On Tue, 2019-05-07 at 12:00 -0700, Daniele Ceraolo Spurio wrote:
>
> On 5/3/19 2:30 PM, Stuart Summers wrote:
> > Currently, the subslice_mask runtime parameter is stored as an
> > array of subslices per slice. Expand the subslice mask array to
> > better match what is presented to userspace through the
> > I915_QUERY_TOPOLOGY_INFO ioctl. The index into this array is
> > then calculated:
> > slice * subslice stride + subslice index / 8
> >
> > v2: fix spacing in set_sseu_info args
> > use set_sseu_info to initialize sseu data when building
> > device status in debugfs
> > rename variables in intel_engine_types.h to avoid checkpatch
> > warnings
> > v3: update headers in intel_sseu.h
> > v4: add const to some sseu_dev_info variables
> > use sseu->eu_stride for EU stride calculations
> > v5: address review comments from Tvrtko and Daniele
> >
> > Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@xxxxxxxxx>
> > Cc: Lionel Landwerlin <lionel.g.landwerlin@xxxxxxxxx>
> > Acked-by: Lionel Landwerlin <lionel.g.landwerlin@xxxxxxxxx>
> > Signed-off-by: Stuart Summers <stuart.summers@xxxxxxxxx>
> > ---
> > drivers/gpu/drm/i915/gt/intel_engine_cs.c | 24 ++-
> > drivers/gpu/drm/i915/gt/intel_engine_types.h | 30 ++--
> > drivers/gpu/drm/i915/gt/intel_hangcheck.c | 3 +-
> > drivers/gpu/drm/i915/gt/intel_sseu.c | 43 ++++-
> > drivers/gpu/drm/i915/gt/intel_sseu.h | 28 +++-
> > drivers/gpu/drm/i915/gt/intel_workarounds.c | 2 +-
> > drivers/gpu/drm/i915/i915_debugfs.c | 40 ++---
> > drivers/gpu/drm/i915/i915_drv.c | 6 +-
> > drivers/gpu/drm/i915/i915_gpu_error.c | 5 +-
> > drivers/gpu/drm/i915/i915_query.c | 10 +-
> > drivers/gpu/drm/i915/intel_device_info.c | 155 ++++++++++--
> > -------
> > 11 files changed, 227 insertions(+), 119 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/gt/intel_engine_cs.c
> > b/drivers/gpu/drm/i915/gt/intel_engine_cs.c
> > index 5907a9613641..290bda5cc82b 100644
> > --- a/drivers/gpu/drm/i915/gt/intel_engine_cs.c
> > +++ b/drivers/gpu/drm/i915/gt/intel_engine_cs.c
> > @@ -909,12 +909,30 @@ const char *i915_cache_level_str(struct
> > drm_i915_private *i915, int type)
> > }
> > }
> >
> > +static inline u32
> > +intel_sseu_fls_subslice(const struct sseu_dev_info *sseu, u32
> > slice)
> > +{
> > + u32 subslice;
> > + int i;
> > +
> > + for (i = sseu->ss_stride - 1; i >= 0; i--) {
> > + subslice = fls(sseu->subslice_mask[slice * sseu-
> > >ss_stride +
> > + i]);
> > + if (subslice) {
> > + subslice += i * BITS_PER_BYTE;
> > + break;
> > + }
> > + }
> > +
> > + return subslice;
> > +}
> > +
> > u32 intel_calculate_mcr_s_ss_select(struct drm_i915_private
> > *dev_priv)
> > {
> > const struct sseu_dev_info *sseu = &RUNTIME_INFO(dev_priv)-
> > >sseu;
> > u32 mcr_s_ss_select;
> > u32 slice = fls(sseu->slice_mask);
> > - u32 subslice = fls(sseu->subslice_mask[slice]);
> > + u32 subslice = intel_sseu_fls_subslice(sseu, slice);
> >
> > if (IS_GEN(dev_priv, 10))
> > mcr_s_ss_select = GEN8_MCR_SLICE(slice) |
> > @@ -990,6 +1008,7 @@ void intel_engine_get_instdone(struct
> > intel_engine_cs *engine,
> > struct intel_instdone *instdone)
> > {
> > struct drm_i915_private *dev_priv = engine->i915;
> > + const struct sseu_dev_info *sseu = &RUNTIME_INFO(dev_priv)-
> > >sseu;
> > struct intel_uncore *uncore = engine->uncore;
> > u32 mmio_base = engine->mmio_base;
> > int slice;
> > @@ -1007,7 +1026,8 @@ void intel_engine_get_instdone(struct
> > intel_engine_cs *engine,
> >
> > instdone->slice_common =
> > intel_uncore_read(uncore, GEN7_SC_INSTDONE);
> > - for_each_instdone_slice_subslice(dev_priv, slice,
> > subslice) {
> > + for_each_instdone_slice_subslice(dev_priv, sseu, slice,
> > + subslice) {
> > instdone->sampler[slice][subslice] =
> > read_subslice_reg(dev_priv, slice,
> > subslice,
> > GEN7_SAMPLER_INSTDONE
> > );
> > diff --git a/drivers/gpu/drm/i915/gt/intel_engine_types.h
> > b/drivers/gpu/drm/i915/gt/intel_engine_types.h
> > index c0ab11b12e14..582340b55144 100644
> > --- a/drivers/gpu/drm/i915/gt/intel_engine_types.h
> > +++ b/drivers/gpu/drm/i915/gt/intel_engine_types.h
> > @@ -535,20 +535,20 @@ intel_engine_needs_breadcrumb_tasklet(const
> > struct intel_engine_cs *engine)
> > return engine->flags & I915_ENGINE_NEEDS_BREADCRUMB_TASKLET;
> > }
> >
> > -#define instdone_slice_mask(dev_priv__) \
> > - (IS_GEN(dev_priv__, 7) ? \
> > - 1 : RUNTIME_INFO(dev_priv__)->sseu.slice_mask)
> > -
> > -#define instdone_subslice_mask(dev_priv__) \
> > - (IS_GEN(dev_priv__, 7) ? \
> > - 1 : RUNTIME_INFO(dev_priv__)->sseu.subslice_mask[0])
> > -
> > -#define for_each_instdone_slice_subslice(dev_priv__, slice__,
> > subslice__) \
> > - for ((slice__) = 0, (subslice__) = 0; \
> > - (slice__) < I915_MAX_SLICES; \
> > - (subslice__) = ((subslice__) + 1) < I915_MAX_SUBSLICES ?
> > (subslice__) + 1 : 0, \
> > - (slice__) += ((subslice__) == 0)) \
> > - for_each_if((BIT(slice__) &
> > instdone_slice_mask(dev_priv__)) && \
> > - (BIT(subslice__) &
> > instdone_subslice_mask(dev_priv__)))
> > +#define instdone_has_slice(dev_priv___, sseu___, slice___) \
> > + ((IS_GEN(dev_priv___, 7) ? 1 : ((sseu___)->slice_mask)) & \
> > + BIT(slice___))
> > +
> > +#define instdone_has_subslice(dev_priv__, sseu__, slice__,
> > subslice__) \
> > + (IS_GEN(dev_priv__, 7) ? 1 : \
>
> This will return true for all ss on gen7, while the original code
> returned 1 as the subslice mask (i.e. only has ss == 0).
True, I'll fix this.
>
> > + intel_sseu_has_subslice(sseu__, slice__, subslice__))
> > +
> > +#define for_each_instdone_slice_subslice(dev_priv_, sseu_, slice_,
> > subslice_) \
> > + for ((slice_) = 0, (subslice_) = 0; (slice_) < I915_MAX_SLICES;
> > \
> > + (subslice_) = ((subslice_) + 1) % I915_MAX_SUBSLICES, \
> > + (slice_) += ((subslice_) == 0)) \
> > + for_each_if((instdone_has_slice(dev_priv_, sseu_,
> > slice_)) && \
> > + (irnstdone_has_subslice(dev_priv_, sseu_,
> > slice_, \
> > + subslice_)))
> >
> > #endif /* __INTEL_ENGINE_TYPES_H__ */
> > diff --git a/drivers/gpu/drm/i915/gt/intel_hangcheck.c
> > b/drivers/gpu/drm/i915/gt/intel_hangcheck.c
> > index 721ab74a382f..10e032c9ab10 100644
> > --- a/drivers/gpu/drm/i915/gt/intel_hangcheck.c
> > +++ b/drivers/gpu/drm/i915/gt/intel_hangcheck.c
> > @@ -51,6 +51,7 @@ static bool instdone_unchanged(u32
> > current_instdone, u32 *old_instdone)
> > static bool subunits_stuck(struct intel_engine_cs *engine)
> > {
> > struct drm_i915_private *dev_priv = engine->i915;
> > + const struct sseu_dev_info *sseu = &RUNTIME_INFO(dev_priv)-
> > >sseu;
> > struct intel_instdone instdone;
> > struct intel_instdone *accu_instdone = &engine-
> > >hangcheck.instdone;
> > bool stuck;
> > @@ -72,7 +73,7 @@ static bool subunits_stuck(struct intel_engine_cs
> > *engine)
> > stuck &= instdone_unchanged(instdone.slice_common,
> > &accu_instdone->slice_common);
> >
