[Intel-gfx] [PATCH v2 3/6] drm/i915: Make shadow tables range-based
Yokoyama, Caz
caz.yokoyama at intel.com
Tue Aug 10 22:02:54 UTC 2021
Reviewed-by: Caz Yokoyama <caz.yokoyama at intel.com>
-caz
On Thu, 2021-07-29 at 14:59 -0700, Matt Roper wrote:
> On Thu, Jul 29, 2021 at 02:55:17PM -0700, Yokoyama, Caz wrote:
> > On Thu, 2021-07-29 at 08:21 -0700, Matt Roper wrote:
> > > Rather than defining our shadow tables as a list of individual
> > > registers, provide them as a list of register ranges; we'll have
> > > some
> > > ranges of multiple registers being added soon (and we already
> > > have a
> > > couple adjacent registers that we can squash into a single range
> > > now).
> > >
> > > This change also defines the table with hex literal values rather
> > > than
> > > symbolic register names; since that's how the tables are defined
> > > in
> > > the
> > > bspec, this change will make it easier to review the tables
> > > overall.
> > >
> > > v2:
> > > - Force signed comparison on range overlap sanity check
> > >
> > > Signed-off-by: Matt Roper <matthew.d.roper at intel.com>
> > > ---
> > > drivers/gpu/drm/i915/gt/intel_workarounds.c | 13 +-
> > > drivers/gpu/drm/i915/intel_uncore.c | 160 +++++++++---
> > > ----
> > > --
> > > drivers/gpu/drm/i915/intel_uncore.h | 6 +
> > > drivers/gpu/drm/i915/selftests/intel_uncore.c | 32 ++--
> > > 4 files changed, 108 insertions(+), 103 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds.c
> > > b/drivers/gpu/drm/i915/gt/intel_workarounds.c
> > > index 9173df59821a..7558414bafb2 100644
> > > --- a/drivers/gpu/drm/i915/gt/intel_workarounds.c
> > > +++ b/drivers/gpu/drm/i915/gt/intel_workarounds.c
> > > @@ -1963,12 +1963,7 @@ void intel_engine_apply_workarounds(struct
> > > intel_engine_cs *engine)
> > > wa_list_apply(engine->gt, &engine->wa_list);
> > > }
> > >
> > > -struct mcr_range {
> > > - u32 start;
> > > - u32 end;
> > > -};
> > > -
> > > -static const struct mcr_range mcr_ranges_gen8[] = {
> > > +static const struct i915_range mcr_ranges_gen8[] = {
> > > { .start = 0x5500, .end = 0x55ff },
> > > { .start = 0x7000, .end = 0x7fff },
> > > { .start = 0x9400, .end = 0x97ff },
> > > @@ -1977,7 +1972,7 @@ static const struct mcr_range
> > > mcr_ranges_gen8[]
> > > = {
> > > {},
> > > };
> > >
> > > -static const struct mcr_range mcr_ranges_gen12[] = {
> > > +static const struct i915_range mcr_ranges_gen12[] = {
> > > { .start = 0x8150, .end = 0x815f },
> > > { .start = 0x9520, .end = 0x955f },
> > > { .start = 0xb100, .end = 0xb3ff },
> > > @@ -1986,7 +1981,7 @@ static const struct mcr_range
> > > mcr_ranges_gen12[] = {
> > > {},
> > > };
> > >
> > > -static const struct mcr_range mcr_ranges_xehp[] = {
> > > +static const struct i915_range mcr_ranges_xehp[] = {
> > > { .start = 0x4000, .end = 0x4aff },
> > > { .start = 0x5200, .end = 0x52ff },
> > > { .start = 0x5400, .end = 0x7fff },
> > > @@ -2004,7 +1999,7 @@ static const struct mcr_range
> > > mcr_ranges_xehp[]
> > > = {
> > >
> > > static bool mcr_range(struct drm_i915_private *i915, u32 offset)
> > > {
> > > - const struct mcr_range *mcr_ranges;
> > > + const struct i915_range *mcr_ranges;
> > > int i;
> > >
> > > if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50))
> > > diff --git a/drivers/gpu/drm/i915/intel_uncore.c
> > > b/drivers/gpu/drm/i915/intel_uncore.c
> > > index ea910f7ee635..2cfbc16f7dee 100644
> > > --- a/drivers/gpu/drm/i915/intel_uncore.c
> > > +++ b/drivers/gpu/drm/i915/intel_uncore.c
> > > @@ -946,101 +946,95 @@ static const struct intel_forcewake_range
> > > __vlv_fw_ranges[] = {
> > > find_fw_domain(uncore, offset)
> > >
> > > /* *Must* be sorted by offset! See intel_shadow_table_check().
> > > */
> > > -static const i915_reg_t gen8_shadowed_regs[] = {
> > > - RING_TAIL(RENDER_RING_BASE), /* 0x2000 (base) */
> > > - GEN6_RPNSWREQ, /* 0xA008 *
> > > - GEN6_RC_VIDEO_FREQ, /* 0xA00C */
> > > - RING_TAIL(GEN6_BSD_RING_BASE), /* 0x12000 (base) */
> > > - RING_TAIL(VEBOX_RING_BASE), /* 0x1a000 (base) */
> > > - RING_TAIL(BLT_RING_BASE), /* 0x22000 (base) */
> > > +static const struct i915_range gen8_shadowed_regs[] = {
> > > + { .start = 0x2030, .end = 0x2030 },
> > > + { .start = 0xA008, .end = 0xA00C },
> > > + { .start = 0x12030, .end = 0x12030 },
> > > + { .start = 0x1a030, .end = 0x1a030 },
> > > + { .start = 0x22030, .end = 0x22030 },
> > You are replacing macro defined values by literal values. These
> > macro
> > defined values are no longer used in this patch, but used in other
> > places such as intel_rps.c:gen6_rps_set(). What plan do you have
> > about
> > the same address is defined in 2 ways, i.e. macro and literal
> > value?
>
> These are ranges of registers offsets (similar to what we already do
> elsewhere in the driver for multicast ranges, forcewake ranges,
> etc.),
> whereas individual registers used throughout the driver are still
> defined in i915_reg.h. Some of these wind up being single-register
> ranges, but that's fine/expected, and some of those single-register
> ranges also get extended to cover a larger range by later patches in
> this series.
>
>
> Matt
>
> > -caz
> >
> > > /* TODO: Other registers are not yet used */
> > > };
> > >
> > > -static const i915_reg_t gen11_shadowed_regs[] = {
> > > - RING_TAIL(RENDER_RING_BASE), /* 0x2000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(RENDER_RING_BASE), /* 0x2550
> > > */
> > > - GEN6_RPNSWREQ, /* 0xA008
> > > */
> > > - GEN6_RC_VIDEO_FREQ, /* 0xA00C
> > > */
> > > - RING_TAIL(BLT_RING_BASE), /* 0x22000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(BLT_RING_BASE), /* 0x22550
> > > */
> > > - RING_TAIL(GEN11_BSD_RING_BASE), /* 0x1C0000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD_RING_BASE), /* 0x1C0550
> > > */
> > > - RING_TAIL(GEN11_BSD2_RING_BASE), /* 0x1C4000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD2_RING_BASE), /* 0x1C4550
> > > */
> > > - RING_TAIL(GEN11_VEBOX_RING_BASE), /* 0x1C8000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_VEBOX_RING_BASE), /* 0x1C8550
> > > */
> > > - RING_TAIL(GEN11_BSD3_RING_BASE), /* 0x1D0000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD3_RING_BASE), /* 0x1D0550
> > > */
> > > - RING_TAIL(GEN11_BSD4_RING_BASE), /* 0x1D4000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD4_RING_BASE), /* 0x1D4550
> > > */
> > > - RING_TAIL(GEN11_VEBOX2_RING_BASE), /* 0x1D8000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_VEBOX2_RING_BASE), /* 0x1D8550
> > > */
> > > +static const struct i915_range gen11_shadowed_regs[] = {
> > > + { .start = 0x2030, .end = 0x2030 },
> > > + { .start = 0x2550, .end = 0x2550 },
> > > + { .start = 0xA008, .end = 0xA00C },
> > > + { .start = 0x22030, .end = 0x22030 },
> > > + { .start = 0x22550, .end = 0x22550 },
> > > + { .start = 0x1C0030, .end = 0x1C0030 },
> > > + { .start = 0x1C0550, .end = 0x1C0550 },
> > > + { .start = 0x1C4030, .end = 0x1C4030 },
> > > + { .start = 0x1C4550, .end = 0x1C4550 },
> > > + { .start = 0x1C8030, .end = 0x1C8030 },
> > > + { .start = 0x1C8550, .end = 0x1C8550 },
> > > + { .start = 0x1D0030, .end = 0x1D0030 },
> > > + { .start = 0x1D0550, .end = 0x1D0550 },
> > > + { .start = 0x1D4030, .end = 0x1D4030 },
> > > + { .start = 0x1D4550, .end = 0x1D4550 },
> > > + { .start = 0x1D8030, .end = 0x1D8030 },
> > > + { .start = 0x1D8550, .end = 0x1D8550 },
> > > /* TODO: Other registers are not yet used */
> > > };
> > >
> > > -static const i915_reg_t gen12_shadowed_regs[] = {
> > > - RING_TAIL(RENDER_RING_BASE), /* 0x2000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(RENDER_RING_BASE), /* 0x2550
> > > */
> > > - GEN6_RPNSWREQ, /* 0xA008
> > > */
> > > - GEN6_RC_VIDEO_FREQ, /* 0xA00C
> > > */
> > > - RING_TAIL(BLT_RING_BASE), /* 0x22000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(BLT_RING_BASE), /* 0x22550
> > > */
> > > - RING_TAIL(GEN11_BSD_RING_BASE), /* 0x1C0000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD_RING_BASE), /* 0x1C0550
> > > */
> > > - RING_TAIL(GEN11_BSD2_RING_BASE), /* 0x1C4000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD2_RING_BASE), /* 0x1C4550
> > > */
> > > - RING_TAIL(GEN11_VEBOX_RING_BASE), /* 0x1C8000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_VEBOX_RING_BASE), /* 0x1C8550
> > > */
> > > - RING_TAIL(GEN11_BSD3_RING_BASE), /* 0x1D0000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD3_RING_BASE), /* 0x1D0550
> > > */
> > > - RING_TAIL(GEN11_BSD4_RING_BASE), /* 0x1D4000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD4_RING_BASE), /* 0x1D4550
> > > */
> > > - RING_TAIL(GEN11_VEBOX2_RING_BASE), /* 0x1D8000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_VEBOX2_RING_BASE), /* 0x1D8550
> > > */
> > > +static const struct i915_range gen12_shadowed_regs[] = {
> > > + { .start = 0x2030, .end = 0x2030 },
> > > + { .start = 0x2550, .end = 0x2550 },
> > > + { .start = 0xA008, .end = 0xA00C },
> > > + { .start = 0x22030, .end = 0x22030 },
> > > + { .start = 0x22550, .end = 0x22550 },
> > > + { .start = 0x1C0030, .end = 0x1C0030 },
> > > + { .start = 0x1C0550, .end = 0x1C0550 },
> > > + { .start = 0x1C4030, .end = 0x1C4030 },
> > > + { .start = 0x1C4550, .end = 0x1C4550 },
> > > + { .start = 0x1C8030, .end = 0x1C8030 },
> > > + { .start = 0x1C8550, .end = 0x1C8550 },
> > > + { .start = 0x1D0030, .end = 0x1D0030 },
> > > + { .start = 0x1D0550, .end = 0x1D0550 },
> > > + { .start = 0x1D4030, .end = 0x1D4030 },
> > > + { .start = 0x1D4550, .end = 0x1D4550 },
> > > + { .start = 0x1D8030, .end = 0x1D8030 },
> > > + { .start = 0x1D8550, .end = 0x1D8550 },
> > > /* TODO: Other registers are not yet used */
> > > };
> > >
> > > -static const i915_reg_t xehp_shadowed_regs[] = {
> > > - RING_TAIL(RENDER_RING_BASE), /* 0x2000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(RENDER_RING_BASE), /* 0x2550
> > > */
> > > - GEN6_RPNSWREQ, /* 0xA008
> > > */
> > > - GEN6_RC_VIDEO_FREQ, /* 0xA00C
> > > */
> > > - RING_TAIL(BLT_RING_BASE), /* 0x22000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(BLT_RING_BASE), /* 0x22550
> > > */
> > > - RING_TAIL(GEN11_BSD_RING_BASE), /* 0x1C0000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD_RING_BASE), /* 0x1C0550
> > > */
> > > - RING_TAIL(GEN11_BSD2_RING_BASE), /* 0x1C4000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD2_RING_BASE), /* 0x1C4550
> > > */
> > > - RING_TAIL(GEN11_VEBOX_RING_BASE), /* 0x1C8000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_VEBOX_RING_BASE), /* 0x1C8550
> > > */
> > > - RING_TAIL(GEN11_BSD3_RING_BASE), /* 0x1D0000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD3_RING_BASE), /* 0x1D0550
> > > */
> > > - RING_TAIL(GEN11_BSD4_RING_BASE), /* 0x1D4000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_BSD4_RING_BASE), /* 0x1D4550
> > > */
> > > - RING_TAIL(GEN11_VEBOX2_RING_BASE), /* 0x1D8000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(GEN11_VEBOX2_RING_BASE), /* 0x1D8550
> > > */
> > > - RING_TAIL(XEHP_BSD5_RING_BASE), /* 0x1E0000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(XEHP_BSD5_RING_BASE), /* 0x1E0550
> > > */
> > > - RING_TAIL(XEHP_BSD6_RING_BASE), /* 0x1E4000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(XEHP_BSD6_RING_BASE), /* 0x1E4550
> > > */
> > > - RING_TAIL(XEHP_VEBOX3_RING_BASE), /* 0x1E8000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(XEHP_VEBOX3_RING_BASE), /* 0x1E8550
> > > */
> > > - RING_TAIL(XEHP_BSD7_RING_BASE), /* 0x1F0000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(XEHP_BSD7_RING_BASE), /* 0x1F0550
> > > */
> > > - RING_TAIL(XEHP_BSD8_RING_BASE), /* 0x1F4000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(XEHP_BSD8_RING_BASE), /* 0x1F4550
> > > */
> > > - RING_TAIL(XEHP_VEBOX4_RING_BASE), /* 0x1F8000
> > > (base) */
> > > - RING_EXECLIST_CONTROL(XEHP_VEBOX4_RING_BASE), /* 0x1F8550
> > > */
> > > +static const struct i915_range xehp_shadowed_regs[] = {
> > > + { .start = 0x2000, .end = 0x2030 },
> > > + { .start = 0x2550, .end = 0x2550 },
> > > + { .start = 0xA008, .end = 0xA00C },
> > > + { .start = 0x22030, .end = 0x22030 },
> > > + { .start = 0x22550, .end = 0x22550 },
> > > + { .start = 0x1C0030, .end = 0x1C0030 },
> > > + { .start = 0x1C0550, .end = 0x1C0550 },
> > > + { .start = 0x1C4030, .end = 0x1C4030 },
> > > + { .start = 0x1C4550, .end = 0x1C4550 },
> > > + { .start = 0x1C8030, .end = 0x1C8030 },
> > > + { .start = 0x1C8550, .end = 0x1C8550 },
> > > + { .start = 0x1D0030, .end = 0x1D0030 },
> > > + { .start = 0x1D0550, .end = 0x1D0550 },
> > > + { .start = 0x1D4030, .end = 0x1D4030 },
> > > + { .start = 0x1D4550, .end = 0x1D4550 },
> > > + { .start = 0x1D8030, .end = 0x1D8030 },
> > > + { .start = 0x1D8550, .end = 0x1D8550 },
> > > + { .start = 0x1E0030, .end = 0x1E0030 },
> > > + { .start = 0x1E0550, .end = 0x1E0550 },
> > > + { .start = 0x1E4030, .end = 0x1E4030 },
> > > + { .start = 0x1E4550, .end = 0x1E4550 },
> > > + { .start = 0x1E8030, .end = 0x1E8030 },
> > > + { .start = 0x1E8550, .end = 0x1E8550 },
> > > + { .start = 0x1F0030, .end = 0x1F0030 },
> > > + { .start = 0x1F0550, .end = 0x1F0550 },
> > > + { .start = 0x1F4030, .end = 0x1F4030 },
> > > + { .start = 0x1F4550, .end = 0x1F4550 },
> > > + { .start = 0x1F8030, .end = 0x1F8030 },
> > > + { .start = 0x1F8550, .end = 0x1F8550 },
> > > /* TODO: Other registers are not yet used */
> > > };
> > >
> > > -static int mmio_reg_cmp(u32 key, const i915_reg_t *reg)
> > > +static int mmio_range_cmp(u32 key, const struct i915_range
> > > *range)
> > > {
> > > - u32 offset = i915_mmio_reg_offset(*reg);
> > > -
> > > - if (key < offset)
> > > + if (key < range->start)
> > > return -1;
> > > - else if (key > offset)
> > > + else if (key > range->end)
> > > return 1;
> > > else
> > > return 0;
> > > @@ -1049,9 +1043,9 @@ static int mmio_reg_cmp(u32 key, const
> > > i915_reg_t *reg)
> > > #define __is_X_shadowed(x) \
> > > static bool is_##x##_shadowed(u32 offset) \
> > > { \
> > > - const i915_reg_t *regs = x##_shadowed_regs; \
> > > + const struct i915_range *regs = x##_shadowed_regs; \
> > > return BSEARCH(offset, regs, ARRAY_SIZE(x##_shadowed_regs),
> > > \
> > > - mmio_reg_cmp); \
> > > + mmio_range_cmp); \
> > > }
> > >
> > > __is_X_shadowed(gen8)
> > > diff --git a/drivers/gpu/drm/i915/intel_uncore.h
> > > b/drivers/gpu/drm/i915/intel_uncore.h
> > > index 3c0b0a8b5250..531665b08039 100644
> > > --- a/drivers/gpu/drm/i915/intel_uncore.h
> > > +++ b/drivers/gpu/drm/i915/intel_uncore.h
> > > @@ -119,6 +119,12 @@ struct intel_forcewake_range {
> > > enum forcewake_domains domains;
> > > };
> > >
> > > +/* Other register ranges (e.g., shadow tables, MCR tables, etc.)
