Bug 11753
Summary: | (No Device to adust brightness) Setting brightness causes crash | ||
---|---|---|---|
Product: | Drivers | Reporter: | Thomas (SMA.TFG) |
Component: | Video(Other) | Assignee: | drivers_video-other |
Status: | REJECTED WILL_NOT_FIX | ||
Severity: | high | CC: | jake.white7485, rui.zhang, SMA.TFG |
Priority: | P1 | ||
Hardware: | All | ||
OS: | Linux | ||
Kernel Version: | 2.6.27-rc7 / still in 2.6.28_rc8 | Subsystem: | |
Regression: | --- | Bisected commit-id: | |
Bug Depends on: | |||
Bug Blocks: | 56331 | ||
Attachments: |
acpidump output
Nvidia-bug-report.log customized DSDT: removed some AML code that may cause the reboot |
Description
Thomas
2008-10-13 14:39:27 UTC
rechner1 backlight # ls /sys/class/backlight/ rechner1 backlight # # is also empty Reply-To: akpm@linux-foundation.org (switched to email. Please respond via emailed reply-to-all, not via the bugzilla web interface). > On Mon, 13 Oct 2008 14:39:31 -0700 (PDT) bugme-daemon@bugzilla.kernel.org > wrote: > http://bugzilla.kernel.org/show_bug.cgi?id=11753 I'm unsure whether this is an fbdev thing, a backlight thing or an acpi video thing. Could people please take a look? Thanks. >CONFIG_ACPI_VIDEO is not set
ACPI video driver is not built.
so thomas, please set this option and try again.
you should see "/proc/acpi/video/" if the video driver is loaded.
First of all thank you for the hint concerning of CONFIG_ACPI_VIDEO. This helped. thomas@rechner1~$ ls /proc/acpi/video/NVID/LCD/ brightness EDID info state If I do a cat state, Systemem reboots. Also if I try to set a supported value to brightness (echo 50 > brightness) --> System hard reboot cat of brightness tells me, that the levels 10 30 40 50 60 70 80 100 are supported and current: 100 In my opinion this could also be the Problem of my Bug Report 11711, but I don't now how to solve this. I can also grant a Kenel developer access to my System via SSH. If a Kernel developer need this access, feel free to contact me. Dear Mr. Morton, i have added a view information to the bugzilla thread. If you need an SSH access to my System, feel free to contact me for a appointment, when i should open a SSH port. I At the moment I updated Nvidia Driver to version 177.80 form Gentoo Portage tree. No change for the problem. Any further idas, how i can collect information, what causes the reboot if i do a echo 50 > /proc/acpi/video/NVID/LCD/brightness a cat at /proc/acpi/video/NVID/LCD/state causes also a reboot. Here is a listing of my proc/acpi/video thomas@rechner1 /proc/acpi/video $ ls -lRt .: insgesamt 0 dr-xr-xr-x 7 root root 0 18. Okt 15:24 GFX0 dr-xr-xr-x 6 root root 0 18. Okt 15:22 NVID ./GFX0: insgesamt 0 dr-xr-xr-x 2 root root 0 18. Okt 15:24 DD01 dr-xr-xr-x 2 root root 0 18. Okt 15:24 DD02 dr-xr-xr-x 2 root root 0 18. Okt 15:24 DD03 dr-xr-xr-x 2 root root 0 18. Okt 15:24 DD04 dr-xr-xr-x 2 root root 0 18. Okt 15:24 DD05 -rw-r--r-- 1 root root 0 18. Okt 15:24 DOS -r--r--r-- 1 root root 0 18. Okt 15:24 info -rw-r--r-- 1 root root 0 18. Okt 15:24 POST -r--r--r-- 1 root root 0 18. Okt 15:24 POST_info -r--r--r-- 1 root root 0 18. Okt 15:24 ROM ./GFX0/DD01: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:24 info -rw-r--r-- 1 root root 0 18. Okt 15:24 state ./GFX0/DD02: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:24 info -rw-r--r-- 1 root root 0 18. Okt 15:24 state ./GFX0/DD03: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:24 info -rw-r--r-- 1 root root 0 18. Okt 15:24 state ./GFX0/DD04: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:24 info -rw-r--r-- 1 root root 0 18. Okt 15:24 state ./GFX0/DD05: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:24 info -rw-r--r-- 1 root root 0 18. Okt 15:24 state ./NVID: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 DOS -r--r--r-- 1 root root 0 18. Okt 15:24 info -rw-r--r-- 1 root root 0 18. Okt 15:24 POST -r--r--r-- 1 root root 