Age | Commit message (Collapse) | Author |
|
Removed VFS dependency.
BAM is unrequired feature for bcm43752, so remove BAM feature.
Also, place kernel_read under DHD_SUPPORT_VFS_CALL to avoid VFS dependency
Bug: 189131064
Test: Verified compile and confirmed kernel_read is gone away from the mod file
Signed-off-by: daniel oh <daniel.oh@broadcom.corp-partner.google.com>
Change-Id: I92ccfde997c1aead7a68cf3eae988e2a58c34d63
|
|
dhd_linux.c:21361:2: error: unterminated function-like macro invocation
DHD_ERROR(("%s: logdump_prsrv_tailsize = %uKB \n",
Signed-off-by: Pat Tjin <pattjin@google.com>
Change-Id: I01c240313b3a2419cbc4ec3eca0f92ce0f95ead9
|
|
Removed VFS dependency because of GKI requirements
Bug: 182959271
Test: verified on slider + 43752 and it is able to connect AP.
Change-Id: I6540024056027fcf89dd83bc2a1782d09ca2d861
Signed-off-by: daniel oh <daniel.oh@broadcom.corp-partner.google.com>
Signed-off-by: Roger Wang <wangroger@google.com>
|
|
Bug: 181819234
Signed-off-by: Petri Gynther <pgynther@google.com>
Change-Id: Id263e1da07b44834c386f9b7e07126a549d4f109
|
|
Bug: 153787390
Test: ls -al /sys/wifi/power_stats
Signed-off-by: Benjamin Schwartz <bsschwar@google.com>
Change-Id: I681e0d45206887d8aa2878116532e334e2e66de1
|
|
This commit has some changes to upgrade dhd to Kernel 5.9
This includes:
1. Replace argument of proc_create_data from file_operations
to proc_ops
2. Replace the wait_queue_head for pno state into a swait_queue_head
3. Remove the duplicate definition of WLAN_AKM_SUITE_FT_8021X_SHA384
4. Replace ioremap_nocache with ioremap (it is no-cache by default)
5. Removed the use of smp_read_barrier_depends()
6. Remove mgmt_frame_register from cfg80211_ops
7. Replace getnstimeofday() with ktime_get_real_ts64()
Bug: 169622740
Test: Build successful in mainline kernel branch
Change-Id: If5f427e2420f4cc28448462343322e36b27ed265
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
This commit replaces the use of the struct timeval and
timespec with the timespec64.
This is part of the effort to switch to kernel 5.9
Bug: 169622740
Test: Build successful in mainline kernel branch
Change-Id: I317b270b3f6e1ca75769030069ee091625772713
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
In this commit we defined BCMPCIE_DISABLE_ASYNC_SUSPEND
which helps to invoked device_disable_async_suspend.
Bug: 168628959
Test: device can boot and wifi can connect
Change-Id: Ib052e9003741967bcd5d8d2018700da78a39a8ab
Signed-off-by: Roger Wang <wangroger@google.com>
|
|
[Issue & Analysis]
kernel panic with msg below
WDT expired - Debugging Information for Hardlockup core(0) - locked CPUs mask (0x7)
This issue occurred due to the use of spin_lock_bh under spin_lock_irqsave. (nested lock)
- dbg_ring_poll_worker() use spin_lock_irqsave()
- osl_malloc() use spin_lock_bh()
When the tasklet (dhd_dpc) is pending, spin_unlock_bh is called in the woker thread(dbg_ring_poll_worker).
and then spin_unlock_bh invokes pending tasklet. it causes unwanted deadlock case.
It should fix to avoid nested spinlock with irqsave and _bh. because the task could be pending due to interrupt.
