Re: [PATCH 4/7] rtlwifi: rtl8821ae: Reverse PM capability exists check

[Ilpo Järvinen]

On Fri, 17 Nov 2023, Bjorn Helgaas wrote:

> On Fri, Nov 17, 2023 at 11:44:22AM +0200, Ilpo Järvinen wrote:
> > Check if PM capability does not exists and return early which follows
> > the usual error handling pattern.
> > 
> > Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@xxxxxxxxxxxxxxx>
> > ---
> >  .../wireless/realtek/rtlwifi/rtl8821ae/hw.c   | 45 ++++++++++---------
> >  1 file changed, 23 insertions(+), 22 deletions(-)
> > 
> > diff --git a/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c b/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
> > index 6ae37d61a2a2..53cfeed0b030 100644
> > --- a/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
> > +++ b/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
> > @@ -2305,30 +2305,31 @@ static void _rtl8821ae_clear_pci_pme_status(struct ieee80211_hw *hw)
> >  		}
> >  	} while (cnt++ < 200);
> >  
> > -	if (cap_id == 0x01) {
> > -		/* Get the PM CSR (Control/Status Register),
> > -		 * The PME_Status is located at PM Capatibility offset 5, bit 7
> > -		 */
> > -		pci_read_config_byte(rtlpci->pdev, cap_pointer + 5, &pmcs_reg);
> > -
> > -		if (pmcs_reg & BIT(7)) {
> > -			/* Clear PME_Status with write */
> > -			pci_write_config_byte(rtlpci->pdev, cap_pointer + 5,
> > -					      pmcs_reg);
> > -			/* Read it back to check */
> > -			pci_read_config_byte(rtlpci->pdev, cap_pointer + 5,
> > -					     &pmcs_reg);
> > -			rtl_dbg(rtlpriv, COMP_INIT, DBG_DMESG,
> > -				"Clear PME status 0x%2x to 0x%2x\n",
> > -				cap_pointer + 5, pmcs_reg);
> > -		} else {
> > -			rtl_dbg(rtlpriv, COMP_INIT, DBG_DMESG,
> > -				"PME status(0x%2x) = 0x%2x\n",
> > -				cap_pointer + 5, pmcs_reg);
> > -		}
> > -	} else {
> > +	if (cap_id != 0x01) {
> >  		rtl_dbg(rtlpriv, COMP_INIT, DBG_WARNING,
> >  			"Cannot find PME Capability\n");
> > +		return;
> > +	}
> > +
> > +	/* Get the PM CSR (Control/Status Register),
> > +	 * The PME_Status is located at PM Capatibility offset 5, bit 7

> But this is PCI_PM_CTRL and PCI_PM_CTRL_PME_STATUS (with a word read),
> right?  No need for a comment then.
> 
> I don't know why the driver needs to do this, but I'm skeptical.  The
> only other drivers that look at PCI_PM_CTRL_PME_STATUS themselves are
> bnx2x and ksz884xp (ksz884x.c), so this is highly suspicious.

I hope you are not asking from me because I'm just the poor guy who tries 
to cleanup this mess. ;-) ...So all I can do is shrug and say, I 
unfortunately don't know the answer.

Hopefully somebody more familiar with the devices could chime in.


-- 
 i.