[Intel-gfx] [PATCH v2] drm/i915: Don't enable IPS when pixel rate exceeds 95% of cdclk
Paulo Zanoni
przanoni at gmail.com
Wed May 20 13:45:50 PDT 2015
2015-05-20 17:40 GMT-03:00 Rodrigo Vivi <rodrigo.vivi at gmail.com>:
> I'm convinced Unclaimed regs Paulo saw came from somewhere else since
> I just got it with pm_rpm rte with no patch applied....
>
> So I'd suggest to merge this patch... Although it also doesn't fix the
> flicker I see here so the bug would still be open...
>
> Anyway,
> Reviewed-by: Rodrigo Vivi <rodrigo.vivi at intel.com>
Nope. The unclaimed register error you see is a recent regression. It
didn't happen back in 2014 when I tested the IPS patch. When I report
regressions like this to the mailing list I always test the revert, so
at that time, with that state on drm-intel-nightly, that patch was
certainly introducing the regression. Maybe something changed now? Did
you retest?
Also, which frequency value are you currently reading? The FCLK one?
If it's the FCLK frequency, it's not really what you want.
>
> On Thu, Sep 18, 2014 at 4:56 AM, Jani Nikula
> <jani.nikula at linux.intel.com> wrote:
>> On Wed, 17 Sep 2014, Ville Syrjälä <ville.syrjala at linux.intel.com> wrote:
>>> On Wed, Sep 17, 2014 at 11:10:48AM -0300, Paulo Zanoni wrote:
>>>> 2014-09-17 11:05 GMT-03:00 Paulo Zanoni <przanoni at gmail.com>:
>>>> > 2014-09-15 4:22 GMT-03:00 Jani Nikula <jani.nikula at linux.intel.com>:
>>>> >> On Fri, 12 Sep 2014, Ville Syrjälä <ville.syrjala at linux.intel.com> wrote:
>>>> >>> On Fri, Sep 12, 2014 at 04:42:33PM +0100, Chris Wilson wrote:
>>>> >>>> On Fri, Sep 12, 2014 at 05:01:57PM +0300, ville.syrjala at linux.intel.com wrote:
>>>> >>>> > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
>>>> >>>> >
>>>> >>>> > Bspec says we shouldn't enable IPS on BDW when the pipe pixel rate
>>>> >>>> > exceeds 95% of the core display clock. Apparently this can cause
>>>> >>>> > underruns.
>>>> >>>> >
>>>> >>>> > There's no similar restriction listed for HSW, so leave that one alone
>>>> >>>> > for now.
>>>> >>>> >
>>>> >>>> > v2: Add pipe_config_supports_ips() (Chris)
>>>> >>>> >
>>>> >>>> > Tested-by: Timo Aaltonen <tjaalton at ubuntu.com>
>>>> >>>> > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=83497
>>>> >>>> > Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
>>>> >>>> > ---
>>>> >>>> > drivers/gpu/drm/i915/intel_display.c | 21 +++++++++++++++++++--
>>>> >>>> > drivers/gpu/drm/i915/intel_drv.h | 1 +
>>>> >>>> > drivers/gpu/drm/i915/intel_pm.c | 16 +++++++---------
>>>> >>>> > 3 files changed, 27 insertions(+), 11 deletions(-)
>>>> >>>> >
>>>> >>>> > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
>>>> >>>> > index 965eb3c..7809177 100644
>>>> >>>> > --- a/drivers/gpu/drm/i915/intel_display.c
>>>> >>>> > +++ b/drivers/gpu/drm/i915/intel_display.c
>>>> >>>> > @@ -5241,12 +5241,29 @@ retry:
>>>> >>>> > return setup_ok ? 0 : -EINVAL;
>>>> >>>> > }
>>>> >>>> >
>>>> >>>> > +static bool pipe_config_supports_ips(struct drm_i915_private *dev_priv,
>>>> >>>> > + struct intel_crtc_config *pipe_config)
>>>> >>>> > +{
>>>> >>>> > + if (pipe_config->pipe_bpp > 24)
>>>> >>>> > + return false;
>>>> >>>> > +
>>>> >>>> > + /* HSW can handle pixel rate up to cdclk? */
>>>> >>>> > + if (IS_HASWELL(dev_priv->dev))
>>>> >>>>
>>>> >>>> This only needs IS_HASWELL(dev_priv)
>>>> >>>
>>>> >>> old habits...
