[PATCH] wifi: ath11k: clear shared SRNG pointer state on restart

[Jeff Johnson]

On 5/13/2026 9:52 PM, Kyle Farnung via B4 Relay wrote:
> From: Kyle Farnung <kfarnung at gmail.com>
> 
> LMAC rings reuse the shared rdp/wrp pointer buffers without going
> through the normal SRNG hw-init path that zeros non-LMAC ring
> pointers. After restart, ath11k_hal_srng_clear() can therefore hand
> stale hp/tp state from the previous firmware instance back to the new
> one.
> 
> Clear the shared pointer buffers while keeping the allocations in
> place so restart still avoids reallocating SRNG DMA memory, but starts
> with fresh ring-pointer state.
> 
> Fixes: 32be3ca4cf78b ("wifi: ath11k: HAL SRNG: don't deinitialize and re-initialize again")
> Cc: stable at vger.kernel.org
> Link: https://lore.kernel.org/all/CAOPSVF04q6uvVdq8GTRLHBrVMdpt9=o9wVcFMc6f-yhmSBcZqQ@mail.gmail.com/

I'm going to change this to a Closes: tag in my pending branch

> Signed-off-by: Kyle Farnung <kfarnung at gmail.com>
> ---
> This patch is the result of investigating suspend/resume failures on a
> Lenovo ThinkPad P14s Gen 5 AMD with ath11k.
> 
> I originally proposed extending the existing ath11k PM quirk for this
> platform, but after discussion in [1] and bisection the issue appears to
> be a regression introduced by [2]. There is also a parallel report in [3]
> that appears consistent with the same root cause. This patch keeps the
> intended no-reallocation behavior from that change, but clears the
> preserved shared SRNG pointer state so restart begins from a clean state.
> 
> Testing so far has been limited to local suspend/resume cycling on the
> affected system. The issue was originally reproduced on v7.0.4, and the
> patch was also built and tested on top of ath-current with repeated
> suspend/resume cycles on a Lenovo ThinkPad P14s Gen 5 AMD.
> 
> [1] https://lore.kernel.org/all/CAOPSVF04q6uvVdq8GTRLHBrVMdpt9=o9wVcFMc6f-yhmSBcZqQ@mail.gmail.com/
> [2] 32be3ca4cf78b ("wifi: ath11k: HAL SRNG: don't deinitialize and re-initialize again")
> [3] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1132343
> ---
>  drivers/net/wireless/ath/ath11k/hal.c | 13 ++++++++++---
>  1 file changed, 10 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/wireless/ath/ath11k/hal.c b/drivers/net/wireless/ath/ath11k/hal.c
> index e821e5a62c1c0..0c0aeb803018e 100644
> --- a/drivers/net/wireless/ath/ath11k/hal.c
> +++ b/drivers/net/wireless/ath/ath11k/hal.c
> @@ -1387,14 +1387,21 @@ EXPORT_SYMBOL(ath11k_hal_srng_deinit);
>  
>  void ath11k_hal_srng_clear(struct ath11k_base *ab)
>  {
> -	/* No need to memset rdp and wrp memory since each individual
> -	 * segment would get cleared in ath11k_hal_srng_src_hw_init()
> -	 * and ath11k_hal_srng_dst_hw_init().
> +	/* Preserve the shared pointer buffers, but clear the previous

Since you are touching it I'm going to change this from the obsolete
networking block comment style to the universal block comment style with
/* on a line by itself

> +	 * firmware instance's hp/tp state before handing them back to FW.
> +	 * LMAC rings reuse this shared memory without going through the
> +	 * normal SRNG hw-init path that zeros non-LMAC ring pointers.
>  	 */
>  	memset(ab->hal.srng_list, 0,
>  	       sizeof(ab->hal.srng_list));
>  	memset(ab->hal.shadow_reg_addr, 0,
>  	       sizeof(ab->hal.shadow_reg_addr));
> +	if (ab->hal.rdp.vaddr)
> +		memset(ab->hal.rdp.vaddr, 0,
> +		       sizeof(*ab->hal.rdp.vaddr) * HAL_SRNG_RING_ID_MAX);
> +	if (ab->hal.wrp.vaddr)
> +		memset(ab->hal.wrp.vaddr, 0,
> +		       sizeof(*ab->hal.wrp.vaddr) * HAL_SRNG_NUM_LMAC_RINGS);
>  	ab->hal.avail_blk_resource = 0;
>  	ab->hal.current_blk_index = 0;
>  	ab->hal.num_shadow_reg_configured = 0;
> 
> ---
> base-commit: 54a5b38e4396530e5b2f12b54d3844e860ab6784
> change-id: 20260513-kfarnung-ath11k-srng-clear-pointer-state-91d8ab07e5e2
> 
> Best regards,