From: Daniel Vetter <daniel@ffwll.ch>
To: Imre Deak <imre.deak@intel.com>
Cc: intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 2/4] drm/i915/bxt: Determine BXT slice/subslice/EU info
Date: Thu, 9 Apr 2015 15:59:20 +0200 [thread overview]
Message-ID: <20150409135920.GP12038@phenom.ffwll.local> (raw)
In-Reply-To: <1428585979.22233.13.camel@ideak-mobl>
On Thu, Apr 09, 2015 at 04:26:19PM +0300, Imre Deak wrote:
> On Fri, 2015-04-03 at 18:13 -0700, jeff.mcgee@intel.com wrote:
> > From: Jeff McGee <jeff.mcgee@intel.com>
> >
> > Modify the Gen9 SSEU info initialization logic to support
> > Broxton. Broxton reuses the SKL fuse registers but has at most
> > 1 slice and 6 EU per subslice.
> >
> > Signed-off-by: Jeff McGee <jeff.mcgee@intel.com>
> > ---
> > drivers/gpu/drm/i915/i915_dma.c | 47 ++++++++++++++++++++++++++---------------
> > drivers/gpu/drm/i915/i915_reg.h | 4 +---
> > 2 files changed, 31 insertions(+), 20 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
> > index 9691f0f..a9b7770 100644
> > --- a/drivers/gpu/drm/i915/i915_dma.c
> > +++ b/drivers/gpu/drm/i915/i915_dma.c
> > @@ -611,9 +611,21 @@ static void gen9_sseu_info_init(struct drm_device *dev)
> > {
> > struct drm_i915_private *dev_priv = dev->dev_private;
> > struct intel_device_info *info;
> > - const int s_max = 3, ss_max = 4, eu_max = 8;
> > + int s_max = 3, ss_max = 4, eu_max = 8;
> > int s, ss;
> > - u32 fuse2, eu_disable[s_max], s_enable, ss_disable;
> > + u32 fuse2, s_enable, ss_disable, eu_disable;
> > + u8 eu_mask = 0xff;
> > +
> > + /*
> > + * BXT has a single slice. BXT also has at most 6 EU per subslice,
> > + * and therefore only the lowest 6 bits of the 8-bit EU disable
> > + * fields are valid.
> > + */
> > + if (IS_BROXTON(dev)) {
> > + s_max = 1;
> > + eu_max = 6;
> > + eu_mask = 0x3f;
> > + }
> >
> > info = (struct intel_device_info *)&dev_priv->info;
> > fuse2 = I915_READ(GEN8_FUSE2);
> > @@ -622,10 +634,6 @@ static void gen9_sseu_info_init(struct drm_device *dev)
> > ss_disable = (fuse2 & GEN9_F2_SS_DIS_MASK) >>
> > GEN9_F2_SS_DIS_SHIFT;
> >
> > - eu_disable[0] = I915_READ(GEN8_EU_DISABLE0);
> > - eu_disable[1] = I915_READ(GEN8_EU_DISABLE1);
> > - eu_disable[2] = I915_READ(GEN8_EU_DISABLE2);
> > -
> > info->slice_total = hweight32(s_enable);
> > /*
> > * The subslice disable field is global, i.e. it applies
> > @@ -644,25 +652,26 @@ static void gen9_sseu_info_init(struct drm_device *dev)
> > /* skip disabled slice */
> > continue;
> >
> > + eu_disable = I915_READ(GEN9_EU_DISABLE(s));
> > for (ss = 0; ss < ss_max; ss++) {
> > - u32 n_disabled;
> > + int eu_per_ss;
> >
> > if (ss_disable & (0x1 << ss))
> > /* skip disabled subslice */
> > continue;
> >
> > - n_disabled = hweight8(eu_disable[s] >>
> > - (ss * eu_max));
> > + eu_per_ss = eu_max - hweight8((eu_disable >> (ss*8)) &
> > + eu_mask);
> >
> > /*
> > * Record which subslice(s) has(have) 7 EUs. we
> > * can tune the hash used to spread work among
> > * subslices if they are unbalanced.
> > */
> > - if (eu_max - n_disabled == 7)
> > + if (eu_per_ss == 7)
> > info->subslice_7eu[s] |= 1 << ss;
> >
> > - info->eu_total += eu_max - n_disabled;
> > + info->eu_total += eu_per_ss;
> > }
> > }
> >
> > @@ -670,7 +679,8 @@ static void gen9_sseu_info_init(struct drm_device *dev)
> > * SKL is expected to always have a uniform distribution
> > * of EU across subslices with the exception that any one
> > * EU in any one subslice may be fused off for die
> > - * recovery.
> > + * recovery. BXT is expected to be perfectly uniform in EU
> > + * distribution.
>
> Nitpick: I would have added here an assertion for BXT about the above.
> The patchset looks otherwise ok to me, so on 1-4:
> Reviewed-by: Imre Deak <imre.deak@intel.com>
Applied to topic/bxt-stage1, thanks.
-Daniel
>
> > */
> > info->eu_per_subslice = info->subslice_total ?
> > DIV_ROUND_UP(info->eu_total,
> > @@ -678,11 +688,14 @@ static void gen9_sseu_info_init(struct drm_device *dev)
> > /*
> > * SKL supports slice power gating on devices with more than
> > * one slice, and supports EU power gating on devices with
> > - * more than one EU pair per subslice.
> > + * more than one EU pair per subslice. BXT supports subslice
> > + * power gating on devices with more than one subslice, and
> > + * supports EU power gating on devices with more than one EU
> > + * pair per subslice.
> > */
> > - info->has_slice_pg = (info->slice_total > 1) ? 1 : 0;
> > - info->has_subslice_pg = 0;
> > - info->has_eu_pg = (info->eu_per_subslice > 2) ? 1 : 0;
> > + info->has_slice_pg = (IS_SKYLAKE(dev) && (info->slice_total > 1));
> > + info->has_subslice_pg = (IS_BROXTON(dev) && (info->subslice_total > 1));
> > + info->has_eu_pg = (info->eu_per_subslice > 2);
> > }
> >
> > /*
> > @@ -747,7 +760,7 @@ static void intel_device_info_runtime_init(struct drm_device *dev)
> > /* Initialize slice/subslice/EU info */
> > if (IS_CHERRYVIEW(dev))
> > cherryview_sseu_info_init(dev);
> > - else if (IS_SKYLAKE(dev))
> > + else if (INTEL_INFO(dev)->gen >= 9)
> > gen9_sseu_info_init(dev);
> >
> > DRM_DEBUG_DRIVER("slice total: %u\n", info->slice_total);
> > diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
> > index b134fa3..69d3689 100644
> > --- a/drivers/gpu/drm/i915/i915_reg.h
> > +++ b/drivers/gpu/drm/i915/i915_reg.h
> > @@ -1554,9 +1554,7 @@ enum skl_disp_power_wells {
> > #define GEN9_F2_SS_DIS_SHIFT 20
> > #define GEN9_F2_SS_DIS_MASK (0xf << GEN9_F2_SS_DIS_SHIFT)
> >
> > -#define GEN8_EU_DISABLE0 0x9134
> > -#define GEN8_EU_DISABLE1 0x9138
> > -#define GEN8_EU_DISABLE2 0x913c
> > +#define GEN9_EU_DISABLE(slice) (0x9134 + (slice)*0x4)
> >
> > #define GEN6_BSD_SLEEP_PSMI_CONTROL 0x12050
> > #define GEN6_BSD_SLEEP_MSG_DISABLE (1 << 0)
>
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2015-04-09 13:57 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-04-04 1:13 [PATCH 0/4 v2] BXT basic slice/subslice/EU stuff jeff.mcgee
2015-04-04 1:13 ` [PATCH 1/4] drm/i915: Split SSEU init into functions by platform jeff.mcgee
2015-04-04 1:13 ` [PATCH 2/4] drm/i915/bxt: Determine BXT slice/subslice/EU info jeff.mcgee
2015-04-09 13:26 ` Imre Deak
2015-04-09 13:59 ` Daniel Vetter [this message]
2015-04-04 1:13 ` [PATCH 3/4] drm/i915: Split-up SSEU device status by platform jeff.mcgee
2015-04-04 1:13 ` [PATCH 4/4] drm/i915/bxt: Support BXT in SSEU device status dump jeff.mcgee
2015-04-09 16:21 ` shuang.he
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20150409135920.GP12038@phenom.ffwll.local \
--to=daniel@ffwll.ch \
--cc=imre.deak@intel.com \
--cc=intel-gfx@lists.freedesktop.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox