- 30 Mar, 2023 40 commits
-
-
Gregory Greenman authored
This field is used only for host TLC, so it can reside inside the corresponding lq_sta struct. Also, TLC lq_sta is cleared in iwl_mvm_rs_rate_init() upon association, but max_agg_bufsize is set earlier in iwl_mvm_sta_init(). Thus, place this field in the persistent part of lq_sta to retain its value. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100040.d55361064e39.Ib79d30f27d94607d097f0192af2aacd455a17958@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
By using the internal station add the station is installed in firmware with zeroed MAC addresses, which is wrong. Use the full installation function instead, to fill all data. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100040.62d5371bb3c7.Ie25b62125a3a022f76a36bae5fed9796c18698aa@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Remove fw_active_links_num counter since we now have a bitmap of active links in vif. Also, update link activation status only when LINK_CONTEXT_MODIFY_ACTIVE bit set in changes parameter. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100040.1ecfb27b6b84.I3a5e0bc32b3728e4caae8a231bc3f04ea1d89cad@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Some mac parameters, such as HE support, can change at this stage. Update mac config before updating links. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100040.d7882a0d6e04.Ie38cd854a237c46cf85fd7143dc757326f30da6e@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Shaul Triebitz authored
For bcase/mcast tx frames use the link queue. Signed-off-by: Shaul Triebitz <shaul.triebitz@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100040.ccd7218e4be2.I40f608a0441190cc26137b039f7cb7b065fd4e0c@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
Here we always have a link ID, not MAC ID, so clean up the naming. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100040.3def62de34b5.I10c9cf5dbfd1fc1e9c9c7d6d4cefcf0c08f1f2da@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
When we remove the station, we call iwl_mvm_sta_del() which returns true if we cannot remove it from the firmware yet, which happens if this is the station ID for the AP station that's still used because the MAC is still associated. However, we still must free the link data as the station is only kept alive in the firmware, in mac80211 and driver the data structures are destroyed. To fix that, we need to make iwl_mvm_mld_free_sta_link() track whether or not the station is still alive in FW, as otherwise we might reuse the station ID in the meantime and iwl_mvm_mld_rm_sta_from_fw() would reject the later delete from the firmware. Add an argument to it for that. Then we can use the return value of iwl_mvm_sta_del() for that to fix the issue, and call iwl_mvm_mld_rm_sta_from_fw() only if we need to not keep the station in FW. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.1d81d4c71f35.I8fc60ac28ffc1147e9b1250e5e6237b3cb5516ac@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Benjamin Berg authored
Initialize max_rc_amsdu_len per-link both on state change and when a new link is added. Signed-off-by: Benjamin Berg <benjamin.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.5bf521fe58b8.I73fe585f0ff75d41b5afd32077e3d6e48c90db2a@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Benjamin Berg authored
We were still using the deflink in most cases, update the code to use the appropriate link. Signed-off-by: Benjamin Berg <benjamin.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.fa1025502fb4.Iaba0c64740fdcf04a521e2f213bd3f3e27862472@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
Use the new macros from mac80211 that do lockdep checking on the RCU dereferences, instead of hard-coding 1 as the argument to rcu_dereference_protected(). Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.112df5c8dec2.I1a1008f5566e509953d988f254d15c9e58630418@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
During link switching there might be a link that's marked active but has no chanctx assigned (so it's not active from our driver's POV), skip such links in power recalculation without any warning. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.c629090bd5d2.If7a680d5e349d454f2122f936c21522b9528a55f@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
If we change only idle, avoid sending the MAC context command since it doesn't depend on idle state. This also fixes an issue with the firmware because without this we send the command during link switching, as we just deactivated the first link, and we cannot send this command when there's no active link. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.f5218f8453ec.I1325ff14ec07a27dd7ea2c1c210a1721d969839f@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
When we remove the AP station, we iterate over the links and remove all the keys, however, the key iteration will return all keys for all links, so skip the ones that we don't need based on the link ID. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.e724878f502e.I66870d4629244b4b309be79e11cbbd384bdf93be@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
When iterating station links, skip the links that are not yet active by checking for mvmvif->link[] existence. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.bd9b4e64c478.Ie21422c3bf2589d22942c3c57d26e6330d2e3afc@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Avraham Stern authored
When looking for a GO on for setting the scan parameters, iterate over all the available links. Signed-off-by: Avraham Stern <avraham.stern@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.4eb25d5655d0.Ie69f7313e4337f78c262a835aea3f707273a4209@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
Add a fw_id_to_link_sta array in mvm to track the link STA for each firmware station ID, and then use that to report the link a frame was received on (since we know the station ID from firmware). Notably, this fixes beacon tracking for the correct link since mac80211 now queues and processes those on the one link identified by the link ID only. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.c7dd3ec18077.I12ef9eb4a5b8b5c2b9d6bcaa1fda73b59eba39d8@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Shaul Triebitz authored
For an MLD AP, use the correct link mcast or bcast station id. Signed-off-by: Shaul Triebitz <shaul.triebitz@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.0cffa6c45242.I342e17e7bca87b7f05939eb2ebd36fc2aff0b49f@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Shaul Triebitz authored
For management frames sent by an AP interface, translate the MLD addresses to LINK addresses (for an MLD AP). AP (non-bufferable) management frames are sent via the broadcast station so the translation cannot be done by the firmware. Signed-off-by: Shaul Triebitz <shaul.triebitz@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.3cb4292f51e8.Ia662c00ff271c70eda927c12ed49b045b1eb8489@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Shaul Triebitz authored
Add/removed from iwl driver and firmware station links. Update the station queues accordingly (which station links are connected to the station queues). Signed-off-by: Shaul Triebitz <shaul.triebitz@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.156d1aae5de1.I32973141be1190222169879f8caf7038c1a8f769@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Shaul Triebitz authored
