]> www.infradead.org Git - users/jedix/linux-maple.git/commit
wifi: ath11k: mark ath11k_dp_rx_mon_mpdu_pop() as noinline
authorJeff Johnson <quic_jjohnson@quicinc.com>
Tue, 19 Nov 2024 15:47:39 +0000 (07:47 -0800)
committerJeff Johnson <quic_jjohnson@quicinc.com>
Thu, 21 Nov 2024 15:59:48 +0000 (07:59 -0800)
commit500d7ec88652ba7316e7fba334754e39e3177e4a
treef362da43ab39bd1a4611e7f65f4bc695effb48ae
parent8ea1d2072ad1a9c24b326b50ebdc2c810c4b2cce
wifi: ath11k: mark ath11k_dp_rx_mon_mpdu_pop() as noinline

When compiling the ath11k driver using clang with KASAN enabled, the
following warning is observed:

drivers/net/wireless/ath/ath11k/dp_rx.c:5244:5: warning: stack frame size (1304) exceeds limit (1024) in 'ath11k_dp_rx_process_mon_status' [-Wframe-larger-than]

This is similar to the issue found in ath12k/qmi.c that was discussed
in [1] and fixed with [2]. The issue is that clang inlining can
explode stack usage.

ath11k_dp_rx_process_mon_status() itself is a pretty
lightweight function, but it dispatches to several other functions
which do the real work:

ath11k_dp_rx_process_mon_status()
ath11k_dp_rx_reap_mon_status_ring()
ath11k_dp_rx_mon_dest_process()
ath11k_dp_rx_mon_mpdu_pop() *
ath11k_dp_rx_mon_deliver()
ath11k_dp_rx_mon_merg_msdus()
ath11k_dp_rx_deliver_msdu()
ath11k_dp_rx_update_peer_stats()

Of these, only ath11k_dp_rx_mon_mpdu_pop() has non-trivial stack
usage, so mark that function as 'noinline_for_stack' to prevent it
from being inlined in ath11k_dp_rx_process_mon_status(), thereby
eliminating the excessive stack usage.

Compile tested only.

Link: https://msgid.link/bc214795-1c51-4cb7-922f-67d6ef98bff2@quicinc.com
Link: https://patch.msgid.link/20241028-ath12k_qmi_driver_event_work-v1-1-0d532eb593fa@quicinc.com
Acked-by: Kalle Valo <kvalo@kernel.org>
Link: https://patch.msgid.link/20241119-ath11k-noinline-v1-2-4ec0a8aa30b2@quicinc.com
Signed-off-by: Jeff Johnson <quic_jjohnson@quicinc.com>
drivers/net/wireless/ath/ath11k/dp_rx.c