp_seq=32 ttl=64 time=4.49 ms
64 bytes from 192.168.0.1: icmp_seq=33 ttl=64 time=148 ms
64 bytes from 192.168.0.1: icmp_seq=34 ttl=64 time=6.83 ms
64 bytes from 192.168.0.1: icmp_seq=35 ttl=64 time=8.66 ms
64 bytes from 192.168.0.1: icmp_seq=36 ttl=64 time=13.0 ms
64 bytes from 192.168.0.1: icmp_seq=37 ttl=64 time=4.78 ms
64 bytes from 192.168.0.1: icmp_seq=38 ttl=64 time=3.85 ms
Cc: <stable@vger.kernel.org>
Signed-off-by: Stanislaw Gruszka <stf_xl@wp.pl>
Link: https://patch.msgid.link/20260908124651.2995-1-stf_xl@wp.pl
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Without this ieee80211_select_key_8023() is never called and wifi never
connects.
Fixes: 0e80db0866 ("wifi: mac80211: fix key selection for encap offload frames")
Signed-off-by: Bert Karwatzki <spasswolf@web.de>
Link: https://patch.msgid.link/20260908094341.12196-1-spasswolf@web.de
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
ieee80211_debugfs_rename_netdev() renames sdata->vif.debugfs_dir
without sharing synchronization with ieee80211_debugfs_recreate_netdev().
When MLO link updates switch an interface to or from MLO, the recreate
path can remove and rebuild that dentry while a concurrent netdev rename
still uses the stale pointer.
Serialize the rename helper with the same wiphy mutex that protects the
recreate path so the dentry cannot be removed while debugfs_change_name()
is operating on it.
Reported-by: Vega <vega@nebusec.ai>
Signed-off-by: Zhiling Zou <zhilinz@nebusec.ai>
Link: https://patch.msgid.link/76eecaf74ca2d4e4aedf46c3ee7505917433468b.1785729184.git.zhilinz@nebusec.ai
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When packet sockets use PACKET_QDISC_BYPASS, mac80211 can parse the
radiotap header while selecting the queue and again before transmit.
If the header includes an FCS, both parses trim the skb. Validate the FCS
on both parses, but trim it only before transmit. This keeps queue
selection read-only and avoids trimming shared skb data twice.
Fixes: cb17ed29a7 ("mac80211: parse radiotap header when selecting Tx queue")
Signed-off-by: Mariano Baragiola <mbaragiola@linux.com>
Link: https://patch.msgid.link/20260817233521.2747027-1-mbaragiola@linux.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
As per IEEE Std 802.11-2024, subclause 26.17.2.2, a 6 GHz AP can transmit
a Beacon frame using a non-HT duplicate PPDU, so that stations scanning
only PSC channels can discover it. Currently, there is no mechanism
to request the driver to transmit non-HT duplicate Beacon frames.
In [1], changes were made to parse the Beacon from user-space to set
the non-HT duplicate flag in the kernel. However, it was suggested that
non-HT transmission be part of the Beacon TX rate settings. Hence, add
NL80211_TXRATE_6GHZ_NON_HT_DUP as a flag attribute under
NL80211_ATTR_TX_RATES to let user-space request the driver to transmit
non-HT duplicate Beacons.
Also add validation to reject the Beacon if non-HT duplicate Beacon
transmission is enabled when legacy rates are not configured.
[1] https://lore.kernel.org/all/1644914581-21682-1-git-send-email-quic_ramess@quicinc.com/
Signed-off-by: Manish Dharanenthiran <manish.dharanenthiran@oss.qualcomm.com>
Link: https://patch.msgid.link/20260805-dup-beacon-v1-1-68574869c896@oss.qualcomm.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The two scnprintf arguments for fq_overlimit and fq_overmemory are in the
wrong order, which results in the aqm debugfs file showing the overmemory
count under the fq_overlimit label and vice versa.
Fixes: 2a4e675d88 ("mac80211: Export fq memory limit information in debugfs")
Reviewed-by: Toke Høiland-Jørgensen <toke@toke.dk>
Assisted-by: Claude:claude-opus-5
Signed-off-by: Julius Bairaktaris <julius@bairaktaris.de>
Link: https://patch.msgid.link/20260904090626.92727-1-julius@bairaktaris.de
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
check_fwfile_format() reads each block header and trusts its data length
before proving that both fit in the firmware image. A truncated header or
oversized block can therefore pass malformed extents to the download path.
Require the remaining image to contain the complete header and payload
before accepting each block.
Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn>
Link: https://patch.msgid.link/20260830140420.26399-1-pengpeng@iscas.ac.cn
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
check_fwfile_format() reads each block header and trusts its data length
before proving that both fit in the firmware image. A truncated header or
oversized block can therefore pass malformed extents to the download path.
Require the remaining image to contain the complete header and payload
before accepting each block.
Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn>
Link: https://patch.msgid.link/20260830124810.97223-1-pengpeng@iscas.ac.cn
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
WRT3200ACM and WRT32x routers always report a fixed US regulatory domain
from the mwifiex ROM, which conflicts with the platform regulatory
configuration on units sold outside the US market. For example:
- phy0 mwlwifi 5 GHz, hard-coded region 98 EU mapped to FR
- phy1 mwlwifi 2.4 GHz, hard-coded region 98 EU mapped to FR
- phy2 mwifiex 2.4 and 5 GHz, hard-coded region US
When the system boots, it detects a conflict between the user selected
region and the radios, e.g. BG FR US, and applies extreme restrictions,
preventing phy0 from starting on any DFS channel. phy2 works.
Add a device-tree flag, which allows affected devices, where the radio
conflicts with the platform to ignore the incorrect regulatory hint in
the ROM of the radio, so the platform regulatory domain can be used.
This does not allow the user to change the regulatory configuration of
any radio. It only allows phy0 to start and operate according to its
certified configuration.
The change affects only platforms which explicitly define this flag, and
helps overcome the incorrect behaviour of proprietary radio firmware.
The change has been tested on OpenWrt and fixes the issue where
phy0 does not start on WRT3200ACM-EU when a DFS channel is used
[1] https://github.com/kaloz/mwlwifi/issues/173#issuecomment-307879699
[2] https://github.com/openwrt/openwrt/issues/9956
Signed-off-by: Georgi Valkov <gvalkov@gmail.com>
Link: https://patch.msgid.link/20260827175850.89395-2-gvalkov@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
WRT3200ACM and WRT32x routers always report a fixed US regulatory domain
from the mwifiex ROM, which conflicts with the platform regulatory
configuration on units sold outside the US market. For example:
- phy0 mwlwifi 5 GHz, hard-coded region 98 EU mapped to FR
- phy1 mwlwifi 2.4 GHz, hard-coded region 98 EU mapped to FR
- phy2 mwifiex 2.4 and 5 GHz, hard-coded region US
When the system boots, it detects a conflict between the user selected
region and the radios, e.g. BG FR US, and applies extreme restrictions,
preventing phy0 from starting on any DFS channel. phy2 works.
Add a device-tree flag, which allows affected devices, where the radio
conflicts with the platform to ignore the incorrect regulatory hint in
the ROM of the radio, so the platform regulatory domain can be used.
This does not allow the user to change the regulatory configuration of
any radio. It only allows phy0 to start and operate according to its
certified configuration.
[1] https://github.com/kaloz/mwlwifi/issues/173#issuecomment-307879699
[2] https://github.com/openwrt/openwrt/issues/9956
Signed-off-by: Georgi Valkov <gvalkov@gmail.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Link: https://patch.msgid.link/20260827175850.89395-1-gvalkov@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The WFxxx device provides an SDIO interface. In addition to the usual
in-band SDIO IRQ, the device also supports an out-of-band IRQ line.
However, using the OOB IRQ with SDIO requires to poll the control
register during the device initialization. This polling is unreliable:
if an IRQ is raised while the host reads the control register, the
device is lost and only a reset allows to recover it.
There is no way to fix this in software. So, warn the user that this
setup should be reserved to debug.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Link: https://patch.msgid.link/20260901083548.113131-3-jerome.pouiller@silabs.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The WFxxx device provides an SDIO interface. In addition to the usual
in-band SDIO IRQ, the device also supports out-of-band IRQ line. OOB IRQ
requires a specific initialization described in commit f00dc1d789
("staging: wfx: poll IRQ during init"):
When the chip starts in SDIO mode, the external IRQ (aka Out-Of-Band
IRQ) cannot be used before to configure it. Therefore, the first
exchanges with the chip have to be done without the OOB IRQ.
This patch allow to poll the data until the OOB IRQ is correctly
setup. In order to keep the code simpler, this patch also poll data
even if OOB IRQ is not used.
Commit 57aa557f11 ("staging: wfx: introduce a way to poll IRQ")
announce some limitation about the data polling mechanism:
This function [wfx_bh_poll_irq()] must used with care: if an IRQ
fires while the host reads control register, the IRQ can be lost.
However, with last analysis, it seems it does not only impact the IRQ.
When this condition happen the hardware is lost and won't reply the next
commands. The only way to recover the error is to pull the reset pin and
bind the device again.
Currently wfx_bh_poll_irq() is only used during the two first exchanges
with the hardware. Therefore, it has been reported to only happen after
a few hours of bind/rebind in a loop. This explain why it has not been
reported before.
Fortunately, wfx_bh_poll_irq() is only required for OOB IRQ, which is
probably very marginal. The In-Band IRQ case also uses this function,
but only to simplify the code.
This patch limits the use of wfx_bh_poll_irq() to the OOB IRQ. The bus
driver now tells wfx_probe() whether polling is necessary. When it is
not (in-band SDIO IRQ and SPI), the IRQ is subscribed before the first
exchange with the device and wfx_bh_poll_irq() is never called. When it
is (SDIO with OOB IRQ), the behavior is unchanged: the control register
is polled until the PDS file has configured the IRQ line.
