Re: [RFC PATCH V3 12/21] mmc: sdhci: UHS-II support, add hooks for additional operations

[AKASHI Takahiro]

Adrian,

Your comments are scattered over various functions, and so
I would like to address them in separate replies.

First, I'd like to discuss sdhci_[add|remove]_host().

On Fri, Aug 21, 2020 at 05:08:32PM +0300, Adrian Hunter wrote:
> On 10/07/20 2:10 pm, Ben Chuang wrote:
> > From: Ben Chuang <ben.chuang@xxxxxxxxxxxxxxxxxxx>
> > 
> > In this commit, UHS-II related operations will be called via a function
> > pointer array, sdhci_uhs2_ops, in order to make UHS-II support as
> > a kernel module.
> > This array will be initialized only if CONFIG_MMC_SDHCI_UHS2 is enabled
> > and when the UHS-II module is loaded. Otherwise, all the functions
> > stay void.
> > 
> > Signed-off-by: Ben Chuang <ben.chuang@xxxxxxxxxxxxxxxxxxx>
> > Signed-off-by: AKASHI Takahiro <takahiro.akashi@xxxxxxxxxx>
> > ---

 (snip)

> >  		if (intmask & (SDHCI_INT_CARD_INSERT | SDHCI_INT_CARD_REMOVE)) {
> >  			u32 present = sdhci_readl(host, SDHCI_PRESENT_STATE) &
> >  				      SDHCI_CARD_PRESENT;
> > @@ -4717,6 +4812,14 @@ int sdhci_setup_host(struct sdhci_host *host)
> >  		/* This may alter mmc->*_blk_* parameters */
> >  		sdhci_allocate_bounce_buffer(host);
> >  
> > +	if (IS_ENABLED(CONFIG_MMC_SDHCI_UHS2) &&
> > +	    host->version >= SDHCI_SPEC_400 &&
> > +	    sdhci_uhs2_ops.add_host) {
> > +		ret = sdhci_uhs2_ops.add_host(host, host->caps1);
> > +		if (ret)
> > +			goto unreg;
> > +	}
> > +
> 
> I think you should look at creating uhs2_add_host() instead
> 
> >  	return 0;
> >  
> >  unreg:
> > @@ -4738,6 +4841,8 @@ void sdhci_cleanup_host(struct sdhci_host *host)
> >  {
> >  	struct mmc_host *mmc = host->mmc;
> >  
> > +	/* FIXME: Do we have to do some cleanup for UHS2 here? */
> > +
> >  	if (!IS_ERR(mmc->supply.vqmmc))
> >  		regulator_disable(mmc->supply.vqmmc);
> >  
> > @@ -4766,6 +4871,14 @@ int __sdhci_add_host(struct sdhci_host *host)
> >  		mmc->cqe_ops = NULL;
> >  	}
> >  
> > +	if ((mmc->caps & MMC_CAP_UHS2) && !host->v4_mode) {
> > +		/* host doesn't want to enable UHS2 support */
> > +		mmc->caps &= ~MMC_CAP_UHS2;
> > +		mmc->flags &= ~MMC_UHS2_SUPPORT;
> > +
> > +		/* FIXME: Do we have to do some cleanup here? */
> > +	}
> > +
> >  	host->complete_wq = alloc_workqueue("sdhci", flags, 0);
> >  	if (!host->complete_wq)
> >  		return -ENOMEM;
> > @@ -4812,6 +4925,9 @@ int __sdhci_add_host(struct sdhci_host *host)
> >  unled:
> >  	sdhci_led_unregister(host);
> >  unirq:
> > +	if (IS_ENABLED(CONFIG_MMC_SDHCI_UHS2) &&
> > +	    sdhci_uhs2_ops.remove_host)
> > +		sdhci_uhs2_ops.remove_host(host, 0);
> >  	sdhci_do_reset(host, SDHCI_RESET_ALL);
> >  	sdhci_writel(host, 0, SDHCI_INT_ENABLE);
> >  	sdhci_writel(host, 0, SDHCI_SIGNAL_ENABLE);
> > @@ -4869,6 +4985,10 @@ void sdhci_remove_host(struct sdhci_host *host, int dead)
> >  
> >  	sdhci_led_unregister(host);
> >  
> > +	if (IS_ENABLED(CONFIG_MMC_SDHCI_UHS2) &&
> > +	    sdhci_uhs2_ops.remove_host)
> > +		sdhci_uhs2_ops.remove_host(host, dead);
> > +
> 
> I think you should look at creating uhs2_remove_host() instead

You suggest that we will have separate sdhci_uhs2_[add|remove]_host(),
but I don't think it's always convenient.

UHS-II capable host will be set to call sdhci_uhs2_add_host() explicitly,
but we can't do that in case of pci and pltfm based drivers as they utilize
common helper functions, sdhci_pci_probe() and sdhci_pltfm_register(),
respectively.
Therefore, we inevitably have to call sdhci_uhs2_add_host() there.

If so, why should we distinguish sdhci_uhs2_add_host from sdhci_uhs_add_host?
I don't see any good reason.
Moreover, as a result, there exists a mixed usage of sdhci_ interfaces
and sdhci_uhs2_ interfaces in sdhci-pci-core.c and sdhci-pltfm.c.

It sounds odd to me.

-Takahiro Akashi


> 
> >  	if (!dead)
> >  		sdhci_do_reset(host, SDHCI_RESET_ALL);
> >  
> > 
>