[Intel-gfx] sysfs: i915_cur_delayinfo is always twice as gt_cur_freq_mhz in 3.8.13.4?

[Ben Widawsky]

On Mon, Aug 26, 2013 at 09:45:50AM +0000, Cui, Dexuan wrote:
> Hi all, I'm using haswell with linux-3.8.13.4:
> http://kernel.ubuntu.com/git?p=ubuntu/linux.git;a=commitdiff;h=d1baa1360260b5a01938674fc518109a4e5a148d
> and I find the CAGF showed by i915_cur_delayinfo is always twice as gt_cur_freq_mhz, e.g., when the system is idle, I can get the below log:
> 
> ----------------------------- Log begins
> # cat /sys/kernel/debug/dri/0/i915_cur_delayinfo; echo;  cat /sys/devices/pci0000:00/0000:00:02.0/drm/card0/gt_cur_freq_mhz; echo ;
> GT_PERF_STATUS: 0x00000800
> RPSTAT1: 0x00040400
> Render p-state ratio: 8
> Render p-state VID: 0
> Render p-state limit: 255
> CAGF: 400MHz
> RP CUR UP EI: 55890us
> RP CUR UP: 276us
> RP PREV UP: 0us
> RP CUR DOWN EI: 94295us
> RP CUR DOWN: 389us
> RP PREV DOWN: 0us
> Lowest (RPN) frequency: 200MHz
> Nominal (RP1) frequency: 400MHz
> Max non-overclocked (RP0) frequency: 1200MHz
> 
> 200
> ----------------------------- Log ends. 
> 
> But with the latest 3.10.0, both values are 200, as expected.
> 
> I guess there is a fix in 3.10.0 I can pick for my 3.8.13.4?
> 
> Thanks,
> -- Dexuan
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx


Please don't use i915_cur_delayinfo unless you have good reason.
However, this is likely what you need:

commit f82855d342b6c8483c56e6d2e200a71731509a39
Author: Ben Widawsky <ben at bwidawsk.net>
Date:   Tue Jan 29 12:00:15 2013 -0800

    drm/i915: Fix CAGF for HSW


-- 
Ben Widawsky, Intel Open Source Technology Center