Bug 35354

As outlined in 80-udisks.rules:

# Set eSATA port type for known eSATA CardBus adapters - first we want to ensure
# the device is on a cardbus controller (upper PCI device) - then we check
# vid/pid (lower PCI device)
#
SUBSYSTEM=="scsi_host", ATTRS{class}=="0x060700", GOTO="ata_port_cardbus"
GOTO="ata_port_cardbus_end"
LABEL="ata_port_cardbus"
# Mass storage controller: Silicon Image, Inc. SiI 3512 [SATALink/SATARaid] Serial ATA Controller (rev 01) 
#
SUBSYSTEMS=="pci", ATTRS{vendor}=="0x1095", ATTRS{device}=="0x3512", ENV{UDISKS_ATA_PORT_CONNECTOR_TYPE}="ata_sata_external"
LABEL="ata_port_cardbus_end"


I try to do the same for a 0x1095 0x3132 SiI ExpressCard 2-Port eSATA card, but it does not work.
This wouldn't work anyhow as my card is no Cardbus card but an ExpressCard, which means the ATTRS{class}=="0x060700" must be set to ATTRS{class}=="0x060400".

This should be fixed as ExpressCard will be more common than Cardbus I guess. Both should therefor be supported.

The problem however remains.
This bug is very similar to https://bugs.freedesktop.org/show_bug.cgi?id=22879 as I am also unable to make e17/udisks see my eSATA disks as they're marked as system internal. The patch mentioned in bug #22879 does work for me.
I just wanted to report the above issue and also that (seemingly correctly) filling in my PCI ids does not change udisks behavior.

In udisks-1.0.2, I took a look into the sources and found src/port.c:

654   /* Third, guess the connector type.
655    *
656    * This can be overriden via the udev property UDISKS_ATA_PORT_CONNECTOR_TYPE -
657    * see the data/80-udisks.rules for an example.
658    */
659   connector_type = g_udev_device_get_property (port->priv->d, "UDISKS_ATA_PORT_CONNECTOR_TYPE");
660   if (connector_type == NULL)
661     {
662       connector_type = "ata";
663       adapter_fabric = adapter_local_get_fabric (adapter);
664       if (g_strcmp0 (adapter_fabric, "ata_pata") == 0)
665         {
666           connector_type = "ata_pata";
667         }
668       else if (g_strcmp0 (adapter_fabric, "ata_sata") == 0)
669         {
670           connector_type = "ata_sata";
671         }
672     }

but it seems to me this "connector_type" never gets used for determining whether a device should be system internal or not.

Therefor, I tried to do the following in src/device.c:

3952   /* devices on certain buses are never system internal */
 3953 //if (device->priv->device_is_drive && device->priv->drive_connection_interface != NULL)
 3954 if (device->priv->device_is_drive)
 3955   {
 3956     if (device->priv->drive_connection_interface != NULL)
 3957     {
 3958       if (strcmp (device->priv->drive_connection_interface, "ata_serial_esata") == 0
 3959           || strcmp (device->priv->drive_connection_interface, "sdio") == 0
 3960           || strcmp (device->priv->drive_connection_interface, "usb") == 0
 3961           || strcmp (device->priv->drive_connection_interface, "firewire") == 0)
 3962         {
 3963           is_system_internal = FALSE;
 3964         }
 3965       else
 3966         {
 3967           is_system_internal = TRUE;
 3968         }
 3969     }
 3970     if (port->priv->d->connector_type != NULL)
 3971     {
 3972       /* this can be set like in 80-udisks.rules: ENV{UDISKS_ATA_PORT_CONNECTOR_TYPE}="ata_sata_external" */
 3973       if (strcmp (port->priv->d->connector_type, "ata_sata_external") == 0)
 3974         {
 3975           is_system_internal = FALSE;
 3976         }
 3977     }
 3978       goto determined;
 3979   }

but the port object is not accessible from within device.c and
a) my C skills somewhat end here (what a shame...) and
b) I don't know if you wanted me to pull in a reference to port at all as I don't know anything about the design philosophy behind udisks.

It however seems to me that connector_type should somewhere be used to set system internal to 0 if it is set to ata_sata_external via udev's UDISKS_ATA_PORT_CONNECTOR_TYPE variable. Currently, only device properties and no _port_ properties are used to determine that.

I hope this was precise enough and I am very sorry my C isn't any better.
I'd really like to fix this myself...

PS: Ah.. If that matters: the disk connected is a naked SATA disk plugged into the eSATA controller. So the kernel and/or udisks can't be put to blame if they do not detect the disk as being external, and I totally understand that I have to customize my setup a little if I want this to work. However, I still think drives connected to a controller with eSATA ports have to be system internal = 0. I can't determine whether my controller advertises having _e_SATA ports and willing to test this if you give me a clue. Still, this should be overridable on a controller or better yet port basis. I have another PC that has eSATA ports connected to 2 of its 6 internal SATA ports - also non-standard, I know, but that's how a lot of people do it - the bracket has already been in the original mobo package after all!