On Wednesday, June 28, 2006 10:23 AM, James Bottomley wrote:
> This one corrects a few errors in the prior version several of which
> caused panics (missing remove children) and updates the port
> add phy api
> (now it's legal to add a phy that's already part of the port). I've
> also fixed the naming scheme.
>
> expanders are now expander-<h>:<e>
>
> where <h> host is the host number and <e> is a unique expander number
> (one per host).
>
> ports are port-<h>:<n> for ports on the hba or port-<h>:<e>:<n> for
> ports on an expander (and phys follow the same naming pattern).
>
Thanks, but can we have a better way in keeping track of the unique
expander numbers than using rphy->scsi_target_id? I get confused when
I see the target id there in the code.
> +/** sas_port_alloc - allocate and initialize a SAS port structure
> + *
> + * @parent: parent device
> + * @port_id: port number
> + *
> + * Allocates a SAS port structure. It will be added to the
> device tree
> + * below the device specified by @parent which must be
> either a Scsi_Host
> + * or a sas_expander_device.
> + *
> + * Returns %NULL on error
> + */
> +struct sas_port *sas_port_alloc(struct device *parent, int port_id)
> +{
> + struct Scsi_Host *shost = dev_to_shost(parent);
> + struct sas_port *port;
Thanks for changing the port_id to int. However the
struct sas_port->port_identifier is still u8. Can you
change this to unsigned int in both places? The port_identifier
is passed to scsi_scan_target as the channel, and its uinit.
That way we are not limited to 255 ports.
>
> @@ -920,7 +1195,7 @@
> break;
> case SAS_EDGE_EXPANDER_DEVICE:
> case SAS_FANOUT_EXPANDER_DEVICE:
> - device_for_each_child(dev, NULL, do_sas_phy_delete);
> + sas_remove_children(dev);
> break;
> default:
> break;
This new patch is not working for cable pull case. Its hanging
right here inside this call to sas_remove_children(). I have
devices attached to across two expanders. In your previous wide port
patch's I don't have this hang, and in the older patch its
calling sas_remove_children.
Are there other changes in the patch is effecting this?
> +struct sas_port {
> + struct device dev;
> +
> + u8 port_identifier;
> + int num_phys;
> +
> + /* the other end of the link */
> + struct sas_rphy *rphy;
> +
> + struct mutex phy_list_mutex;
> + struct list_head phy_list;
> +};
Pls change port_identifier to uint or u32.
> @@ -141,13 +160,21 @@
> extern void sas_phy_delete(struct sas_phy *);
> extern int scsi_is_sas_phy(const struct device *);
>
> -extern struct sas_rphy *sas_end_device_alloc(struct sas_phy *);
> -extern struct sas_rphy *sas_expander_alloc(struct sas_phy *,
> enum sas_device_type);
> +extern struct sas_rphy *sas_end_device_alloc(struct sas_port *);
> +extern struct sas_rphy *sas_expander_alloc(struct sas_port
> *, enum sas_device_type);
> void sas_rphy_free(struct sas_rphy *);
> extern int sas_rphy_add(struct sas_rphy *);
> extern void sas_rphy_delete(struct sas_rphy *);
> extern int scsi_is_sas_rphy(const struct device *);
>
> +struct sas_port *sas_port_alloc(struct device *, int);
> +int sas_port_add(struct sas_port *);
> +void sas_port_free(struct sas_port *);
> +void sas_port_delete(struct sas_port *);
> +void sas_port_add_phy(struct sas_port *, struct sas_phy *);
> +void sas_port_delete_phy(struct sas_port *, struct sas_phy *);
> +int scsi_is_sas_port(const struct device *);
> +
Pls add extern in front of all the calls that do EXPORT_SYMBOL; e.g
sas_port_alloc, sas_port_add, sas_port_delete, sas_port_add_phy, and
sas_port_delete_phy.
-
: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
