platform/kernel/linux-rpi.git
21 months agowifi: ath11k: avoid deadlock during regulatory update in ath11k_regd_update()
Wen Gong [Thu, 6 Oct 2022 15:17:47 +0000 (18:17 +0300)]
wifi: ath11k: avoid deadlock during regulatory update in ath11k_regd_update()

Running this test in a loop it is easy to reproduce an rtnl deadlock:

iw reg set FI
ifconfig wlan0 down

What happens is that thread A (workqueue) tries to update the regulatory:

    try to acquire the rtnl_lock of ar->regd_update_work

    rtnl_lock+0x17/0x20
    ath11k_regd_update+0x15a/0x260 [ath11k]
    ath11k_regd_update_work+0x15/0x20 [ath11k]
    process_one_work+0x228/0x670
    worker_thread+0x4d/0x440
    kthread+0x16d/0x1b0
    ret_from_fork+0x22/0x30

And thread B (ifconfig) tries to stop the interface:

    try to cancel_work_sync(&ar->regd_update_work) in ath11k_mac_op_stop().
    ifconfig  3109 [003]  2414.232506: probe:

    ath11k_mac_op_stop: (ffffffffc14187a0)
    drv_stop+0x30 ([mac80211])
    ieee80211_do_stop+0x5d2 ([mac80211])
    ieee80211_stop+0x3e ([mac80211])
    __dev_close_many+0x9e ([kernel.kallsyms])
    __dev_change_flags+0xbe ([kernel.kallsyms])
    dev_change_flags+0x23 ([kernel.kallsyms])
    devinet_ioctl+0x5e3 ([kernel.kallsyms])
    inet_ioctl+0x197 ([kernel.kallsyms])
    sock_do_ioctl+0x4d ([kernel.kallsyms])
    sock_ioctl+0x264 ([kernel.kallsyms])
    __x64_sys_ioctl+0x92 ([kernel.kallsyms])
    do_syscall_64+0x3a ([kernel.kallsyms])
    entry_SYSCALL_64_after_hwframe+0x63 ([kernel.kallsyms])
    __GI___ioctl+0x7 (/lib/x86_64-linux-gnu/libc-2.23.so)

The sequence of deadlock is:

1. Thread B calls rtnl_lock().

2. Thread A starts to run and calls rtnl_lock() from within
   ath11k_regd_update_work(), then enters wait state because the lock is owned by
   thread B.

3. Thread B continues to run and tries to call
   cancel_work_sync(&ar->regd_update_work), but thread A is in
   ath11k_regd_update_work() waiting for rtnl_lock(). So cancel_work_sync()
   forever waits for ath11k_regd_update_work() to finish and we have a deadlock.

Fix this by switching from using regulatory_set_wiphy_regd_sync() to
regulatory_set_wiphy_regd(). Now cfg80211 will schedule another workqueue which
handles the locking on it's own. So the ath11k workqueue can simply exit without
taking any locks, avoiding the deadlock.

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Signed-off-by: Wen Gong <quic_wgong@quicinc.com>
[kvalo: improve commit log]
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20221006151747.13757-1-kvalo@kernel.org
21 months agowifi: ath11k: fix warning in dma_free_coherent() of memory chunks while recovery
Wen Gong [Wed, 28 Sep 2022 07:38:32 +0000 (03:38 -0400)]
wifi: ath11k: fix warning in dma_free_coherent() of memory chunks while recovery

Commit 26f3a021b37c ("ath11k: allocate smaller chunks of memory for
firmware") and commit f6f92968e1e5 ("ath11k: qmi: try to allocate a
big block of DMA memory first") change ath11k to allocate the memory
chunks for target twice while wlan load. It fails for the 1st time
because of large memory and then changed to allocate many small chunks
for the 2nd time sometimes as below log.

1st time failed:
[10411.640620] ath11k_pci 0000:05:00.0: qmi firmware request memory request
[10411.640625] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 6881280
[10411.640630] ath11k_pci 0000:05:00.0: qmi mem seg type 4 size 3784704
[10411.640658] ath11k_pci 0000:05:00.0: qmi dma allocation failed (6881280 B type 1), will try later with small size
[10411.640671] ath11k_pci 0000:05:00.0: qmi delays mem_request 2
[10411.640677] ath11k_pci 0000:05:00.0: qmi respond memory request delayed 1
2nd time success:
[10411.642004] ath11k_pci 0000:05:00.0: qmi firmware request memory request
[10411.642008] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642012] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642014] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642016] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642018] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642020] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642022] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642024] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642027] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642029] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288
[10411.642031] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 458752
[10411.642033] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 131072
[10411.642035] ath11k_pci 0000:05:00.0: qmi mem seg type 4 size 524288
[10411.642037] ath11k_pci 0000:05:00.0: qmi mem seg type 4 size 524288
[10411.642039] ath11k_pci 0000:05:00.0: qmi mem seg type 4 size 524288
[10411.642041] ath11k_pci 0000:05:00.0: qmi mem seg type 4 size 524288
[10411.642043] ath11k_pci 0000:05:00.0: qmi mem seg type 4 size 524288
[10411.642045] ath11k_pci 0000:05:00.0: qmi mem seg type 4 size 524288
[10411.642047] ath11k_pci 0000:05:00.0: qmi mem seg type 4 size 491520
[10411.642049] ath11k_pci 0000:05:00.0: qmi mem seg type 1 size 524288

And then commit 5962f370ce41 ("ath11k: Reuse the available memory after
firmware reload") skip the ath11k_qmi_free_resource() which frees the
memory chunks while recovery, after that, when run recovery test on
WCN6855, a warning happened every time as below and finally leads fail
for recovery.

[  159.570318] BUG: Bad page state in process kworker/u16:5  pfn:33300
[  159.570320] page:0000000096ffdbb9 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x33300
[  159.570324] flags: 0xfffffc0000000(node=0|zone=1|lastcpupid=0x1fffff)
[  159.570329] raw: 000fffffc0000000 0000000000000000 dead000000000122 0000000000000000
[  159.570332] raw: 0000000000000000 0000000000000000 00000001ffffffff 0000000000000000
[  159.570334] page dumped because: nonzero _refcount
[  159.570440]  firewire_ohci syscopyarea sysfillrect psmouse sdhci_pci ahci sysimgblt firewire_core fb_sys_fops libahci crc_itu_t cqhci drm sdhci e1000e wmi video
[  159.570460] CPU: 2 PID: 217 Comm: kworker/u16:5 Kdump: loaded Tainted: G    B             5.19.0-rc1-wt-ath+ #3
[  159.570465] Hardware name: LENOVO 418065C/418065C, BIOS 83ET63WW (1.33 ) 07/29/2011
[  159.570467] Workqueue: qmi_msg_handler qmi_data_ready_work [qmi_helpers]
[  159.570475] Call Trace:
[  159.570476]  <TASK>
[  159.570478]  dump_stack_lvl+0x49/0x5f
[  159.570486]  dump_stack+0x10/0x12
[  159.570493]  bad_page+0xab/0xf0
[  159.570502]  check_free_page_bad+0x66/0x70
[  159.570511]  __free_pages_ok+0x530/0x9a0
[  159.570517]  ? __dev_printk+0x58/0x6b
[  159.570525]  ? _dev_printk+0x56/0x72
[  159.570534]  ? qmi_decode+0x119/0x470 [qmi_helpers]
[  159.570543]  __free_pages+0x91/0xd0
[  159.570548]  dma_free_contiguous+0x50/0x60
[  159.570556]  dma_direct_free+0xe5/0x140
[  159.570564]  dma_free_attrs+0x35/0x50
[  159.570570]  ath11k_qmi_msg_mem_request_cb+0x2ae/0x3c0 [ath11k]
[  159.570620]  qmi_invoke_handler+0xac/0xe0 [qmi_helpers]
[  159.570630]  qmi_handle_message+0x6d/0x180 [qmi_helpers]
[  159.570643]  qmi_data_ready_work+0x2ca/0x440 [qmi_helpers]
[  159.570656]  process_one_work+0x227/0x440
[  159.570667]  worker_thread+0x31/0x3d0
[  159.570676]  ? process_one_work+0x440/0x440
[  159.570685]  kthread+0xfe/0x130
[  159.570692]  ? kthread_complete_and_exit+0x20/0x20
[  159.570701]  ret_from_fork+0x22/0x30
[  159.570712]  </TASK>

The reason is because when wlan start to recovery, the type, size and
count is not same for the 1st and 2nd QMI_WLFW_REQUEST_MEM_IND message,
Then it leads the parameter size is not correct for the dma_free_coherent().
For the chunk[1], the actual dma size is 524288 which allocate in the
2nd time of the initial wlan load phase, and the size which pass to
dma_free_coherent() is 3784704 which is got in the 1st time of recovery
phase, then warning above happened.

Change to use prev_size of struct target_mem_chunk for the paramter of
dma_free_coherent() since prev_size is the real size of last load/recovery.
Also change to check both type and size of struct target_mem_chunk to
reuse the memory to avoid mismatch buffer size for target. Then the
warning disappear and recovery success. When the 1st QMI_WLFW_REQUEST_MEM_IND
for recovery arrived, the trunk[0] is freed in ath11k_qmi_alloc_target_mem_chunk()
and then dma_alloc_coherent() failed caused by large size, and then
trunk[1] is freed in ath11k_qmi_free_target_mem_chunk(), the left 18
trunks will be reuse for the 2nd QMI_WLFW_REQUEST_MEM_IND message.

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Fixes: 5962f370ce41 ("ath11k: Reuse the available memory after firmware reload")
Signed-off-by: Wen Gong <quic_wgong@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220928073832.16251-1-quic_wgong@quicinc.com
21 months agowifi: ath11k: Don't exit on wakeup failure
Baochen Qiang [Wed, 28 Sep 2022 01:51:40 +0000 (09:51 +0800)]
wifi: ath11k: Don't exit on wakeup failure

Currently, ath11k_pcic_read() returns an error if wakeup()
fails, this makes firmware crash debug quite hard because we can
get nothing.

Change to go ahead on wakeup failure, in that case we still may
get something valid to check. There should be no mislead due
to incorrect content because we are aware of the failure with the
log printed.

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-01720.1-QCAHSPSWPL_V1_V2_SILICONZ_LITE-1

Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220928015140.5431-1-quic_bqiang@quicinc.com
21 months agocarl9170: Replace zero-length array with DECLARE_FLEX_ARRAY() helper
Gustavo A. R. Silva [Mon, 26 Sep 2022 21:44:57 +0000 (16:44 -0500)]
carl9170: Replace zero-length array with DECLARE_FLEX_ARRAY() helper

Zero-length arrays are deprecated and we are moving towards adopting
C99 flexible-array members, instead. So, replace zero-length arrays
declarations in anonymous union with the new DECLARE_FLEX_ARRAY()
helper macro.

This helper allows for flexible-array members in unions.

Link: https://github.com/KSPP/linux/issues/193
Link: https://github.com/KSPP/linux/issues/215
Link: https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/YzIdWc8QSdZFHBYg@work
21 months agowifi: ath10k: Replace zero-length arrays with DECLARE_FLEX_ARRAY() helper
Gustavo A. R. Silva [Tue, 27 Sep 2022 06:18:55 +0000 (09:18 +0300)]
wifi: ath10k: Replace zero-length arrays with DECLARE_FLEX_ARRAY() helper

Zero-length arrays are deprecated and we are moving towards adopting
C99 flexible-array members, instead. So, replace zero-length arrays
declarations in anonymous union with the new DECLARE_FLEX_ARRAY()
helper macro.

This helper allows for flexible-array members in unions.

Link: https://github.com/KSPP/linux/issues/193
Link: https://github.com/KSPP/linux/issues/212
Link: https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/YzIcoloIQBDqGlgc@work
21 months agowifi: wil6210: debugfs: use DEFINE_SHOW_ATTRIBUTE to simplify fw_capabilities/fw_version
Liu Shixin [Tue, 27 Sep 2022 06:18:54 +0000 (09:18 +0300)]
wifi: wil6210: debugfs: use DEFINE_SHOW_ATTRIBUTE to simplify fw_capabilities/fw_version

Use DEFINE_SHOW_ATTRIBUTE helper macro to simplify the code.
No functional change.

Signed-off-by: Liu Shixin <liushixin2@huawei.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220922142858.3250469-1-liushixin2@huawei.com
21 months agowifi: ath11k: stop tx queues immediately upon firmware exit
Aditya Kumar Singh [Tue, 27 Sep 2022 06:18:54 +0000 (09:18 +0300)]
wifi: ath11k: stop tx queues immediately upon firmware exit

Currently, recovery flag is set immediately upon firmware
exit but tx queues are stopped once firmware arrives back
and is ready which is during ath11k_core_restart. Once
ieee80211 hw restart is completed, tx queues are resumed.
If during the time delta between firmware exit and firmware
ready, mac80211 send packets, currently ath11k will drop it
since recovery flag will be set. But warning prints will
come -
  "ath11k c000000.wifi: failed to transmit frame -108"

If more tx packets are there, this could lead to flooding
of above print.

However, actually tx queues should be stopped immediately
when firmware leaves. This will prevent packets to get
dropped when firmware is recovering.

Add fix to stop tx queues immediately after firmware exit.

Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.5.0.1-01100-QCAHKSWPL_SILICONZ-1

Signed-off-by: Aditya Kumar Singh <quic_adisi@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220923170235.18873-1-quic_adisi@quicinc.com
21 months agoMerge ath-next from git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git
Kalle Valo [Tue, 27 Sep 2022 06:17:01 +0000 (09:17 +0300)]
Merge ath-next from git://git./linux/kernel/git/kvalo/ath.git

ath.git patches for v6.1. Major changes:

ath11k

* cold boot calibration support on WCN6750

* Target Wake Time (TWT) debugfs support for STA interface

* support to connect to a non-transmit MBSSID AP profile

* enable remain-on-channel support on WCN6750

* implement SRAM dump debugfs interface

* enable threaded NAPI on all hardware

* WoW support for WCN6750

* support to provide transmit power from firmware via nl80211

* support to get power save duration for each client

* spectral scan support for 160 MHz

wcn36xx

* add SNR from a received frame as a source of system entropy

21 months agobrcmfmac: Remove the call to "dtim_assoc" IOVAR
Ramesh Rangavittal [Thu, 22 Sep 2022 10:41:40 +0000 (05:41 -0500)]
brcmfmac: Remove the call to "dtim_assoc" IOVAR

When STA roams from one AP to another, after roam is complete, host
driver tries to get TIM information from firmware. This is no longer
supported in the firmware & hence, this call will always fail.
This failure results in the below message being displayed on the
console all the time when roam is done.

ieee80211 phy0: brcmf_update_bss_info: wl dtim_assoc failed (-52)

Changes ensure that the host driver will no longer try to get TIM
information from firmware.

Signed-off-by: Ramesh Rangavittal <ramesh.rangavittal@infineon.com>
Signed-off-by: Chi-hsien Lin <chi-hsien.lin@infineon.com>
Signed-off-by: Ian Lin <ian.lin@infineon.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922104140.11889-5-ian.lin@infineon.com
21 months agobrcmfmac: increase dcmd maximum buffer size
Lo(Double)Hsiang Lo [Thu, 22 Sep 2022 10:41:39 +0000 (05:41 -0500)]
brcmfmac: increase dcmd maximum buffer size

Increase dcmd maximum buffer size to match firmware
configuration for new chips.

Signed-off-by: Lo(Double)Hsiang Lo <double.lo@cypress.com>
Signed-off-by: Chi-Hsien Lin <chi-hsien.lin@cypress.com>
Signed-off-by: Ian Lin <ian.lin@infineon.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922104140.11889-4-ian.lin@infineon.com
21 months agobrcmfmac: Support 89459 pcie
Alexander Prutskov [Thu, 22 Sep 2022 10:41:38 +0000 (05:41 -0500)]
brcmfmac: Support 89459 pcie

Adds support of 89459 chip pcie device and save restore support.

Signed-off-by: Alexander Prutskov <alep@cypress.com>
Signed-off-by: Joseph chuang <jiac@cypress.com>
Signed-off-by: Chi-Hsien Lin <chi-hsien.lin@cypress.com>
Signed-off-by: Ian Lin <ian.lin@infineon.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922104140.11889-3-ian.lin@infineon.com
21 months agobrcmfmac: increase default max WOWL patterns to 16
Ryohei Kondo [Thu, 22 Sep 2022 10:41:37 +0000 (05:41 -0500)]
brcmfmac: increase default max WOWL patterns to 16

4373 has support of 16 WOWL patterns thus increasing the default value

Signed-off-by: Ryohei Kondo <ryohei.kondo@cypress.com>
Signed-off-by: Chi-hsien Lin <chi-hsien.lin@cypress.com>
Signed-off-by: Ian Lin <ian.lin@infineon.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922104140.11889-2-ian.lin@infineon.com
21 months agocw1200: fix incorrect check to determine if no element is found in list
Xiaomeng Tong [Wed, 13 Apr 2022 09:17:23 +0000 (17:17 +0800)]
cw1200: fix incorrect check to determine if no element is found in list

The bug is here: "} else if (item) {".

The list iterator value will *always* be set and non-NULL by
list_for_each_entry(), so it is incorrect to assume that the iterator
value will be NULL if the list is empty or no element is found in list.

Use a new value 'iter' as the list iterator, while use the old value
'item' as a dedicated pointer to point to the found element, which
1. can fix this bug, due to now 'item' is NULL only if it's not found.
2. do not need to change all the uses of 'item' after the loop.
3. can also limit the scope of the list iterator 'iter' *only inside*
   the traversal loop by simply declaring 'iter' inside the loop in the
   future, as usage of the iterator outside of the list_for_each_entry
   is considered harmful. https://lkml.org/lkml/2022/2/17/1032

Fixes: a910e4a94f692 ("cw1200: add driver for the ST-E CW1100 & CW1200 WLAN chipsets")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220413091723.17596-1-xiam0nd.tong@gmail.com
21 months agowifi: ath11k: Fix deadlock during WoWLAN suspend
Baochen Qiang [Sat, 24 Sep 2022 13:48:29 +0000 (16:48 +0300)]
wifi: ath11k: Fix deadlock during WoWLAN suspend

We are seeing system hangs during WoWLAN suspend, and get below
two stacks:

Stack1:
[ffffb02cc1557b20] __schedule at ffffffff8bb10860
[ffffb02cc1557ba8] schedule at ffffffff8bb10f24
[ffffb02cc1557bb8] schedule_timeout at ffffffff8bb16d88
[ffffb02cc1557c30] wait_for_completion at ffffffff8bb11778
[ffffb02cc1557c78] __flush_work at ffffffff8b0b30cd
[ffffb02cc1557cf0] __cancel_work_timer at ffffffff8b0b33ad
[ffffb02cc1557d60] ath11k_mac_drain_tx at ffffffffc0c1f0ca [ath11k]
[ffffb02cc1557d70] ath11k_wow_op_suspend at ffffffffc0c5201e [ath11k]
[ffffb02cc1557da8] __ieee80211_suspend at ffffffffc11e2bd3 [mac80211]
[ffffb02cc1557dd8] wiphy_suspend at ffffffffc0f901ac [cfg80211]
[ffffb02cc1557e08] dpm_run_callback at ffffffff8b75118a
[ffffb02cc1557e38] __device_suspend at ffffffff8b751630
[ffffb02cc1557e70] async_suspend at ffffffff8b7519ea
[ffffb02cc1557e88] async_run_entry_fn at ffffffff8b0bf4ce
[ffffb02cc1557ea8] process_one_work at ffffffff8b0b1a24
[ffffb02cc1557ee0] worker_thread at ffffffff8b0b1c4a
[ffffb02cc1557f18] kthread at ffffffff8b0b9cb8
[ffffb02cc1557f50] ret_from_fork at ffffffff8b001d32

Stack2:
[ffffb02cc00b7d18] __schedule at ffffffff8bb10860
[ffffb02cc00b7da0] schedule at ffffffff8bb10f24
[ffffb02cc00b7db0] schedule_preempt_disabled at ffffffff8bb112b4
[ffffb02cc00b7db8] __mutex_lock at ffffffff8bb127ea
[ffffb02cc00b7e38] ath11k_mgmt_over_wmi_tx_work at ffffffffc0c1aa44 [ath11k]
[ffffb02cc00b7ea8] process_one_work at ffffffff8b0b1a24
[ffffb02cc00b7ee0] worker_thread at ffffffff8b0b1c4a
[ffffb02cc00b7f18] kthread at ffffffff8b0b9cb8
[ffffb02cc00b7f50] ret_from_fork at ffffffff8b001d32

From the first stack, ath11k_mac_drain_tx calls
cancel_work_sync(&ar->wmi_mgmt_tx_work) and waits all packets to be sent
out or dropped. However, we find from Stack2 that this work item is blocked
because ar->conf_mutex is already held by ath11k_wow_op_suspend.

Fix this issue by moving ath11k_mac_wait_tx_complete to the start of
ath11k_wow_op_suspend where ar->conf_mutex has not been acquired. And
this change also makes the logic in ath11k_wow_op_suspend match the
logic in ath11k_mac_op_start and ath11k_mac_op_stop.

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220919021435.2459-1-quic_bqiang@quicinc.com
21 months agowifi: ath11k: Remove redundant ath11k_mac_drain_tx
Baochen Qiang [Sat, 24 Sep 2022 13:48:18 +0000 (16:48 +0300)]
wifi: ath11k: Remove redundant ath11k_mac_drain_tx

ath11k_mac_drain_tx is already called in ath11k_mac_wait_tx_complete, no need to call it again. So remove it.

This is found in code review.

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220919020259.1746-1-quic_bqiang@quicinc.com
21 months agowifi: ath11k: Add spectral scan support for 160 MHz
Tamizh Chelvam Raja [Sat, 24 Sep 2022 13:47:12 +0000 (16:47 +0300)]
wifi: ath11k: Add spectral scan support for 160 MHz

There are two types of 160 MHz spectral scan support mentioned below

1. Fragmented approach
2. Single event approach

In this fragmented approach, single 160 MHz will be split as two
80 MHz buffer. First fft sample buffer will contain spectral scan
result of primary 80 MHz and the second fft sample buffer will contain
secondary 80 MHz and here cfreq1 and cfreq2 will be mentioned.
In case of 160 MHz on 36th channel will contain cfreq1 as 5210 and
cfreq2 as 5290. Chipsets which support this approach are IPQ8074/IPQ6018.

Replacing freq1 with freq2 in every secondary sepctral scan event to
distinguish between two different 80 MHz spectral event data.

In the 2nd approach each fft sample buffer will contain spectral scan
result for whole 160 MHz by mentioning cfreq1 as 5250 which is center
frequency of whole 160 MHz. Chipset which support this approach is QCN9074.

Host will receive spectral event from target for every 5 fft samples.

Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.5.0.1-01120-QCAHKSWPL-1
Tested-on: QCN9074 hw1.0 PCI WLAN.HK.2.5.0.1-01120-QCAHKSWP

Signed-off-by: Tamizh Chelvam Raja <quic_tamizhr@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220725055001.15194-1-quic_tamizhr@quicinc.com
21 months agowifi: ath11k: Add support to get power save duration for each client
Venkateswara Naralasetty [Sat, 24 Sep 2022 13:47:12 +0000 (16:47 +0300)]
wifi: ath11k: Add support to get power save duration for each client

Add support to get the following power save information through debugfs interface,

 * Current ps state of the peer
 * Time duration since the peer is in power save
 * Total duration of the peer spent in power save

Above information is helpful in debugging the issues with power save clients.

This patch also add trace log support for PS timekeeper to track the PS state
change of the peers alongs with the peer MAC address and timestamp.

Use the below commands to get the above power save information,

To know the time_since_station_in_power_save:
cat /sys/kernel/debug/ieee80211/phyX/netdev:wlanX/stations/
XX:XX:XX:XX:XX:XX/current_ps_duration

To know power_save_duration:
cat /sys/kernel/debug/ieee80211/phyX/netdev:wlanX/stations/
XX:XX:XX:XX:XX:XX/total_ps_duration

To reset the power_save_duration of all stations connected to AP:
echo 1 > /sys/kernel/debug/ieee80211/phyX/ath11k/reset_ps_duration

To enable/disable the ps_timekeeper:
echo Y > /sys/kernel/debug/ieee80211/phyX/ath11k/ps_timekeeper_enable
Y = 1 to enable and Y = 0 to disable.

To record PS timekeeer logs after enabling ps_timekeeper:
trace-cmd record -e ath11k_ps_timekeeper

Tested-on: Tested-on: IPQ8074 WLAN.HK.2.5.0.1-00991-QCAHKSWPL_SILICONZ-1

Signed-off-by: Venkateswara Naralasetty <quic_vnaralas@quicinc.com>
Signed-off-by: Tamizh Chelvam Raja <quic_tamizhr@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220725054601.14719-1-quic_tamizhr@quicinc.com
21 months agowifi: ath11k: fix peer addition/deletion error on sta band migration
Christian 'Ansuel' Marangi [Thu, 22 Sep 2022 07:35:14 +0000 (10:35 +0300)]
wifi: ath11k: fix peer addition/deletion error on sta band migration

This patch try to fix the following error.

Wed Jun  1 22:19:30 2022 kern.warn kernel: [  119.561227] ath11k c000000.wifi: peer already added vdev id 0 req, vdev id 1 present
Wed Jun  1 22:19:30 2022 kern.warn kernel: [  119.561282] ath11k c000000.wifi: Failed to add peer: 28:c2:1f:xx:xx:xx for VDEV: 0
Wed Jun  1 22:19:30 2022 kern.warn kernel: [  119.568053] ath11k c000000.wifi: Failed to add station: 28:c2:1f:xx:xx:xx for VDEV: 0
Wed Jun  1 22:19:31 2022 daemon.notice hostapd: wlan2: STA 28:c2:1f:xx:xx:xx IEEE 802.11: Could not add STA to kernel driver
Wed Jun  1 22:19:31 2022 daemon.notice hostapd: wlan2: STA 28:c2:1f:xx:xx:xx IEEE 802.11: did not acknowledge authentication response
Wed Jun  1 22:19:31 2022 daemon.notice hostapd: wlan1: AP-STA-DISCONNECTED 28:c2:1f:xx:xx:xx
Wed Jun  1 22:19:31 2022 daemon.info hostapd: wlan1: STA 28:c2:1f:xx:xx:xx IEEE 802.11: disassociated due to inactivity
Wed Jun  1 22:19:32 2022 daemon.info hostapd: wlan1: STA 28:c2:1f:xx:xx:xx IEEE 802.11: deauthenticated due to inactivity (timer DEAUTH/REMOVE)

To repro this:
- Have 2 Wifi with the same bssid and pass on different band (2.4 and
5GHz)
- Enable 802.11r Fast Transaction with same mobility domain
- FT Protocol: FT over the Air
From a openwrt system issue the command (with the correct mac)
ubus call hostapd.wlan1 wnm_disassoc_imminent '{"addr":"28:C2:1F:xx:xx:xx"}'
Notice the log printing the errors.

The cause of this error has been investigated and we found that this is
related to the WiFi Fast Transaction feature. We observed that this is
triggered when the router tells the device to change band. In this case
the device first auth to the other band and then the disconnect path
from the prev band is triggered.
This is problematic with the current rhash implementation since the
addrs is used as key and the logic of "adding first, delete later"
conflicts with the rhash logic.
In fact peer addition will fail since the peer is already added and with
that fixed a peer deletion will cause unitended effect by removing the
peer just added.

Current solution to this is to add additional logic to the peer delete,
make sure we are deleting the correct peer taken from the rhash
table (and fallback to the peer list) and for the peer add logic delete
the peer entry for the rhash list before adding the new one (counting as
an error only when a peer with the same vlan_id is asked to be added).

With this change, a sta can correctly transition from 2.4GHz and 5GHZ
with no drop and no error are printed.

Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.5.0.1-01208-QCAHKSWPL_SILICONZ-1

Fixes: 7b0c70d92a43 ("ath11k: Add peer rhash table support")
Signed-off-by: Christian 'Ansuel' Marangi <ansuelsmth@gmail.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220603164559.27769-1-ansuelsmth@gmail.com
21 months agowifi: rtw89: support for enable/disable MSDU aggregation
Kuan-Chung Chen [Thu, 22 Sep 2022 01:27:37 +0000 (09:27 +0800)]
wifi: rtw89: support for enable/disable MSDU aggregation

To enable/disable amsdu in set_tid_config, and currently only support to
configure all tids (==0xff) for AMSDU, not individual tid.

The command example is:
  iw wlan0 set tidconf tids 0xff amsdu off
  iw wlan0 set tidconf peer xx:xx:xx:xx:xx:xx tids 0xff amsdu on

Signed-off-by: Kuan-Chung Chen <damon.chen@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922012737.15091-1-pkshih@realtek.com
21 months agowifi: rtw89: disable 26-tone RU HE TB PPDU transmissions
Kuan-Chung Chen [Thu, 22 Sep 2022 01:27:19 +0000 (09:27 +0800)]
wifi: rtw89: disable 26-tone RU HE TB PPDU transmissions

Align with the spec of 802.11ax, follow the conditions for not responding
with an HE TB PPDU. When there are OBSS that cannot interpret 26-tone RU
transmissions, we should disable such transmissions.

Signed-off-by: Kuan-Chung Chen <damon.chen@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922012719.15037-1-pkshih@realtek.com
21 months agowifi: rtw89: support for processing P2P power saving
Dian-Syuan Yang [Thu, 22 Sep 2022 01:04:35 +0000 (09:04 +0800)]
wifi: rtw89: support for processing P2P power saving

Support P2P client to process Notice of Absence (NoA) mechanism when it
connects with P2P GO applying an NoA schedule. We define some action
types including init, update, remove and terminate in h2c function to
enable/disable NoA schedule.

Signed-off-by: Dian-Syuan Yang <dian_syuan0116@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922010435.12167-6-pkshih@realtek.com
21 months agowifi: rtw89: support WMM-PS in P2P GO mode
Dian-Syuan Yang [Thu, 22 Sep 2022 01:04:34 +0000 (09:04 +0800)]
wifi: rtw89: support WMM-PS in P2P GO mode

To handle a connected client using WMM-PS, the P2P GO also need to enable
this power save mechanism.

We add WIPHY_FLAG_AP_UAPSD flag to support it and define maximum number
of buffered frames the WMM GO may deliver to the WMM client.

Signed-off-by: Dian-Syuan Yang <dian_syuan0116@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922010435.12167-5-pkshih@realtek.com
21 months agowifi: rtw89: set wifi_role of P2P
Ping-Ke Shih [Thu, 22 Sep 2022 01:04:33 +0000 (09:04 +0800)]
wifi: rtw89: set wifi_role of P2P

Consider vif->p2p to set wifi_role to let firmware know current vif is
running as GC or GO. And, allow GC to enter PS mode, but disallow to enter
deep PS for now.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922010435.12167-4-pkshih@realtek.com
21 months agowifi: rtw89: send OFDM rate only in P2P mode
Dian-Syuan Yang [Thu, 22 Sep 2022 01:04:32 +0000 (09:04 +0800)]
wifi: rtw89: send OFDM rate only in P2P mode

Check IEEE80211_TX_CTL_NO_CCK_RATE flag to avoid sending frames with
CCK rates in 2GHz band.

In TX flow, add IEEE80211_TX_CTL_NO_CCK_RATE flag to check and get its
lowest rate without CCK rates if the TX type is mgmt frames or data frames.
Besides, the decision of phy rate and retry rate in P2P mode are also
be handled.

In P2P GO mode, it should send beacon of no CCK rates in its frame rate.
Therefore, We add a condition to decide which rate is added to beacon
content.

Moreover, we avoid setting a mask of rates to be used for rate control
selection before and after connection in P2P mode.

Signed-off-by: Dian-Syuan Yang <dian_syuan0116@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922010435.12167-3-pkshih@realtek.com
21 months agowifi: rtw89: support P2P
Po Hao Huang [Thu, 22 Sep 2022 01:04:31 +0000 (09:04 +0800)]
wifi: rtw89: support P2P

To support P2P in driver, we set P2P interface mode to the wiphy
allocated for 802.11 PHY and add a change interface function to
switch the interface type to P2P.

Signed-off-by: Po Hao Huang <phhuang@realtek.com>
Signed-off-by: Dian-Syuan Yang <dian_syuan0116@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220922010435.12167-2-pkshih@realtek.com
21 months agowifi: rt2x00: correctly set BBP register 86 for MT7620
Daniel Golle [Sat, 17 Sep 2022 20:30:09 +0000 (21:30 +0100)]
wifi: rt2x00: correctly set BBP register 86 for MT7620

Instead of 0 set the correct value for BBP register 86 for MT7620.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/257267247ee4fa7ebc6a5d0c4948b3f8119c0d77.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: set SoC wmac clock register
Daniel Golle [Sat, 17 Sep 2022 20:29:55 +0000 (21:29 +0100)]
wifi: rt2x00: set SoC wmac clock register

Instead of using the default value 33 (pci), set US_CYC_CNT init based
on Programming guide:
If available, set chipset bus clock with fallback to cpu clock/3.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/3e275d259f476f597dab91a9c395015ef3fe3284.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: set VGC gain for both chains of MT7620
Daniel Golle [Sat, 17 Sep 2022 20:29:40 +0000 (21:29 +0100)]
wifi: rt2x00: set VGC gain for both chains of MT7620

Set bbp66 for all chains of the MT7620.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/29e161397e5c9d9399da0fe87d44458aa2b90a78.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: set correct TX_SW_CFG1 MAC register for MT7620
Daniel Golle [Sat, 17 Sep 2022 20:29:26 +0000 (21:29 +0100)]
wifi: rt2x00: set correct TX_SW_CFG1 MAC register for MT7620

Set correct TX_SW_CFG1 MAC register as it is done also in v3 of the
vendor driver[1].

[1]: https://gitlab.com/dm38/padavan-ng/-/blob/master/trunk/proprietary/rt_wifi/rtpci/3.0.X.X/mt76x2/chips/rt6352.c#L531
Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/4be38975ce600a34249e12d09a3cb758c6e71071.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: fix HT20/HT40 bandwidth switch on MT7620
Daniel Golle [Sat, 17 Sep 2022 20:29:13 +0000 (21:29 +0100)]
wifi: rt2x00: fix HT20/HT40 bandwidth switch on MT7620

Add missing configuration of the channel bandwidth filter to the
channel setup function for MT7620.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/1664d89ba149f7b0bcec18b2a2abaedf49654507.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: move helper functions up in file
Daniel Golle [Sat, 17 Sep 2022 20:28:58 +0000 (21:28 +0100)]
wifi: rt2x00: move helper functions up in file

Move register access helper functions up to the head of the file so
they can be used in all functions.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/c27baa8efd5c29e2bcb2432925d9cdc5c913a125.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: add TX LOFT calibration for MT7620
Tomislav Požega [Sat, 17 Sep 2022 20:28:43 +0000 (21:28 +0100)]
wifi: rt2x00: add TX LOFT calibration for MT7620

Add TX LOFT calibration from mtk driver.

Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/d9295a9138a1f552b648aacb84e1419d38f5c896.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: don't run Rt5592 IQ calibration on MT7620
Daniel Golle [Sat, 17 Sep 2022 20:28:29 +0000 (21:28 +0100)]
wifi: rt2x00: don't run Rt5592 IQ calibration on MT7620

The function rt2800_iq_calibrate is intended for Rt5592 only.
Don't call it for MT7620 which has it's own calibration functions.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/31a1c34ddbd296b82f38c18c9ae7339059215fdc.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: add RXIQ calibration for MT7620
Tomislav Požega [Sat, 17 Sep 2022 20:28:10 +0000 (21:28 +0100)]
wifi: rt2x00: add RXIQ calibration for MT7620

Add RXIQ calibration found in mtk driver. With old openwrt builds this
gets us ~8Mbps more of RX bandwidth (test with iPA/eLNA layout).

Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/033a39a697d51f6df258acea4c33608e0944fe4c.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: add RXDCOC calibration for MT7620
Tomislav Požega [Sat, 17 Sep 2022 20:27:56 +0000 (21:27 +0100)]
wifi: rt2x00: add RXDCOC calibration for MT7620

Add RXDCOC calibration code from mtk driver.

Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
[fixed typo reported by Serge Vasilugin <vasilugin@yandex.ru>]
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/850b30f652e88de30d79e968af4eb47aa5bc2511.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: add r calibration for MT7620
Tomislav Požega [Sat, 17 Sep 2022 20:27:41 +0000 (21:27 +0100)]
wifi: rt2x00: add r calibration for MT7620

Add r calibration code as found in mtk driver.

Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/e0c34f233089bec4eb73826bc4f512166ee25934.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: add RF self TXDC calibration for MT7620
Tomislav Požega [Sat, 17 Sep 2022 20:27:26 +0000 (21:27 +0100)]
wifi: rt2x00: add RF self TXDC calibration for MT7620

Add TX self calibration based on mtk driver.

Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/dbb6e5a0c12d6101477bd09e83253091d21512c9.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: move up and reuse busy wait functions
Daniel Golle [Sat, 17 Sep 2022 20:27:10 +0000 (21:27 +0100)]
wifi: rt2x00: move up and reuse busy wait functions

Move bbp_ready and rf_ready busy wait functions up in the code so they
can more easily be used. Allow specifying register mask in rf_ready
function which is useful for calibration routines which will be added
in follow-up commits.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/3fdb9dc15e76a9f9c1948b4a3a1308a7a5677bb8.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: add support for external PA on MT7620
Daniel Golle [Sat, 17 Sep 2022 20:26:55 +0000 (21:26 +0100)]
wifi: rt2x00: add support for external PA on MT7620

Implement support for external PA connected to MT7620A.

Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
[pozega.tomislav@gmail.com: use chanreg and dccal helpers.]
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/af2c68ff831816a86fc39b0c10911c129a1f03dc.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: add throughput LED trigger
David Bauer [Sat, 17 Sep 2022 20:26:40 +0000 (21:26 +0100)]
wifi: rt2x00: add throughput LED trigger

This adds a (currently missing) throughput LED trigger for the rt2x00
driver. Previously, LED triggers had to be assigned to the netdev, which
was limited to a single VAP.

Tested-by: Christoph Krapp <achterin@googlemail.com>
Signed-off-by: David Bauer <mail@david-bauer.net>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/73f5ba4134e621462a26186449400cf0c1ac1730.1663445157.git.daniel@makrotopia.org
21 months agowifi: rt2x00: define RF5592 in init_eeprom routine
Tomislav Požega [Sat, 17 Sep 2022 20:26:27 +0000 (21:26 +0100)]
wifi: rt2x00: define RF5592 in init_eeprom routine

Fix incorrect RF value encoded in EEPROM on devices with Ralink Rt5592
PCIe radio (a single chip 2T2R 802.11abgn solution).

Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/d7eccb2c7b8ec4cd360fa2007796abffc35abb0d.1663445157.git.daniel@makrotopia.org
21 months agowifi: ath11k: fix number of VHT beamformee spatial streams
Jesus Fernandez Manzano [Thu, 22 Sep 2022 07:35:14 +0000 (10:35 +0300)]
wifi: ath11k: fix number of VHT beamformee spatial streams

The number of spatial streams used when acting as a beamformee in VHT
mode are reported by the firmware as 7 (8 sts - 1) both in IPQ6018 and
IPQ8074 which respectively have 2 and 4 sts each. So the firmware should
report 1 (2 - 1) and 3 (4 - 1).

Fix this by checking that the number of VHT beamformee sts reported by
the firmware is not greater than the number of receiving antennas - 1.
The fix is based on the same approach used in this same function for
sanitizing the number of sounding dimensions reported by the firmware.

Without this change, acting as a beamformee in VHT mode is not working
properly.

Tested-on: IPQ6018 hw1.0 AHB WLAN.HK.2.5.0.1-01208-QCAHKSWPL_SILICONZ-1
Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.5.0.1-01208-QCAHKSWPL_SILICONZ-1

Fixes: d5c65159f289 ("ath11k: driver for Qualcomm IEEE 802.11ax devices")
Signed-off-by: Jesus Fernandez Manzano <jesus.manzano@galgus.net>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220616173947.21901-1-jesus.manzano@galgus.net
21 months agowifi: ath11k: retrieve MAC address from system firmware if provided
Jun Yu [Tue, 20 Sep 2022 00:31:17 +0000 (00:31 +0000)]
wifi: ath11k: retrieve MAC address from system firmware if provided

Devices may provide their own MAC address via system firmware (e.g.,
device tree), especially in the case where the device doesn't have a
useful EEPROM on which to store its MAC address (e.g., for integrated
ahb WCN6750).

Use the generic device helper to retrieve the MAC address, and (if
present) honor it above the MAC address advertised by the card.

Tested-on: WCN6750 hw1.0 AHB WLAN.MSL.1.0.1-00887-QCAMSLSWPLZ-1

Signed-off-by: Jun Yu <junyuu@chromium.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220920003117.841442-1-junyuu@chromium.org
21 months agowifi: ath11k: add get_txpower mac ops
Aditya Kumar Singh [Tue, 20 Sep 2022 15:24:21 +0000 (18:24 +0300)]
wifi: ath11k: add get_txpower mac ops

Driver does not support get_txpower mac ops because of which
cfg80211 returns vif->bss_conf.txpower to user space. bss_conf.txpower
gets its value from ieee80211_channel->max_reg_power. However, the final
txpower is dependent on few other parameters apart from max regulatory
supported power. It is the firmware which knows about all these
parameters and considers the minimum for each packet transmission.

All ath11k firmware reports the final tx power in firmware pdev stats
which falls under fw_stats.

Add get_txpower mac ops to get the tx power from firmware leveraging
fw_stats and return it accordingly.

Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.5.0.1-01100-QCAHKSWPL_SILICONZ-1
Tested-on: QCN9074 hw1.0 PCI WLAN.HK.2.5.0.1-01100-QCAHKSWPL_SILICONZ-1
Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Signed-off-by: Aditya Kumar Singh <quic_adisi@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220603082814.31466-3-quic_adisi@quicinc.com
21 months agowifi: ath11k: move firmware stats out of debugfs
Aditya Kumar Singh [Tue, 20 Sep 2022 15:24:12 +0000 (18:24 +0300)]
wifi: ath11k: move firmware stats out of debugfs

Currently, firmware stats, comprising pdev, vdev and beacon stats are
part of debugfs. In firmware pdev stats, firmware reports the final
Tx power used to transmit each packet. If driver wants to know the
final Tx power being used at firmware level, it can leverage from
firmware pdev stats.

Move firmware stats out of debugfs context in order to leverage
the final Tx power reported in it even when debugfs is disabled.

Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.5.0.1-01100-QCAHKSWPL_SILICONZ-1
Tested-on: QCN9074 hw1.0 PCI WLAN.HK.2.5.0.1-01100-QCAHKSWPL_SILICONZ-1
Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Signed-off-by: Aditya Kumar Singh <quic_adisi@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220603082814.31466-2-quic_adisi@quicinc.com
21 months agowifi: ath9k: fix repeated the words in a comment
Jilin Yuan [Thu, 15 Sep 2022 03:08:59 +0000 (11:08 +0800)]
wifi: ath9k: fix repeated the words in a comment

Delete the redundant word 'the'.

Signed-off-by: Jilin Yuan <yuanjilin@cdjrlc.com>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220915030859.45384-1-yuanjilin@cdjrlc.com
21 months agowifi: ath9k: fix repeated to words in a comment
Jilin Yuan [Thu, 15 Sep 2022 03:05:59 +0000 (11:05 +0800)]
wifi: ath9k: fix repeated to words in a comment

Delete the redundant word 'to'.

Signed-off-by: Jilin Yuan <yuanjilin@cdjrlc.com>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220915030559.42371-1-yuanjilin@cdjrlc.com
21 months agowifi: ath10k: reset pointer after memory free to avoid potential use-after-free
Wen Gong [Tue, 20 Sep 2022 15:23:54 +0000 (18:23 +0300)]
wifi: ath10k: reset pointer after memory free to avoid potential use-after-free

When running suspend test, kernel crash happened in ath10k, and it is
fixed by commit b72a4aff947b ("ath10k: skip ath10k_halt during suspend
for driver state RESTARTING").

Currently the crash is fixed, but as a common code style, it is better
to set the pointer to NULL after memory is free.

This is to address the code style and it will avoid potential bug of
use-after-free.

Tested-on: QCA6174 hw3.2 PCI WLAN.RM.4.4.1-00110-QCARMSWP-1
Signed-off-by: Wen Gong <quic_wgong@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220505092248.787-1-quic_wgong@quicinc.com
21 months agowifi: ath11k: fix failed to find the peer with peer_id 0 when disconnected
Wen Gong [Tue, 20 Sep 2022 15:23:41 +0000 (18:23 +0300)]
wifi: ath11k: fix failed to find the peer with peer_id 0 when disconnected

It has a fail log which is ath11k_dbg in ath11k_dp_rx_process_mon_status(),
as below, it will not print when debug_mask is not set ATH11K_DBG_DATA.
ath11k_dbg(ab, ATH11K_DBG_DATA,
  "failed to find the peer with peer_id %d\n",
   ppdu_info.peer_id);

When run scan with station disconnected, the peer_id is 0 for case
HAL_RX_MPDU_START in ath11k_hal_rx_parse_mon_status_tlv() which called
from ath11k_dp_rx_process_mon_status(), and the peer_id of ppdu_info is
reset to 0 in the while loop, so it does not match condition of the
check "if (ppdu_info->peer_id == HAL_INVALID_PEERID" in the loop, and
then the log "failed to find the peer with peer_id 0" print after the
check in the loop, it is below call stack when debug_mask is set
ATH11K_DBG_DATA.

The reason is this commit 01d2f285e3e5 ("ath11k: decode HE status tlv")
add "memset(ppdu_info, 0, sizeof(struct hal_rx_mon_ppdu_info))" in
ath11k_dp_rx_process_mon_status(), but the commit does not initialize
the peer_id to HAL_INVALID_PEERID, then lead the check mis-match.

Callstack of the failed log:
[12335.689072] RIP: 0010:ath11k_dp_rx_process_mon_status+0x9ea/0x1020 [ath11k]
[12335.689157] Code: 89 ff e8 f9 10 00 00 be 01 00 00 00 4c 89 f7 e8 dc 4b 4e de 48 8b 85 38 ff ff ff c7 80 e4 07 00 00 01 00 00 00 e9 20 f8 ff ff <0f> 0b 41 0f b7 96 be 06 00 00 48 c7 c6 b8 50 44 c1 4c 89 ff e8 fd
[12335.689180] RSP: 0018:ffffb874001a4ca0 EFLAGS: 00010246
[12335.689210] RAX: 0000000000000000 RBX: ffff995642cbd100 RCX: 0000000000000000
[12335.689229] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff99564212cd18
[12335.689248] RBP: ffffb874001a4dc0 R08: 0000000000000001 R09: 0000000000000000
[12335.689268] R10: 0000000000000220 R11: ffffb874001a48e8 R12: ffff995642473d40
[12335.689286] R13: ffff99564212c5b8 R14: ffff9956424736a0 R15: ffff995642120000
[12335.689303] FS:  0000000000000000(0000) GS:ffff995739000000(0000) knlGS:0000000000000000
[12335.689323] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[12335.689341] CR2: 00007f43c5d5e039 CR3: 000000011c012005 CR4: 00000000000606e0
[12335.689360] Call Trace:
[12335.689377]  <IRQ>
[12335.689418]  ? rcu_read_lock_held_common+0x12/0x50
[12335.689447]  ? rcu_read_lock_sched_held+0x25/0x80
[12335.689471]  ? rcu_read_lock_held_common+0x12/0x50
[12335.689504]  ath11k_dp_rx_process_mon_rings+0x8d/0x4f0 [ath11k]
[12335.689578]  ? ath11k_dp_rx_process_mon_rings+0x8d/0x4f0 [ath11k]
[12335.689653]  ? lock_acquire+0xef/0x360
[12335.689681]  ? rcu_read_lock_sched_held+0x25/0x80
[12335.689713]  ath11k_dp_service_mon_ring+0x38/0x60 [ath11k]
[12335.689784]  ? ath11k_dp_rx_process_mon_rings+0x4f0/0x4f0 [ath11k]
[12335.689860]  call_timer_fn+0xb2/0x2f0
[12335.689897]  ? ath11k_dp_rx_process_mon_rings+0x4f0/0x4f0 [ath11k]
[12335.689970]  run_timer_softirq+0x21f/0x540
[12335.689999]  ? ktime_get+0xad/0x160
[12335.690025]  ? lapic_next_deadline+0x2c/0x40
[12335.690053]  ? clockevents_program_event+0x82/0x100
[12335.690093]  __do_softirq+0x151/0x4a8
[12335.690135]  irq_exit_rcu+0xc9/0x100
[12335.690165]  sysvec_apic_timer_interrupt+0xa8/0xd0
[12335.690189]  </IRQ>
[12335.690204]  <TASK>
[12335.690225]  asm_sysvec_apic_timer_interrupt+0x12/0x20

Reset the default value to HAL_INVALID_PEERID each time after memset
of ppdu_info as well as others memset which existed in function
ath11k_dp_rx_process_mon_status(), then the failed log disappeared.

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Fixes: 01d2f285e3e5 ("ath11k: decode HE status tlv")
Signed-off-by: Wen Gong <quic_wgong@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220518033556.31940-1-quic_wgong@quicinc.com
21 months agowifi: ath11k: change complete() to complete_all() for scan.completed
Wen Gong [Mon, 19 Sep 2022 13:06:03 +0000 (16:06 +0300)]
wifi: ath11k: change complete() to complete_all() for scan.completed

Currently commit 1f682dc9fb37 ("ath11k: reduce the wait time of 11d scan
and hw scan while add interface") introduced a wait_for_completion_timeout
operation for ar->scan.completed, another one is existed in ath11k_scan_stop(),
then ath11k has two places to wait for the ar->scan.completed and they
run in different thread, thus it is possible to happend that the two
thread both enter wait status. To handle this scenario, ath11k should
change the complete() to complete_all() for the ar->scan.completed. This
also work well when it is only one thread wait for ar->scan.completed.

