I have had a Toshiba Satellite PSPADU-033026 for about 4 months and have been having the same problem: The fan won't turn on unless I disable ACPI--not really an option on a laptop. Without the fan, the card will get _very_ hot even during casual use: upwards of 75 Celsius. I wouldn't dare try to use Beryl or anything intensive until this is fixed for fear of melting the card, or worse: catching fire!
Toshiba ships the system with their Satellite v3.5 BIOS, but still hasn't posted this code on their support site (only v3.3 is available). According to the numerous posts about this issue, it seems that v3.3 suffers from similar problems. It is irrelevant anyhow, as the Phoenix BIOS flash utility (pflash16.exe) will not allow a downgrade of the BIOS to 3.3 or 2.4, regardless of whether you use the switches to force a downgrade. Toshiba tech. support has confirmed this as well. Apparently they have tricked out the BIOS so that you can only go to newer releases. Nice job! As another angry aside, they have also removed other useful BIOS menu items, such as enabling / disabling Intel VT extensions. Of course, the machine ships with VT _disabled_ and don't offer a utility to toggle this bit. Thanks! Apparently you can only use their hardware as _they_ intend you to use it. Their technical support team has more or less confirmed this as well. Use Vista, or else. So, I am stuck with this "Vista-ready" v3.5 BIOS and apparently broken DSDT.
I am running OpenSuse 10.2 with the vanilla kernel for 10.2: 126.96.36.199-34-default #1 SMP. I have also tried newer kernels such as 188.8.131.52-86-default from the OpenSuse release factory. This got me no further with my issues or endless investigation (obsession?) with this problem.
I have contacted Toshiba multiple times, but they are _zero_ assistance, becuase I am not running Vista.
I extracted the DSDT tables from Vista and compared them to /proc/acpi/dsdt under Linux, but of course they are the same, because the DSDT is read from BIOS when the OS boots.
I used the hand-dandy DSDT extract, repair, compile, override method, as recommended in the Gentoo wiki (http://gentoo-wiki.com/HOWTO_Fix_Common_ACPI_Problems)
. That made my sound card work, but the Nvidia fan still doesn't turn on. In is interesting to note that the fans are on *prior* to booting any OS, but as soon as grub hands off control to Linux, the fans shut off and the cooking begins.
So, while reading all of this, you are probably thinking, "why are you posting here? This is a _Toshiba_ problem!" Read on... there are a few more interesting details...
I believe this to be an _Nvidia_ driver issue, because of observed behavior under Vista. Under Vista, the Toshiba-supplied Nvidia 7600 driver turns the fan on as needed, changes fan speed in response to temperature, etc. Keep in mind, that under Vista you are using the same _broken_ DSDT / ACPI information with the same v3.5 BIOS, but using whatever code Nvidia has supplied to Toshiba to make the card work properly. Under Linux, even using the latest Nvidia Linux driver release: 100.14.09, the fan _does not_ come on.
I suppose it is possible that this issue is related to ACPI support in the Linux kernel or perhaps there is a communications problem between the Maxim 1617 chipset on the Nvidia 7600 and the Linux kernel, but yet I _am_ able to get temperature reads using nvidia-settings and nvclock... So Nvidia has code to communicate with this card. Why no fan control?!? Hell, I’d even be satisfied with an rc script forcing the fan on at all times at this point. Does anyone at Nvidia have any suggestions on how to further troubleshoot this thing?
I contacted Nvidia customer service explaining my situation in hopes of reaching a technician or Linux developer, but after pondering about it for a couple of weeks they directed me to this forum, stating, "We're sorry, but we are not able to address Linux installation, usage and troubleshooting questions... Please refer to Linux support (http://www.nvnews.net/vbulletin/forumdisplay.php?s=&forumid=14)."
Is this an official Nvidia forum that is monitored by Nvidia developers and technicians? I was under the impression that it is an independently run forum.
Any help that anyone can offer would be appreciated. Let me know if you need (or want) any further info... (See attached nvidia-bug-report.log as well).
-Justin P. Linenkohl