Re: [PATCH 1/2] [libata] sata_mv: Remove PCI dependency

[Jeff Garzik <jeff@garzik.org>]

saeed.bishara@gmail.com wrote:
> From: Saeed Bishara <saeed@marvell.com>
> 
> The integrated SATA controller is connected directly to the SoC's
> internal bus, not via PCI interface. this patch removes the dependency
> on the PCI interface.
> 
> Signed-off-by: Saeed Bishara <saeed@marvell.com>
> ---
>  drivers/ata/Kconfig   |    2 +-
>  drivers/ata/sata_mv.c |  113 ++++++++++++++++++++++++++++++-------------------
>  2 files changed, 71 insertions(+), 44 deletions(-)


Overall, a good patch, though some minor revisions are needed (comments 
inline).



> diff --git a/drivers/ata/Kconfig b/drivers/ata/Kconfig
> index ba63619..c60842b 100644
> --- a/drivers/ata/Kconfig
> +++ b/drivers/ata/Kconfig
> @@ -69,7 +69,7 @@ config ATA_PIIX
>  
>  config SATA_MV
>  	tristate "Marvell SATA support (HIGHLY EXPERIMENTAL)"
> -	depends on PCI && EXPERIMENTAL
> +	depends on EXPERIMENTAL
>  	help
>  	  This option enables support for the Marvell Serial ATA family.
>  	  Currently supports 88SX[56]0[48][01] chips.
> diff --git a/drivers/ata/sata_mv.c b/drivers/ata/sata_mv.c
> index 97c3e11..cbd3f1b 100644
> --- a/drivers/ata/sata_mv.c
> +++ b/drivers/ata/sata_mv.c
> @@ -421,7 +421,9 @@ static void mv_error_handler(struct ata_port *ap);
>  static void mv_post_int_cmd(struct ata_queued_cmd *qc);
>  static void mv_eh_freeze(struct ata_port *ap);
>  static void mv_eh_thaw(struct ata_port *ap);
> +#ifdef CONFIG_PCI
>  static int mv_init_one(struct pci_dev *pdev, const struct pci_device_id *ent);
> +#endif

with the proper ordering (see comments below), you can remove this 
prototype altogether


>  static void mv5_phy_errata(struct mv_host_priv *hpriv, void __iomem *mmio,
>  			   unsigned int port);
> @@ -638,14 +640,14 @@ static const struct pci_device_id mv_pci_tbl[] = {
>  
>  	{ }			/* terminate list */
>  };
> -
> +#ifdef CONFIG_PCI
>  static struct pci_driver mv_pci_driver = {
>  	.name			= DRV_NAME,
>  	.id_table		= mv_pci_tbl,
>  	.probe			= mv_init_one,
>  	.remove			= ata_pci_remove_one,
>  };
> -
> +#endif

I would prefer that you move this, and the pci_device_id table, down to 
the bottom of the file with the rest of the #CONFIG_PCI section of code 
that you have nicely arranged.


> @@ -2616,6 +2581,47 @@ done:
>  	return rc;
>  }
>  
> +#ifdef CONFIG_PCI
> +
> +/*
> + * module options
> + */
> +static int msi;	      /* Use PCI msi; either zero (off, default) or non-zero */
> +
> +
> +/* move to PCI layer or libata core? */
> +static int pci_go_64(struct pci_dev *pdev)
> +{
> +	int rc;
> +
> +	if (!pci_set_dma_mask(pdev, DMA_64BIT_MASK)) {
> +		rc = pci_set_consistent_dma_mask(pdev, DMA_64BIT_MASK);
> +		if (rc) {
> +			rc = pci_set_consistent_dma_mask(pdev, DMA_32BIT_MASK);
> +			if (rc) {
> +				dev_printk(KERN_ERR, &pdev->dev,
> +					   "64-bit DMA enable failed\n");
> +				return rc;
> +			}
> +		}
> +	} else {
> +		rc = pci_set_dma_mask(pdev, DMA_32BIT_MASK);
> +		if (rc) {
> +			dev_printk(KERN_ERR, &pdev->dev,
> +				   "32-bit DMA enable failed\n");
> +			return rc;
> +		}
> +		rc = pci_set_consistent_dma_mask(pdev, DMA_32BIT_MASK);
> +		if (rc) {
> +			dev_printk(KERN_ERR, &pdev->dev,
> +				   "32-bit consistent DMA enable failed\n");
> +			return rc;
> +		}
> +	}
> +
> +	return rc;
> +}
> +
>  /**
>   *      mv_print_info - Dump key info to kernel log for perusal.
>   *      @host: ATA host to print info about
> @@ -2721,15 +2727,34 @@ static int mv_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
>  				 IS_GEN_I(hpriv) ? &mv5_sht : &mv6_sht);
>  }
>  

nicely done, this [once my comments are taken] arranges everything into 
a single PCI-specific area.


> -static int __init mv_init(void)
> +static int __init mv_pci_register_driver(void)
>  {
>  	return pci_register_driver(&mv_pci_driver);
>  }
>  
> -static void __exit mv_exit(void)
> +static void __exit mv_pci_unregister_driver(void)
>  {
>  	pci_unregister_driver(&mv_pci_driver);
>  }
> +#else
> +static int __init mv_pci_register_driver(void)
> +{
> +	return 0;
> +}
> +
> +static void __exit mv_pci_unregister_driver(void)
> +{
> +}
> +#endif
> +static int __init mv_init(void)
> +{
> +	return mv_pci_register_driver();
> +}
> +
> +static void __exit mv_exit(void)
> +{
> +	mv_pci_unregister_driver();
> +}

The way we do multi-bus drivers that you do something like

static int mv_init(void)
{
	#ifdef CONFIG_PCI
	rc = pci_register_driver(&mv_driver);
	if (rc)
		return rc;
	#endif

	#ifdef MY_MARVELL_SOC_EMBEDDED_CHIP
	rc = my_soc_register_platform_driver(&platform_driver);
	if (rc)
		goto err_out;
	#endif

	return 0;

err_out:
#ifdef CONFIG_PCI
	pci_unregister_driver(&mv_driver);
#endif
	return rc;
}

The module unregister function mv_exit() should do something similar, in 
reverse order.

Even embedded SoC controllers need to register a driver (usually a 
platform_driver with platform_devices).



>  MODULE_AUTHOR("Brett Russ");
>  MODULE_DESCRIPTION("SCSI low-level driver for Marvell SATA controllers");
> @@ -2737,8 +2762,10 @@ MODULE_LICENSE("GPL");
>  MODULE_DEVICE_TABLE(pci, mv_pci_tbl);
>  MODULE_VERSION(DRV_VERSION);
>  
> +#ifdef CONFIG_PCI
>  module_param(msi, int, 0444);
>  MODULE_PARM_DESC(msi, "Enable use of PCI MSI (0=off, 1=on)");
> +#endif

agreed -- in contrast with my comments above, please keep this module 
option right where it is, and add the #ifdef CONFIG_PCI precisely as you 
have done.  do /not/ move it into the CONFIG_PCI section referred to in 
my other comments.

	Jeff