0 18. Okt 15:24 POST_info -r--r--r-- 1 root root 0 18. Okt 15:24 ROM dr-xr-xr-x 2 root root 0 18. Okt 15:22 CRT1 dr-xr-xr-x 2 root root 0 18. Okt 15:22 DVI dr-xr-xr-x 2 root root 0 18. Okt 15:22 LCD dr-xr-xr-x 2 root root 0 18. Okt 15:22 TV ./NVID/CRT1: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:22 info -rw-r--r-- 1 root root 0 18. Okt 15:22 state ./NVID/DVI: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:22 info -rw-r--r-- 1 root root 0 18. Okt 15:22 state ./NVID/LCD: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:22 info -rw-r--r-- 1 root root 0 18. Okt 15:22 state ./NVID/TV: insgesamt 0 -rw-r--r-- 1 root root 0 18. Okt 15:24 brightness -r--r--r-- 1 root root 0 18. Okt 15:24 EDID -r--r--r-- 1 root root 0 18. Okt 15:22 info -rw-r--r-- 1 root root 0 18. Okt 15:22 state New Info on this Bugreport please have a look at bugzilla Andrew Morton schrieb: > (switched to email. Please respond via emailed reply-to-all, not via the > bugzilla web interface). > > >> On Mon, 13 Oct 2008 14:39:31 -0700 (PDT) bugme-daemon@bugzilla.kernel.org >> wrote: >> http://bugzilla.kernel.org/show_bug.cgi?id=11753 >> > > I'm unsure whether this is an fbdev thing, a backlight thing or an acpi > video thing. Could people please take a look? > > Thanks. > > New thing recognized. If I write a Value to the brightness device, the System has the value, after the reboot. But System reboots and I have no idea what the reason for this problem is. Will you please attach the output of acpidump? Created attachment 18430 [details]
acpidump output
output of acpidump as root
If you need any further information or access to my system, feel free to contact me. I could open a SSH Console for you. Thank you for your help. Best wishes from Bavaria. Thanks for the acpidump info. From the acpidump info it seems that there exists two ACPI video devices on the laptop. One is intel integrated vidoe(GFX0) and another is external Nvidia video device. There is no _BCL/_BCM object understand the scope of GFX0. If the GFX0 ACPI video device is used, the brightness can't be changed by ACPI video interface. If the nvidia ACPI device is used, the brightness can be changed by ACPI video interface. Will you please confirm which graphics device is used? At the same time it is described in comment #8 that the system will be rebooted by cat /proc/acpi/video/NVID/LCD/state. Will you please try the following test and see whether the system will be rebooted? a. cat /proc/acpi/video/NVID/CRT1/state b. cat /proc/acpi/video/NVID/TV/state If the system is still rebooted, IMO the crash is related with the BIOS. When catting /proc/acpi/video/NVID/LCD/state the _DCS object will be evaluated, in which the SMI operation is executed. >Device (LCD) { > Name (_ADR, 0x0110) > Method (_DCS, 0, NotSerialized) { > STRP (0x30, One) //SMI operation Thanks. In my opinion, there should be only one Graphic Card. - The Nvidia. But I don't know, the Intel could be onboard, but isn't used if it exists. The Test: cat /proc/acpi/video/NVID/LCD/state --> reboot cat /proc/acpi/video/NVID/CRT1/state --> reboot cat /proc/acpi/video/NVID/TV/state --> reboot (tested one minute ago) Please have a look at Comment 10, I wrote to NVID/LCD, so I'm nearly sure, that the Nvidia Card is used. Thank you for your help. can you let Nvidia guys look at this? Linux ACPI just calls some AML code, and in you machine, the call will enter into SMM mode and all are done in BIOS. Do you know any person, witch I could look over this thread ? I posted a Info to the Nvidia Forum. http://www.nvnews.net/vbulletin/showthread.php?t=121410 Created attachment 18498 [details]
Nvidia-bug-report.log
Created attachment 18508 [details]
customized DSDT: removed some AML code that may cause the reboot
I remove the SMI call part, the method may not work any more but it should not bring any reboot either. will you please give it a try?
