Mathieu Bérard [Sat, 24 Mar 2007 17:52:20 +0000 (17:52 +0000)]
* Fix cooling feature actual state output
* Added dmi signature for P25 M60 and another flavor of Tecra S1
* Updated Changelog
* Added support for upcoming 2.6.21 blacklight API changes, there is now
3 API versions supported (2.6.16 to 2.6.19 ; 2.6.20 only and 2.6.21 to ...)
which of course leads to more #ifdef uglyness ...
Mathieu Bérard [Sun, 4 Mar 2007 17:13:45 +0000 (17:13 +0000)]
* backlight API change in 2.6.21-rc : disable backlight binding for now
* Update version string
* Fix procfs read funtion: for a reason I don't really understand all read methods
were called twice.
Mathieu Bérard [Wed, 31 Jan 2007 21:05:36 +0000 (21:05 +0000)]
* Split TSM30X (ectype 12) in:
-TSM70 (ectype 12) => new implementation (compal.c & acpi.c backends)
-TSM30X (ectype 15) => old implementation (legacy backends)
This is done by renaming TSM30X to TSM40 and restore old TSM30X
This sucks but is the only way to fix bugs 1617818 and 1605278
Holger Nelson [Fri, 12 Jan 2007 14:41:16 +0000 (14:41 +0000)]
Sometimes I miss multimedia-keys on my laptop. This small patch maps some
Fn-combos to play etc. with nbsmi-backend. It might be of some use to
others, but I don't think this is something that should go into trunk by
default.
Holger Nelson [Fri, 12 Jan 2007 02:02:10 +0000 (02:02 +0000)]
nbsmi.c Fixed NULL-pointer-dereference in omnibook_handle_fnkey: Pass priv_data instead of NULL to INIT_WORK for kernels<=2.6.19
debian/rules compatibility-fix for Ubuntu: Use DEPMOD="/bin/true" instead of DEPMOD="" during install.
Mathieu Bérard [Sun, 12 Nov 2006 23:29:39 +0000 (23:29 +0000)]
* Cleanup in fan_pollicy, dump, battery
* Kill legacy_ec_read/write
* nbsmi.c: make use of backend private data pointer by folding all global variables in a struct
remove smi_spinlock and use preempt_* and local_irq_* instead
try to improve hotkeys reliability
* acpi.c : hopefuly finish bluetooth work
* A few other small cleanup all over the place
* Fix bugs in wifi/bluetooth set funtions in nbsmi and acpi
I am actually wondering how this could have worked as
the raw_value computation was busted.
* Add debug printks all over the place.
* TSM30X bluetooth update: bind to the Toshiba's Bluetooth ACPI
device and hook to ACPI methods. Seems to be required for the
bt adapter to be detected (on the USB bus) and activated.
Testing by R.M. Albrecht.
* Update Changelog
* Lindent run
* Check all struct omnibook_operations created on the stack:
-Remove those who are clearly avoidable
-Correctly init the others
* Yet another nbsmi asm implementation:
-use spinlock_irqsave/restore to save/load eflags
-don't pusha/popa but put all registers in clobber list
Should be more stack friendly
Has been runtime checked by replacing smi call by
"lets clobber the registers" asm instructions.
* Yet more fix by Holger Nelson.
* Fix bug in init.c: If specific feature init function
fail we have to decrement kref on backend
* Fix forgotten TSM40 table entry in display.c
* Add printk in case of smi command error
Mathieu Bérard [Thu, 17 Aug 2006 16:55:41 +0000 (16:55 +0000)]
M40 vs. M40X issue hopefully resolved:
dmi_matched now return a non null result which means that we stop on the FIRST fully
matched entry of omnibook_ids[] which means that we MUST store records in that list from
the MORE specific to the LEAST specific.
As M40X is now before M40 this sould works.
And that was the behaviour of the old (pre fork) DMI matching system and
thus the intended one as the list order was not modified.
Mathieu Bérard [Fri, 11 Aug 2006 14:45:07 +0000 (14:45 +0000)]
* omnibook.h: Fix #if define to correctly compile backlight
device support if it's enabled in kernel (or as a module)
* Makefile: Include explicit depmod in load target and clean
Modules.symvers as well as Module.symvers (seems to be kernel
version dependent)