Commit 424321be authored by Ben Skeggs's avatar Ben Skeggs

drm/nouveau/gr/gp102: initial support

Differences from GP100:
- 3 PPCs/GPC.
- Another random reg to calculate/write.
- Attrib CB setup a little different.
- PascalB
- PascalComputeB
Signed-off-by: default avatarBen Skeggs <bskeggs@redhat.com>
parent 4eb3390e
...@@ -125,6 +125,7 @@ ...@@ -125,6 +125,7 @@
#define MAXWELL_B /* cl9097.h */ 0x0000b197 #define MAXWELL_B /* cl9097.h */ 0x0000b197
#define PASCAL_A /* cl9097.h */ 0x0000c097 #define PASCAL_A /* cl9097.h */ 0x0000c097
#define PASCAL_B /* cl9097.h */ 0x0000c197
#define NV74_BSP 0x000074b0 #define NV74_BSP 0x000074b0
...@@ -163,6 +164,7 @@ ...@@ -163,6 +164,7 @@
#define MAXWELL_COMPUTE_A 0x0000b0c0 #define MAXWELL_COMPUTE_A 0x0000b0c0
#define MAXWELL_COMPUTE_B 0x0000b1c0 #define MAXWELL_COMPUTE_B 0x0000b1c0
#define PASCAL_COMPUTE_A 0x0000c0c0 #define PASCAL_COMPUTE_A 0x0000c0c0
#define PASCAL_COMPUTE_B 0x0000c1c0
#define NV74_CIPHER 0x000074c1 #define NV74_CIPHER 0x000074c1
#endif #endif
...@@ -43,4 +43,5 @@ int gm107_gr_new(struct nvkm_device *, int, struct nvkm_gr **); ...@@ -43,4 +43,5 @@ int gm107_gr_new(struct nvkm_device *, int, struct nvkm_gr **);
int gm200_gr_new(struct nvkm_device *, int, struct nvkm_gr **); int gm200_gr_new(struct nvkm_device *, int, struct nvkm_gr **);
int gm20b_gr_new(struct nvkm_device *, int, struct nvkm_gr **); int gm20b_gr_new(struct nvkm_device *, int, struct nvkm_gr **);
int gp100_gr_new(struct nvkm_device *, int, struct nvkm_gr **); int gp100_gr_new(struct nvkm_device *, int, struct nvkm_gr **);
int gp102_gr_new(struct nvkm_device *, int, struct nvkm_gr **);
#endif #endif
...@@ -32,6 +32,7 @@ nvkm-y += nvkm/engine/gr/gm107.o ...@@ -32,6 +32,7 @@ nvkm-y += nvkm/engine/gr/gm107.o
nvkm-y += nvkm/engine/gr/gm200.o nvkm-y += nvkm/engine/gr/gm200.o
nvkm-y += nvkm/engine/gr/gm20b.o nvkm-y += nvkm/engine/gr/gm20b.o
nvkm-y += nvkm/engine/gr/gp100.o nvkm-y += nvkm/engine/gr/gp100.o
nvkm-y += nvkm/engine/gr/gp102.o
nvkm-y += nvkm/engine/gr/ctxnv40.o nvkm-y += nvkm/engine/gr/ctxnv40.o
nvkm-y += nvkm/engine/gr/ctxnv50.o nvkm-y += nvkm/engine/gr/ctxnv50.o
...@@ -50,3 +51,4 @@ nvkm-y += nvkm/engine/gr/ctxgm107.o ...@@ -50,3 +51,4 @@ nvkm-y += nvkm/engine/gr/ctxgm107.o
nvkm-y += nvkm/engine/gr/ctxgm200.o nvkm-y += nvkm/engine/gr/ctxgm200.o
nvkm-y += nvkm/engine/gr/ctxgm20b.o nvkm-y += nvkm/engine/gr/ctxgm20b.o
nvkm-y += nvkm/engine/gr/ctxgp100.o nvkm-y += nvkm/engine/gr/ctxgp100.o
nvkm-y += nvkm/engine/gr/ctxgp102.o
...@@ -102,6 +102,10 @@ void gm200_grctx_generate_405b60(struct gf100_gr *); ...@@ -102,6 +102,10 @@ void gm200_grctx_generate_405b60(struct gf100_gr *);
extern const struct gf100_grctx_func gm20b_grctx; extern const struct gf100_grctx_func gm20b_grctx;
extern const struct gf100_grctx_func gp100_grctx; extern const struct gf100_grctx_func gp100_grctx;
void gp100_grctx_generate_main(struct gf100_gr *, struct gf100_grctx *);
void gp100_grctx_generate_pagepool(struct gf100_grctx *);
extern const struct gf100_grctx_func gp102_grctx;
/* context init value lists */ /* context init value lists */
......
