Closed amitarvind closed 10 years ago
actual url to tarball instead of my goofy comment on the bugtracker: https://bugs.launchpad.net/lpbugreporter/+bug/752542/+attachment/3946298/+files/Dell_Inc.-Inspiron_7720.tar.gz
Either you guys are very busy or I'm missing something from this. If you could, please let me know what I'm missing from my info and I can supply it. Thanks.
@Lekensteyn is indeed very busy with exams, and I have no idea of what is wrong here.
Yes, I saw this issue and anticipated that it would take some time. However, seeing the line "nvidia 01:00.0" just below "bbswitch" immediately reveals what is wrong: nvidia should not be loaded when bbswitch turned the device off. You must blacklist nvidia with kmod to ensure that it does not get loaded automatically (e.g. while bbswitch has disabled the device).
The results seem promising, actually. I'm a little surprised because I can't remember needing to blacklist nvidia. Either way, before testing, the output of dmesg:
... [ 27.333427] bbswitch: version 0.8 [ 27.333436] bbswitch: Found integrated VGA device 0000:00:02.0: SB.PCI0.GFX0 [ 27.333441] bbswitch: Found discrete VGA device 0000:01:00.0: SB.PCI0.PEG0.PEGP [ 27.333453] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 27.333787] bbswitch: detected an Optimus _DSM function [ 27.333800] pci 0000:01:00.0: enabling device (0006 -> 0007) [ 27.333845] bbswitch: Succesfully loaded. Discrete card 0000:01:00.0 is on [ 27.336466] bbswitch: disabling discrete graphics [ 27.336476] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 40.317159] bbswitch: enabling discrete graphics [ 41.730885] nvidia: module license 'NVIDIA' taints kernel. [ 41.730889] Disabling lock debugging due to kernel taint [ 41.735710] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=io+mem,decodes=none:owns=none [ 41.735832] [drm] Initialized nvidia-drm 0.0.0 20130102 for 0000:01:00.0 on minor 1 [ 41.735838] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 331.38 Wed Jan 8 19:32:30 PST 2014 [ 45.037535] nvidia 0000:01:00.0: irq 47 for MSI/MSI-X [ 45.042859] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 45.042895] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 45.042908] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 45.042920] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 45.042931] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 45.043237] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 45.043382] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 45.043394] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 49.295105] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 49.320392] NVRM: GPU at 0000:01:00.0 has fallen off the bus. [ 49.320397] NVRM: os_pci_init_handle: invalid context! [ 49.320399] NVRM: os_pci_init_handle: invalid context! [ 49.320403] NVRM: GPU at 0000:01:00.0 has fallen off the bus. [ 49.320405] NVRM: os_pci_init_handle: invalid context! [ 49.320406] NVRM: os_pci_init_handle: invalid context! [ 49.334919] NVRM: RmInitAdapter failed! (0x25:0x28:1156) [ 49.334926] NVRM: rm_init_adapter failed for device bearing minor number 0 [ 49.334941] NVRM: nvidia_frontend_open: minor 0, module->open() failed, error -5 [ 49.335282] NVRM: request_irq() failed (-22) [ 49.335285] NVRM: nvidia_frontend_open: minor 0, module->open() failed, error -22
$ optirun cat /proc/acpi/bbswitch [ 283.930317] [ERROR]Cannot access secondary GPU - error: XORG NVIDIA(GPU-0): Failed to initialize the NVIDIA GPU at PCI:1:0:0. Please
[ 283.930404] [ERROR]Aborting because fallback start is disabled.
... [ 66.510212] bbswitch: enabling discrete graphics [ 66.522502] nvidia 0000:01:00.0: Refused to change power state, currently in D3
So, back to square one again.
The "fallen off the bus" etc. is probably due to nvidia - kernel 3.10 incompatibility. See https://github.com/Bumblebee-Project/Bumblebee/issues/455
Yeah, I saw that particular ticket but the last I saw of it, it hadn't been reopened yet. If I'm understanding the other ticket correctly, you're not necessarily suggesting a rollback to pre-3.10, right?
It seems worth mentioning that past 3.10, I was still running the 325.15 driver, and it worked for a bit, and then stopped. Nvidia's compatibility issues with our kernels makes my head hurt.
That ticked has always been opened, except for a few hours a month ago.
The workaround described in the first comment there seems to work for many people.
Yeah, you can see in my grub cmdline cfg, I tried it. But! Not to worry, I finally figured out the problem. See here: http://forums.gentoo.org/viewtopic-t-935316-start-0.html . Look at the post timestamped Wed Sep 19, 2012 9:37 pm.
I'm still on 3.12.7 but I recompiled with the following options compiled: CONFIG_NO_HZ CONFIG_RCU_FAST_NO_HZ CONFIG_CALGARY_IOMMU CONFIG_CALGARY_IOMMU_ENABLED_BY_DEFAULT
... 23.526664] bbswitch: version 0.8 [ 23.526673] bbswitch: Found integrated VGA device 0000:00:02.0: SB.PCI0.GFX0 [ 23.526680] bbswitch: Found discrete VGA device 0000:01:00.0: SB.PCI0.PEG0.PEGP [ 23.526691] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 23.527053] bbswitch: detected an Optimus _DSM function [ 23.527066] pci 0000:01:00.0: enabling device (0006 -> 0007) [ 23.527110] bbswitch: Succesfully loaded. Discrete card 0000:01:00.0 is on [ 23.529386] bbswitch: disabling discrete graphics [ 23.529394] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package [ 37.299238] bbswitch: enabling discrete graphics [ 38.190770] nvidia: module license 'NVIDIA' taints kernel. [ 38.190773] Disabling lock debugging due to kernel taint [ 38.195798] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=io+mem,decodes=none:owns=none [ 38.195928] [drm] Initialized nvidia-drm 0.0.0 20130102 for 0000:01:00.0 on minor 1 [ 38.195933] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 331.38 Wed Jan 8 19:32:30 PST 2014 [ 38.751129] nvidia 0000:01:00.0: irq 47 for MSI/MSI-X [ 38.756391] ACPI Warning: SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires Package
That last line likes to spam a lot, but that's a known nuisance and frankly, a pet peeve I have with Dell's ACPI coding practices.
It works!
$ optirun glxspheres Polygons in scene: 62464 Visual ID of window: 0x20 Context is Direct OpenGL Renderer: GeForce GT 650M/PCIe/SSE2 253.578307 frames/sec - 282.993391 Mpixels/sec 288.744940 frames/sec - 322.239353 Mpixels/sec 255.175457 frames/sec - 345.011200 Mpixels/sec 191.765979 frames/sec - 379.236401 Mpixels/sec
This used to work before for this laptop, so I'm not sure at what point it changed. I'm only just now getting time to work on the issue and tried upgrading the kernel and the driver today. /proc/acpi/bbswitch shows the gpu as OFF and it refuses to turn on, showing the message: nvidia 0000:01:00.0: Refused to change power state, currently in D3
sending "_PS0" to acpi_call seems to go through without error but never successfully turns the card on. All the requested info is below.
URL to requested tarball: https://bugs.launchpad.net/lpbugreporter/+bug/752542/comments/675