| Summary: | [BAT][HSW] igt@tools_test@tools_test - Unclaimed read from register 0x[4c]400c | ||
|---|---|---|---|
| Product: | DRI | Reporter: | Martin Peres <martin.peres> |
| Component: | IGT | Assignee: | Default DRI bug account <dri-devel> |
| Status: | CLOSED FIXED | QA Contact: | |
| Severity: | critical | ||
| Priority: | highest | CC: | intel-gfx-bugs |
| Version: | XOrg git | ||
| Hardware: | Other | ||
| OS: | All | ||
| Whiteboard: | ReadyForDev | ||
| i915 platform: | HSW | i915 features: | |
|
Description
Martin Peres
2017-09-05 09:27:03 UTC
That looks entirely to be a test bug, intel_reg_read reading unknown registers which is then caught by the kernel on its next mmio. Since the kernel is not aware of a third party messing around with registers, it takes the blame upon itself. The test has been passing recently. Has anything changed? From our pov, no. The highlighted commit was just the merge to 4.13. But the failure does depend upon i915 doing something during the test, but the test itself bypasses i915 for direct hw access. What we could do is to disable automatic mmio checking if user forcewake is taken. Then clear any errors upon release. commit d7a133d886b45651e36e7065998b1413d379ac1f Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Thu Sep 7 14:44:41 2017 +0100 drm/i915: Disable mmio debugging during user access If the user bypasses i915 and accesses mmio directly, that easily confuses our automatic mmio debugging (any error we then detect is likely to be as a result of the user). Since we expect userspace to open debugfs/i915_forcewake_user if i915.ko is loaded and they want mmio access, that makes the opportune time to disable our debugging for duration of the bypass. v2: Move the fiddling of uncore internals to uncore.c The issue in intel_reg_dump is still there, just not triggering a kernel warning. Thanks Chris! I will close this bug. What about https://bugs.freedesktop.org/show_bug.cgi?id=102249 though? Could the patch you sent fix this too? (In reply to Martin Peres from comment #6) > Thanks Chris! I will close this bug. > > What about https://bugs.freedesktop.org/show_bug.cgi?id=102249 though? Could > the patch you sent fix this too? It's not intended to have any effect for any test except for tools_test; but drv_suspend/forcewake, pm_rpm/debugfs-forcewake-user, gem_exec_latency, gem_exec_parse, gem_workarounds and pm_lsps also poke around with registers under forcewake - any may end up polluting the mmio debugger. #102249 looks clear of outside influence. |
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.