Replace the deflink.addr with the proper link address for setting the peer_link_address in the station command. For a non-MLD station, it will be the deflink. Signed-off-by: Shaul Triebitz <shaul.triebitz@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.03ab287da0ae.I88fb5ab4e3ea9c886a3fac7ce09c4791469c3c8e@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Rate scale initialization needs some parameters stored separately for each link. Pass link_conf and link_sta pointers and adjust the relevant code accordingly. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.e2f458198844.I98bf3ea807dd8ae8d703915ce9c01e7b7d5ccb42@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Make the decision about radar detection status depending on each active MLO link. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.a61b1d3f681a.I27bb8726d8806ad725c22bc7256856b1b6636745@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
It has to be done per link. We still don't support keys configuration for several links, but the single vif link can still point to a link different from deflink. For now handle the removal of keys for the default link. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.19d729cc4654.I4ebe8e3eb5fc00a994761f7c0ad40107382705ca@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
We're hitting these while starting to enable MLO in the driver, but getting them each and every time isn't very useful one way or the other. Make these warnings trigger only once. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.f333741d3dcf.If063d4cfe8a583f0f980a1b0ae4e63e17ba4ddc9@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
Iterate all the links here and check which ones are active, then adjust them accordingly. This is still wrong as far as the RSSI event is concerned (calling iwl_mvm_bt_coex_enable_rssi_event) since we call that now multiple times, which could overwrite the data; we need to either make that per link or call it only once, but need to sort out first what the firmware will be doing for beacon filtering/beacon abort in MLO. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.fe813fd8de5d.I216236717876510e51e4c21c7fac7691925443f1@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
This is more of a workaround, with MLO we can get here with the deflink not assigned. It's not critical right now that we have this right, so WARN_ON_ONCE() and don't crash until we can fix this area of the code. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.2b0d31bdb60b.I39d23c76eec16ac49f6ae3a6d5f7652041bde855@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
For now we only support a single link, so just use the deflink's address for the link address, instead of the STA's (MLD) address, but use the link address anyway in order to facilitate MLO connections. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.c853c8ced3ba.I2e1915d1090c526e6a4c718440b45a7192bbbb03@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
For now, if using MLO, skip the MEI update. It probably needs new API for multiple channels. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.aeaa746b14e7.I42b084f3cdc20bb52cf867700a3c2e1f5e7b4ecb@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
This needs to apply to the link, not deflink. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.0347fa77d522.Ibdf7a40cbe86e64ce086f243330fc27f43aeb3aa@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
With newer MLD firmware API we no longer need beacon timing information from the host, so disable TIMING_BEACON_ONLY. If MLO is enabled (currently only for testing), then we must not have DEAUTH_NEED_MGD_TX_PREP because mac80211 doesn't support it yet, we'll have to fix that later. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230329100039.28ec631487bf.I6eb27c3d4c0289a5ec3682f573aae3424f45619f@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
Implement the link change ops for links and stations. Note that the stations one is empty for now as we only have support for a single link so far, and then the stations are created with the first link as deflink by mac80211, so right now we don't really need anything. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.6186c5a37e99.Ifd00d3ee93356ddef273aa18f1e081cd8f2c84ae@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Clean up data for all allocated links of a vif. Invalidate all link sta pointers. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.948a7bee8b09.Ic6b202959c158258d391ccce77f66e1281879690@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Move code handling specific state transitions into separate handlers and adjust them for MLO. Adjust relevant callbacks as well. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.5a3f8a849723.I0670d20436858a1cd3c055e03c7528db81292811@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
vif->bss_conf is used in this function only when TLC is not offloaded, so not in MLO flow. Simplify the related "if" condition and call iwl_mvm_hwrate_to_tx_status() only for driver rate scale. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.c6826d5b5477.Ib56ec6025c0da3a381aaf88e71085ce9b96a9e65@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Pass link_conf when it's necessary and change the code accordingly. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.dc1074239c5b.I30db722af9009f1b274fa929f276fd2b78520dea@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Driver uses link_id as an index in the array. FW currently can support only 2 concurrently active links per vif with the ids in the range 0-3. Add a mapping of dirver link ids to fw link id and track the number of active link ids. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.a53e5df49c33.I02b25648d2d5ca370c0697bf19d0d34724eae8a1@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Internal stations are mcast/bcast and sniffer. They need some link specific data like bss id/phy id. Add link_conf as a parameter. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.25b013fd30de.I7f9f5906123acae05040cceec470c9acb07b9d42@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
These are two new handlers for MLO. As the configurations done in bss_info_changed() are now split into two separate flows, use MLO specific implementation instead of common functions with the non-MLO code. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.7b238cae0895.Ieb87f204787fb1c7cb7562e1cbf54ef518d87123@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Gregory Greenman authored
Add link_conf parameter and change all the relevant calls accordingly. Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.374015eed2e3.Icbf15a18e2599b53f4fa1c92fe3db64b551b84b1@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-
Johannes Berg authored
In newer firmware APIs the firmware is responsible for tracking the DTIM period and other beacon timing, so we don't need to wait with setting associated. In real MLO operation, mac80211 isn't tracking this anyway, and connections wouldn't work. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Gregory Greenman <gregory.greenman@intel.com> Link: https://lore.kernel.org/r/20230328104949.02354241fede.Id957bed3851fdf1fe902d79a1b0338c6d80bc0e1@changeidSigned-off-by: Johannes Berg <johannes.berg@intel.com>
-