https://bugzilla.novell.com/show_bug.cgi?id=302482#c1 Thomas Renninger changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |astarikovskiy@novell.com Status|NEW |ASSIGNED --- Comment #1 from Thomas Renninger 2007-08-22 13:09:31 MST --- Outch, dock device is totally mis-designed, it's even not an ACPI driver. I have an idea. We could do it similar like the bay device and add a ACPI_DOCK_HID id to the CID list of devices with a _DCK function. Not sure whether I can finish this up before holidays... I try to have a short look at that tomorrow. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=302482#c3 --- Comment #3 from Thomas Renninger 2007-08-25 00:30:12 MST --- Patch does not work, needs more investigation... I am on holidays for two weeks. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=302482#c6 Kay Sievers changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #159429|0 |1 is obsolete| | --- Comment #6 from Kay Sievers 2007-08-25 15:39:53 MST --- Created an attachment (id=159868) --> (https://bugzilla.novell.com/attachment.cgi?id=159868) enable dock device in acpi tree and add LNXDOCK alias You are right, it doesn't work because the devices not in "present" state are completely ignored: if (ACPI_FAILURE(result) || !device->status.present) { result = -ENOENT; ... } This patch disables not present devices only when they have a "static" state: if (!device->flags.dynamic_status && !device->status.present) { result = -ENODEV; ... } Here is the new sysfs entry: $ grep . /sys/bus/acpi/devices/IBM0079:00/* /sys/bus/acpi/devices/IBM0079:00/hid:IBM0079 /sys/bus/acpi/devices/IBM0079:00/modalias:acpi:IBM0079:PNP0C15:LNXDOCK: /sys/bus/acpi/devices/IBM0079:00/path:\_SB_.GDCK ... And it loads the dock module: $ modprobe --first-time -a -v acpi:IBM0079:PNP0C15:LNXDOCK: insmod /lib/modules/2.6.23-rc3-g1a8f4610-dirty/kernel/drivers/acpi/dock.ko The string-add didn't work, because the variable sized info buffer can not be partly realloc'd. The array extension logic got replaced by a simple copy to the final data in acpi_device_set_id(). I'm not sure if we can just enable the non "present" devices like this, someone with deeper knowledge of acpi bus implementation needs to have look at this. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=302482#c8 --- Comment #8 from Alexey Starikovskiy 2007-08-28 13:31:28 MST --- Frank, I already discussed this patch with Thomas. It changes ACPI hot-plug functionality a lot, e.g. I have an previously absent BAT1 now. This is not bad by itself, but all other hot-pluggable devices like processors and memory should be checked for breakage against this patch. The only questionable place seems to be this check: if (!device->flags.dynamic_status && !device->status.present) { dynamic_status means presence of _STA method for a given device. status is an output from this method, if it is present, or copy of parent status, if _STA does not exist. So if child of such device does not have _STA (dynamic_status = 0, present = parent->present = 0), it will be left outside, while device itself gets driver attached. Me thinks, this is not desirable behavior, and this check should be removed altogether as suggested by Thomas earlier. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=302482#c10 --- Comment #10 from Alexey Starikovskiy 2007-08-29 14:03:40 MST --- Frank, This one looks much safer, but again, _all_ static devices under the dock or bay will not be handled automatically. This patch is better than nothing, but some time later we should remove this check completely. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=302482#c11 Frank Seidel changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #160747|0 |1 is obsolete| | --- Comment #11 from Frank Seidel 2007-08-29 16:17:05 MST --- Created an attachment (id=160841) --> (https://bugzilla.novell.com/attachment.cgi?id=160841) v2 of last patch Alexey, i absolutely agree. I considered this as a temporary workaround (at best) for this problem. Nevertheless perhaps you find this version a bit better, as it also will handle any device below a dock- or bay-device. Do you think we should use this one for now? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=302482#c12 --- Comment #12 from Alexey Starikovskiy 2007-08-30 00:01:10 MST --- Don't have any other objections. :) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=302482#c14 Frank Seidel changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #160841|0 |1 is obsolete| | --- Comment #14 from Frank Seidel 2007-09-03 14:23:17 MST --- Created an attachment (id=161569) --> (https://bugzilla.novell.com/attachment.cgi?id=161569) working and tested version of last patch Ok, finally this one seems to be ok. I tested it on many machines (with and without dockingsupport in their dsdt) and it seems to work as expected. There was a problem when cid_list was not existing already (as it was for a dell and a toshiba notebook here), it wasn't created correctly and the dock driver still wasn't loaded although expected (_DCK present). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.