[Solution]
1. Split the critical section that poll from ring buffer to HAL to avoid grap spin_lock_bh under spin_lock_irqsave
2. Eliminate ASSERT in dhd_dbg_ring_pull_single() if buf_len < rlen
- this is acceptable case and does not need to be ASSERT())
3. Add check to prevent the case of accessing ring buffer from HardIRQ context
4. Change logging macro from DHD_ERROR to DHD_DBGIF in critical section
BUG: 165487892
Change-Id: Ibd7dd796e024573d3921d2f1d7dde1e92dbdfec9
|
|
issues)
1. Didn't clear SCANNING bit after scan timeout.
2. Didn't clear DHD_BUS_BUSY_IN_NAPI bit.
fixed)
1. Added clear SCANNING bit in wl_scan_timeout.
2. Added clear DHD_BUS_BUSY_IN_NAPI in interface down routin.
Becuase, somtimes didn't schedule npi_poll deu to disable npi so cleariing bit after disable npai.
Bug: 166182383
Test: WiFi basic function works fine.
Signed-off-by: Thomson Kang <thomson.kang@broadcom.corp-partner.google.com>
Change-Id: I46a09c4fe6234695926a866179e4ad6d73c60f53
|
|
This is tempotary solution for loading bcmdhd43752
in first initail stage.
Bug: 163302778
Test: device can boot successfully.
Change-Id: I5d3acd58ed0c213e0458c599e732fad3daddd7c7
Signed-off-by: Roger Wang <wangroger@google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
This commit makes the include of linuxwakelock.h dependent
on CONFIG_HAS_WAKELOCK.
This change is required to be able to build with Kernel 5.4
Bug: 163302778
Test: dhd builds successfully with kernel 5.4
Change-Id: I8df76bcc731db4b4250cf3e904a62571a3e44a5a
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
In this commit we add changes to solve build break
and build warning in kernel 5.4 when building bcmdhd43752.
- Check GS101 definition first
- Fix function name to avoid duplicate naming
- Fix wiphy register failed
- Fix chi_check failure on start_xmit
- Fix chi_check failure on dhd_mem_dump
- Adjust order of sscd registr
- Remove DHD_NOTIFY_MAC_CHANGED
Bug: 163302778
Test: Build bcmdhd43752 with kernel 5.4 successfully.
Test: Device can boot into home even no WiFi cards.
Test: WiFi is workable when 43752/4389 card is plugged
Change-Id: Idf471ab9f2214c785381acdc3cebc58a7e36a27c
Signed-off-by: Roger Wang <wangroger@google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
issues)
The TX_PKT_FATE_FW_PKT_FREE type was added for a different project.
When packet is freed by the dongle after its real status is sent to host.
This status is not applicable to Android project.
fix)
This commit ignores this packet fate type.
Then, if WLFC_CTL_PKTFLAG_MKTFREE of txstatus happen,
then the sent fate type is TX_PKT_FATE_FW_DROP_OTHER.
Bug: 162393066
Test: Manual
Time usec Walltime Direction Fate Protocol Type Result
191127565 06:43:31.857 TX firmware dropped (other) ICMPv6 MLDv2 report N/A
Signed-off-by: Thomson Kang <thomson.kang@broadcom.corp-partner.google.com>
Change-Id: I59bab864307068b055b18060d0f0af8353447e7b
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
issues)
1. event_data is allocated to 38 bytes, payload_len is 50.
So, copy to the debug ring buffer exceed the event_data size in dhd_dbg_ring_push.
2. event_data->tlvs is allocated to 12 bytes, but copy to tlvs greater than 24 (SSID up to 32, rssi(2) and channel(2)).
fixed)
1. Chnaged the allocated event_data size from 38 to 50.
2. Changed event_data->tlvs type from pointer to zero value array for making continuous array.
Note :
Added define LINUX for log_conn_event_t because of compile error happened for Windows driver of all compile verification.
Bug: 161418541
Test: Manual
Signed-off-by: Thomson Kang <thomson.kang@broadcom.corp-partner.google.com>
Change-Id: I8270584adb922357f77d57e37fcd7a5faf344f97
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
This commit changes the devices supported by PCI from PCI_ANY_ID to
only the device id for BCM43752 chip
Bug: 162985533
Test: Make sure dhd only loads when BCM43752 chip is connected
Change-Id: I7c4169f7318438c384df3b6ab96b5ac2795f4346
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
This commit changes the name of the resulting module into bcmdhd43752.ko
Bug: 162985533
Test: Build and verify resulting module name
Change-Id: I9cce0a7837c06a3e1fc85442763eba9b0e4caffe
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Initialize the size of tlv_data->value to prevent from illegal memory access.
There are 3 cases for tlv_data: ssid, channel, rssi.
And ssid is the max size.
BUG: 160760185
Test: Hikey960 + Android 11 + BCM43752PCIe
Signed-off-by: Winnie Chen <winnie.chen@broadcom.corp-partner.google.com>
Change-Id: Ib200c51c2db7d887ccdf53bfc252d190c304285f
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
In this commit, we add sysfs for knowing current
running chip. dumpstate will base on this
information to dump correct data.
Bug: 161334127
Test: check running chip infomation is correct.
$ cat /sys/module/bcmdhd/parameters/chip_info
43752
$ cat /sys/module/bcmdhd/parameters/chip_info
4389
Change-Id: I0c5f2656bada018be5e0141050b7dc29aa139247
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Before invoke DHD_ERROR, we need to unlock
spin lock first. otherwise it will cause
recursive locking.
Bug: 160750236
Test: Wifi can connect and browse.
Running Wifi ON/OFF testing
Change-Id: Ie1f75cd261695abf70583e1d1f344af9db34aac5
Signed-off-by: Roger Wang <wangroger@google.com>
|
|
Enable 11d only when framework sets CC = 00
If CC = any valid country, 11d will disable
Bug: 153770652
Test: halutil -country XX to check country and 11d enabled/disabled
Signed-off-by: sanfran <sanfran.shen@broadcom.corp-partner.google.com>
Change-Id: Id17f583a8b778e4a48450dfb095a1fec2629e9aa
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Use freq list from hostapd to do channel select.
Finally send vendor event with selected channel info to hostapd.
Bug: 149885852
Test: ACS works on Hikey
Change-Id: I284314ccb1e031f55eaad47bb48ccef715b02b6e
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
1. Make NAPI weight as 64 with RX traffic flow control
2. Schedule delayed DPC to fix iovar resume on timeout
3. Clean up unused completion tasklet
4. Export sysfs entries for load balancing tuning
Bug: 158468691
Test:
UDP throughput at 6G/160 OTA with Skylake x86_64 FC25
Orig - TX: 1936M, RX: 1846M
RXFC - TX: 1948M, RX: 1757M
Signed-off-by: Stephen Chu <stephen.chu@broadcom.corp-partner.google.com>
Change-Id: I5fb9616f862cfde80e2c8482055195adcf1c6bdc
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Malloc buf for sdtc request instead of using bcm_iov_buf
Bug: 160739430
Test: Local build, load and connect ok
Signed-off-by: Stephen Chu <stephen.chu@broadcom.corp-partner.google.com>
Change-Id: Ife5ff2bd1f43b8292b3b97d44da9ca4fbed6f763
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Enable the support of WPA3 softap mode with hostapd. Not dongle
offload, we use hostapd as external authenticator.
Since the Android 11 PDK has no WPA3 softap support so far, we test
this function in command line with wpa_cli.