To conclude, In-Band users are now safe. Users of OOB IRQ will be still
impacted by the bug, but there is nothing we can do (and nobody
complained during the last 6 years).
Fixes: f00dc1d789 ("staging: wfx: poll IRQ during init")
Reported-by: Gerard Salvatella <gerard.salvatella@siemens.com>
Reported-by: Lukas Stockmann <lukas.stockmann@siemens.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Reviewed-by: Alexander Sverdlin <alexander.sverdlin@siemens.com>
Tested-by: Alexander Sverdlin <alexander.sverdlin@siemens.com>
Link: https://patch.msgid.link/20260901083548.113131-2-jerome.pouiller@silabs.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The ack skb has to be stored while the frame is still in 802.3
format, since that is what is reported back to userspace. It's
currently done in ieee80211_build_hdr(), but that complicates
it a bit.
Thanks the the previous ieee80211_lookup_ra_sta() change, it's
now known whether or not the frame will be group addressed on
the air, which is needed for this condition.
Also simplify callers of ieee80211_store_ack_skb() by filling
the TX info directly and fix the free in ieee80211_build_hdr()
to use ieee80211_free_txskb(), reporting the drop to userspace
rather than being quiet about it.
This also lets use validate the link ID for control port TX in
that code itself, rather than interleaving it into build_hdr().
Link: https://patch.msgid.link/20260803141309.292711-16-johannes@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
ieee80211_build_hdr() takes the info and control flags to use as
arguments, but that leads to some messy code and missed cleanup.
Fill them in the SKB before calling ieee80211_build_hdr().
Link: https://patch.msgid.link/20260803141309.292711-15-johannes@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
ieee80211_lookup_ra_sta() returns ERR_PTR(-ENOENT) both when the
802.11 frame will be is group addressed and when it's going to
an unknown station.
Keep reporting an ERR_PTR() for group addressed but set to NULL
for individually addressed frames so the callers can use it to
distinguish later. This is already explicitly the case for all
interface types except mesh, and callers already tolerate (but
don't distinguish) both cases.
Also change the ieee80211_lookup_ra_sta() to be static.
Link: https://patch.msgid.link/20260803141309.292711-14-johannes@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
ieee80211_build_hdr() unshares the skb halfway through, after it
has already looked at it but before it writes anything. That's not
a problem, but complicates the callers and requires passing the
flags to the function, etc.
Move the unsharing to the only caller that matters, the new one
from the previous patch has a fresh SKB and doesn't need it and
the TDLS callers via ieee80211_tdls_build_mgmt_packet_data() are
also fresh SKBs.
Link: https://patch.msgid.link/20260803141309.292711-13-johannes@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
ieee80211_tx_control_port() already does most of the necessary work,
much of which is repeated by __ieee80211_subif_start_xmit(). Most
work in the latter function is completely unnecessary since these
are special frames we built directly here (long enough, etc.)
Call ieee80211_build_hdr() and ieee80211_xmit() directly instead.
It also means some errors can now be reported to userspace rather
than frames silently getting dropped, but that's a good thing.
Link: https://patch.msgid.link/20260803141309.292711-11-johannes@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
In ieee80211_build_hdr(), IEEE80211_TX_CTL_REQ_TX_STATUS is checked
in ctrl_flags, which is wrong - IEEE80211_TX_CTL_REQ_TX_STATUS is in
info->flags, so the check is for IEEE80211_TX_CTRL_PORT_CTRL_PROTO
instead. This works out OK because the only place setting _that_ at
a point before this gets called is in fact ieee80211_tx_control_port()
which was intended here. This is actually needed for correctness, if
jt were checking info->flags instead, TDLS would fill/leak the status
frames created here.
However, we can just fix it by using the cookie instead, that'll be
set for PREAUTH frames via ieee80211_tx_control_port() as well, and
no other frames.
Additionally, the code also erroneously set it, but that just caused
preauth frames to get it set, since storing the ACK SKB will also set
the correct flag. Drop this assignment entirely.
Fixes: a7528198ad ("mac80211: support control port TX status reporting")
Link: https://patch.msgid.link/20260803141309.292711-10-johannes@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
This function isn't used by any driver, and we may want to
change it in the future. Don't expose it. Drivers may still
do the same determination based on the flags, but if we'd
ever change the encapsulation internally etc. (which I'm
considering) then the use in mac80211 may need to change,
or the function itself.
Link: https://patch.msgid.link/20260803135248.cd23ccac65d7.I1c238c6abe39f5a1ff1b11d8556599d01dd03ff0@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
For 802.11 frames, ieee80211_tx_h_select_key() drops a data frame
that ended up without a key if the station requires encryption, so
that traffic can never leak out unprotected. Encapsulation offload
erroneously didn't get this code path. Fix that.
Fixes: 50ff477a86 ("mac80211: add 802.11 encapsulation offloading support")
Link: https://patch.msgid.link/20260802085818.1336616-7-johannes@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
ieee80211_tx_h_select_key() assumes the frame is in 802.11 format
for picking the key, at least when there's no pairwise TK for the
STA to transmit with.