>>>> >>
>>>> >> Thanks to your old habits I didn't have to make any changes when pushing
>>>> >> this to drm-intel-fixes which is still old habit land. Thanks for the
>>>> >> patch and review.
>>>>
>>>> And another detail: even if the register read worked, we'd have
>>>> printed a bad value since the cdclock is disabled at the moment we run
>>>> the function, so we'd return the FCLK frequency. So maybe we need to
>>>> cache the value of the "cdclk under normal display operation" and
>>>> reuse it here.
>>>
>>> Yeah. We already do that on VLV/CHV to some extent. We should just do
>>> it for all platforms. Although for this case what we'd need to do is
>>> precalculate the optimal cdclk during compute_config phase and use
>>> that value for these checks, and then during global_resources we
>>> actually change the cdclk to match the precomputed value.
>>>
>>> Also if someone takes this on they could also unify the ddi cdclk
>>> code to look more like the other platforms, and perhaps even implement
>>> support for changing the cdclk frequency on HSW/BDW ;)
>>
>> I've dropped the commit from drm-intel-fixes and reopened the bug. We
>> need someone(tm) to look at this now.
>>
>> BR,
>> Jani.
>>
>>
>>
>>>
>>>>
>>>> >
>>>> > This patches causes a regression on igt/pm_rpm/rte:
>>>> >
>>>> > [ 44.871662] [drm:intel_dp_compute_config] DP link bw required
>>>> > 650358 available 1728000
>>>> > [ 44.871666] ------------[ cut here ]------------
>>>> > [ 44.871691] WARNING: CPU: 0 PID: 1745 at
>>>> > drivers/gpu/drm/i915/intel_uncore.c:47
>>>> > assert_device_not_suspended.isra.8+0x43/0x50 [i915]()
>>>> > [ 44.871692] Device suspended
>>>> > [ 44.871694] Modules linked in: hid_multitouch hid_sensor_hub
>>>> > snd_hda_codec_hdmi intel_rapl x86_pkg_temp_thermal intel_powerclamp
>>>> > efivars snd_hda_intel btusb snd_hda_controller snd_hda_codec snd_hwdep
>>>> > iwlmvm snd_pcm_oss snd_mixer_oss iwlwifi mei_me mei snd_pcm snd_timer
>>>> > int3403_thermal i2c_designware_platform i2c_designware_core acpi_pad
>>>> > fuse nls_utf8 nls_cp437 vfat fat i915 sdhci_pci drm_kms_helper e1000e
>>>> > drm sdhci_acpi sdhci
>>>> > [ 44.871726] CPU: 0 PID: 1745 Comm: pm_rpm Not tainted
>>>> > 3.17.0-rc5.1409171019pz+ #1120
>>>> > [ 44.871728] Hardware name: Intel Corporation Broadwell Client
>>>> > platform/Wilson Beach SDS, BIOS BDW-E2R1.86C.0072.R03.1405072127
>>>> > 05/07/2014
>>>> > [ 44.871730] 0000000000000009 ffff8800aa763a90 ffffffff816ec803
>>>> > ffff8800aa763ad8
>>>> > [ 44.871735] ffff8800aa763ac8 ffffffff8107a0c8 ffff88023ec20068
>>>> > 0000000000130040
>>>> > [ 44.871739] ffff88023f12b3d8 0000000000130040 ffff88023ec20000
>>>> > ffff8800aa763b28
>>>> > [ 44.871743] Call Trace:
>>>> > [ 44.871749] [<ffffffff816ec803>] dump_stack+0x4d/0x66
>>>> > [ 44.871754] [<ffffffff8107a0c8>] warn_slowpath_common+0x78/0xa0
>>>> > [ 44.871758] [<ffffffff8107a137>] warn_slowpath_fmt+0x47/0x50
>>>> > [ 44.871783] [<ffffffffa015a1f4>] ?