...@@ -29,7 +29,7 @@ ...@@ -29,7 +29,7 @@
* PGRAPH context implementation * PGRAPH context implementation
******************************************************************************/ ******************************************************************************/
static void void
gp100_grctx_generate_pagepool(struct gf100_grctx *info) gp100_grctx_generate_pagepool(struct gf100_grctx *info)
{ {
const struct gf100_grctx_func *grctx = info->gr->func->grctx; const struct gf100_grctx_func *grctx = info->gr->func->grctx;
...@@ -123,7 +123,7 @@ gp100_grctx_generate_405b60(struct gf100_gr *gr) ...@@ -123,7 +123,7 @@ gp100_grctx_generate_405b60(struct gf100_gr *gr)
nvkm_wr32(device, 0x405ba0 + (i * 4), gpcs[i]); nvkm_wr32(device, 0x405ba0 + (i * 4), gpcs[i]);
} }
static void void
gp100_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) gp100_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info)
{ {
struct nvkm_device *device = gr->base.engine.subdev.device; struct nvkm_device *device = gr->base.engine.subdev.device;
......
/*
* Copyright 2016 Red Hat Inc.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
* Authors: Ben Skeggs <bskeggs@redhat.com>
*/
#include "ctxgf100.h"
#include <subdev/fb.h>
/*******************************************************************************
* PGRAPH context implementation
******************************************************************************/
static void
gp102_grctx_generate_attrib(struct gf100_grctx *info)
{
struct gf100_gr *gr = info->gr;
const struct gf100_grctx_func *grctx = gr->func->grctx;
const u32 alpha = grctx->alpha_nr;
const u32 attrib = grctx->attrib_nr;
const u32 pertpc = 0x20 * (grctx->attrib_nr_max + grctx->alpha_nr_max);
const u32 size = roundup(gr->tpc_total * pertpc, 0x80);
const u32 access = NV_MEM_ACCESS_RW;
const int s = 12;
const int b = mmio_vram(info, size, (1 << s), access);
const int max_batches = 0xffff;
u32 ao = 0;
u32 bo = ao + grctx->alpha_nr_max * gr->tpc_total;
int gpc, ppc, n = 0;
mmio_refn(info, 0x418810, 0x80000000, s, b);
mmio_refn(info, 0x419848, 0x10000000, s, b);
mmio_refn(info, 0x419c2c, 0x10000000, s, b);
mmio_refn(info, 0x419b00, 0x00000000, s, b);
mmio_wr32(info, 0x419b04, 0x80000000 | size >> 7);
mmio_wr32(info, 0x405830, attrib);
mmio_wr32(info, 0x40585c, alpha);
mmio_wr32(info, 0x4064c4, ((alpha / 4) << 16) | max_batches);
for (gpc = 0; gpc < gr->gpc_nr; gpc++) {
for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++, n++) {
const u32 as = alpha * gr->ppc_tpc_nr[gpc][ppc];
const u32 bs = attrib * gr->ppc_tpc_nr[gpc][ppc];
const u32 u = 0x418ea0 + (n * 0x04);
const u32 o = PPC_UNIT(gpc, ppc, 0);
const u32 p = GPC_UNIT(gpc, 0xc44 + (ppc * 4));
if (!(gr->ppc_mask[gpc] & (1 << ppc)))
continue;
mmio_wr32(info, o + 0xc0, bs);
mmio_wr32(info, p, bs);
mmio_wr32(info, o + 0xf4, bo);
mmio_wr32(info, o + 0xf0, bs);
bo += grctx->attrib_nr_max * gr->ppc_tpc_nr[gpc][ppc];
mmio_wr32(info, o + 0xe4, as);
mmio_wr32(info, o + 0xf8, ao);
ao += grctx->alpha_nr_max * gr->ppc_tpc_nr[gpc][ppc];
mmio_wr32(info, u, bs);
}
}
mmio_wr32(info, 0x4181e4, 0x00000100);
mmio_wr32(info, 0x41befc, 0x00000100);
}
const struct gf100_grctx_func
gp102_grctx = {
.main = gp100_grctx_generate_main,
.unkn = gk104_grctx_generate_unkn,
.bundle = gm107_grctx_generate_bundle,
.bundle_size = 0x3000,
.bundle_min_gpm_fifo_depth = 0x180,
.bundle_token_limit = 0x900,
.pagepool = gp100_grctx_generate_pagepool,
.pagepool_size = 0x20000,
.attrib = gp102_grctx_generate_attrib,
.attrib_nr_max = 0x5d4,
.attrib_nr = 0x320,
.alpha_nr_max = 0xc00,
.alpha_nr = 0x800,
};
...@@ -124,6 +124,7 @@ struct gf100_gr_func { ...@@ -124,6 +124,7 @@ struct gf100_gr_func {
void (*init_gpc_mmu)(struct gf100_gr *); void (*init_gpc_mmu)(struct gf100_gr *);
void (*init_rop_active_fbps)(struct gf100_gr *); void (*init_rop_active_fbps)(struct gf100_gr *);
void (*init_ppc_exceptions)(struct gf100_gr *); void (*init_ppc_exceptions)(struct gf100_gr *);
void (*init_swdx_pes_mask)(struct gf100_gr *);
void (*set_hww_esr_report_mask)(struct gf100_gr *); void (*set_hww_esr_report_mask)(struct gf100_gr *);
const struct gf100_gr_pack *mmio; const struct gf100_gr_pack *mmio;
struct { struct {
...@@ -150,6 +151,9 @@ int gk20a_gr_init(struct gf100_gr *); ...@@ -150,6 +151,9 @@ int gk20a_gr_init(struct gf100_gr *);
int gm200_gr_init(struct gf100_gr *); int gm200_gr_init(struct gf100_gr *);
int gm200_gr_rops(struct gf100_gr *); int gm200_gr_rops(struct gf100_gr *);
int gp100_gr_init(struct gf100_gr *);
void gp100_gr_init_rop_active_fbps(struct gf100_gr *);
#define gf100_gr_chan(p) container_of((p), struct gf100_gr_chan, object) #define gf100_gr_chan(p) container_of((p), struct gf100_gr_chan, object)
struct gf100_gr_chan { struct gf100_gr_chan {
......