For unicast this is likely not relevant because if there's no
pairwise TK for the STA there's probably not going to be any other
key either. But multicast encapsulation offload frames have no
STA, so it looks at the ethernet header as if it was 802.11. This
may not matter very much since multicast encapsulation offload is
likely to ignore the key selection, but it's still wrong.
Create a separate selection function that doesn't (need to) look
at the frame header - it's data anyway, so either unicast to a
station (with sta pointer) or multicast. For MLO, the driver has
to duplicate the frame and select the key anyway, so just skip
it in that case entirely.
Fixes: 50ff477a86 ("mac80211: add 802.11 encapsulation offloading support")
Link: https://patch.msgid.link/20260802105818.0feb7ad61047.Ia506ca211176a3c466bc60e4cc3e506df34a4a67@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The check in ieee80211_tx_dequeue() reads frame_control, addr1 and
addr2 out of the skb, but for encapsulation offload that's just a
random part of the ethernet addresses, so dropping depends on bits
in the destination address. This is obviously wrong.
Refactor the check and short-circuit for ethernet format frames,
it only needs the port-control flag check since those are locally
generated and not forwarded frames.
Fixes: 50ff477a86 ("mac80211: add 802.11 encapsulation offloading support")
Link: https://patch.msgid.link/20260802105818.dd5da579813d.I51ed0121154bb652bebc97bb0597b1fae9ac978f@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Store the per-descriptor metadata in the DMA ring allocation instead of
allocating it separately.
This ties the metadata lifetime directly to the ring, removes a separate
allocation failure path, and keeps the descriptor count available for
__counted_by() bounds checking.
Assisted-by: Codex:GPT-5.5
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://patch.msgid.link/20260511041934.378209-1-rosenp@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Embed the RX reorder pointer array in struct mwifiex_rx_reorder_tbl
instead of allocating it separately.
This ties the array to the reorder table lifetime and removes a
separate allocation and cleanup path.
Change the void pointer type to the proper one for clarity.
Assisted-by: Codex:GPT-5.5
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://patch.msgid.link/20260727003035.118047-1-rosenp@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When initialising the rate control system, ieee80211_s1g_sta_rate_init()
is called for S1G stations during association where it initialises a
last rate of type STA_STATS_RATE_TYPE_S1G but does not initialise
nss, mcs bw etc. Since the last_rate no longer has a value equivalent
to STA_STATS_RATE_INVALID, the invalid rate will be decoded and
the NL80211_STA_INFO_RX_BITRATE flag will be set.
As a result, there is a period during association before the first
data frame is rx'ed where the seeded value is overwritten by
ieee80211_rx_h_sta_process() while this invalid last_rate lives. If
the sta is desroyed during this process, nl80211_put_sta_rate() will
call into cfg80211_calculate_bitrate() and attempt to calculate
the bitrate using the invalid last_rate leading to a WARN:
4,336,675067400,-;------------[ cut here ]------------
4,337,675067524,-;WARNING: net/wireless/util.c:1960 at cfg80211_calculate_bitrate+0x4d8/0x8b0 [cfg80211], CPU#3: kworker/u16:3/874
4,338,675068176,-;invalid rate bw=0, mcs=0, nss=0
[snip loaded modules]
4,341,675068954,-;Hardware name: BCM2711
4,342,675068961,-;Workqueue: events_unbound cfg80211_wiphy_work [cfg80211]
4,343,675069386,-;Call trace:
4,344,675069392,-; unwind_backtrace from show_stack+0x10/0x14
4,345,675069440,-; show_stack from dump_stack_lvl+0xa8/0xb8
4,346,675069475,-; dump_stack_lvl from __warn+0x90/0x224
4,347,675069517,-; __warn from warn_slowpath_fmt+0x130/0x1a8
4,348,675069559,-; warn_slowpath_fmt from cfg80211_calculate_bitrate+0x4d8/0x8b0 [cfg80211]
4,349,675070005,-; cfg80211_calculate_bitrate [cfg80211] from nl80211_put_sta_rate+0x48/0x590 [cfg80211]
4,350,675070819,-; nl80211_put_sta_rate [cfg80211] from nl80211_put_sta_info_common+0x4f4/0xd50 [cfg80211]
