On Wed, 2008-07-16 at 09:25 +0930, Jonathan Woithe wrote: > > > It is likely that everything works out with video.ko then. > > > If not, with a patch I sent recently you should switch off Windows osi strings > > > with: > > > acpi_osi=windows_false (preferred) > > > or (already works) > > > acpi_osi="!Windows 2006" > > > > disabling windows_osi disables backlight through video.ko and everything > > works fine with fujitsu_laptop (also brightness-buttons since ACPI then > > generates events on the fujitsu device) but some other ACPI stuff is > > disabled too (extended battery status etc...) > > As per my previous email, the fact that this works is probably because in > this instance acpi_video_backlight_support() returns false and allows > fujitsu-laptop to set up its backlight control functions. Peter, can you attach the acpidump output please? you can use the latest pmtools at http://www.lesswatts.org/patches/linux_acpi/ > > > > If it still does not work, then there is a driver missing which was a > > > proprietary one on XP and which has not been fully re-engeeniered for Linux. > > > > > > Using ACPI video events and pass them to the Fujitsu driver to change > > > backlight (is this how it works? Is this done through userspace?), mixes up > > > everything. > > > > without acpi_osi="!Windows 2006" the buttons just produce ACPI video > > events (recieved by video.ko) and hal does the adjustment through > > fujitsu_laptop.ko > > I guess video.ko receives them because they are ACPI video events. However, > as previously determined it doesn't do anything useful with them. I'm a > little surprised that the physical brightness adjustment comes from hal > though. On my S7020 the brightness changes brought about by the brightness > buttons are seemingly done in hardware - the buttons have an effect even > when fujitsu-laptop isn't loaded. The brightness control functions in > fujitsu-laptop were added primarily to allow *software* control of the > brightness - something which definitely doesn't work without fujitsu-laptop. > The generation of ACPI video brightness keypress events was added to > fujitsu-laptop so userspace could keep track of changes instigated by the > hardware buttons. > > > The Problem is that video.ko only touches an otherwise unused ACPI > > register but does not change the brightness... (you can just store a > > number between 0 and 7 there) > > Yes, which is why I'm fairly sure there's no risk of double-touching the > hardware. video.ko and fujitsu-laptop.ko are driving different hardware. > > > Perhaps there is a way to disable the dysfunctional backlight stuff in > > video.ko but retain the ACPI video event handling?! > > The problem with doing this is that it's only the platform-specific drivers > which have the knowledge as to whether this is necessary or not. video.ko > would therefore have to query each loaded platform-specific module which > seems messy to me. If video.ko was loaded before the relevant > platform-specific module it gets even messier. I guess we could add a > video.ko function like acpi_video_backlight_disable() which platform modules > could call if they choose to handle the backlight. However, this seems > suboptimal to since it means video.ko sets up a backlight interface only to > have to tear it down again soon after. I guess it's no worse than setting > up a backlight interface which never gets used though. > > The fundamental problem as I see at this point it is that the new > acpi_video_backlight_support() function returns true on Fujitsu S-series > laptops because the ACPI environment looks "right" from video.ko's point of > view. However, it's clear that the desired registers/methods - although > present - are completely non-functional in these laptops so in reality > video.ko cannot handle the backlight in this case. Unless video.ko > hooks into the Fujitsu-specific registers (which defeats the modularity of > hardware-specific modules) this won't change. IMO, In this case, we should add a dmi check in video_detect.c and set acpi_backlight=vendor for this laptop, both before acpi video and fujitsu laptop driver is loaded. thanks, rui -- To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
