Skip to content

3.1.0 wi-fi upmerge #3045

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 41 commits into from
Jul 21, 2025
Merged

3.1.0 wi-fi upmerge #3045

merged 41 commits into from
Jul 21, 2025

Conversation

rado17
Copy link
Contributor

@rado17 rado17 commented Jul 14, 2025

Wi-Fi only Upmerge for 3.1.0

manifest-pr-skip

@krish2718
Copy link
Contributor

Why do we still need this noup, it was done last time as were in the release rush. If the certs from upstream don't work out-of-box then we can submit a variant that works out-of-box for nRF platforms.

@rado17
Copy link
Contributor Author

rado17 commented Jul 14, 2025

Why do we still need this noup, it was done last time as were in the release rush. If the certs from upstream don't work out-of-box then we can submit a variant that works out-of-box for nRF platforms.

Nordic specific directory in upstream certs then? RSA-3k certs work fine, but we may have to enable additional configs for RSA-2k certs from upstream.
Also, for MBEDTLS_HEAP_SIZE, the increase is required only with downstream. So, I think, noup is justified for that commit.

@krish2718
Copy link
Contributor

Nordic specific directory in upstream certs then? RSA-3k certs work fine, but we may have to enable additional configs for RSA-2k certs from upstream.

I would put a algo specific directory e.g., rsa2k-no-ebc3, this way it can be reused in case other platforms have similar constraints.

Also, for MBEDTLS_HEAP_SIZE, the increase is required only with downstream. So, I think, noup is justified for that commit.

We don't want to rely on noup unless its absolutely must, why can't we add <increased_heap> if SOC_FAMILY_NORDIC_NRF to upstream?

@rado17
Copy link
Contributor Author

rado17 commented Jul 14, 2025

Nordic specific directory in upstream certs then? RSA-3k certs work fine, but we may have to enable additional configs for RSA-2k certs from upstream.

I would put a algo specific directory e.g., rsa2k-no-ebc3, this way it can be reused in case other platforms have similar constraints.

Also, for MBEDTLS_HEAP_SIZE, the increase is required only with downstream. So, I think, noup is justified for that commit.

We don't want to rely on noup unless its absolutely must, why can't we add <increased_heap> if SOC_FAMILY_NORDIC_NRF to upstream?

Let me check this.

Copy link

github-actions bot commented Jul 14, 2025

The following west manifest projects have changed revision in this Pull Request:

Name Old Revision New Revision Diff
hostap zephyrproject-rtos/hostap@8412f4b zephyrproject-rtos/hostap@e942f86 (main) zephyrproject-rtos/[email protected]
nrf_wifi zephyrproject-rtos/nrf_wifi@936b502 zephyrproject-rtos/nrf_wifi@5f59c23 (main) zephyrproject-rtos/[email protected]

All manifest checks OK

Note: This message is automatically posted and updated by the Manifest GitHub Action.

@krish2718 krish2718 force-pushed the certs_update branch 2 times, most recently from 6697ba5 to 380a770 Compare July 14, 2025 20:11
@kapbh kapbh force-pushed the certs_update branch 2 times, most recently from 6a2a279 to 359ba2e Compare July 15, 2025 10:21
@rado17 rado17 force-pushed the certs_update branch 4 times, most recently from 882d7e2 to 245e1c7 Compare July 17, 2025 07:07
@rado17 rado17 changed the title Update certificates for RSA-3072 support with NRF security Wi-Fi upmerge: Update certificates for RSA-3072 support with NRF security Jul 17, 2025
@krish2718 krish2718 changed the title Wi-Fi upmerge: Update certificates for RSA-3072 support with NRF security 3.1.0 wi-fi upmerge Jul 20, 2025
@krish2718 krish2718 requested a review from D-Triveni July 20, 2025 09:06
krish2718 and others added 27 commits July 21, 2025 13:54
In case FMAC API fails, increment the TX drop counter, helpful in
debug.

Signed-off-by: Chaitanya Tata <[email protected]>
(cherry picked from commit e728f80)
…ull check

Deleted a redundant check for 'rpu_ctx_zep' pointer after it was already
dereferenced.

Clarifies code logic in nrf_wifi_get_power_save_config function.

Signed-off-by: Gaetan Perrot <[email protected]>
(cherry picked from commit bac20e419eabf36625c073a366008f1384fcc272)
Return proper error codes for xmit instead of generic -1.

Signed-off-by: Chaitanya Tata <[email protected]>
(cherry picked from commit 001f34d84c647d15d76480cb64fc6efce6ea736f)
Rpu recovey debug stats are stored in hal_dev_ctx which is not
persistent in case of interface is brought down/up. Need to add
in nrf_wifi_ctx_zep and update before interface goes down.

Signed-off-by: Kapil Bhatt <[email protected]>
(cherry picked from commit dd7413d269d69e2d0c8ee10c3d03929703e7efb1)
Automatically hide the nRF7x interface from the connection manager if
the TX path is disabled (scan only mode). This prevents function calls
like `conn_mgr_all_if_up(true)` from bringing up the interface which
can never result in a connection.

Signed-off-by: Jordan Yates <[email protected]>
(cherry picked from commit 79edfc012803572b1c75a4667f03bfbb4fa9b6b0)
Monitor mode doesn't require station mode. Disabling station mode
require necessary changes to work monitor mode.

Upstream PR #: 92226

Signed-off-by: Kapil Bhatt <[email protected]>
…de from monitor"

This reverts commit 926fafb.

Signed-off-by: Chaitanya Tata <[email protected]>
Fixes memory leak seen during continuous raw TX transmission.

Signed-off-by: Chaitanya Tata <[email protected]>
(cherry picked from commit d5c7cb171b21049758f2907ec4b5a85ac73b43eb)
…port"

This reverts commit e2f9c7b.

Signed-off-by: Chaitanya Tata <[email protected]>
Add a snippet for Wi-Fi credentials support. Update the heap sizes as
required for enterprise mode.

Signed-off-by: Ravi Dondaputi <[email protected]>
(cherry picked from commit 4c14c618f69e8ae0d5ee95a6abbc8a1d5f89afac)
…lity issue.

1. When external station connects to soft ap, zephyr
   hostap will fill sta_info before sending mgmt event
   to l2 wifi layer.
   sta_info.twt_capable should be filled with external
   station capability rather than soft ap self.
2. Rename hapd_is_twt_capable to
   hapd_get_sta_he_twt_capable.
   Try to get twt_capa form capability of external station
   rather than soft ap.

Signed-off-by: Qiankun Li <[email protected]>
(cherry picked from commit 7ffc538)
Earlier we had socketpair to pass wifi event information from
wpa_supplicant side to zephyr adaption layer. This is now replaced
by k_fifo to save some RAM.

Signed-off-by: Jukka Rissanen <[email protected]>
(cherry picked from commit 96818f45a91472cb0eabd5cfe42b5b256860f6a5)
In case anyone enabled EAP_*_ALL explicitly without enabling the
Enterprise then it leads to a build error.

Fix by adding the dependency.

Signed-off-by: Chaitanya Tata <[email protected]>
(cherry picked from commit 014dee7)
If the disconnect event is raised before the network has been connected,
report the connection result as `WIFI_STATUS_CONN_FAIL`, instead of as
`WIFI_REASON_DISCONN_SUCCESS`, which is interpretted as
`WIFI_STATUS_CONN_SUCCESS`.

Signed-off-by: Jordan Yates <[email protected]>
(cherry picked from commit 9245f58b4ccf32b261341fdfc09e993b7eba74f0)
Set enterprise crypto insecure because certifcate validation
is disabled.

Signed-off-by: Flavio Ceolin <[email protected]>
(cherry picked from commit 979fcc0)
Ensure 'params' is not NULL before accessing its fields.

Prevents possible null pointer dereference when calling
strlen(params->ssid).

Delay access to ssid->ssid and ssid->ssid_len until after null check.

Prevents potential crash if wpa_s->current_ssid is NULL.

Signed-off-by: Gaetan Perrot <[email protected]>
(cherry picked from commit b6a5202)
Add a new promptless Kconfig symbol (INSECURE). This symbols must
be selected by any setting which is used to enable an insecure
feature.

Signed-off-by: Flavio Ceolin <[email protected]>
(cherry picked from commit 674fd094aa67b31d1d54b1f318adce8ed2faddd6)
Move usage of ap_ctx pointers after null checks to prevent potential
crashes.

Signed-off-by: Gaetan Perrot <[email protected]>
(cherry picked from commit cd85510)
Move usage of ap_ctx pointers after null checks to prevent potential
crashes.

Signed-off-by: Gaetan Perrot <[email protected]>
(cherry picked from commit 8a565c0a2b8c12b1ea0298367a0ff525c1ae20b8)
This change is needed to handle to stack overflow issues
when using SPIM.

Signed-off-by: Triveni Danda <[email protected]>
(cherry picked from commit 0dcb5a4923f1bf07229e1695d726edb990064572)
… SAE mode

When the external AP is in WPA3 SAE mode, the group cipher uses
TKIP and the pairwise cipher uses TKIP and CCMP. Should not
connect to the AP.
For WPA2 & WPA3 SAE & WPA2/WPA3 mixed mode, set group cipher to
CCMP, pairwise cipher to CCMP.

Signed-off-by: Gang Li <[email protected]>
(cherry picked from commit c8fa5e5899575fe5c0a0a6c3013fdc4527d2cdb4)
sync for coverity fix in hostap

Signed-off-by: Qingling Wu <[email protected]>
(cherry picked from commit f089959)
Make sure to use k_fifo for IPC in hostap to save memory.

Signed-off-by: Jukka Rissanen <[email protected]>
(cherry picked from commit 03ad3a32a27da098fc697376e2932c36b85c3509)
Fix a warning seen during interface down.

Signed-off-by: Chaitanya Tata <[email protected]>
(cherry picked from commit a869c3e3a31afa893a8c88f8761f8f206769ee27)
Commit 5e25283821a("drivers: wifi: Create dedicated mem pool for Wi-Fi
driver") introduced OSAL dependecy in the Zephyr QSPI driver for HL
read, but in bustest we don't enable nrf_wifi OS module, so, it crashes
here. And we should not be using OSAL APIs in Zephyr code anyway.

And in this case we don't even need to use the heap, so, move the rx
buffer to stack.

Upstream PR #: 93303

Signed-off-by: Chaitanya Tata <[email protected]>
…covery

When running Zperf traffic + scan in the background eventual we hit a
deadlock:

 * sysworkq: recovery->stop_zep->vif_lock->hal_disable->wait lock_rx
 * nrf70_bh_wq: event_tasklet->lock_rx->disp_scan_done->
             disp_scan_res_get_zep-> waiting on vif_lock

The traffic triggers recovery (another bug) and conflicts with display
scan.

Fix by moving scan results processing to system workqueue instead of
doing it in the FMAC event callback context, this is how supplicant scan
also works.

Upstream PR #: 93377

Signed-off-by: Chaitanya Tata <[email protected]>
hostap needs more eventfd to function so set a proper default value.

Signed-off-by: Jukka Rissanen <[email protected]>
(cherry picked from commit 410afba)
@rlubos rlubos merged commit cebbc7c into nrfconnect:main Jul 21, 2025
24 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.