Bug:149448589
Test:
Firmware: fw.bcm43752.18.35.387.wpa3.hostapd.bin(Test firmware only)
Hikey960 + Android 11 PDK
Change-Id: I1b14071cf3b36953e94e64cdd908e188cc5d1d5a
Signed-off-by: Anthony Huang <anthony.huang@broadcom.corp-partner.google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Bug: 151283039
Test: wl bus:disocnnect 102, send hang event with memdump_reason string
[cfg80211][wlan] wl_cfg80211_hang: In : HANG , reason=0x8002
[cfg80211][wlan] wl_cfgvendor_send_hang_event: wl_cfgvendor_send_fw_dump_event 0x8002
[cfg80211][wlan] wl_cfgvendor_send_hang_event: hang reason: Dongle_Trap_0003fd14_0019194f
Signed-off-by: Roger Wang <wangroger@google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
Change-Id: I8d345881d869ecdcc975faf6bc200bfc161324d6
|
|
Support ANQP random SA from wpa_supplicant
Bug: 155491082
Test: Check if anqp sa address is from wpa_supplicant
Signed-off-by: sanfran <sanfran.shen@broadcom.corp-partner.google.com>
Change-Id: Ibc550d330f262a945a68ab042161a6a0b073d9f0
Signed-off-by: sanfran <sanfran.shen@broadcom.corp-partner.google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Bug: 159298287
Test: Hikey960
[dhd][wlan] MACEVENT: WLC_E_AUTH, MAC 00:11:32:ab:26:3d, SAE, FAILURE, status 1 reason 1
[dhd][wlan] MACEVENT: WLC_E_SET_SSID, failed status 1
[dhd][wlan] wl_process_host_event: Link event 11, flags 0, status 0, role 0, del_sta 0
wpa_supplicant: wlan0: CTRL-EVENT-ASSOC-REJECT bssid=00:11:32:ab:26:3d status_code=1
Change-Id: If506178d201d763d558fe2004def8219b3cf3c12
Signed-off-by: Roger Wang <wangroger@google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
For support wpa_s->enable_oce, dhd needs to advertise other ext_features
Bug: 154372567
Test: Check if wpa_s->enable_oce is true in
getWpaDriverCapabilitiesInternal
Signed-off-by: sanfran <sanfran.shen@broadcom.corp-partner.google.com>
Change-Id: I11d20369eaf468ef8a5d0d5cb1588f1ade3ce46b
Signed-off-by: sanfran <sanfran.shen@broadcom.corp-partner.google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Include only data frames with non-fixed rate in the counts.
Bug: 151425271
Test: Hikey960 + Android 11 + BCM43752PCIe
Fw needs patch and will be included in next release.
Signed-off-by: Winnie Chen <winnie.chen@broadcom.corp-partner.google.com>
Change-Id: I908b7708055ab479086656309ebc0f25705a68f2
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
[Issue] When fw trap, driver status don't clean, keeps
in SCANNING state. Pcie dev wake think it take
too much time before enter DeepSleep. And driver
trigger ASSERT KP for debug purpose.
[Fix] When fw trap, driver don't enter DeepSleep,
and don't need to does DeepSleep threshold time check,
add dhd_query_bus_erros check before enter threshold check.
Bug: 159276663
Test: use "wl bus:disconnect 102" cmd verify on hikey960.
Change-Id: Ia3ebffc0df02c3f72bf0ecfcdea017668ae19909
Signed-off-by: Roger Wang <wangroger@google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
In this comment, we support loading different map file
for 43752 and 4389.
Bug: 158628077
Test: Trigger trap and get dump reason on 4389 and 43752 chips
Change-Id: I71017030017630318b28e62339e2111492460e05
Signed-off-by: Roger Wang <wangroger@google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Look up function symbol in map file to get function name and offset of PC/LR.
Bug: 158628077
Test: Trigger trap and get dump reason
[724119.696872] dhd_mem_dump: dump reason: Dongle_Trap_traptest+0x4_pcidev_handle_user_disconnect+0x5b
Signed-off-by: Stephen Chu <stephen.chu@broadcom.corp-partner.google.com>
Change-Id: Idc5fa88f045cb24319ecf0f019e811b470637eb8
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Add CUSTOM_SCAN_UNASSOC_ACTIVE_TIME in Kbuild to set scan dwell time.
Bug: 157503497
Test: After wlan0 up
$ wl scan_unassoc_time
40
Signed-off-by: Stephen Chu <stephen.chu@broadcom.corp-partner.google.com>
Change-Id: Iea76b2419ffc1916a1827c6b8a24b3f698461e8e
|
|
Provide onTimeInMs/ccaBusyTimeInMs/channel information needed in
HAL V1.3 wifi_get_link_stats
Bug: 151425271
Test: Fw update and will be included in next release (May 29th)
Provide test fw first.
Change-Id: Ia82b3f133efcb76f8a951bd7df014a74b38ec6c6
Signed-off-by: Winnie Chen <winnie.chen@broadcom.corp-partner.google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Observed 14 bytes offset in tcpdump any device capturing. The capturing
is Linux cooked mode (LINUX_SLL linke type) using SOCK_DGRAM which has no
link-layer header. Push ether header length in monitor skb to keep
complete radiotap content.
Bug: 158203741
Test: tcpdump -i radiotap0 and tcpdump -i any can get correct capture.
