| Summary: | Virtualbox with 3D acceleration and with installed vbox-additions crashes | ||
|---|---|---|---|
| Product: | Mesa | Reporter: | Tomasz Paweł Gajc <tpgxyz> |
| Component: | Drivers/DRI/nouveau | Assignee: | Nouveau Project <nouveau> |
| Status: | RESOLVED MOVED | QA Contact: | Nouveau Project <nouveau> |
| Severity: | normal | ||
| Priority: | medium | ||
| Version: | 10.6 | ||
| Hardware: | Other | ||
| OS: | All | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Bug Depends on: | 92077 | ||
| Bug Blocks: | |||
|
Description
Tomasz Paweł Gajc
2015-10-17 23:35:42 UTC
I'm running this with quite old GeForce 9800 GT Does vbox have multiple contexts concurrently active in multiple threads? That's the only thing I can think of that would cause these types of errors. If you search around, the 406040 thing is something that's been plaguing nv50 for quite some time, with no progress made on figuring out how an invalid command is making it into the fifo stream. (In reply to Ilia Mirkin from comment #2) > Does vbox have multiple contexts concurrently active in multiple threads? Ilia it is hard for me to give you and answer, because simply i do not understand what you are asking me for. How can i check this ? I'm still having this issue with mesa-10.6.9 and kernel-4.1.12 and libdrm-2.4.65 with vbox-5.0.10 I'm trying to run Plasma 5 inside vbox. Nowadays Plasma 5 and SDDM uses heavily OGL features throug Qt5Quick. Tricks like QT_XCB_FORCE_SOFTWARE_OPENGL=1 or LIBGL_ALWAYS_SOFTWARE=1 gives black screen. Below is the output from host system when , in start Plasma 5 inside vbox with 3D accel enabled. gru 13 22:20:12 lazur kernel: vboxdrv: ffffffffc087c020 VMMR0.r0 gru 13 22:20:12 lazur kernel: vboxdrv: ffffffffc0975020 VBoxDDR0.r0 gru 13 22:20:12 lazur kernel: vboxdrv: ffffffffc007d020 VBoxDD2R0.r0 gru 13 22:20:12 lazur kernel: PS2K: Selected scan set 2 gru 13 22:20:21 lazur kernel: HPET: Using timer above configured range: 3 gru 13 22:20:21 lazur kernel: HPET: Using timer above configured range: 3 gru 13 22:20:21 lazur kernel: HPET: Using timer above configured range: 3 gru 13 22:24:08 lazur kernel: nouveau E[ PFIFO][0000:01:00.0] DMA_PUSHER - ch 6 [VBoxCrWinCmd[26409]] get 0x002001cf60 put 0x002001d6f4 ib_get 0x00000045 ib_put 0x00000046 state 0x40000004 (... push 0x00406040 gru 13 22:24:08 lazur kernel: nouveau E[ PGRAPH][0000:01:00.0] DATA_ERROR INVALID_VALUE gru 13 22:24:08 lazur kernel: nouveau E[ PGRAPH][0000:01:00.0] ch 6 [0x001f5dd000 VBoxCrWinCmd[26409]] subc 4 class 0x502d mthd 0x0214 data 0x00086e04 gru 13 22:24:08 lazur kernel: nouveau E[ PGRAPH][0000:01:00.0] DATA_ERROR INVALID_VALUE gru 13 22:24:08 lazur kernel: nouveau E[ PGRAPH][0000:01:00.0] ch 6 [0x001f5dd000 VBoxCrWinCmd[26409]] subc 4 class 0x502d mthd 0x0218 data 0x02800000 gru 13 22:24:08 lazur kernel: nouveau E[ PGRAPH][0000:01:00.0] DATA_ERROR INVALID_VALUE gru 13 22:24:08 lazur kernel: nouveau E[ PGRAPH][0000:01:00.0] ch 6 [0x001f5dd000 VBoxCrWinCmd[26409]] subc 4 class 0x502d mthd 0x021c data 0x01e00000 gru 13 22:24:08 lazur kernel: nouveau E[ PGRAPH][0000:01:00.0] DATA_ERROR INVALID_BITFIELD gru 13 22:24:08 lazur kernel: nouveau E[ PGRAPH][0000:01:00.0] ch 6 [0x001f5dd000 VBoxCrWinCmd[26409]] subc 4 class 0x502d mthd 0x0220 data 0x00086e14 gru 13 22:24:08 lazur kernel: nouveau E[ PFIFO][0000:01:00.0] CACHE_ERROR - ch 6 [VBoxCrWinCmd[26409]] subc 0 mthd 0x0000 data 0x00086e24 gru 13 22:26:05 lazur kernel: vboxdrv: ffffffffc0993020 VMMR0.r0 gru 13 22:26:06 lazur kernel: vboxdrv: ffffffffc0a8c020 VBoxDDR0.r0 gru 13 22:26:06 lazur kernel: vboxdrv: ffffffffc0098020 VBoxDD2R0.r0 I think it's been pretty much established that vbox uses opengl from multiple threads at the same time. This is likely to lead to crashes with nouveau... no amount of upgrading will fix this. [However it should be noted that 10.6 series was end-of-lifed a long time ago... 11.0.x series is the supported one, with 11.1 coming out soon.] Resolving this issue is very low on my todo list. Right now I'm aware of two applications this affects -- vbox, a proprietary closed source app which includes a kernel module that taints the kernel, and warsow 2.0, which just came out, and includes a workaround to disable the multithreading. And this wouldn't matter, but it's also a giant pain to fix in a way that doesn't break perf for 99.9999% of the users. Looks like this is related to broken multi-threading in nouveau, see linked bugs. -- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/1088. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.