Summary: | radeon_pm_info should be in sysfs instead of debugfs | ||
---|---|---|---|
Product: | DRI | Reporter: | almos <aaalmosss> |
Component: | DRM/Radeon | Assignee: | Default DRI bug account <dri-devel> |
Status: | RESOLVED MOVED | QA Contact: | |
Severity: | normal | ||
Priority: | medium | CC: | alexandre.f.demers |
Version: | XOrg git | ||
Hardware: | Other | ||
OS: | All | ||
Whiteboard: | |||
i915 platform: | i915 features: |
Description
almos
2013-12-26 14:25:12 UTC
radeon_pm_info is for debugging. OK, then how can I get information about the clock speeds of the GPU when not debugging? I guess the same thing applies to vgaswitcheroo. I'm not a kernel developer, but I'm quite sure these are not just for debugging. We had the discussion a couple of times now. The power management infos radeon_pm_info provides are definitely only for debugging. (In reply to comment #4) > We had the discussion a couple of times now. The power management infos > radeon_pm_info provides are definitely only for debugging. Are the minutes of those discussions available somewhere? I'm curious about the arguments. BTW have you considered moving /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq to debugfs for consistency's sake? How about removing lm-sensors support? These are definitely for debugging only. (In reply to comment #5) > Are the minutes of those discussions available somewhere? I'm curious about > the arguments. radeon_pm_info is not the real hardware state, but rather the state the kernel driver thinks the hardware is in. So it's actually not very useful to an end user, except if you want to debug the different states the kernel driver can be into (UVD/performance/battery etc...) I agree that we need a proper sysfs interface to make certain PM functions and informations available to userspace, but it's definitely not what's inside radeon_pm_info. I'm about to close this bug since, as Christian said, there should be a way to expose some PM functions to the user, but not through radeon_pm_info. But before, I'd like to know if there is something coming this way to expose these functionality? I'd like the RADEON_INFO ioctl to be able to return the clocks. (In reply to comment #8) > I'd like the RADEON_INFO ioctl to be able to return the clocks. That sounds like a pretty good idea to me and we should actually query the hardware for the clocks and return that instead of the software state. I'd like to have RADEON_INFO support for these: - engine and memory clocks - temperatures - status regs (for radeontop without root and Gallium HUD) Just send out a patch set to implement this: http://lists.freedesktop.org/archives/dri-devel/2014-October/069410.html -- 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/drm/amd/issues/420. |
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.