4,351,675071616,-; nl80211_put_sta_info_common [cfg80211] from nl80211_send_station.constprop.0+0x114/0x424 [cfg80211]
4,352,675072427,-; nl80211_send_station.constprop.0 [cfg80211] from cfg80211_del_sta_sinfo+0xc4/0x368 [cfg80211]
4,353,675073222,-; cfg80211_del_sta_sinfo [cfg80211] from __sta_info_destroy_part2+0x108/0x2c0 [mac80211]
4,354,675074178,-; __sta_info_destroy_part2 [mac80211] from sta_info_destroy_addr+0x4c/0x88 [mac80211]
4,355,675075243,-; sta_info_destroy_addr [mac80211] from ieee80211_destroy_assoc_data+0x194/0x2d8 [mac80211]
4,356,675075920,-; ieee80211_destroy_assoc_data [mac80211] from ieee80211_rx_mgmt_assoc_resp+0x220/0x1d4c [mac80211]
4,357,675076558,-; ieee80211_rx_mgmt_assoc_resp [mac80211] from ieee80211_iface_work+0x1c0/0x5d0 [mac80211]
4,358,675077187,-; ieee80211_iface_work [mac80211] from cfg80211_wiphy_work+0x110/0x368 [cfg80211]
4,359,675077763,-; cfg80211_wiphy_work [cfg80211] from process_one_work+0x260/0x868
4,360,675078022,-; process_one_work from worker_thread+0x1bc/0x3ec
4,361,675078042,-; worker_thread from kthread+0x120/0x154
4,362,675078060,-; kthread from ret_from_fork+0x14/0x20
4,363,675078074,-;Exception stack(0xf0ac1fb0 to 0xf0ac1ff8)
4,364,675078081,-;1fa0: 00000000 00000000 00000000 00000000
4,365,675078088,-;1fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
4,366,675078093,-;1fe0: 00000000 00000000 00000000 00000000 00000013 00000000
4,367,675080226,-;irq event stamp: 15780967
4,368,675080287,-;hardirqs last enabled at (15780975): [<c050c154>] __up_console_sem+0x50/0x60
4,369,675080395,-;hardirqs last disabled at (15780982): [<c050c140>] __up_console_sem+0x3c/0x60
4,370,675080497,-;softirqs last enabled at (15780966): [<c0476f14>] handle_softirqs+0x398/0x650
4,371,675080608,-;softirqs last disabled at (15780939): [<c0477388>] __irq_exit_rcu+0x150/0x1f0
4,372,675080715,-;---[ end trace 0000000000000000 ]---
To fix, we can simply not seed this psuedo rate. It was originally
implemented to avoid errornous paths that S1G shouldn't take, but those
have since been fixed / cleaned up. While we are here, we also don't
need to initialise the tx side's last rate with
IEEE80211_TX_RC_S1G_MCS (which is equivalent to
IEEE80211_TX_RC_VHT_MCS). The entire tx reporting side does not
support S1G widths and is assumed to be handled by the driver so
we can just remove the entire function.
Fixes: 12bf8fad4c ("mac80211: initialize last_rate for S1G STAs")
Signed-off-by: Lachlan Hodges <lachlan.hodges@morsemicro.com>
Link: https://patch.msgid.link/20260827054456.254573-1-lachlan.hodges@morsemicro.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Previous releases - regressions:
- page_pool: keep frag_offset aligned for odd-sized requests
- sched: fix u32 duplicate handle when node ID pool is exhausted
- udp: create exceptions before socket matching
- igmp: convert struct ip_sf_list to RCU
- ip6_gre: check tunnel info before xmit in ip6gre_tunnel_xmit
- rds: acquire the fastpath locks in rds_conn_shutdown()
- tipc:
- protect node reset trace dump with node lock
- fix NULL deref in tipc_named_node_up() on empty publication list
- bluetooth:
L2CAP: fix out-of-bounds write in l2cap_ecred_connect
hci_core: fix race condition during device registration
- eth: mlx5e: prevent stale XSK buffer release on refill retries
- eth: bridge: don't truncate the port group walk on teardown
Previous releases - always broken:
- gro: fix nesting of TCP GSO SKBs in skb_gro_receive_list()
- sched: fix skb sizing and action leak on reoffload delete
- tcp: fix use-after-free in do_tcp_getsockopt()
- af_packet: don't cast tpacket_hdr.tp_len to int in tpacket_parse_header().
- sctp: fix soft lockup from unpadded ASCONF-ACK parameter iteration
- iptunnel: fix stale transport header during tunnel decapsulation
- eth: vxlan: fix use-after-free in vxlan_mdb_remote_src_del()
- eth: bonding: fix uninitialized transport header access in alb_determine_nd()
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
-----BEGIN PGP SIGNATURE-----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=0sv8
-----END PGP SIGNATURE-----
Merge tag 'net-7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Pull networking fixes from Paolo Abeni:
"Including fixes from bluetooth.
Previous releases - regressions:
- page_pool: keep frag_offset aligned for odd-sized requests
- sched: fix u32 duplicate handle when node ID pool is exhausted
- udp: create exceptions before socket matching
- igmp: convert struct ip_sf_list to RCU
- ip6_gre: check tunnel info before xmit in ip6gre_tunnel_xmit
- rds: acquire the fastpath locks in rds_conn_shutdown()
- tipc:
- protect node reset trace dump with node lock
- fix NULL deref in tipc_named_node_up() on empty publication
list
- bluetooth:
- L2CAP: fix out-of-bounds write in l2cap_ecred_connect
- hci_core: fix race condition during device registration
- eth:
- mlx5e: prevent stale XSK buffer release on refill retries
- bridge: don't truncate the port group walk on teardown
Previous releases - always broken:
- gro: fix nesting of TCP GSO SKBs in skb_gro_receive_list()
- sched: fix skb sizing and action leak on reoffload delete
- tcp: fix use-after-free in do_tcp_getsockopt()
- af_packet: don't cast tpacket_hdr.tp_len to int in
tpacket_parse_header()
- sctp: fix soft lockup from unpadded ASCONF-ACK parameter iteration
- iptunnel: fix stale transport header during tunnel decapsulation
- eth:
- vxlan: fix use-after-free in vxlan_mdb_remote_src_del()
- bonding: fix uninitialized transport header access in
alb_determine_nd()"
* tag 'net-7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (83 commits)
net: gro: Fix nesting of TCP GSO SKBs in skb_gro_receive_list()
net: stmmac: reconfigure RX packet parser table in stmmac_hw_setup() after reset
net: airoha: enable RX_DONE interrupt for RX queue 31
net/rds: don't let rds_conn_shutdown() consume a concurrent drop
net/rds: acquire the fastpath locks in rds_conn_shutdown()
net/rds: acquire RDS_IN_XMIT in rds_tcp_reset_callbacks()
net/rds: tcp: don't force RDS_CONN_RESETTING over a concurrent shutdown
net/rds: clear cp_flags bits individually in rds_conn_path_reset()
net/rds: use clear_bit_unlock() in release_refill()
net/rds: use wq_has_sleeper() in release_in_xmit()
net: usb: qmi_wwan: add Compal EXM-G1x support
net: macb: exclude software FCS from TX byte statistics
net: Remove conflicting altnames for dying netns in __dev_change_net_namespace().
net: bridge: mcast: don't truncate the port group walk on teardown
bonding: do not clear curr_active_slave prematurely when releasing all slaves
net: qrtr: Send HELLO message on endpoint register
octeontx2-af: Fix limiting SRIOV VF count logic
bonding: alb: fix uninitialized transport header access in alb_determine_nd()
s390/ctcm: Prevent XID null dereference
net: psp: do not inherit the Rx association on clone
...
- dm-cache: fix rwsem being locked and unlocked from different processes
- dm-integrity: set the 'stable writes' flag
- dm-integrity: fix a buffer overflow introduced in this merge window
- dm-integrity: fix an infinite loop if tag size is greater than 64
- dm-cache: fix demotion statistics
- dm-integrity: fix NULL pointer dereference in the data-recovery mode
- dm-ebs: remove a bogus restriction on the starting sector offset
-----BEGIN PGP SIGNATURE-----
iIoEABYIADIWIQRnH8MwLyZDhyYfesYTAyx9YGnhbQUCapmHXxQcbXBhdG9ja2FA
cmVkaGF0LmNvbQAKCRATAyx9YGnhbXwlAP9TGvtNybtjhdQJmo3t528XRdwYrYuJ
TZTRcQ+KMXEAkgEAxG1l4vvECTwSkUhZXub2own7+IypiY8N5cFOWCkdVwc=
=hVzb
-----END PGP SIGNATURE-----
Merge tag 'for-7.3/dm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm
Pull device mapper fixes from Mikulas Patocka:
- fix a dm-crypt race condition that could make errors not being reported
- dm-cache:
- fix rwsem being locked and unlocked from different processes
- fix demotion statistics
- dm-integrity:
- set the 'stable writes' flag
- fix a buffer overflow introduced in this merge window
- fix an infinite loop if tag size is greater than 64
- fix NULL pointer dereference in dm-integrity data-recovery mode
- remove a bogus restriction on the dm-ebs starting sector offset
* tag 'for-7.3/dm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
dm-ebs: fix incorrect device offset check in ebs_ctr()
dm-integrity: fix NULL pointer dereference when the 'R' flag is used
dm cache: fix demotion stats in passthrough mode
dm-integrity: fix infinite loop on discard with large tag size
dm-integrity: fix buffer overflow with keyed discard
dm-integrity: require stable writes for internal hash modes
dm cache: fix issue with background work locking
dm-crypt: fix a tiny race condition in crypt_dec_pending
invalidate_lock, avoiding mmap failures during concurrent size changes
and exposure of uninitialized data during allocation.
- Correct fallocate signal and zeroing error handling.
- Fix FITRIM range alignment to prevent discard requests from extending
into allocated clusters.
- Fix free-cluster accounting when cluster-freeing rollback or bitmap
clearing fails.
- Keep volumes marked dirty when ntfs errors have been recorded.
- Compute bi_sector in 512-byte units, preventing silent corruption on
4Kn devices.
- Validate sectors_per_cluster values and prevent undefined shifts when
parsing MFT and index record sizes.
- Bound $AttrDef traversal to the loaded table size.
- Fix MFT record resizing, memmove overlap, and kmap_local cleanup issues.
- Improve error propagation across attribute, EA, and reparse operations,
including returning -ERANGE for undersized xattr buffers.