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Signed-off-by: Wen Gong <quic_wgong@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220919024413.25083-1-quic_wgong@quicinc.com
21 months agoMerge tag 'iwlwifi-next-for-kalle-2022-09-18' of git://git.kernel.org/pub/scm/linux...
Kalle Valo [Thu, 22 Sep 2022 06:14:33 +0000 (09:14 +0300)]
Merge tag 'iwlwifi-next-for-kalle-2022-09-18' of git://git./linux/kernel/git/iwlwifi/iwlwifi-next

iwlwifi patches for v6.1

21 months agoMerge tag 'mt76-for-kvalo-2022-09-15' of https://github.com/nbd168/wireless
Kalle Valo [Thu, 22 Sep 2022 06:13:17 +0000 (09:13 +0300)]
Merge tag 'mt76-for-kvalo-2022-09-15' of https://github.com/nbd168/wireless

mt76 patches for 6.1

- fixes
- suspend/resume improvements
- tx status reporting improvements

21 months agowifi: rtw89: coex: update coexistence to 6.3.0
Ching-Te Ku [Tue, 20 Sep 2022 01:09:39 +0000 (09:09 +0800)]
wifi: rtw89: coex: update coexistence to 6.3.0

Since we maintain coexistence as shared code, so move coexistence version
from chip specific attribute to a definition.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-10-pkshih@realtek.com
21 months agowifi: rtw89: coex: add logic to control BT scan priority
Ching-Te Ku [Tue, 20 Sep 2022 01:09:38 +0000 (09:09 +0800)]
wifi: rtw89: coex: add logic to control BT scan priority

Add control logic to operate Wi-Fi to BT scoreboard to control BT scan
priority. And patch mechanism parameter to enhance Wi-Fi throughput while
coexisting with BT profile & BT scan. Set PTA priority let Wi-Fi BT can RX
at the same time.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-9-pkshih@realtek.com
21 months agowifi: rtw89: coex: summarize Wi-Fi to BT scoreboard and inform BT one time a cycle
Ching-Te Ku [Tue, 20 Sep 2022 01:09:37 +0000 (09:09 +0800)]
wifi: rtw89: coex: summarize Wi-Fi to BT scoreboard and inform BT one time a cycle

If Wi-Fi driver send Wi-Fi status to BT via scoreboard too frequent in a
short moment, BT will loss some of them because of hardware response time.
To avoid this issue, change the code flow. Summarize the scoreboard changes
and if the value have changed, send the scoreboard to BT only once in
a coexistence processing cycle. It also can help to reduce driver I/O.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-8-pkshih@realtek.com
21 months agowifi: rtw89: coex: modify LNA2 setting to avoid BT destroyed Wi-Fi aggregation
Ching-Te Ku [Tue, 20 Sep 2022 01:09:36 +0000 (09:09 +0800)]
wifi: rtw89: coex: modify LNA2 setting to avoid BT destroyed Wi-Fi aggregation

To prevent LNA2 change its gain during a Wi-Fi aggregation packet while
GNT_BT pull high. Otherwise, changes of this gain will destroy the whole
aggregation when Wi-Fi RX.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-7-pkshih@realtek.com
21 months agowifi: rtw89: coex: add WL_S0 hardware TX/RX mask to allow WL_S0 TX/RX during GNT_BT
Ching-Te Ku [Tue, 20 Sep 2022 01:09:35 +0000 (09:09 +0800)]
wifi: rtw89: coex: add WL_S0 hardware TX/RX mask to allow WL_S0 TX/RX during GNT_BT

WiFi/BT combo module could only have two antenna, namely WL_S0 and WL_S1.
WiFi can use two antenna to TX/RX 2SS data, and BT can share one of the
antenna. This patch is to allow WiFi to TX/RX 1SS data like ACK/RTS/CTS to
improve Wi-Fi performance while coexisting with Bluetooth.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-6-pkshih@realtek.com
21 months agowifi: rtw89: coex: add v1 Wi-Fi firmware steps report
Ching-Te Ku [Tue, 20 Sep 2022 01:09:34 +0000 (09:09 +0800)]
wifi: rtw89: coex: add v1 Wi-Fi firmware steps report

This report is to record firmware call flow like notify events, and take
actions. This can help to address if firmware flow is in expectation.
Implement v1 parser to support 8852CE firmware report.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-5-pkshih@realtek.com
21 months agowifi: rtw89: coex: add v1 summary info to parse the traffic status from firmware
Ching-Te Ku [Tue, 20 Sep 2022 01:09:33 +0000 (09:09 +0800)]
wifi: rtw89: coex: add v1 summary info to parse the traffic status from firmware

This debug entry is to summarize important messages to quickly address
problem types, such as firmware hang, C2H/H2C stuck, or too much
occupation of BT A2DP. If unexpected something is addressed, we can dig
the problem via other debug messages that provide more detail information.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-4-pkshih@realtek.com
21 months agowifi: rtw89: coex: translate slot ID to readable name
Ching-Te Ku [Tue, 20 Sep 2022 01:09:32 +0000 (09:09 +0800)]
wifi: rtw89: coex: translate slot ID to readable name

To analyze debug log quickly, use readable name in string format instead.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-3-pkshih@realtek.com
21 months agowifi: rtw89: coex: add v1 cycle report to parsing Bluetooth A2DP status
Ching-Te Ku [Tue, 20 Sep 2022 01:09:31 +0000 (09:09 +0800)]
wifi: rtw89: coex: add v1 cycle report to parsing Bluetooth A2DP status

'cysta' is short for statistics for cycles. That is a circular buffer to
record snapshot status including beacon count, RX count, TX count etc.
Since 8852CE Wi-Fi firmware report this statistics in different format,
add v1 parser by this patch.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220920010939.12173-2-pkshih@realtek.com
21 months agowifi: mwifiex: fix array of flexible structures warnings
Johannes Berg [Sun, 4 Sep 2022 19:29:07 +0000 (21:29 +0200)]
wifi: mwifiex: fix array of flexible structures warnings

There are two, just change them to have a "u8 data[]" type
member, and add casts where needed. No binary changes.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220904212910.2c885310ebee.If7177ea588b56c405eee6e6df595e9efccdfb99a@changeid
21 months agowifi: rndis_wlan: fix array of flexible structures warning
Johannes Berg [Sun, 4 Sep 2022 19:29:04 +0000 (21:29 +0200)]
wifi: rndis_wlan: fix array of flexible structures warning

Use "u8 bssid_data[]" with an appropriate cast. No binary
changes.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220904212910.0e7ce5fdbcfb.I972158734def3f93b93a3858a087cbc2cca6337e@changeid
21 months agowifi: ipw2x00: fix array of flexible structures warnings
Johannes Berg [Sun, 4 Sep 2022 19:29:02 +0000 (21:29 +0200)]
wifi: ipw2x00: fix array of flexible structures warnings

There are a number of these here, fix them by using
appropriate casts. No binary changes.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220904212910.645346411660.I471e8fadce54ea262920828f25b8e84545bcd07e@changeid
21 months agomwifiex: fix sleep in atomic context bugs caused by dev_coredumpv
Duoming Zhou [Tue, 23 Aug 2022 11:21:27 +0000 (19:21 +0800)]
mwifiex: fix sleep in atomic context bugs caused by dev_coredumpv

There are sleep in atomic context bugs when uploading device dump
data in mwifiex. The root cause is that dev_coredumpv could not
be used in atomic contexts, because it calls dev_set_name which
include operations that may sleep. The call tree shows execution
paths that could lead to bugs:

   (Interrupt context)
fw_dump_timer_fn
  mwifiex_upload_device_dump
    dev_coredumpv(..., GFP_KERNEL)
      dev_coredumpm()
        kzalloc(sizeof(*devcd), gfp); //may sleep
        dev_set_name
          kobject_set_name_vargs
            kvasprintf_const(GFP_KERNEL, ...); //may sleep
            kstrdup(s, GFP_KERNEL); //may sleep

The corresponding fail log is shown below:

[  135.275938] usb 1-1: == mwifiex dump information to /sys/class/devcoredump start
[  135.281029] BUG: sleeping function called from invalid context at include/linux/sched/mm.h:265
...
[  135.293613] Call Trace:
[  135.293613]  <IRQ>
[  135.293613]  dump_stack_lvl+0x57/0x7d
[  135.293613]  __might_resched.cold+0x138/0x173
[  135.293613]  ? dev_coredumpm+0xca/0x2e0
[  135.293613]  kmem_cache_alloc_trace+0x189/0x1f0
[  135.293613]  ? devcd_match_failing+0x30/0x30
[  135.293613]  dev_coredumpm+0xca/0x2e0
[  135.293613]  ? devcd_freev+0x10/0x10
[  135.293613]  dev_coredumpv+0x1c/0x20
[  135.293613]  ? devcd_match_failing+0x30/0x30
[  135.293613]  mwifiex_upload_device_dump+0x65/0xb0
[  135.293613]  ? mwifiex_dnld_fw+0x1b0/0x1b0
[  135.293613]  call_timer_fn+0x122/0x3d0
[  135.293613]  ? msleep_interruptible+0xb0/0xb0
[  135.293613]  ? lock_downgrade+0x3c0/0x3c0
[  135.293613]  ? __next_timer_interrupt+0x13c/0x160
[  135.293613]  ? lockdep_hardirqs_on_prepare+0xe/0x220
[  135.293613]  ? mwifiex_dnld_fw+0x1b0/0x1b0
[  135.293613]  __run_timers.part.0+0x3f8/0x540
[  135.293613]  ? call_timer_fn+0x3d0/0x3d0
[  135.293613]  ? arch_restore_msi_irqs+0x10/0x10
[  135.293613]  ? lapic_next_event+0x31/0x40
[  135.293613]  run_timer_softirq+0x4f/0xb0
[  135.293613]  __do_softirq+0x1c2/0x651
...
[  135.293613] RIP: 0010:default_idle+0xb/0x10
[  135.293613] RSP: 0018:ffff888006317e68 EFLAGS: 00000246
[  135.293613] RAX: ffffffff82ad8d10 RBX: ffff888006301cc0 RCX: ffffffff82ac90e1
[  135.293613] RDX: ffffed100d9ff1b4 RSI: ffffffff831ad140 RDI: ffffffff82ad8f20
[  135.293613] RBP: 0000000000000003 R08: 0000000000000000 R09: ffff88806cff8d9b
[  135.293613] R10: ffffed100d9ff1b3 R11: 0000000000000001 R12: ffffffff84593410
[  135.293613] R13: 0000000000000000 R14: 0000000000000000 R15: 1ffff11000c62fd2
...
[  135.389205] usb 1-1: == mwifiex dump information to /sys/class/devcoredump end

This patch uses delayed work to replace timer and moves the operations
that may sleep into a delayed work in order to mitigate bugs, it was
tested on Marvell 88W8801 chip whose port is usb and the firmware is
usb8801_uapsta.bin. The following is the result after using delayed
work to replace timer.

[  134.936453] usb 1-1: == mwifiex dump information to /sys/class/devcoredump start
[  135.043344] usb 1-1: == mwifiex dump information to /sys/class/devcoredump end

As we can see, there is no bug now.

Fixes: f5ecd02a8b20 ("mwifiex: device dump support for usb interface")
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Reviewed-by: Brian Norris <briannorris@chromium.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/5cfa5c473ff6d069cb67760ffa04a2f84ef450a8.1661252818.git.duoming@zju.edu.cn
22 months agowifi: ath11k: Add WoW support for WCN6750
Manikanta Pubbisetty [Mon, 19 Sep 2022 12:47:14 +0000 (15:47 +0300)]
wifi: ath11k: Add WoW support for WCN6750

Add support for WoW on WCN6750 chipset.

Unlike other chips where WoW exit happens after sending WoW wakeup
WMI command, exit from WoW suspend in the case of WCN6750 happens
upon sending a WoW exit SMP2P (Shared memory point to point) message
to the firmware.

Tested-on: WCN6750 hw1.0 AHB WLAN.MSL.1.0.1-00887-QCAMSLSWPLZ-1

Signed-off-by: Manikanta Pubbisetty <quic_mpubbise@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220902112520.24804-3-quic_mpubbise@quicinc.com
22 months agodt: bindings: net: add bindings to add WoW support on WCN6750
Manikanta Pubbisetty [Mon, 19 Sep 2022 12:47:13 +0000 (15:47 +0300)]
dt: bindings: net: add bindings to add WoW support on WCN6750

Add required bindings to support WoW (Wake on Wireless) on
WCN6750 which is based on ath11k driver.

Signed-off-by: Manikanta Pubbisetty <quic_mpubbise@quicinc.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220902112520.24804-2-quic_mpubbise@quicinc.com
22 months agowifi: ath11k: Include STA_KEEPALIVE_ARP_RESPONSE TLV header by default
Baochen Qiang [Tue, 13 Sep 2022 04:43:58 +0000 (12:43 +0800)]
wifi: ath11k: Include STA_KEEPALIVE_ARP_RESPONSE TLV header by default

In current code STA_KEEPALIVE_ARP_RESPONSE TLV header is included only
when ARP method is used, this causes firmware always to crash when wowlan
is enabled because firmware needs it to be present no matter ARP method
is used or not.

Fix this issue by including STA_KEEPALIVE_ARP_RESPONSE TLV header by
default.

Also fix below typo:
  s/WMI_TAG_STA_KEEPALVE_ARP_RESPONSE/WMI_TAG_STA_KEEPALIVE_ARP_RESPONSE/

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Fixes: 0f84a156aa3b ("ath11k: Handle keepalive during WoWLAN suspend and resume")
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220913044358.2037-1-quic_bqiang@quicinc.com
22 months agowifi: wcn36xx: fix repeated words in comments
Jilin Yuan [Thu, 15 Sep 2022 03:04:28 +0000 (11:04 +0800)]
wifi: wcn36xx: fix repeated words in comments

Delete the redundant word 'that'.

Signed-off-by: Jilin Yuan <yuanjilin@cdjrlc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220915030428.38510-1-yuanjilin@cdjrlc.com
22 months agowifi: wcn36xx: Add RX frame SNR as a source of system entropy
Bryan O'Donoghue [Thu, 15 Sep 2022 00:41:17 +0000 (01:41 +0100)]
wifi: wcn36xx: Add RX frame SNR as a source of system entropy

The signal-to-noise-ratio SNR is returned by the wcn36xx firmware for each
received frame. SNR represents all of the unwanted interference signal
after filtering out the fundamental frequency and harmonics of the
frequency.

Noise can come from various electromagnetic sources, from temperature
affecting the performance hardware components or quantization effects
converting from analog to digital domains.

The SNR value returned by the WiFi firmware then is a good source of
entropy.

Other WiFi drivers offer up the noise component of the FFT as an entropy
source for the random pool e.g.

commit 2aa56cca3571 ("ath9k: Mix the received FFT bins to the random pool")

I attended Jason's talk on sources of randomness at Plumbers and it
occurred to me that SNR is a reasonable candidate to add.

Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220915004117.1562703-2-bryan.odonoghue@linaro.org
22 months agowifi: rtw89: uninitialized variable on error in rtw89_early_fw_feature_recognize()
Dan Carpenter [Thu, 15 Sep 2022 14:13:34 +0000 (17:13 +0300)]
wifi: rtw89: uninitialized variable on error in rtw89_early_fw_feature_recognize()

If request_partial_firmware_into_buf() fails then "firmware" is not
initialized and the release_firmware(firmware) will crash.

Fixes: deebea35d699 ("wifi: rtw89: early recognize FW feature to decide if chanctx")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/YyMzDtX/3fUBnonC@kili
22 months agowifi: rt2x00: Fix "Error - Attempt to send packet over invalid queue 2"
Hans de Goede [Thu, 8 Sep 2022 17:36:18 +0000 (19:36 +0200)]
wifi: rt2x00: Fix "Error - Attempt to send packet over invalid queue 2"

Even though ieee80211_hw.queues is set to 2, the ralink rt2x00 driver
is seeing tx skbs submitted to it with the queue-id set to 2 / set to
IEEE80211_AC_BE on a rt2500 card when associating with an access-point.

This causes rt2x00queue_get_tx_queue() to return NULL and the following
error to be logged: "ieee80211 phy0: rt2x00mac_tx: Error - Attempt to
send packet over invalid queue 2", after which association with the AP
fails.

This patch works around this by mapping QID_AC_BE and QID_AC_BK
to QID_AC_VI when there are only 2 tx_queues.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220908173618.155291-2-hdegoede@redhat.com
22 months agowifi: rtw89: 8852c: add multi-port ID to TX descriptor
Po-Hao Huang [Fri, 16 Sep 2022 03:38:11 +0000 (11:38 +0800)]
wifi: rtw89: 8852c: add multi-port ID to TX descriptor

Update tx descriptor settings so broadcast packets on other ports can be
issued properly when DTIM count is 0. Before this, all broadcast packets
are sent via port 0 and won't be transmitted correctly.

Signed-off-by: Po-Hao Huang <phhuang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220916033811.13862-8-pkshih@realtek.com
22 months agowifi: rtw89: fix rx filter after scan
Po-Hao Huang [Fri, 16 Sep 2022 03:38:10 +0000 (11:38 +0800)]
wifi: rtw89: fix rx filter after scan

In monitor mode we should be able to received all packets even if it's not
destined to us. But after scan, the configuration was wrongly set, so we
fix it.

Signed-off-by: Po-Hao Huang <phhuang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220916033811.13862-7-pkshih@realtek.com
22 months agowifi: rtw89: free unused skb to prevent memory leak
Po-Hao Huang [Fri, 16 Sep 2022 03:38:09 +0000 (11:38 +0800)]
wifi: rtw89: free unused skb to prevent memory leak

This avoid potential memory leak under power saving mode.

Signed-off-by: Po-Hao Huang <phhuang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220916033811.13862-6-pkshih@realtek.com
22 months agowifi: rtw89: reset halt registers before turn on wifi CPU
Ping-Ke Shih [Fri, 16 Sep 2022 03:38:08 +0000 (11:38 +0800)]
wifi: rtw89: reset halt registers before turn on wifi CPU

Reset these registers to prevent firmware get false alarm after wifi CPU
is running.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220916033811.13862-5-pkshih@realtek.com
22 months agowifi: rtw89: pci: update LTR settings
Ping-Ke Shih [Fri, 16 Sep 2022 03:38:07 +0000 (11:38 +0800)]
wifi: rtw89: pci: update LTR settings

Modify PCI LTR control flow and LTR idle latency to improve power save
efficiency.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220916033811.13862-4-pkshih@realtek.com
22 months agowifi: rtw89: mac: set NAV upper to 25ms
Ping-Ke Shih [Fri, 16 Sep 2022 03:38:06 +0000 (11:38 +0800)]
wifi: rtw89: mac: set NAV upper to 25ms

NAV upper register is to limit the maximum NAV value to prevent
unexpected NAV, but the old setting is too small to reflect NAV from
AP transmiting big MPDU at once.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220916033811.13862-3-pkshih@realtek.com
22 months agowifi: rtw89: initialize DMA of CMAC
Ping-Ke Shih [Fri, 16 Sep 2022 03:38:05 +0000 (11:38 +0800)]
wifi: rtw89: initialize DMA of CMAC

8852C needs this to change CMAC dma to full mode to keep receiving
packets after RX full event being resolved.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220916033811.13862-2-pkshih@realtek.com
22 months agowifi: rtw89: support SER L1 simulation
Zong-Zhe Yang [Wed, 14 Sep 2022 03:50:34 +0000 (11:50 +0800)]
wifi: rtw89: support SER L1 simulation

SER (system error recovery) can deal with different crash types by
different levels of processes. Previous FW crash simulation triggers
a CPU exception which is one kind of SER L2 type. It can verify SER L2
flow which includes HW/FW restart.

Now, we want to increase crash simulation types. A debug function is added
to trigger control error in purpose for SER L1 simulation/verification.
And, debugfs fw_crash is extended to accept different parameters.

echo 1 > fw_crash:
simulate CPU exception as before
(keep 1 for compatibility with previous)

It will be catched and handled by SER L2.
(this requires HW/FW restart)

echo 2 > fw_crash:
simulate control error

It will be catched and handled by SER L1.
(driver and FW cooperate to recover this)

Besides, in order to apply to the above two cases,
rename RTW89_FLAG_RESTART_TRIGGER to RTW89_FLAG_CRASH_SIMULATING
and adjust where SER flow clears this bit for both L1 and L2.

Signed-off-by: Zong-Zhe Yang <kevin_yang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220914035034.14521-5-pkshih@realtek.com
22 months agowifi: rtw89: 8852c: support fw crash simulation
Zong-Zhe Yang [Wed, 14 Sep 2022 03:50:33 +0000 (11:50 +0800)]
wifi: rtw89: 8852c: support fw crash simulation

With FW >= v0.27.40.0, 8852C FW has feature to handle crash simulation.
Besides, use RTW89_WCPU_BASE_MASK to replace use of RTW89_WCPU_BASE_ADDR
and work for both 8852A and 8852C.

