- 22 Sep, 2020 21 commits
-
-
Kalle Valo authored
Some of the function declarations are for functions in debugfs_htt_stats.c and debugfs_sta.c, move them to corresponding header files. As debugfs_sta.h didn't exist create it. Also in debugfs_htt_stats.h move dunction declarations to the end of the file. No functional changes. Compile tested only. Signed-off-by:
Kalle Valo <kvalo@codeaurora.org> Link: https://lore.kernel.org/r/1600264523-12939-4-git-send-email-kvalo@codeaurora.org
-
Kalle Valo authored
For consistency with debugfs.c rename debug_htt_stats files and functions to debugfs_htt_stats. No functional changes. Compile tested only. Signed-off-by:
-
Kalle Valo authored
As these functions are now defined in debugfs.c change the prefix to use ath11k_debugfs_ as well. No functional changes. Compile tested only. Signed-off-by:
-
Kalle Valo authored
If CONFIG_ATH11K_DEBUGFS is disabled there are warnings debug.c: drivers/net/wireless/ath/ath11k/debug.c:36:20: warning: 'htt_bp_lmac_ring' defined but not used [-Wunused-variable] 36 | static const char *htt_bp_lmac_ring[HTT_SW_LMAC_RING_IDX_MAX] = { | ^~~~~~~~~~~~~~~~ drivers/net/wireless/ath/ath11k/debug.c:15:20: warning: 'htt_bp_umac_ring' defined but not used [-Wunused-variable] 15 | static const char *htt_bp_umac_ring[HTT_SW_UMAC_RING_IDX_MAX] = { | ^~~~~~~~~~~~~~~~ Fix this by refactoring debugfs code to debugfs.c. This also reduces the number of ifdefs in debug.c and makes it easier to maintain the code. No functional changes. Compile tested only. Reported-by:YueHaibing <yuehaibing@huawei.com> Signed-off-by:
-
Bo YU authored
Return value form wait_for_completion_timeout should to be checked. This is detected by Coverity: #CID:1464479 (CHECKED_RETURN) Fixes: d5c65159 ("ath11k: driver for Qualcomm IEEE 802.11ax devices") Signed-off-by:
Bo YU <tsu.yubo@gmail.com> Signed-off-by:
-
Bryan O'Donoghue authored
This is a small update to fix an error I saw where a few functions do not have a blank line in between them. Affects smd.c and main.c - no logic is affected by this change. Signed-off-by:
Bryan O'Donoghue <bryan.odonoghue@linaro.org> Signed-off-by:
-
Bryan O'Donoghue authored
This commit marks all smd.c functions that are only used inside of smd.c as static. Previous commits added some VHT specific setup functions non-static which is the right thing to do in terms of having granular git commits that compile warning free. What we really want is for local not global scope on those functions. This patch makes the conversion from global to local scope. Signed-off-by:
-
Bryan O'Donoghue authored
This patch adds ieee802.11 VHT flags for the wcn3680b. - RX_STBC1 - SU Beamformee - MU Beamformee - VHT80 SGI - Single spatial stream RX LDPC is declared as supported in the datasheet but not enabled at this time. After this patch is applied an AP should see the wcn3680 as an 802.11ac capable device. Signed-off-by:
-
Bryan O'Donoghue authored
This commit adds VHT rates to the wcn36xx_update_allowed_rates() routine. Thus allowing the driver to latch the declared rates and transmit them to the firmware in the same way as other 80211.n rates are. Signed-off-by:
-
Bryan O'Donoghue authored
In order to send VHT parameters to wcn3680 we need to pass the extended V1 parameter structures to the firmware. These commands need to have the version number set to 1. This patch makes the conversion. The conversion consists of 1. Setting the version number for wcn3680 or leaving it at 0 otherwise 2. Setting the size of the packet header lower for wcn3620 and wcn3660 Once done all three chips can continue to use the same code to pass parameters to their respective firmware. In the case of the wcn3680 the passed structures will be slightly larger to accommodate communication of VHT descriptors. Signed-off-by:
-
Bryan O'Donoghue authored
In order to pass 802.11ac VHT parameters from the SoC to wcn36xx we need to use the V1 data structures associated with BSS and STA parameters. The means of identifying a V1 data-structure is via the SMD version field. This patch defines a INIT_HAL_MSG_V1() which operates the same way as INIT_HAL_MSG() with the exception that it defines VERSION1 as opposed to VERSION0. Signed-off-by:
-
Bryan O'Donoghue authored
This commit makes use of wcn36xx_smd_set_bss_vht_params() to extract VHT parameters from the 80211_sta structure and latch appropriate bits in the bss_params_v1 structure for transmission to the wcnss firmware. Signed-off-by:
-
Bryan O'Donoghue authored
This commit modifies wcn36xx_smd_start() so that it can download wcn3680 specific firmware parameters if we are talking to the wcn3680. If not the original generic firmware parameter table should continue to be used for wcn3620 and wcn3660. Signed-off-by:
-
Bryan O'Donoghue authored
This commit defines a firmware configuration for the wcn3680 which represents a working downstream configuration. This configuration has been successfully applied to the upstream driver with antecedent patches resulting in the same or better through-put in comparison to the downstream driver on the test hardware. Signed-off-by:
-
Bryan O'Donoghue authored
In order to get 802.11ac working the way we want, additional parameters need to be passed down to the firmware. This patch takes the full remaining set of parameters defined in the downstream riva/inc/wlan_hal_cfg.h and imports them into hal.h with some minor name length adjustments. This addition will allow us to pass a larger firmware configuration set later on. Signed-off-by:
-
Bryan O'Donoghue authored
For the 80MHz channel we need to set the PHY mode to one of four PHY modes that span the 80MHz range. This patch latches the hw_value PHY field previously defined for 5GHz channels directly to the parameter passed to the firmware. Signed-off-by:
-
Bryan O'Donoghue authored
This commit encodes the 802.11ac PHY mode for a given channel in the upper bits of the hw_value field. This allows for a neat read-out and application of the relevant PHY setting. Signed-off-by:
-
Bryan O'Donoghue authored
Adds HW_VALUE_PHY(hw_value) an access macro that will be used to extract a hardware specific PHY setting for a given channel. Signed-off-by:
-
Bryan O'Donoghue authored
Uses HW_VALUE_CHANNEL() to extract the channel number from a struct ieee80211_channel->hw_value. Once done we can use the upper bits of the hw_value to encode PHY related data. Signed-off-by:
-
Bryan O'Donoghue authored
Adds HW_VALUE_CHANNEL(hw_value) an access macro that will be used to extract the channel number from struct ieee80211_channel->hw_value in preparation for also storing PHY settings for 802.11ac in the upper bits of hw_value. Signed-off-by:
-
Rakesh Pillai authored
Board Data File (BDF) is loaded upon driver boot-up procedure. The right board data file is identified using bus and qmi-board-id. The problem, however, can occur when the (default) board data file cannot fulfill with the vendor requirements and it is necessary to use a different board data file. Also using the chip_id for identifying the board data helps in dealing with different variants of the board data file based on the RF card. If the chip_id is not programmed, a default value of 0xff will be used for parsing the board data file. Add the support to get the variant field from DTSI and use this information along with the chip_id to load the vendor specific BDF. The device tree requires addition strings to define the variant name wifi@a000000 { status = "okay"; qcom,ath10k-calibration-variant = "xyz-v2"; }; wifi@a800000 { status = "okay"; qcom,ath10k-calibration-variant = "xyz-v1"; }; This would create the boarddata identifiers for the board-2.bin search * bus=snoc,qmi-board-id=16,qmi-chip-id=0,variant=xyz-v1 * bus=snoc,qmi-board-id=17,qmi-chip-id=0,variant=xyz-v2 Tested-on: WCN3990 hw1.0 SNOC WLAN.HL.3.1-01040-QCAHLSWMTPLZ-1 Signed-off-by:Rakesh Pillai <pillair@codeaurora.org> Signed-off-by:
-
- 21 Sep, 2020 18 commits
-
-
Bryan O'Donoghue authored
wcn36xx_smd_config_bss_v0() and wcn36xx_smd_config_bss_v1() have been designed to operate in standalone fashion. As a result we can drop the dead code now present in wcn36xx_smd_config_bss() and happily remove one kzalloc from the BSS config path as we do so. Signed-off-by:
Loic Poulain <loic.poulain@linaro.org> Signed-off-by:
-
Bryan O'Donoghue authored
A previous patch added wcn36xx_smd_config_bss_v0() this patch converts the version 0 data-path in wcn36xx_smd_config_bss() to use wcn36xx_smd_config_bss_v0(). Signed-off-by:
-
Bryan O'Donoghue authored
This commit adds wcn36xx_smd_config_bss_v0() as a step along the road of functionally decomposing wcn36xx_smd_config_bss(). Signed-off-by:
-
Bryan O'Donoghue authored
This patch updates wcn36xx_smd_config_bss_v1() to update on internally derived parameters only, specifically making use of STA v1 wrapper routines previously added. Once done we no longer need to pass a struct wcn36xx_hal_config_bss_req_msg which gives us options in later patches to eliminate the kzalloc() in wcn36xx_smd_config_bss entirely. Signed-off-by:
-
Bryan O'Donoghue authored
This commit moves BSS parameter setup to a separate function wcn36xx_smd_set_bss_params(). This will allow for further functional decomposition and fewer kzalloc() operations in subsequent patches. Signed-off-by:
-
Bryan O'Donoghue authored
In order to facilitate functional decomposition of wcn36xx_smd_config_bss() we need to move wcn36xx_smd_set_sta_params() later in function. Signed-off-by:
-
Bryan O'Donoghue authored
This commit functionally decomposes wcn36xx_smd_config_sta into a clearly defined wcn36xx_smd_config_sta_v0 and wcn36xx_smd_config_sta_v1 path. Signed-off-by:
-
Bryan O'Donoghue authored
This commit adds a wrapper function wcn36xx_smd_set_sta_params_v1() which calls into wcn36xx_smd_set_sta_params() and then subsequently sets version-1 specific parameters. Signed-off-by:
-
Bryan O'Donoghue authored
This commit adds wcn36xx_smd_set_bss_vht_params(). The job of this function is to decide if the BSS is VHT capable and if so set the appropriate bit in the BSS parameter structure for passing to the firmware. VHT Channel width set is not set since we don't support 160MHz. Signed-off-by:
-
Bryan O'Donoghue authored
Adds a routine to allow setting the LDPC bit for HT parameter passing inside the version 1 STA parameters data structure. Signed-off-by:
-
Bryan O'Donoghue authored
This commit adds support for setting VHT parameters based on the declared VHT capability bits in the VHT capability structure. We cannot do 160MHz so VHT Channel width set should be zero. Signed-off-by:
-
Bryan O'Donoghue authored
Toggling the LDPC enabled bit is possible only via the extended V1 data-structure. This function provides a means of setting the default depending on chip-type. Signed-off-by:
-
Bryan O'Donoghue authored
This commit adds support for setting default VHT parameters, which are exposed by the extended version 1 STA parameter type. Signed-off-by:
-
Bryan O'Donoghue authored
Add a routine to set some additional default parameters associated with the V1 data structure. Signed-off-by:
-
Bryan O'Donoghue authored
This patch converts the internal data structure used to store data-rates from version 0 to version 1. This allows us to extend out the internal storage to represent VHT parameters. Using the extended version 1 data-structure allows us to avoid a whole raft of version 1 specific fixup functions. Signed-off-by:
-
Bryan O'Donoghue authored
In order to pass VHT parameters to wcn3680 we need to use a super-set of the V1 data-structures with additional VHT parameters tacked on. This patch adds the additional fields to the STA and BSS parameter structures. Since neither wcn3620 nor wcn3660 support VHT the size of the passed message is fixed to the previous message length. Subsequent changes will differentiate between wcn3620/wcn3660 and wcn3680 which does use the larger message size. Signed-off-by:
-
Govind Singh authored
IPQ8074 HW uses rproc with AHB as underlying hif layer. Move rproc references from common core layer to target hif layer. Remove IS_ENABLED check for CONFIG_REMOTEPROC as it's not required anymore. No functional changes. Compile tested only. Signed-off-by:
Govind Singh <govinds@codeaurora.org> Signed-off-by:
-
Brooke Basile authored
Calls to usb_kill_anchored_urbs() after usb_kill_urb() on multiprocessor systems create a race condition in which usb_kill_anchored_urbs() deallocates the URB before the completer callback is called in usb_kill_urb(), resulting in a use-after-free. To fix this, add proper lock protection to usb_kill_urb() calls that can possibly run concurrently with usb_kill_anchored_urbs(). Reported-by: syzbot+89bd486af9427a9fc605@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?id=cabffad18eb74197f84871802fd2c5117b61febfSigned-off-by:
Brooke Basile <brookebasile@gmail.com> Signed-off-by:
-
- 12 Sep, 2020 1 commit
-
-
Barry Song authored
Rather than doing request_irq and then disabling the irq immediately, it should be safer to use IRQ_NOAUTOEN flag for the irq. It removes any gap between request_irq() and disable_irq(). Cc: Salil Mehta <salil.mehta@huawei.com> Reviewed-by:
Yunsheng Lin <linyunsheng@huawei.com> Signed-off-by:
Barry Song <song.bao.hua@hisilicon.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