Signed-off-by: Stephen Chu <stephen.chu@broadcom.corp-partner.google.com>
Change-Id: I6af37ef0898bfd4645f67ca967ed381fe403757b
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Create one ring buffer 'roam_stats' for store roam statistics log
Bug: 149451043
Test: hikey960
hikey960:/ # cat /data/ring-roam_stats-0.bin
[00:02:29.039384][ 143.194208] roam reason: LOW_RSSI
[00:02:29.039397][ 143.194220] roam status: 0
[00:02:29.039420][ 143.194243] roam scan channels: count(2) 6 8
[00:02:29.039432][ 143.194255] Candidate AP number: 3
[00:02:29.039452][ 143.194274] ROAM_AP_CANDIDATE 0: 70:4d:7b:0c:20:50 rssi:-42 score:0 channel:8
[00:02:29.039470][ 143.194293] ROAM_AP_CANDIDATE 1: 78:24:af:68:a1:a1 rssi:-22 score:65513 channel:6 [connected]
[00:02:29.039489][ 143.194311] ROAM_AP_CANDIDATE 2: 02:09:0b:71:3f:d5 rssi:-27 score:65508 channel:6
[00:05:50.845605][ 345.000429] roam reason: RECEIVED DEAUTHENTICATION
[00:05:50.845629][ 345.000452] roam status: 0
[00:05:50.845653][ 345.000476] roam scan channels: count(2) 6 8
[00:05:50.845664][ 345.000487] Candidate AP number: 3
[00:05:50.845684][ 345.000506] ROAM_AP_CANDIDATE 0: 78:24:af:68:a1:a1 rssi:-18 score:65517 channel:6 [connected]
[00:05:50.845703][ 345.000525] ROAM_AP_CANDIDATE 1: 02:09:0b:71:3f:d5 rssi:-31 score:65504 channel:6
[00:05:50.845721][ 345.000543] ROAM_AP_CANDIDATE 2: 70:4d:7b:0c:20:50 rssi:-41 score:65494 channel:8
[00:05:54.090990][ 348.245814] roam reason: BEACONS LOST
[00:05:54.091014][ 348.245837] roam status: 0
[00:05:54.091037][ 348.245860] roam scan channels: count(2) 6 8
[00:05:54.091048][ 348.245871] Candidate AP number: 2
[00:05:54.091068][ 348.245891] ROAM_AP_CANDIDATE 0: 02:09:0b:71:3f:d5 rssi:-30 score:65505 channel:6
[00:05:54.091116][ 348.245939] ROAM_AP_CANDIDATE 1: 70:4d:7b:0c:20:50 rssi:-40 score:65495 channel:8 [connected]
Change-Id: Ifb0e00a0af8d51f9763164739f68983217d6569e
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Because we have enabled the in-band device wake in default, it will
have many debug logs from dhd_bus_handle_mb_data() if the device
requested to deep-sleep but host is still active.
[11579.895858] [10:46:11.768358][dhd][wlan] dhd_bus_handle_mb_data:
Failed to send DS-ACK, host_active_cnt is 1
[11580.565439] [10:46:12.437923][dhd][wlan] dhd_bus_handle_mb_data:
Failed to send DS-ACK, host_active_cnt is 1
[11581.541735] [10:46:13.414224][dhd][wlan] dhd_bus_handle_mb_data:
Failed to send DS-ACK, host_active_cnt is 1
Since it is just a normal case but not an error. Lower this message
level from DHD_ERROR to DHD_TRACE.
Bug: 158535681
Change-Id: I866b70152d6f6d943416e81e7ae533e14c24cc1e
Signed-off-by: Anthony Huang <anthony.huang@broadcom.corp-partner.google.com>
|
|
Add mac_cap_info and phy_cap_info to ieee80211_sband_iftype_data
Bug: 153995393
Test: Check if is_80211ax_supported is true
Change-Id: I6b618b56584fbb6773e8a531a71dd5667ac105c8
Signed-off-by: sanfran <sanfran.shen@broadcom.corp-partner.google.com>
|
|
1. When softap mode, before set mac address, driver
execute dev_close, then driver delete the virtual
interface (wlan1). Then interface become NULL, cause
driver crash. So before delete the virtual interface,
driver needs to check dhd_allow_stop, to determine
delete interface or not.
2. Add error handler when set mac address fail.
Bug: 158318360
Test: Roger help to verify on Slider
Change-Id: I7ccc223ee7172b9545edd7f7d05f38b9fd474e8a
Signed-off-by: Roger Wang <wangroger@google.com>
|
|
There is no additional pcie info log in ring buffer,
only keep the pcie error log.
Bug: 158535681
Test: verify on hikey960.
Change-Id: I1e8f91527765c9ce2e319190736dfe2ccb3d0970
Signed-off-by: Roger Wang <wangroger@google.com>
|
|
Extend the idletime of PCIe runtime PM from 500ms to 1100ms.
It is because the runtime PM will suspend/resume too often after
connected to AP.
Bug: 158286446
Test: Android 11 + Hikey960 + BCM43752 PCIe
Change-Id: I9793acc95d707866830a300a028a0319ed48ac5f
Signed-off-by: Anthony Huang <anthony.huang@broadcom.corp-partner.google.com>
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Support SAR scenario for HEAD, GRIP and BT
Support 5G sub-band
Support regulatory
BUG: 153199259
Change-Id: I1e6d25aa93cbe843dc5680d150b1adae36cc777c
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Provide Wifi chip AWAKE/ASLEEP stats table along with PCIe link stats table.
Bug: 153787390
Test: Fw patch is formal release v18.35.387.23
Change-Id: I1439645be07426c026b63549b6f24c2a2691d244
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
|
|
- Update the IOVARs changed for saving firmware memory
- Add the retry if the "conn_owner_nlportid" of cfg80211 is not ready
Change-Id: I41b2b3eb0b09b9d752d7e99151a35be113e802b7
Bug: 149448388
Test:
firmware: fw.bcm43752.18.35.387.23.bin
Platform: Hikey960 + BCM43752 PCIe
OS: Android 11 + wpa_supplicant 2.8
WPA3-Personal: Verified with Synology MR2200AC AP.
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
For dbg ring level :
- dhd dbgring level : Add DHD_EVENT_VAL/DHD_PKT_MON_VAL/DHD_IOVAR_MEM_VAL/DHD_INFO_VAL as default level
- wl dbgring level : Add WL_DBG_P2P_ACTION/WL_DBG_INFO as default level
For driver log level
- dhd log level : Remove DHD_FWLOG_VAL/DHD_IOVAR_MEM_VAL/DHD_INFO_VAL
- wl log level : No change
For fw event log tag
- remove BEACON_LOG/PCI_WARN/PCI_INFO/PCI_DEBUG fw event log
- add more for debug fw issue
Bug: 151785383
Test: hikey960
Change-Id: If36147c2947dd34171080f662000438209de5922
Signed-off-by: Roger Wang <wangroger@google.com>
|
|
1. Because this problem cannot be reproduced. We enable the
exynos_pcie_l1_exit(0) in R_REG and W_REG.
2. "exynos_pcie_l1_exit" can make sure the PCIe link is in L0 before
we accessing it. Then keep monitoring this issue will happened again
or not.
Change-Id: I6b399bc7c72efeec2cedf7ec3355cf5758b1fe15
Bug: 157711644
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
1. Move the dhdpcie_cto_init after the si_attach(). In case the chip
may be not ready yet.
2. This issue cannot be reproduced. We need to keep monitoring the
SError happened again or not.
Bug: 157711644
Test:
firmware: fw.bcm43752.18.35.387.23.bin
Platform: Hikey960 + BCM43752 PCIe
Change-Id: Ie65bb841a056863879b5812bb21c0911c69ad4a5
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|
|
Enable the new feature "In-band device wake" to avoid the latency caused
by deep sleep of wifi chip. We also need to enable this in wifi
firmware.
Also include the debug check for latency of PCIe read.
Bug: 155212501
Test: Android 10 + Hikey960 + BCM43752 PCIe
Change-Id: Ia7a63059ced37e40ae2fc610170812ad1ae58589
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
|