48-bit PPGTT
Subsystems (3)
v2 cover 48-bit PPGTT
These are the rebased patches, after Mika's ppgtt clean-up series (and reusing the macros added). New functions also follow these changes.
In order expand the GPU address space, a 4th level translation is added, the Page Map Level 4 (PML4). This PML4 has 256 PML4 Entries (PML4E), PML4[0-255], each pointing to a PDP. All the existing "dynamic alloc ppgtt" functions are used, only adding the 4th level changes. I also updated some remaining variables that were 32b only.
There are 2 hardware workarounds needed to allow correct operation with 48b addresses (Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset). I added a flag (I915_EXEC_SUPPORT_48BADDRESS) that will indicate if a given object can be allocated outside the first 4 PDPs; if that's the case, insert will use the DRM_MM_CREATE_TOP flag. I'm also including an igt test for this change.
This feature is only available in BDW and Gen9, requires LRC submission mode (execlists) and setting i915.enable_ppgtt=3.
Also note that this expanded address space is only available for full PPGTT, aliasing PPGTT remains 32b.
Michel Thierry (18):
drm/i915/lrc: Update PDPx registers with lri commands drm/i915/gtt: Switch gen8_free_page_tables params drm/i915: Remove unnecessary gen8_clamp_pd
drm/i915/gen8: Make pdp allocation more dynamic drm/i915/gen8: Abstract PDP usage
drm/i915/gen8: Add dynamic page trace events drm/i915/gen8: implement alloc/free for 4lvl drm/i915/gen8: Add 4 level switching infrastructure and lrc support drm/i915/gen8: Generalize PTE writing for GEN8 PPGTT drm/i915/gen8: Pass sg_iter through pte inserts drm/i915/gen8: Add 4 level support in insert_entries and clear_range drm/i915/gen8: Initialize PDPs
drm/i915: Expand error state's address width to 64b drm/i915/gen8: Add ppgtt info and debug_dump
drm/i915: object size needs to be u64
drm/i915: Check against correct user_size limit in 48b ppgtt mode drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset drm/i915/gen8: Flip the 48b switch
drivers/gpu/drm/i915/i915_debugfs.c | 18 +- drivers/gpu/drm/i915/i915_drv.h | 12 +- drivers/gpu/drm/i915/i915_gem.c | 24 +- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 36 +- drivers/gpu/drm/i915/i915_gem_gtt.c | 674 ++++++++++++++++++++++++----- drivers/gpu/drm/i915/i915_gem_gtt.h | 66 ++- drivers/gpu/drm/i915/i915_gem_userptr.c | 12 +- drivers/gpu/drm/i915/i915_gpu_error.c | 17 +- drivers/gpu/drm/i915/i915_params.c | 2 +- drivers/gpu/drm/i915/i915_reg.h | 5 +- drivers/gpu/drm/i915/i915_trace.h | 16 + drivers/gpu/drm/i915/intel_lrc.c | 93 +++- include/uapi/drm/i915_drm.h | 4 +- 13 files changed, 809 insertions(+), 170 deletions(-)
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 1/18 drm/i915/lrc: Update PDPx registers with lri commands
A safer way to update the PDPx registers is sending lri commands, added in the ring before the batchbuffer start. Otherwise, the ctx must be idle before trying to change anything (but the ring-tail) in the ctx image. An example where the ctx won't be idle is lite-restore.
This patch depends on [1], and has the advantage that it doesn't require to pre-allocate the top pdps like here [2].
[1] http://mid.gmane.org/[email protected]
[2] http://mid.gmane.org/[email protected]
v2: Combine lri writes (and save 8 commands). (Mika)
--- drivers/gpu/drm/i915/intel_lrc.c | 43 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+)
drivers/gpu/drm/i915/intel_lrc.c +43 -0
@@ -1116,13 +1116,56 @@ static int gen9_init_render_ring(struct intel_engine_cs *ring)return init_workarounds_ring (ring); } +static int intel_logical_ring_emit_pdps (struct intel_engine_cs *ring, +struct intel_context *ctx) +{ +struct i915_hw_ppgtt *ppgtt = ctx->ppgtt; +struct intel_ringbuffer *ringbuf = ctx->engine[ring->id].ringbuf; +const int num_lri_cmds =GEN8_LEGACY_PDPES *2 ; +int i, ret; + + ret =intel_logical_ring_begin (ringbuf, ctx, num_lri_cmds *2 +2 ); +if (ret) +return ret; + +intel_logical_ring_emit (ringbuf,MI_LOAD_REGISTER_IMM (num_lri_cmds)); +for (i =GEN8_LEGACY_PDPES -1 ; i >=0 ; i--) { +const dma_addr_t pd_daddr =i915_page_dir_dma_addr (ppgtt, i); + +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_UDW (ring, i)); +intel_logical_ring_emit (ringbuf,upper_32_bits (pd_daddr)); +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_LDW (ring, i)); +intel_logical_ring_emit (ringbuf,lower_32_bits (pd_daddr)); + } + +intel_logical_ring_emit (ringbuf,MI_NOOP ); +intel_logical_ring_advance (ringbuf); + +return 0 ; +} +static int gen8_emit_bb_start (struct intel_ringbuffer *ringbuf,struct intel_context *ctx,u64 offset,unsigned dispatch_flags) { +struct intel_engine_cs *ring = ringbuf->ring;bool ppgtt = !(dispatch_flags &I915_DISPATCH_SECURE );int ret; +/* Don't rely in hw updating PDPs, specially in lite-restore. +* Ideally, we should set Force PD Restore in ctx descriptor, +* but we can't. Force Restore would be a second option, but +* it is unsafe in case of lite-restore (because the ctx is +* not idle). */ +if (ctx->ppgtt && + (intel_ring_flag (ring) & ctx->ppgtt->pd_dirty_rings)) { + ret =intel_logical_ring_emit_pdps (ring, ctx); +if (ret) +return ret; + + ctx->ppgtt->pd_dirty_rings &= ~intel_ring_flag (ring); + } + ret =intel_logical_ring_begin (ringbuf, ctx,4 );if (ret)return ret;
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 01/18] drm/i915/lrc: Update PDPx registers with lri commands
Quoted 60 lines · Michel Thierry
A safer way to update the PDPx registers is sending lri commands, added in the ring before the batchbuffer start. Otherwise, the ctx must be idle before trying to change anything (but the ring-tail) in the ctx image. An example where the ctx won't be idle is lite-restore.