- Avoid modifying the HasEA flag when setxattr fails and return
DT_UNKNOWN when directory inode lookup fails.
- Reduce contention in WOF decompression by performing block reads outside
the decompression lock.
-----BEGIN PGP SIGNATURE-----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=gAcy
-----END PGP SIGNATURE-----
Merge tag 'ntfs-for-7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/linkinjeon/ntfs
Pull ntfs fixes from Namjae Jeon:
- Serialize truncate, fallocate, and mmap fault paths with
invalidate_lock, avoiding mmap failures during concurrent size
changes and exposure of uninitialized data during allocation
- Correct fallocate signal and zeroing error handling
- Fix FITRIM range alignment to prevent discard requests from extending
into allocated clusters
- Fix free-cluster accounting when cluster-freeing rollback or bitmap
clearing fails
- Keep volumes marked dirty when ntfs errors have been recorded
- Compute bi_sector in 512-byte units, preventing silent corruption on
4Kn devices
- Validate sectors_per_cluster values and prevent undefined shifts when
parsing MFT and index record sizes
- Bound $AttrDef traversal to the loaded table size
- Fix MFT record resizing, memmove overlap, and kmap_local cleanup
issues
- Improve error propagation across attribute, EA, and reparse
operations, including returning -ERANGE for undersized xattr buffers
- Avoid modifying the HasEA flag when setxattr fails and return
DT_UNKNOWN when directory inode lookup fails
- Reduce contention in WOF decompression by performing block reads
outside the decompression lock
* tag 'ntfs-for-7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/linkinjeon/ntfs: (23 commits)
ntfs: take invalidate_lock in ntfs_filemap_page_mkwrite()
ntfs: take invalidate_lock in ntfs_setattr_size()
ntfs: handle signal interruption in fallocate
ntfs: fix FITRIM range alignment
ntfs: read WOF chunks outside the decompression lock
ntfs: leave HasEA flag untouched on setxattr failure
ntfs: fix race between fallocate and mmap reads
ntfs: fix memmove overlap in ntfs_new_attr_flags
ntfs: compute bi_sector in 512-byte units
ntfs: reject invalid sectors_per_cluster in the boot sector
ntfs: bound $AttrDef table walk to the loaded table size
ntfs: fix undefined behavior in mft/index record size calculation
ntfs: treat any nonzero dio zero-range return as an error
ntfs: fix incorrect MFT record pointer passed to ntfs_attr_record_resize
ntfs: do not mark the volume clean in sync_fs when errors were recorded
ntfs: skip free cluster decrement when rollback fails
ntfs: only count successfully cleared runs when freeing clusters
ntfs: fix kmap_local leak in write_mft_record_nolock() error paths
ntfs: return real error from ntfs_non_resident_attr_record_add()
ntfs: preserve error code in ntfs_resident_attr_record_add()
...
The flags defined in `enum wx_pf_flags` and its terminator
`WX_PF_FLAGS_NBITS` were originally named with a "PF" (Physical Function)
prefix. However, these flags are not exclusive to PFs and are also
utilized by Virtual Functions (VFs).
Rename the enumeration to `enum wx_flags` and the bit count to
`WX_FLAGS_NBITS` to accurately reflect their shared usage across both PF
and VF drivers, and to prevent potential confusion.
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Reviewed-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/9479B0506F544556+20260901065659.41273-1-jiawenwu@trustnetic.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
vzalloc() returns zeroed memory, so the memset after the allocation is
redundant.
No functional change.
Signed-off-by: Sang-Heon Jeon <ekffu200098@gmail.com>
Reviewed-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
Link: https://patch.msgid.link/20260901110913.977745-1-ekffu200098@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Register and handle the NIX LF error (NIX_LF_ERR_INT) and poison/RAS
(NIX_LF_RAS) interrupt vectors, The handlers acknowledge the interrupt
and log the status (ratelimited) to help debug NIX LF faults.
Signed-off-by: Nitin Shetty J <nshettyj@marvell.com>
Signed-off-by: Naveen Mamindlapalli <naveenm@marvell.com>
Link: https://patch.msgid.link/20260831062600.839203-1-nshettyj@marvell.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Fraglist GRO and hardware GRO can create an fraglist of
HW-GRO packets. This cannot be segmented back into
the original form on TCP tethering scenario.
Avoid constructing such a GSO packet, by flushing an already
built fraglist GRO packet if a hardware GRO packet arrives.
Scenario (Tethering/Forwarding):
1.Driver submits a single TCP packet, P1. P1 is kept in the
gro_list as the first packet.
2. The driver submits a TCP GSO skb, P2. P2 has already aggregated
multiple TCP packets by HW_GRO, and its non-linear data is stored in
frags[].
3. P1 and P2 match the GRO rules, and since there is no local socket,
they are aggregated by skb_gro_receive_list(). The resulting skb,
P3, has a frag_list entry that still contains frags[]:
P3: [ Linear Data ] -> frag_list -> [ Linear Data ]
[ frag[1] ]
[ frag[2] ]
...