...@@ -30,7 +30,7 @@ ...@@ -30,7 +30,7 @@
* PGRAPH engine/subdev functions * PGRAPH engine/subdev functions
******************************************************************************/ ******************************************************************************/
static void void
gp100_gr_init_rop_active_fbps(struct gf100_gr *gr) gp100_gr_init_rop_active_fbps(struct gf100_gr *gr)
{ {
struct nvkm_device *device = gr->base.engine.subdev.device; struct nvkm_device *device = gr->base.engine.subdev.device;
...@@ -40,7 +40,7 @@ gp100_gr_init_rop_active_fbps(struct gf100_gr *gr) ...@@ -40,7 +40,7 @@ gp100_gr_init_rop_active_fbps(struct gf100_gr *gr)
nvkm_mask(device, 0x408958, 0x0000000f, fbp_count); /* crop */ nvkm_mask(device, 0x408958, 0x0000000f, fbp_count); /* crop */
} }
static int int
gp100_gr_init(struct gf100_gr *gr) gp100_gr_init(struct gf100_gr *gr)
{ {
struct nvkm_device *device = gr->base.engine.subdev.device; struct nvkm_device *device = gr->base.engine.subdev.device;
...@@ -85,6 +85,8 @@ gp100_gr_init(struct gf100_gr *gr) ...@@ -85,6 +85,8 @@ gp100_gr_init(struct gf100_gr *gr)
nvkm_wr32(device, GPC_BCAST(0x033c), nvkm_rd32(device, 0x100804)); nvkm_wr32(device, GPC_BCAST(0x033c), nvkm_rd32(device, 0x100804));
gr->func->init_rop_active_fbps(gr); gr->func->init_rop_active_fbps(gr);
if (gr->func->init_swdx_pes_mask)
gr->func->init_swdx_pes_mask(gr);
nvkm_wr32(device, 0x400500, 0x00010001); nvkm_wr32(device, 0x400500, 0x00010001);
nvkm_wr32(device, 0x400100, 0xffffffff); nvkm_wr32(device, 0x400100, 0xffffffff);
......
/*
* Copyright 2016 Red Hat Inc.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
* Authors: Ben Skeggs <bskeggs@redhat.com>
*/
#include "gf100.h"
#include "ctxgf100.h"
#include <nvif/class.h>
static void
gp102_gr_init_swdx_pes_mask(struct gf100_gr *gr)
{
struct nvkm_device *device = gr->base.engine.subdev.device;
u32 mask = 0, data, gpc;
for (gpc = 0; gpc < gr->gpc_nr; gpc++) {
data = nvkm_rd32(device, GPC_UNIT(gpc, 0x0c50)) & 0x0000000f;
mask |= data << (gpc * 4);
}
nvkm_wr32(device, 0x4181d0, mask);
}
static const struct gf100_gr_func
gp102_gr = {
.init = gp100_gr_init,
.init_gpc_mmu = gm200_gr_init_gpc_mmu,
.init_rop_active_fbps = gp100_gr_init_rop_active_fbps,
.init_ppc_exceptions = gk104_gr_init_ppc_exceptions,
.init_swdx_pes_mask = gp102_gr_init_swdx_pes_mask,
.rops = gm200_gr_rops,
.ppc_nr = 3,
.grctx = &gp102_grctx,
.sclass = {
{ -1, -1, FERMI_TWOD_A },
{ -1, -1, KEPLER_INLINE_TO_MEMORY_B },
{ -1, -1, PASCAL_B, &gf100_fermi },
{ -1, -1, PASCAL_COMPUTE_B },
{}
}
};
int
gp102_gr_new(struct nvkm_device *device, int index, struct nvkm_gr **pgr)
{
return gm200_gr_new_(&gp102_gr, device, index, pgr);
}
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment