- 22 Mar, 2023 24 commits
-
-
Felix Fietkau authored
MT7996 hardware supports mesh A-MSDU subframes in hardware, but uses a big-endian length field Signed-off-by:
Felix Fietkau <nbd@nbd.name> Signed-off-by:
Ryder Lee <ryder.lee@mediatek.com> Link: https://lore.kernel.org/r/20230314095956.62085-7-nbd@nbd.nameSigned-off-by:
Johannes Berg <johannes.berg@intel.com>
-
Felix Fietkau authored
This helps bring down rx CPU usage by avoiding calls to the rx handlers in the slow path. Supports forwarding and local rx, including A-MSDU. Signed-off-by:
-
Felix Fietkau authored
Significantly reduces mesh forwarding path CPU usage and enables the direct use of iTXQ. Signed-off-by:
-
Felix Fietkau authored
Previously, fast xmit only worked on interface types where initially a sta lookup is performed, and a cached header can be attached to the sta, requiring only some fields to be updated at runtime. This technique is not directly applicable for a mesh device type due to the dynamic nature of the topology and protocol. There are more addresses that need to be filled, and there is an extra header with a dynamic length based on the addressing mode. Change the code to cache entries contain a copy of the mesh subframe header + bridge tunnel header, as well as an embedded struct ieee80211_fast_tx, which contains the information for building the 802.11 header. Add a mesh specific early fast xmit call, which looks up a cached entry and adds only the mesh subframe header, before passing it over to the generic fast xmit code. To ensure the changes in network are reflected in these cached headers, flush affected cached entries on path changes, as well as other conditions that currently trigger a fast xmit check in other modes (key changes etc.) This code is loosely based on a previous implementation by: Sriram R <quic_srirrama@quicinc.com> Cc: Sriram R <quic_srirrama@quicinc.com> Signed-off-by:
-
Felix Fietkau authored
Since the sequence number is shared across different tx queues, it needs to be atomic in order to avoid accidental duplicate assignment Signed-off-by:
-
Felix Fietkau authored
On newer MediaTek SoCs (e.g. MT7986), WLAN->WLAN or WLAN->Ethernet flows can be offloaded by the SoC. In order to support that, the .ndo_setup_tc op is needed. Signed-off-by:
-
Abhishek Naik authored
Add debugfs file in mvm to retrieve TAS status per LMAC, TAS block list, current mcc, OEM name and OEM allowed list. Signed-off-by:
Abhishek Naik <abhishek.naik@intel.com> Signed-off-by:
Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230320122330.8efc8c41efae.I94e1a6efb9c33e2cdbcf4bf3ed2384005397dee9@changeidSigned-off-by:
-
Avraham Stern authored
Instead of enabling HW timestamping by default every time a station is connected, disable it by default and enable it only upon request for a specific station. HW timestamping can be enabled for only one peer at a time. Signed-off-by:
Avraham Stern <avraham.stern@intel.com> Signed-off-by:
-
Mukesh Sisodiya authored
Add missing rf support for spacific rf for BnJ device Signed-off-by:
Mukesh Sisodiya <mukesh.sisodiya@intel.com> Signed-off-by:
-
Miri Korenblit authored
As we have a new MLD STA cmd, there will be a different function to add/remove a station in MLD mode. But both functions will share a common part. Put this part in a separate function which will be used later in adding/removing a sta in MLD mode. Signed-off-by:
Miri Korenblit <miriam.rachel.korenblit@intel.com> Signed-off-by:
-
Miri Korenblit authored
In iwl_mvm_add_sta() we're initializing the new allocated mvm_sta. We are setting some fields to zero even though it was allocated with kzalloc, for the case of HW restart. But in such a case we will never get to this initializations due to the goto statement that we have in this case. Remove these initializations. Signed-off-by:
-
Miri Korenblit authored
We have a new STA cmd as part of the new MLD API. There are some parts of sending the STA cmd which are common to both the old and the new one. Put this parts in functions which will later be used to send the new STA cmd. Signed-off-by:
-
Miri Korenblit authored
As the new MLD API is introduced, we have a new enumeration for the different station types. Since struct iwl_mvm_sta's sta_type field will be in use for both new and old API, change its type to a unified one, so it can hold both. Signed-off-by:
-
Miri Korenblit authored
This function sends the STA_HE_CTXT_CMD, which won't be used in the new MLD API. Instead, a part of the fields of this command will be in the new STA cmd. Put the parts that are common to both commands in functions, which will later be used in sending the new STA cmd. Signed-off-by:
-
Miri Korenblit authored
We used to send a MAC CTXT cmd to ask the FW to not pass MCAST frames if we're associated but not authorized, because we don't have the keys in that stage, and after authorization - we sent the cmd again to ask the FW to pass MCAST, as we have the keys now. The patch linked below was changing this strategy to always allow MCAST frames, and if we're not authorized - the driver will drop them. But we're still sending the MAC CTXT cmd after deaouthorization even though we don't tell the FW to not pass MCAST frames anymore. Basically we don't tell the FW anything new with this cmd. Fix this by not sending MAC CTXT command after deauthorization. For authorization we're sending the cmd to configure other changes too, so keep it. Signed-off-by:
-
Miri Korenblit authored
This is another patch in the series adding all the ops for the new MLD ieee80211_ops. The callbacks added here use the new MLD FW API Signed-off-by:
-
Miri Korenblit authored
This is another patch in the series adding all the ops for the new MLD ieee80211_ops. The callbacks added here use the new MLD FW API Signed-off-by:
-
Avraham Stern authored
iwl_mvm_get_sync_time() reads the gp2 from the device and then reads the system clock. Since the two reads are not done atomically, unexpected delays may happen between the two reads (e.g. context switch) which make it inaccurate. In order to improve the accuracy of the cross timestamp, call iwl_mvm_get_sync_time() multiple times in a loop and take the result in which the difference between the two clock is the smallest. Signed-off-by:
-
Avraham Stern authored
Implement the following PHC operations: 1. adjtime - for adding an offset to the PHC clock time. 2. adjfine - for adjusting the PHC clock frequency. 3. gettime64 - for getting the PHC clock time. This function returns the time as adjusted by adjtime and adjfine. The adjusted time will also be used for time sync frames timestamping. Signed-off-by:
-
Johannes Berg authored
Check hardware/firmware support and enable TX beacon protection as well if supported, programming the key into the firmware as usual. Signed-off-by:
-
Krishnanand Prabhu authored
Add support for timing measurement in extended capabilities, used for time synchronization. Signed-off-by:
Krishnanand Prabhu <krishnanand.prabhu@intel.com> Signed-off-by:
-
Avraham Stern authored
For TM/FTM frames, report the hardware timestamps reported by the fw as part of the RX/TX status. Since the fw reports the timestamps in a dedicated notification (and not as part of the RX/TX status), hold the frame until the fw timestamps notification is received. Timestamping is enabled when a station is connected and disabled when disconnected. For AP interface, only the first station will have timestamping enabled since the fw only supports timestamping for one peer. Signed-off-by:
-
Avraham Stern authored
If the firmware supports reading synced GP2/ATRB timestamps, read the synced timestamps from firmware instead of reading the GP2 register and the system time separately. Reading the synced time from firmware should be more accurate. Signed-off-by:
-
Krishnanand Prabhu authored
Add support to enable/disable PHC clock. The PHC clock includes support for fetching the cross timestamp i.e. a non-atomic snapshot of the current time from the hardware (WiFi device) clock and system clock (wall-clock) simultaneously. Signed-off-by:
Luca Coelho <luciano.coelho@intel.com> Signed-off-by:
-
- 20 Mar, 2023 2 commits
-
-
Bitterblue Smith authored
This chip is found in cheap "free driver" USB adapters from Aliexpress. Initially they pretend to be a CD-ROM containing the driver for Windows. "Ejecting" switches the device to wifi mode. Features: 2.4 GHz, b/g/n mode, 1T1R, 150 Mbps. This chip is more unique than other Realtek chips: * The registers at addresses 0x0-0xff, which all the other chips use, can't be used here. New registers at 0x8000-0x80ff must be used instead. And it's not a simple matter of adding 0x8000: 0x2 (REG_SYS_FUNC) became 0x8004, 0x80 (REG_MCU_FW_DL) became 0x8090, etc. * Also there are a few new registers which must be accessed indirectly because their addresses don't fit in 16 bits. No other chips seem to have these. * The vendor driver compiles to 8188gu.ko, but the code calls the chip RTL8710B(U) pretty much everywhere, including messages visible to the user. Another difference compared to the other chips supported by rtl8xxxu is that it has a new PHY status struct, or three of them actually, from which we extract the RSSI, among other things. This is not unique, though, just new. The chips supported by rtw88 also use it. Signed-off-by:
Bitterblue Smith <rtl8821cerfe2@gmail.com> Reviewed-by:
Ping-Ke Shih <pkshih@realtek.com> Tested-by: Philipp Hortmann <philipp.g.hortmann@gmail.com> # Edimax N150 Signed-off-by:
Kalle Valo <kvalo@kernel.org> Link: https://lore.kernel.org/r/4edbe29f-00b9-8eef-9789-20bed0b141e2@gmail.com
-
Bitterblue Smith authored
Always run the entire init sequence (rtl8xxxu_init_device()) for RTL8192EU. It's what the vendor driver does too. This fixes a bug where the device is unable to connect after rebooting: wlp3s0f3u2: send auth to ... (try 1/3) wlp3s0f3u2: send auth to ... (try 2/3) wlp3s0f3u2: send auth to ... (try 3/3) wlp3s0f3u2: authentication with ... timed out Rebooting leaves the device powered on (partially? at least the firmware is still running), but not really in a working state. Cc: stable@vger.kernel.org Signed-off-by:
Jes Sorensen <jes@trained-monkey.org> Signed-off-by:
-
- 16 Mar, 2023 1 commit
-
-
Nathan Chancellor authored
Clang errors: drivers/net/wireless/intel/iwlwifi/iwl-devtrace.c:15:32: error: unknown warning group '-Wsuggest-attribute=format', ignored [-Werror,-Wunknown-warning-option] #pragma GCC diagnostic ignored "-Wsuggest-attribute=format" ^ 1 error generated. The warning being disabled by this pragma is GCC specific. Guard its use with CONFIG_CC_IS_GCC so that it is not used with clang to clear up the error. Fixes: 4eca8cbf ("wifi: iwlwifi: suppress printf warnings in tracing") Link: https://github.com/ClangBuiltLinux/linux/issues/1818Signed-off-by:Nathan Chancellor <nathan@kernel.org> Link: https://lore.kernel.org/r/20230315-iwlwifi-fix-pragma-v1-1-ad23f92c4739@kernel.orgSigned-off-by:
-
- 15 Mar, 2023 13 commits
-
-
Johannes Berg authored
This can't really be fixed due to the macro layout of tracepoints (you'd need a special tracepoint macro for when this is needed), so just suppress the warnings. Signed-off-by: -
Johannes Berg authored
There are a number of issues here: * if trans->reduce_power_loaded is already true, we call iwl_trans_set_reduce_power() with an uninitialized len value * in this case we also clobber a previous load/setting * if iwl_uefi_get_reduced_power() returns an ERR_PTR() we try to kfree() it Move the iwl_trans_set_reduce_power() call into the success case only to fix these issues. Signed-off-by:
-
Mukesh Sisodiya authored
Add missing rf support for Bz device and B-Step for BnJ device Signed-off-by:
-
Johannes Berg authored
If there's a rate->index that maps inside the range but to an uninitialized value, then that's also a bad rate, avoid printing "(nil)" in that case and rather print the "BAD_RATE" string instead as in the else branch. Signed-off-by:
-
Solomon Tan authored
This patch addresses the checkpatch.pl warning that code indent should use tabs. Signed-off-by:
Solomon Tan <solomonbstoner@protonmail.ch> Signed-off-by:
-
Solomon Tan authored
This patch addresses the error from checkpatch.pl that a space is required before an open parenthesis. Signed-off-by:
-
Solomon Tan authored
This patch addresses the error from checkpatch.pl regarding the presence of prohibited spaces. Signed-off-by:
-
Julia Lawall authored
Spelling mistakes (triple letters) in comment. Detected with the help of Coccinelle. Signed-off-by:
Julia Lawall <Julia.Lawall@inria.fr> Signed-off-by:
-
Tom Rix authored
cppcheck reports [drivers/net/wireless/intel/iwlwifi/mvm/rs.c:2686]: (warning) Assignment of function parameter has no effect outside the function. Did you forget dereferencing it? The setting of the 'sta' parameter is not needed. In the if-check that sets it to NULL, mvm_sta is also set to NULL. Then the next statement checks if mvm_sta is NULL and does an early return. So remove setting sta. Signed-off-by:
Tom Rix <trix@redhat.com> Signed-off-by:
-
Miri Korenblit authored
This is another patch in the series adding all the ops for the new MLD ieee80211_ops. The callback added here uses the new MLD FW API Signed-off-by:
-
Miri Korenblit authored
Since parts of the functionality of this function is going to be used also by the MLD version of it, put in a separate function the parts that are common for both MLD and non-MLD modes. The common function will later be used in the MLD ops. Signed-off-by:
-
Miri Korenblit authored
This is another patch in the series adding all the ops for the new MLD ieee80211_ops. The callback added here uses the new MLD FW API. Signed-off-by:
-
Miri Korenblit authored
Since parts of the functionality of this function is going to be used also by the MLD version of it, put in a separate function the parts that are common for both MLD and non-MLD modes. The common function will later be used in the MLD ops. Signed-off-by:
-