Note:
with the customized DSDT,
poking /proc/.../brightness should not reboot.
poking /proc/.../LCD(DVI)/state should not reboot
poking /proc/.../TV(CRT1)/state should reboot
Please can you give my short briefing how to apply this patch, because I can compile a kernel, but I don't know how to patch the kernel. How to use the custom DSDT can be found in http://www.lesswatts.org/projects/acpi/faq.php Thanks. cat /proc/acpi/video/NVID/LCD/state state: 0x1d query: 0x00 cd /proc/acpi/video/NVID/LCD/ cat brightness levels: 10 30 40 50 60 70 80 100 current: 100 echo 50 > brightness #No Reboot, but also no change of brightness -> cat tells now, that current is 50 cat /proc/acpi/video/NVID/DVI/state --> no reboot cat /proc/acpi/video/NVID/TV/state --> reboot cat /proc/acpi/video/NVID/CRT1/state --> reboot PS: The next test can be at Tuesday because I'm not at home the next 5 Days. First, as you have a nvidia graphics card on your laptop, I suggest you not use the ACPI video driver. And the test result in comment #32 shows that the computer reboots when executing an SMI call which is transparent to OS. So this is apparently a BIOS bug(buggy BIOS calls) to me. and this also means that the ACPI Backlight interface for nvidia card doesn't work. For this issue, I think you can 1. blacklist the ACPI video driver to prevent the computer from reboot 2. use some nvidia specific tools for backlight control.(I can not help you on this because I know little about nvidia) so I think this is really a "how to control backlight on nvidia platform" problem, right? I think we need some help from the nvidia guys. :) re-assign to the video driver category. OK thank you for you answer. Is this a bug in the Nvida Bios ? I hope, that this bug can be fixed. Because I like to control the backlight to save energy. But one further question, if this is a bug in the nvidia bios, how can the w**dows control the backlight wihout a crash. Could you send my a DSDT.hex file without all SMI callas ? Then I could test, if there is any influence to my other Bug report (11711) no, this is not a bug of Nvdia BIOS. there are some platform specific methods to control the backlight on Nvdia graphics. as the generic method (ACPI) doesn't work, we'd better try the Nvdia platform specific method. something like this: http://ubuntuforums.org/showthread.php?p=2702790 Is there any chance that the functionality for acpi will be included in one of the next kernels ? I tested smartdimmer, but this tool does not do anything on my laptop. :-( Do you have any other ideas. My preferred thing would be acpi (In reply to comment #26) > Is there any chance that the functionality for acpi will be included in one > of > the next kernels ? > which functionality? the backlight functionality is exported by BIOS, and OS can control the backlight *via* ACPI. so if the BIOS doesn't support backlight control via ACPI, I'm afraid we can do nothing here now. (In reply to comment #27) > I tested smartdimmer, but this tool does not do anything on my laptop. :-( > Do you have any other ideas. well, not yet. I'm not familiar with Nvidia, maybe someone here could help you. And I think you may also ask help in bugs.freedesktop.org (In reply to comment #28) > (In reply to comment #26) > > Is there any chance that the functionality for acpi will be included in one > of > > the next kernels ? > > > which functionality? > the backlight functionality is exported by BIOS, and OS can control the > backlight *via* ACPI. so if the BIOS doesn't support backlight control via > ACPI, I'm afraid we can do nothing here now. > > (In reply to comment #27) > > I tested smartdimmer, but this tool does not do anything on my laptop. :-( > > Do you have any other ideas. > > well, not yet. I'm not familiar with Nvidia, > maybe someone here could help you. > And I think you may also ask help in bugs.freedesktop.org > Hm, but on W*****s backlight is working correctly so I think, its not directly a Bios bug. Or how does W*****s control the Backlight. ACPI is a generic solution that can work on all platforms. W*****s can control the backlight via nvidia platform specific drivers. they don't need ACPI. :) and that's why I think this is a nvidia driver issue. :p But if I control the Backlight on the Grub Boot prompt, with the Keys on my Laptop, no W****s driver is loaded. Can I tell the Kernel not to capture the FN Keyborad codes ? Maybe I'am able to controll the Backlight via FN + Brightnes in Linux if the Kernel isn't working against this. If I set acpi=off apm=off at Grub, I can controll backlight. Without any reboot. If I boot Linux Kernel and the boot process has passed IO Scheduler System Crashes If I try to controll the Backlight |