2024-02-27 03:05:48

by Kevin Lo

Subject: [PATCH] wifi: ath11k: adjust a comment to reflect reality

In ath11k_mhi_set_mhictrl_reset(), I observed on QCA6390/QCN9074/WCN6855,
MHISTATUS has SYSERR bit always been set after SOC_GLOBAL_RESET.

Signed-off-by: Kevin Lo <[email protected]>
---
diff --git a/drivers/net/wireless/ath/ath11k/mhi.c b/drivers/net/wireless/ath/ath11k/mhi.c
index 3de7fa6f88d0..1198e80d9dff 100644
--- a/drivers/net/wireless/ath/ath11k/mhi.c
+++ b/drivers/net/wireless/ath/ath11k/mhi.c
@@ -158,8 +158,8 @@ void ath11k_mhi_set_mhictrl_reset(struct ath11k_base *ab)

ath11k_dbg(ab, ATH11K_DBG_PCI, "mhistatus 0x%x\n", val);

- /* Observed on QCA6390 that after SOC_GLOBAL_RESET, MHISTATUS
- * has SYSERR bit set and thus need to set MHICTRL_RESET
+ /* After SOC_GLOBAL_RESET, MHISTATUS has SYSERR bit
+ * always been set and thus need to set MHICTRL_RESET
* to clear SYSERR.
*/
ath11k_pcic_write32(ab, MHICTRL, MHICTRL_RESET_MASK);

2024-03-01 16:16:47

by Jeff Johnson

Subject: Re: [PATCH] wifi: ath11k: adjust a comment to reflect reality

On 2/26/2024 6:45 PM, Kevin Lo wrote:
> In ath11k_mhi_set_mhictrl_reset(), I observed on QCA6390/QCN9074/WCN6855,
> MHISTATUS has SYSERR bit always been set after SOC_GLOBAL_RESET.
>
> Signed-off-by: Kevin Lo <[email protected]>
> ---
> diff --git a/drivers/net/wireless/ath/ath11k/mhi.c b/drivers/net/wireless/ath/ath11k/mhi.c
> index 3de7fa6f88d0..1198e80d9dff 100644
> --- a/drivers/net/wireless/ath/ath11k/mhi.c
> +++ b/drivers/net/wireless/ath/ath11k/mhi.c
> @@ -158,8 +158,8 @@ void ath11k_mhi_set_mhictrl_reset(struct ath11k_base *ab)
>
> ath11k_dbg(ab, ATH11K_DBG_PCI, "mhistatus 0x%x\n", val);
>
> - /* Observed on QCA6390 that after SOC_GLOBAL_RESET, MHISTATUS
> - * has SYSERR bit set and thus need to set MHICTRL_RESET
> + /* After SOC_GLOBAL_RESET, MHISTATUS has SYSERR bit
> + * always been set and thus need to set MHICTRL_RESET
> * to clear SYSERR.
> */
> ath11k_pcic_write32(ab, MHICTRL, MHICTRL_RESET_MASK);

Unless you've verified this is always true for every supported chipset
I'd rather more accurately say something like:
After SOC_GLOBAL_RESET, MHISTATUS may still have SYSERR bit set and thus...

2024-03-04 07:57:57

by Kevin Lo

Subject: Re: [PATCH] wifi: ath11k: adjust a comment to reflect reality

On Fri, Mar 01, 2024 at 08:16:16AM -0800, Jeff Johnson wrote:
>
> On 2/26/2024 6:45 PM, Kevin Lo wrote:
> > In ath11k_mhi_set_mhictrl_reset(), I observed on QCA6390/QCN9074/WCN6855,
> > MHISTATUS has SYSERR bit always been set after SOC_GLOBAL_RESET.
> >
> > Signed-off-by: Kevin Lo <[email protected]>
> > ---
> > diff --git a/drivers/net/wireless/ath/ath11k/mhi.c b/drivers/net/wireless/ath/ath11k/mhi.c
> > index 3de7fa6f88d0..1198e80d9dff 100644
> > --- a/drivers/net/wireless/ath/ath11k/mhi.c
> > +++ b/drivers/net/wireless/ath/ath11k/mhi.c
> > @@ -158,8 +158,8 @@ void ath11k_mhi_set_mhictrl_reset(struct ath11k_base *ab)
> >
> > ath11k_dbg(ab, ATH11K_DBG_PCI, "mhistatus 0x%x\n", val);
> >
> > - /* Observed on QCA6390 that after SOC_GLOBAL_RESET, MHISTATUS
> > - * has SYSERR bit set and thus need to set MHICTRL_RESET
> > + /* After SOC_GLOBAL_RESET, MHISTATUS has SYSERR bit
> > + * always been set and thus need to set MHICTRL_RESET
> > * to clear SYSERR.
> > */
> > ath11k_pcic_write32(ab, MHICTRL, MHICTRL_RESET_MASK);
>
> Unless you've verified this is always true for every supported chipset

Only PCIe devices trigger the call to ath11k_mhi_set_mhictrl_reset(),
and I've verified this on the QCA6390, QCN9074, and WCN6855 chips.

> I'd rather more accurately say something like:
> After SOC_GLOBAL_RESET, MHISTATUS may still have SYSERR bit set and thus...

Okay, I'll update the comments as per your suggestion, thanks.