> > - for_each_instdone_slice_subslice(dev_priv, slice, subslice) {
> > + for_each_instdone_slice_subslice(dev_priv, sseu, slice,
> > subslice) {
> > stuck &=
> > instdone_unchanged(instdone.sampler[slice][subslice],
> > &accu_instdone-
> > >sampler[slice][subslice]);
> > stuck &=
> > instdone_unchanged(instdone.row[slice][subslice],
> > diff --git a/drivers/gpu/drm/i915/gt/intel_sseu.c
> > b/drivers/gpu/drm/i915/gt/intel_sseu.c
> > index a0756f006f5f..a8b98b0266b7 100644
> > --- a/drivers/gpu/drm/i915/gt/intel_sseu.c
> > +++ b/drivers/gpu/drm/i915/gt/intel_sseu.c
> > @@ -8,6 +8,17 @@
> > #include "intel_lrc_reg.h"
> > #include "intel_sseu.h"
> >
> > +void intel_sseu_set_info(struct sseu_dev_info *sseu, u8
> > max_slices,
> > + u8 max_subslices, u8 max_eus_per_subslice)
> > +{
> > + sseu->max_slices = max_slices;
> > + sseu->max_subslices = max_subslices;
> > + sseu->max_eus_per_subslice = max_eus_per_subslice;
> > +
> > + sseu->ss_stride = GEN_SSEU_STRIDE(sseu->max_subslices);
> > + sseu->eu_stride = GEN_SSEU_STRIDE(sseu->max_eus_per_subslice);
> > +}
> > +
> > unsigned int
> > intel_sseu_subslice_total(const struct sseu_dev_info *sseu)
> > {
> > @@ -19,10 +30,40 @@ intel_sseu_subslice_total(const struct
> > sseu_dev_info *sseu)
> > return total;
> > }
> >
> > +void intel_sseu_copy_subslices(const struct sseu_dev_info *sseu,
> > int slice,
> > + u8 *to_mask)
> > +{
> > + int offset = slice * sseu->ss_stride;
> > +
> > + memcpy(&to_mask[offset], &sseu->subslice_mask[offset], sseu-
> > >ss_stride);
> > +}
> > +
> > +u32 intel_sseu_get_subslices(const struct sseu_dev_info *sseu, u8
> > slice)
>
> nitpick: extra space
Ok.
>
> > +{
> > + int i, offset = slice * sseu->ss_stride;
> > + u32 mask;
> > +
>
> GEM_BUG_ON(sseu->ss_stride > size(u32)) to catch the unlikely case we
> go
> over and need to update the mask size?
Makes sense. I'll add this.
>
> > + for (i = 0; i < sseu->ss_stride; i++)
> > + mask |= (u32)sseu->subslice_mask[offset + i] <<
> > + i * BITS_PER_BYTE;
> > +
> > + return mask;
> > +}
> > +
> > +void intel_sseu_set_subslices(struct sseu_dev_info *sseu, int
> > slice,
> > + u32 ss_mask)
> > +{
> > + int i, offset = slice * sseu->ss_stride;
> > +
> > + for (i = 0; i < sseu->ss_stride; i++)
> > + sseu->subslice_mask[offset + i] =
> > + (ss_mask >> (BITS_PER_BYTE * i)) & 0xff;
> > +}
> > +
> > unsigned int
> > intel_sseu_subslices_per_slice(const struct sseu_dev_info *sseu,
> > u8 slice)
> > {
> > - return hweight8(sseu->subslice_mask[slice]);
> > + return hweight32(intel_sseu_get_subslices(sseu, slice));
> > }
> >
> > u32 intel_sseu_make_rpcs(struct drm_i915_private *i915,
>
> <snip>
>
> > --- a/drivers/gpu/drm/i915/intel_device_info.c
> > +++ b/drivers/gpu/drm/i915/intel_device_info.c
> > @@ -84,17 +84,46 @@ void intel_device_info_dump_flags(const struct
> > intel_device_info *info,
> > #undef PRINT_FLAG
> > }
> >
> > +#define SS_STR_MAX_SIZE (GEN_MAX_SUBSLICE_STRIDE * 2 + 1)
> > +
> > +static char *
> > +subslice_per_slice_str(char *buf, u8 size, const struct
> > sseu_dev_info *sseu,
> > + u8 slice)
> > +{
> > + int i;
> > + u8 ss_offset = slice * sseu->ss_stride;
> > +
> > + GEM_BUG_ON(slice >= sseu->max_slices);
> > +
> > + /* Two ASCII character hex plus null terminator */
> > + GEM_BUG_ON(size < sseu->ss_stride * 2 + 1);
> > +
> > + memset(buf, 0, size);
> > +
> > + /*
> > + * Print subslice information in reverse order to match
> > + * userspace expectations.