4. Later, tcp4_gso_segment() or tcp6_gso_segment() calls
skb_segment_list() to segment P3. However, skb_segment_list() only
segments the entries in frag_list. It does not segment the frags[]
inside P2, so P3 is not restored to the original packets, which leads
to IP fragmentation or packet drop in the following path.
Check skb_is_gso(skb) and current GRO method, make sure fraglist GRO
applies to consecutive non-GSO skb, others adopt regular GRO path.
Fixes: 8d95dc474f ("net: add code for TCP fraglist GRO")
Signed-off-by: Zhaoping Shu <zhaoping.shu@mediatek.com>
Signed-off-by: HW He <hw.he@mediatek.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20260901082312.14596-1-zhaoping.shu@mediatek.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
stmmac_select_queue() forces every GSO frame (TCPv4/TCPv6/UDP-L4) onto
TX queue 0, relying on the legacy assumption that at least queue 0 is
always TSO/USO capable. This is no longer needed since
stmmac_features_check() already validates each GSO frame against the
capabilities of the queue it is mapped to, clearing the checksum and GSO
features for queues that cannot perform TSO (e.g. TBS queues or queues
without TX checksum offload support).
Remove stmmac_select_queue() so GSO frames are selected through the
regular queue selection path (XPS, flow hashing and priority mapping)
and can be spread across all TX queues. This also fixes the case where
queue 0 is not TSO-capable while other queues are, which previously
forced all TSO traffic into software GSO. Non-GSO traffic is unaffected,
since the core already falls back to netdev_pick_tx() when ndo_select_queue
is not set.
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
Link: https://patch.msgid.link/20260831-stmmac_select_queue-tso-fix-v3-1-d9c215fe7a85@oss.qualcomm.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The PF and VF drivers warn and return -EPROBE_DEFER when the AF has
not programmed the RVUM revision or does not answer the mailbox READY
message. That is expected until rvu_af has finished initializing, so
the warning is repeated on every deferred retry.
Use dev_err_probe() instead of dev_warn() so the expected deferral is
not printed at warning level. -EPROBE_DEFER is logged at debug level
and the reason is recorded in devices_deferred.
Signed-off-by: Juuso Rinta <juuso.rinta@nokia.com>
Acked-by: Ratheesh Kannoth <rkannoth@marvell.com>
Link: https://patch.msgid.link/20260901-octeontx2-dev-err-probe-v1-1-1b5e1affb994@nokia.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The core software reset issued in stmmac_init_dma_engine() during
ndo_open() callback clears the MTL RX packet parser registers, but
stmmac_rxp_config() is only invoked from the cls_u32 add/delete paths.
After an ifdown/ifup cycle the hardware therefore runs with the default
all-pass table while priv->tc_entries still reports the filters as
installed. Re-apply the RX packet parser table from priv->tc_entries in
stmmac_hw_setup(), right after the software reset, so the filters are
restored when the interface is brought up again.
Fixes: 4dbbe8dde8 ("net: stmmac: Add support for U32 TC filter using Flexible RX Parser")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
Link: https://patch.msgid.link/20260831-stmmac_tc_cls32_reconfigure-v1-1-21cb459e64ae@oss.qualcomm.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
RX queue 31 has always been allocated and filled by airoha_qdma_init_rx()
since RX_DONE_INT_MASK spans queues 0-31, but none of the RX_IRQ*
_BANK_PIN_MASK values covered BIT(31). As a consequence the RX_DONE
interrupt for queue 31 was never enabled, airoha_qdma_rx_process() never
ran on that queue and its buffers were never reaped.
Route RX queue 31's RX_DONE interrupt to IRQ bank 1 so that the queue
is drained and its buffers returned to the page pool.
Fixes: f252493e18 ("net: airoha: Enable multiple IRQ lines support in airoha_eth driver.")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260830-airoha-rxdone-rxq31-v1-1-830a91503f2f@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Maxime Chevallier says:
====================
net: stmmac: dwmac4: Auto-discover UC filter size
Following my attempts to get clean ethtool selftests runs on stmmac [1],
several issues were found with the unicast filtering.
[1]: https://lore.kernel.org/netdev/20260827134004.45ffb57e@kernel.org/
This lead to the discovery that on dwmac4 platforms, the unicast filter
size is always falling back to the default value of 1 entry. On older
variants, the size comes from devicetree.
With a single-entry filter, we directly fallback to unicast promisc as
soon as any entry is added in the dev->uc list.
dwmac4 exposes through the HW_Features0 register the actual size of the
unicast filter, which is made of 3 banks (details in patch 2). Let's use
this feature to grab the filter size.
Results on actual hardware :
- imx8mp : goes from 1 entry to 64
- stm32mp1 : goes from 1 entry to 4
- jh7110 : goes from 1 entry to 9
- yt6801 : still 1 single entry, on extra implemented in HW
====================
Link: https://patch.msgid.link/20260831070121.349778-1-maxime.chevallier@bootlin.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