> > > */
> > > +struct i915_range {
> > > + u32 start;
> > > + u32 end;
> > > +};
> > > +
> > > struct intel_uncore {
> > > void __iomem *regs;
> > >
> > > diff --git a/drivers/gpu/drm/i915/selftests/intel_uncore.c
> > > b/drivers/gpu/drm/i915/selftests/intel_uncore.c
> > > index 720b60853f8b..d6a9c11afa23 100644
> > > --- a/drivers/gpu/drm/i915/selftests/intel_uncore.c
> > > +++ b/drivers/gpu/drm/i915/selftests/intel_uncore.c
> > > @@ -62,30 +62,40 @@ static int intel_fw_table_check(const struct
> > > intel_forcewake_range *ranges,
> > > static int intel_shadow_table_check(void)
> > > {
> > > struct {
> > > - const i915_reg_t *regs;
> > > + const struct i915_range *regs;
> > > unsigned int size;
> > > - } reg_lists[] = {
> > > + } range_lists[] = {
> > > { gen8_shadowed_regs,
> > > ARRAY_SIZE(gen8_shadowed_regs) },
> > > { gen11_shadowed_regs,
> > > ARRAY_SIZE(gen11_shadowed_regs)
> > > },
> > > { gen12_shadowed_regs,
> > > ARRAY_SIZE(gen12_shadowed_regs)
> > > },
> > > { xehp_shadowed_regs,
> > > ARRAY_SIZE(xehp_shadowed_regs) },
> > > };
> > > - const i915_reg_t *reg;
> > > + const struct i915_range *range;
> > > unsigned int i, j;
> > > s32 prev;
> > >
> > > - for (j = 0; j < ARRAY_SIZE(reg_lists); ++j) {
> > > - reg = reg_lists[j].regs;
> > > - for (i = 0, prev = -1; i < reg_lists[j].size; i++,
> > > reg++) {
> > > - u32 offset = i915_mmio_reg_offset(*reg);
> > > + for (j = 0; j < ARRAY_SIZE(range_lists); ++j) {
> > > + range = range_lists[j].regs;
> > > + for (i = 0, prev = -1; i < range_lists[j].size;
> > > i++,
> > > range++) {
> > > + if (range->end < range->start) {
> > > + pr_err("%s: range[%d]:(%06x-%06x)
> > > has
> > > end before start\n",
> > > + __func__, i, range->start,
> > > range->end);
> > > + return -EINVAL;
> > > + }
> > > +
> > > + if (prev >= (s32)range->start) {
> > > + pr_err("%s: range[%d]:(%06x-%06x)
> > > is
> > > before end of previous (%06x)\n",
> > > + __func__, i, range->start,
> > > range->end, prev);
> > > + return -EINVAL;
> > > + }
> > >
> > > - if (prev >= (s32)offset) {
> > > - pr_err("%s: entry[%d]:(%x) is
> > > before
> > > previous (%x)\n",
> > > - __func__, i, offset, prev);
> > > + if (range->start % 4) {
> > > + pr_err("%s: range[%d]:(%06x-%06x)
> > > has
> > > non-dword-aligned start\n",
> > > + __func__, i, range->start,
> > > range->end);
> > > return -EINVAL;
> > > }
> > >
> > > - prev = offset;
> > > + prev = range->end;
> > > }
> > > }
> > >
More information about the Intel-gfx
mailing list