> > + */
> > + for (i = 0; i < sseu->ss_stride; i++)
> > + sprintf(&buf[i * 2], "%02x",
> > + sseu->subslice_mask[ss_offset + sseu->ss_stride
> > -
> > + (i + 1)]);
> > +
> > + return buf;
> > +}
> > +
> > static void sseu_dump(const struct sseu_dev_info *sseu, struct
> > drm_printer *p)
> > {
> > int s;
> > + char buf[SS_STR_MAX_SIZE];
> >
> > drm_printf(p, "slice total: %u, mask=%04x\n",
> > hweight8(sseu->slice_mask), sseu->slice_mask);
> > drm_printf(p, "subslice total: %u\n",
> > intel_sseu_subslice_total(sseu));
> > for (s = 0; s < sseu->max_slices; s++) {
> > - drm_printf(p, "slice%d: %u subslices, mask=%04x\n",
> > + drm_printf(p, "slice%d: %u subslices, mask=%s\n",
> > s, intel_sseu_subslices_per_slice(sseu, s),
> > - sseu->subslice_mask[s]);
> > + subslice_per_slice_str(buf, ARRAY_SIZE(buf),
> > sseu, s));
>
> Now that we have intel_sseu_get_subslices() can't we just print the
> return from that instead of using the buffer?
I personally would prefer we keep the stringify function as it gives a
little more flexibility. Do you have a strong preference to move to a
direct printk formatted string?
>
>
> > }
> > drm_printf(p, "EU total: %u\n", sseu->eu_total);
> > drm_printf(p, "EU per subslice: %u\n", sseu->eu_per_subslice);
>
> <snip>
>
> > @@ -555,6 +570,7 @@ static void haswell_sseu_info_init(struct
> > drm_i915_private *dev_priv)
> > struct sseu_dev_info *sseu = &RUNTIME_INFO(dev_priv)->sseu;
> > u32 fuse1;
> > int s, ss;
> > + u32 subslice_mask;
> >
> > /*
> > * There isn't a register to tell us how many slices/subslices.
> > We
> > @@ -566,22 +582,18 @@ static void haswell_sseu_info_init(struct
> > drm_i915_private *dev_priv)
> > /* fall through */
> > case 1:
> > sseu->slice_mask = BIT(0);
> > - sseu->subslice_mask[0] = BIT(0);
> > + subslice_mask = BIT(0);
> > break;
> > case 2:
> > sseu->slice_mask = BIT(0);
> > - sseu->subslice_mask[0] = BIT(0) | BIT(1);
> > + subslice_mask = BIT(0) | BIT(1);
> > break;
> > case 3:
> > sseu->slice_mask = BIT(0) | BIT(1);
> > - sseu->subslice_mask[0] = BIT(0) | BIT(1);
> > - sseu->subslice_mask[1] = BIT(0) | BIT(1);
> > + subslice_mask = BIT(0) | BIT(1);
> > break;
> > }
> >
> > - sseu->max_slices = hweight8(sseu->slice_mask);
> > - sseu->max_subslices = hweight8(sseu->subslice_mask[0]);
> > -
> > fuse1 = I915_READ(HSW_PAVP_FUSE1);
> > switch ((fuse1 & HSW_F1_EU_DIS_MASK) >> HSW_F1_EU_DIS_SHIFT) {
> > default:
> > @@ -598,9 +610,14 @@ static void haswell_sseu_info_init(struct
> > drm_i915_private *dev_priv)
> > sseu->eu_per_subslice = 6;
> > break;
> > }
> > - sseu->max_eus_per_subslice = sseu->eu_per_subslice;
> > +
> > + intel_sseu_set_info(sseu, hweight8(sseu->slice_mask),
> > + hweight8(subslice_mask),
> > + sseu->eu_per_subslice);
>
> I'd still prefer this to use a local variable so that we always only
> set
> sseu->eu_per_subslice from within intel_sseu_set_info.
So the reason I kept this is in intel_sseu_set_info we are really just
setting the max_eus_per_subslice, not the eu_per_subslice. Are you
saying you'd also like to move the code that sets eu_per_subslice in
each generation's handler to local variables and/or just passed
directly as an argument to intel_sseu_set_info?
I.e. should we use intel_sseu_set_info to set most or all of the
members of the intel_sseu structure? Or is it OK to keep the current
implementation of only using this to set default maximums per platform?
-Stuart
>
> Daniele
>
> >
> > for (s = 0; s < sseu->max_slices; s++) {
> > + intel_sseu_set_subslices(sseu, s, subslice_mask);
> > +
> > for (ss = 0; ss < sseu->max_subslices; ss++) {
> > intel_sseu_set_eus(sseu, s, ss,
> > (1UL << sseu-
> > >eu_per_subslice) - 1);
> >
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
