Summary: | [HSW] power domain refcount underrun | ||
---|---|---|---|
Product: | DRI | Reporter: | Jonathan McDowell <noodles> |
Component: | DRM/Intel | Assignee: | Intel GFX Bugs mailing list <intel-gfx-bugs> |
Status: | CLOSED FIXED | QA Contact: | Intel GFX Bugs mailing list <intel-gfx-bugs> |
Severity: | normal | ||
Priority: | medium | CC: | intel-gfx-bugs |
Version: | unspecified | ||
Hardware: | x86-64 (AMD64) | ||
OS: | Linux (All) | ||
Whiteboard: | |||
i915 platform: | i915 features: |
Description
Jonathan McDowell
2014-11-25 13:52:45 UTC
It's an inbalance in the power domain reference counts. Once we managed to get there only a report will stop these. We have a few other reports, but thus far no one has a good way to quickly reproduce these. Which is what we need to be able to understand where that refcount goes amiss - all the backtraces here are way past the point of the failure unfortunately. So if you could try to figure out what sequence is required to get there (really needs some good luck I guess) that might help. If you manage to reproduce the failure, could you dump /sys/kernel/debug/dri/0/i915_power_domain_info? this will have some useful info to confine the problem within a certain area. I failed to see a reproduction all week with a fresh boot into 3.18-rc6; I have rebooted today into 3.18-rc7 and will continue to keep an eye out, grabbing that file if/when I see the problem again. I didn't manage to catch this at the first occurrence but hopefully it's still helpful: Power well/domain Use count always-on -2 PIPE_A 1 TRANSCODER_EDP 1 PORT_DDI_A_2_LANES 0 PORT_DDI_A_4_LANES -4 PORT_DDI_B_2_LANES 0 PORT_DDI_B_4_LANES 0 PORT_DDI_C_2_LANES 0 PORT_DDI_C_4_LANES 0 PORT_DDI_D_2_LANES 0 PORT_DDI_D_4_LANES 0 PORT_CRT 0 PLLS 0 INIT 0 display 1 PIPE_B 0 PIPE_C 0 PIPE_A_PANEL_FITTER 0 PIPE_B_PANEL_FITTER 0 PIPE_C_PANEL_FITTER 0 TRANSCODER_A 0 TRANSCODER_B 0 TRANSCODER_C 0 PORT_DSI 0 PORT_OTHER 0 VGA 0 AUDIO 1 INIT 0 Still an issue with current bits? The power domain handling has seen some fixes since then iirc. I haven't seen this in over a month (that's as far back as logs go). Currently running 4.0-rc3 and seeing a warning in intel_check_page_flip() but it looks like that's fixed by 6c51d46f135b00c00373fcd029786ccef2b02b5b so I will update to 4.0-rc6. Feel free to close this and I can re-open if I see a power_put problem again? Thanks for the report, don't hesitate to reopen if the problem reappears. |
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.