Signed-off-by: Zong-Zhe Yang <kevin_yang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220914035034.14521-4-pkshih@realtek.com
22 months agowifi: rtw89: introudce functions to drop packets
Zong-Zhe Yang [Wed, 14 Sep 2022 03:50:32 +0000 (11:50 +0800)]
wifi: rtw89: introudce functions to drop packets

Introudce a H2C feature to drop packets according to given parameters.
And, we implement instances to drop packets from BE, BK, VI, VO queues
by vif or sta. Then, we refine our callback of ieee80211_ops::flush to
deal with the case of drop=true via this feature.

Signed-off-by: Zong-Zhe Yang <kevin_yang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220914035034.14521-3-pkshih@realtek.com
22 months agowifi: rtw89: unify use of rtw89_h2c_tx()
Ping-Ke Shih [Wed, 14 Sep 2022 03:50:31 +0000 (11:50 +0800)]
wifi: rtw89: unify use of rtw89_h2c_tx()

Seaprate calling of rtw89_h2c_tx() out of if-expression, and bypass the
return value to upper caller.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220914035034.14521-2-pkshih@realtek.com
22 months agowifi: rtw89: pci: concentrate control function of TX DMA channel
Chin-Yen Lee [Mon, 12 Sep 2022 07:17:06 +0000 (15:17 +0800)]
wifi: rtw89: pci: concentrate control function of TX DMA channel

Different chips use different register and mask for
tx dma channels, so concentrate them.

Signed-off-by: Chin-Yen Lee <timlee@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220912071706.13619-4-pkshih@realtek.com
22 months agowifi: rtw89: correct enable functions of HCI/PCI DMA
Chin-Yen Lee [Mon, 12 Sep 2022 07:17:05 +0000 (15:17 +0800)]
wifi: rtw89: correct enable functions of HCI/PCI DMA

Some PCI and MAC registers are changed for different
chips and correct them accordingly. And HCI MAD functions
belongs to MAC core, so move it to mac.h/.c.

Signed-off-by: Chin-Yen Lee <timlee@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220912071706.13619-3-pkshih@realtek.com
22 months agowifi: rtw89: 8852c: L1 DMA reset has offloaded to FW
Zong-Zhe Yang [Mon, 12 Sep 2022 07:17:04 +0000 (15:17 +0800)]
wifi: rtw89: 8852c: L1 DMA reset has offloaded to FW

For 8852C, rtw89_pci_lv1rst_stop_dma() and rtw89_pci_lv1rst_start_dma()
are offloaded to FW L1 reset flow. So, driver no longer needs to do them.

Signed-off-by: Zong-Zhe Yang <kevin_yang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220912071706.13619-2-pkshih@realtek.com
22 months agowifi: rtw89: support for setting TID specific configuration
Kuan-Chung Chen [Mon, 12 Sep 2022 07:00:14 +0000 (15:00 +0800)]
wifi: rtw89: support for setting TID specific configuration

Add ops set_tid_config to support TID specific configuration.
We currently only support ampdu setting.

The command example is:
  iw wlan0 set tidconf tids 0x3 ampdu off
  iw wlan0 set tidconf peer xx:xx:xx:xx:xx:xx tids 0x2 ampdu on

Signed-off-by: Kuan-Chung Chen <damon.chen@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220912070014.10018-3-pkshih@realtek.com
22 months agowifi: rtw89: support for setting HE GI and LTF
Kuan-Chung Chen [Mon, 12 Sep 2022 07:00:13 +0000 (15:00 +0800)]
wifi: rtw89: support for setting HE GI and LTF

Support setting HE GI and LTF values to the kernel via nl80211.
We currently only support some GI and LTF values settings.

The command example is:
  iw wlan0 set bitrates he-gi-2.4 0.8 he-ltf-2.4 2

Signed-off-by: Kuan-Chung Chen <damon.chen@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220912070014.10018-2-pkshih@realtek.com
22 months agowifi: rtw89: coex: show connecting state in debug message
Ping-Ke Shih [Mon, 12 Sep 2022 02:10:09 +0000 (10:10 +0800)]
wifi: rtw89: coex: show connecting state in debug message

The variable cnt_connecting is to indicate if we are connecting to an AP.
This is an important clue for coexistence to assign more time slot to WiFi
side in this situation to ensure WiFi can establish connection.

Without this patch, compiler warns:
  drivers/net/wireless/realtek/rtw89/coex.c:3244:25: warning: variable
    'cnt_connecting' set but not used [-Wunused-but-set-variable]

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220912021009.6011-1-pkshih@realtek.com
22 months agowifi: rtw89: coex: Combine set grant WL/BT and correct the debug log
Ching-Te Ku [Tue, 13 Sep 2022 09:25:46 +0000 (17:25 +0800)]
wifi: rtw89: coex: Combine set grant WL/BT and correct the debug log

To reduce register IO, combine set_gnt_wl/set_gnt_bt to set the same
register one time. Because RTL8852C use different register to control
antenna path, so make correction of path control related debug logs.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220913092546.43722-8-pkshih@realtek.com
22 months agowifi: rtw89: coex: Remove trace_step at COEX-MECH control structure for RTL8852C
Ching-Te Ku [Tue, 13 Sep 2022 09:25:45 +0000 (17:25 +0800)]
wifi: rtw89: coex: Remove trace_step at COEX-MECH control structure for RTL8852C

RTL8852C don't need to send the data trace_step which used to tell
firmware how many TDMA steps should record. Remove the member.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220913092546.43722-7-pkshih@realtek.com
22 months agowifi: rtw89: coex: Parsing Wi-Fi firmware TDMA info from reports
Ching-Te Ku [Tue, 13 Sep 2022 09:25:44 +0000 (17:25 +0800)]
wifi: rtw89: coex: Parsing Wi-Fi firmware TDMA info from reports

Show TDMA information containing TDMA policy and time slot of Wi-Fi/BT in
debug message to check things are in expected. The v1 format contains
additional header, and remaining part is the same as original. So 8852CE
selects v1 version, and then everything like original.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220913092546.43722-6-pkshih@realtek.com
22 months agowifi: rtw89: coex: Parsing Wi-Fi firmware error message from reports
Ching-Te Ku [Tue, 13 Sep 2022 09:25:43 +0000 (17:25 +0800)]
wifi: rtw89: coex: Parsing Wi-Fi firmware error message from reports

Parsing firmware error message from original version and v1 reports to
show up exception counter of commands from firmware in debug message.
Then, we can make sure exchange commands are correct totally.

In the later version Wi-Fi firmware(v1), the report format was changed.
With this update, we can yield correct report from proper struct.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220913092546.43722-5-pkshih@realtek.com
22 months agowifi: rtw89: coex: Move coexistence firmware buffer size parameter to chip info
Ching-Te Ku [Tue, 13 Sep 2022 09:25:42 +0000 (17:25 +0800)]
wifi: rtw89: coex: Move coexistence firmware buffer size parameter to chip info

Because RTL8852A/RTL8852C use different firmware buffer size to
send C2H packet, it's necessary to use different size to parse C2H report.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220913092546.43722-4-pkshih@realtek.com
22 months agowifi: rtw89: coex: Add v1 Wi-Fi firmware power-saving null data report
Ching-Te Ku [Tue, 13 Sep 2022 09:25:41 +0000 (17:25 +0800)]
wifi: rtw89: coex: Add v1 Wi-Fi firmware power-saving null data report

The later version Wi-Fi firmware will report null data TX times,
so the structure is different from before.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220913092546.43722-3-pkshih@realtek.com
22 months agowifi: rtw89: coex: use void pointer as temporal type to copy report
Ping-Ke Shih [Tue, 13 Sep 2022 09:25:40 +0000 (17:25 +0800)]
wifi: rtw89: coex: use void pointer as temporal type to copy report

With void pointer, we don't need to cast to 'u8 *' by one by.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220913092546.43722-2-pkshih@realtek.com
22 months agoarm64: dts: apple: Add WiFi module and antenna properties
Hector Martin [Fri, 16 Sep 2022 16:03:22 +0000 (17:03 +0100)]
arm64: dts: apple: Add WiFi module and antenna properties

Add the new module-instance/antenna-sku properties required to select
WiFi firmwares properly to all board device trees.

Signed-off-by: Hector Martin <marcan@marcan.st>
Reviewed-by: Mark Kettenis <kettenis@openbsd.org>
Acked-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/E1oZDoI-0077b3-Dd@rmk-PC.armlinux.org.uk
22 months agowifi: brcmfmac: pcie: Add IDs/properties for BCM4378
Hector Martin [Fri, 16 Sep 2022 16:03:17 +0000 (17:03 +0100)]
wifi: brcmfmac: pcie: Add IDs/properties for BCM4378

This chip is present on Apple M1 (t8103) platforms:

* atlantisb (apple,j274): Mac mini (M1, 2020)
* honshu    (apple,j293): MacBook Pro (13-inch, M1, 2020)
* shikoku   (apple,j313): MacBook Air (M1, 2020)
* capri     (apple,j456): iMac (24-inch, 4x USB-C, M1, 2020)
* santorini (apple,j457): iMac (24-inch, 2x USB-C, M1, 2020)

Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Hector Martin <marcan@marcan.st>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/E1oZDoD-0077ax-AI@rmk-PC.armlinux.org.uk
22 months agowifi: brcmfmac: pcie: Support PCIe core revisions >= 64
Hector Martin [Fri, 16 Sep 2022 16:03:12 +0000 (17:03 +0100)]
wifi: brcmfmac: pcie: Support PCIe core revisions >= 64

These newer PCIe core revisions include new sets of registers that must
be used instead of the legacy ones. Introduce a brcmf_pcie_reginfo to
hold the specific register offsets and values to use for a given
platform, and change all the register accesses to indirect through it.

Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/E1oZDo8-0077aq-6I@rmk-PC.armlinux.org.uk
22 months agowifi: brcmfmac: msgbuf: Increase RX ring sizes to 1024
Hector Martin [Fri, 16 Sep 2022 16:03:07 +0000 (17:03 +0100)]
wifi: brcmfmac: msgbuf: Increase RX ring sizes to 1024

Newer chips used on Apple platforms have a max_rxbufpost greater than
512, which causes warnings when brcmf_msgbuf_rxbuf_data_fill tries to
put more entries in the ring than will fit. Increase the ring sizes
to 1024.

Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Hector Martin <marcan@marcan.st>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/E1oZDo3-0077ak-2h@rmk-PC.armlinux.org.uk