>>>> > intel_dp_compute_config+0x274/0x600 [i915]
>>>> > [ 44.871803] [<ffffffffa0122cf3>]
>>>> > assert_device_not_suspended.isra.8+0x43/0x50 [i915]
>>>> > [ 44.871822] [<ffffffffa0125a20>] gen6_read32+0x30/0x150 [i915]
>>>> > [ 44.871844] [<ffffffffa0154636>] intel_ddi_get_cdclk_freq+0xd6/0x140 [i915]
>>>> > [ 44.871865] [<ffffffffa0139a92>] __intel_set_mode+0x1672/0x1750 [i915]
>>>> > [ 44.871886] [<ffffffffa01418e1>] intel_set_mode+0x11/0x30 [i915]
>>>> > [ 44.871906] [<ffffffffa01429ed>] intel_crtc_set_config+0xa9d/0xeb0 [i915]
>>>> > [ 44.871923] [<ffffffffa0026283>]
>>>> > drm_mode_set_config_internal+0x63/0x100 [drm]
>>>> > [ 44.871938] [<ffffffffa002a7e3>] drm_mode_setcrtc+0x283/0x580 [drm]
>>>> > [ 44.871948] [<ffffffffa001c81f>] drm_ioctl+0x1df/0x6a0 [drm]
>>>> > [ 44.871954] [<ffffffff811e07b0>] do_vfs_ioctl+0x2e0/0x4e0
>>>> > [ 44.871957] [<ffffffff816f6af7>] ? sysret_check+0x1b/0x56
>>>> > [ 44.871962] [<ffffffff810bfb0d>] ? trace_hardirqs_on_caller+0x15d/0x200
>>>> > [ 44.871966] [<ffffffff811e0a31>] SyS_ioctl+0x81/0xa0
>>>> > [ 44.871969] [<ffffffff816f6ad2>] system_call_fastpath+0x16/0x1b
>>>> > [ 44.871972] ---[ end trace 91bf8920a2c6b824 ]---
>>>> > [ 44.871975] ------------[ cut here ]------------
>>>> > [ 44.871994] WARNING: CPU: 0 PID: 1745 at
>>>> > drivers/gpu/drm/i915/intel_uncore.c:528
>>>> > hsw_unclaimed_reg_debug+0x6d/0x80 [i915]()
>>>> > [ 44.871996] Unclaimed register detected before reading register 0x130040
>>>> > [ 44.871997] Modules linked in: hid_multitouch hid_sensor_hub
>>>> > snd_hda_codec_hdmi intel_rapl x86_pkg_temp_thermal intel_powerclamp
>>>> > efivars snd_hda_intel btusb snd_hda_controller snd_hda_codec snd_hwdep
>>>> > iwlmvm snd_pcm_oss snd_mixer_oss iwlwifi mei_me mei snd_pcm snd_timer
>>>> > int3403_thermal i2c_designware_platform i2c_designware_core acpi_pad
>>>> > fuse nls_utf8 nls_cp437 vfat fat i915 sdhci_pci drm_kms_helper e1000e
>>>> > drm sdhci_acpi sdhci
>>>> > [ 44.872025] CPU: 0 PID: 1745 Comm: pm_rpm Tainted: G W
>>>> > 3.17.0-rc5.1409171019pz+ #1120
>>>> > [ 44.872026] Hardware name: Intel Corporation Broadwell Client
>>>> > platform/Wilson Beach SDS, BIOS BDW-E2R1.86C.0072.R03.1405072127
>>>> > 05/07/2014
>>>> > [ 44.872028] 0000000000000009 ffff8800aa763a88 ffffffff816ec803
>>>> > ffff8800aa763ad0
>>>> > [ 44.872032] ffff8800aa763ac0 ffffffff8107a0c8 ffff88023ec20000
>>>> > 0000000000130040
>>>> > [ 44.872035] ffff88023f12b3d8 0000000000130040 ffff88023ec20000
>>>> > ffff8800aa763b20
>>>> > [ 44.872039] Call Trace:
>>>> > [ 44.872042] [<ffffffff816ec803>] dump_stack+0x4d/0x66
>>>> > [ 44.872046] [<ffffffff8107a0c8>] warn_slowpath_common+0x78/0xa0
