The rtl8723bs driver has its own set of macros and inline functions in
wifi.h for working with 802.11 frame headers. They access the header
through hardcoded byte offsets and pointer casts, and they duplicate
struct ieee80211_hdr and the helpers in <linux/ieee80211.h> that
mac80211, cfg80211 and other wireless drivers already use.

This series switches the driver over to the generic versions and
deletes the private ones from wifi.h. Each patch converts one group of
macros and removes their definitions once nothing uses them anymore,
so the driver builds after every patch.

  Patches 1 to 3 convert header address access and the Frame Control
  flag checks and updates.

  Patches 4 and 5 convert the frame type and subtype checks. Single
  comparisons use the ieee80211_is_*() helpers, and switch statements
  mask frame_control with IEEE80211_FCTL_* and use IEEE80211_STYPE_*
  case labels, the same way mac80211 does.

  Patch 6 removes an unused struct member. Its initialisers were the
  last users of a few private subtype constants that have no generic
  equivalent.

  Patch 7 converts the frame subtype setter and removes the private
  WIFI_* frame type enum.

  Patches 8 and 9 convert the Sequence Control, Duration/ID and QoS
  Control fields.

  Patch 10 replaces the driver's destination and source address
  helpers with ieee80211_get_DA() and ieee80211_get_SA().

None of this should change behaviour. In the few places where a
generic helper checks more than the code it replaces, or where the old
macro kept bits that a plain assignment would overwrite, the commit
message explains why the result is the same.

Compile tested with W=1 after each patch, and checked with checkpatch
and smatch. I don't have a device with this chip, so it has not been
tested on hardware.

Rishik Yechuri (10):
  staging: rtl8723bs: use struct ieee80211_hdr for header addresses
  staging: rtl8723bs: use ieee80211_has_*() to test frame control flags
  staging: rtl8723bs: set frame control flags with IEEE80211_FCTL_*
    masks
  staging: rtl8723bs: use ieee80211_is_*() for frame type tests
  staging: rtl8723bs: use IEEE80211_FCTL_* masks for frame type dispatch
  staging: rtl8723bs: remove unused num member from struct mlme_handler
  staging: rtl8723bs: use IEEE80211_FTYPE_* and IEEE80211_STYPE_* values
  staging: rtl8723bs: use struct ieee80211_hdr for seq_ctrl and
    duration_id
  staging: rtl8723bs: use IEEE80211_QOS_CTL_* masks for QoS control
  staging: rtl8723bs: use ieee80211_get_DA() and ieee80211_get_SA()

 drivers/staging/rtl8723bs/core/rtw_mlme_ext.c | 254 ++++++++---------
 drivers/staging/rtl8723bs/core/rtw_recv.c     | 134 ++++-----
 drivers/staging/rtl8723bs/core/rtw_security.c |  41 +--
 .../staging/rtl8723bs/core/rtw_wlan_util.c    |   7 +-
 drivers/staging/rtl8723bs/core/rtw_xmit.c     |  75 +++---
 drivers/staging/rtl8723bs/hal/rtl8723b_cmd.c  |  42 +--
 .../staging/rtl8723bs/hal/rtl8723bs_recv.c    |   9 +-
 .../staging/rtl8723bs/hal/rtl8723bs_xmit.c    |   3 +-
 .../staging/rtl8723bs/include/rtw_mlme_ext.h  |   1 -
 drivers/staging/rtl8723bs/include/wifi.h      | 255 +-----------------
 .../staging/rtl8723bs/os_dep/ioctl_cfg80211.c |  19 +-
 11 files changed, 319 insertions(+), 521 deletions(-)

-- 
2.43.0

On Tue, Oct 13, 2026 at 03:15:28PM -0500, Rishik Yechuri wrote:
> The rtl8723bs driver has its own set of macros and inline functions in
> wifi.h for working with 802.11 frame headers. They access the header
> through hardcoded byte offsets and pointer casts, and they duplicate
> struct ieee80211_hdr and the helpers in <linux/ieee80211.h> that
> mac80211, cfg80211 and other wireless drivers already use.
> 
> This series switches the driver over to the generic versions and
> deletes the private ones from wifi.h. Each patch converts one group of
> macros and removes their definitions once nothing uses them anymore,
> so the driver builds after every patch.
> 
>   Patches 1 to 3 convert header address access and the Frame Control
>   flag checks and updates.
> 
>   Patches 4 and 5 convert the frame type and subtype checks. Single
>   comparisons use the ieee80211_is_*() helpers, and switch statements
>   mask frame_control with IEEE80211_FCTL_* and use IEEE80211_STYPE_*
>   case labels, the same way mac80211 does.
> 
>   Patch 6 removes an unused struct member. Its initialisers were the
>   last users of a few private subtype constants that have no generic
>   equivalent.
> 
>   Patch 7 converts the frame subtype setter and removes the private
>   WIFI_* frame type enum.
> 
>   Patches 8 and 9 convert the Sequence Control, Duration/ID and QoS
>   Control fields.
> 
>   Patch 10 replaces the driver's destination and source address
>   helpers with ieee80211_get_DA() and ieee80211_get_SA().
> 
> None of this should change behaviour. In the few places where a
> generic helper checks more than the code it replaces, or where the old
> macro kept bits that a plain assignment would overwrite, the commit
> message explains why the result is the same.
> 
> Compile tested with W=1 after each patch, and checked with checkpatch
> and smatch. I don't have a device with this chip, so it has not been
> tested on hardware.

Did you forget the assisted-by tag for this series or did you really do
this "by hand"?

thanks,

greg k-h