>>>> > [ 44.872049] [<ffffffff8107a137>] warn_slowpath_fmt+0x47/0x50
>>>> > [ 44.872068] [<ffffffffa0122bbd>] hsw_unclaimed_reg_debug+0x6d/0x80 [i915]
>>>> > [ 44.872087] [<ffffffffa0125a41>] gen6_read32+0x51/0x150 [i915]
>>>> > [ 44.872109] [<ffffffffa0154636>] intel_ddi_get_cdclk_freq+0xd6/0x140 [i915]
>>>> > [ 44.872129] [<ffffffffa0139a92>] __intel_set_mode+0x1672/0x1750 [i915]
>>>> > [ 44.872149] [<ffffffffa01418e1>] intel_set_mode+0x11/0x30 [i915]
>>>> > [ 44.872168] [<ffffffffa01429ed>] intel_crtc_set_config+0xa9d/0xeb0 [i915]
>>>> > [ 44.872182] [<ffffffffa0026283>]
>>>> > drm_mode_set_config_internal+0x63/0x100 [drm]
>>>> > [ 44.872196] [<ffffffffa002a7e3>] drm_mode_setcrtc+0x283/0x580 [drm]
>>>> > [ 44.872205] [<ffffffffa001c81f>] drm_ioctl+0x1df/0x6a0 [drm]
>>>> > [ 44.872210] [<ffffffff811e07b0>] do_vfs_ioctl+0x2e0/0x4e0
>>>> > [ 44.872213] [<ffffffff816f6af7>] ? sysret_check+0x1b/0x56
>>>> > [ 44.872218] [<ffffffff810bfb0d>] ? trace_hardirqs_on_caller+0x15d/0x200
>>>> > [ 44.872221] [<ffffffff811e0a31>] SyS_ioctl+0x81/0xa0
>>>> > [ 44.872224] [<ffffffff816f6ad2>] system_call_fastpath+0x16/0x1b
>>>> > [ 44.872226] ---[ end trace 91bf8920a2c6b825 ]---
>>>> > [ 44.872228] ------------[ cut here ]------------
>>>> > [ 44.872246] WARNING: CPU: 0 PID: 1745 at
>>>> > drivers/gpu/drm/i915/intel_uncore.c:528
>>>> > hsw_unclaimed_reg_debug+0x6d/0x80 [i915]()
>>>> > [ 44.872248] Unclaimed register detected after reading register 0x130040
>>>> > [ 44.872249] Modules linked in: hid_multitouch hid_sensor_hub
>>>> > snd_hda_codec_hdmi intel_rapl x86_pkg_temp_thermal intel_powerclamp
>>>> > efivars snd_hda_intel btusb snd_hda_controller snd_hda_codec snd_hwdep
>>>> > iwlmvm snd_pcm_oss snd_mixer_oss iwlwifi mei_me mei snd_pcm snd_timer
>>>> > int3403_thermal i2c_designware_platform i2c_designware_core acpi_pad
>>>> > fuse nls_utf8 nls_cp437 vfat fat i915 sdhci_pci drm_kms_helper e1000e
>>>> > drm sdhci_acpi sdhci
>>>> > [ 44.872276] CPU: 0 PID: 1745 Comm: pm_rpm Tainted: G W
>>>> > 3.17.0-rc5.1409171019pz+ #1120
>>>> > [ 44.872277] Hardware name: Intel Corporation Broadwell Client
>>>> > platform/Wilson Beach SDS, BIOS BDW-E2R1.86C.0072.R03.1405072127
>>>> > 05/07/2014
>>>> > [ 44.872279] 0000000000000009 ffff8800aa763a88 ffffffff816ec803
>>>> > ffff8800aa763ad0
>>>> > [ 44.872282] ffff8800aa763ac0 ffffffff8107a0c8 ffff88023ec20000
>>>> > 0000000000130040
>>>> > [ 44.872286] 00000000ffffffff 0000000000130040 ffff88023ec20000
>>>> > ffff8800aa763b20
>>>> > [ 44.872289] Call Trace:
>>>> > [ 44.872293] [<ffffffff816ec803>] dump_stack+0x4d/0x66
>>>> > [ 44.872296] [<ffffffff8107a0c8>] warn_slowpath_common+0x78/0xa0
>>>> > [ 44.872299] [<ffffffff8107a137>] warn_slowpath_fmt+0x47/0x50
>>>> > [ 44.872318] [<ffffffffa0122bbd>] hsw_unclaimed_reg_debug+0x6d/0x80 [i915]
>>>> > [ 44.872335] [<ffffffffa0125a7e>] gen6_read32+0x8e/0x150 [i915]
>>>> > [ 44.872356] [<ffffffffa0154636>] intel_ddi_get_cdclk_freq+0xd6/0x140 [i915]
>>>> > [ 44.872376] [<ffffffffa0139a92>] __intel_set_mode+0x1672/0x1750 [i915]
>>>> > [ 44.872396] [<ffffffffa01418e1>] intel_set_mode+0x11/0x30 [i915]
>>>> > [ 44.872416] [<ffffffffa01429ed>] intel_crtc_set_config+0xa9d/0xeb0 [i915]
>>>> > [ 44.872429] [<ffffffffa0026283>]
>>>> > drm_mode_set_config_internal+0x63/0x100 [drm]
>>>> > [ 44.872443] [<ffffffffa002a7e3>] drm_mode_setcrtc+0x283/0x580 [drm]
>>>> > [ 44.872453] [<ffffffffa001c81f>] drm_ioctl+0x1df/0x6a0 [drm]
>>>> > [ 44.872458] [<ffffffff811e07b0>] do_vfs_ioctl+0x2e0/0x4e0
>>>> > [ 44.872461] [<ffffffff816f6af7>] ? sysret_check+0x1b/0x56
>>>> > [ 44.872465] [<ffffffff810bfb0d>] ? trace_hardirqs_on_caller+0x15d/0x200
>>>> > [ 44.872468] [<ffffffff811e0a31>] SyS_ioctl+0x81/0xa0
>>>> > [ 44.872471] [<ffffffff816f6ad2>] system_call_fastpath+0x16/0x1b
>>>> > [ 44.872473] ---[ end trace 91bf8920a2c6b826 ]---
>>>> > [ 44.872512] [drm:intel_modeset_pipe_config] plane bpp: 24, pipe
>>>> > bpp: 18, dithering: 1
>>>> >
>>>> >>
>>>> >> BR,
>>>> >> Jani.
>>>> >>
>>>> >>
>>>> >>>
>>>> >>>>
>>>> >>>> > + return true;
>>>> >>>> > +
>>>> >>>> > + return ilk_pipe_pixel_rate(pipe_config) <=
>>>> >>>> > + intel_ddi_get_cdclk_freq(dev_priv) * 95 / 100;
>>>> >>>>
>>>> >>>> Otherwise
>>>> >>>> Acked-by: Chris Wilson <chris at chris-wilson.co.uk>
>>>> >>>> -Chris
>>>> >>>>
>>>> >>>> --
>>>> >>>> Chris Wilson, Intel Open Source Technology Centre
>>>> >>>
>>>> >>> --
>>>> >>> Ville Syrjälä
>>>> >>> Intel OTC
>>>> >>> _______________________________________________
>>>> >>> Intel-gfx mailing list
>>>> >>> Intel-gfx at lists.freedesktop.org
>>>> >>> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>>>> >>
>>>> >> --
>>>> >> Jani Nikula, Intel Open Source Technology Center
>>>> >> _______________________________________________
>>>> >> Intel-gfx mailing list
>>>> >> Intel-gfx at lists.freedesktop.org
>>>> >> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>>>> >
>>>> >
>>>> >
>>>> > --
>>>> > Paulo Zanoni
>>>>
>>>>
>>>>
>>>> --
>>>> Paulo Zanoni
>>>
>>> --
>>> Ville Syrjälä
>>> Intel OTC
>>
>> --
>> Jani Nikula, Intel Open Source Technology Center
>> _______________________________________________
>> Intel-gfx mailing list
>> Intel-gfx at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>
>
>
> --
> Rodrigo Vivi
> Blog: http://blog.vivi.eng.br
--
Paulo Zanoni
More information about the Intel-gfx
mailing list