This patch depends on [1], and has the advantage that it doesn't require to pre-allocate the top pdps like here [2].
[1] http://mid.gmane.org/[email protected]
[2] http://mid.gmane.org/[email protected]v2: Combine lri writes (and save 8 commands). (Mika)
Cc Dave Gordon <[email protected]>Cc Mika Kuoppala <[email protected]>Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/intel_lrc.c | 43 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+)drivers/gpu/drm/i915/intel_lrc.c +28
-0@@ -1116,13 +1116,56 @@ static int gen9_init_render_ring(struct intel_engine_cs *ring)return init_workarounds_ring (ring); } +static int intel_logical_ring_emit_pdps (struct intel_engine_cs *ring, +struct intel_context *ctx) +{ +struct i915_hw_ppgtt *ppgtt = ctx->ppgtt; +struct intel_ringbuffer *ringbuf = ctx->engine[ring->id].ringbuf; +const int num_lri_cmds =GEN8_LEGACY_PDPES *2 ; +int i, ret; + + ret =intel_logical_ring_begin (ringbuf, ctx, num_lri_cmds *2 +2 ); +if (ret) +return ret; + +intel_logical_ring_emit (ringbuf,MI_LOAD_REGISTER_IMM (num_lri_cmds)); +for (i =GEN8_LEGACY_PDPES -1 ; i >=0 ; i--) { +const dma_addr_t pd_daddr =i915_page_dir_dma_addr (ppgtt, i); + +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_UDW (ring, i)); +intel_logical_ring_emit (ringbuf,upper_32_bits (pd_daddr)); +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_LDW (ring, i)); +intel_logical_ring_emit (ringbuf,lower_32_bits (pd_daddr)); + } + +intel_logical_ring_emit (ringbuf,MI_NOOP ); +intel_logical_ring_advance (ringbuf); + +return 0 ; +} +static int gen8_emit_bb_start (struct intel_ringbuffer *ringbuf,struct intel_context *ctx,u64 offset,unsigned dispatch_flags) {
drivers/gpu/drm/i915/intel_lrc.c static int gen9_init_render_ring(struct intel_engine_cs *ring) · line 1160
Quoted 10 lines
drivers/gpu/drm/i915/intel_lrc.c +7
-0+struct intel_engine_cs *ring = ringbuf->ring;bool ppgtt = !(dispatch_flags &I915_DISPATCH_SECURE );int ret; +/* Don't rely in hw updating PDPs, specially in lite-restore. +* Ideally, we should set Force PD Restore in ctx descriptor, +* but we can't. Force Restore would be a second option, but +* it is unsafe in case of lite-restore (because the ctx is +* not idle). */ + if (ctx->ppgtt &&
Is this superfluous? Can the ctx->ppgtt ever be null with execlists?
-Mika
Quoted 18 lines
drivers/gpu/drm/i915/intel_lrc.c +8
-0+ (intel_ring_flag(ring) & ctx->ppgtt->pd_dirty_rings)) { + ret =intel_logical_ring_emit_pdps (ring, ctx); +if (ret) +return ret; + + ctx->ppgtt->pd_dirty_rings &= ~intel_ring_flag (ring); + } + ret =intel_logical_ring_begin (ringbuf, ctx,4 );if (ret)return ret;Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 01/18] drm/i915/lrc: Update PDPx registers with lri commands
Quoted 66 lines · Mika Kuoppala
Quoted 64 lines · Michel Thierry
A safer way to update the PDPx registers is sending lri commands, added in the ring before the batchbuffer start. Otherwise, the ctx must be idle before trying to change anything (but the ring-tail) in the ctx image. An example where the ctx won't be idle is lite-restore.
This patch depends on [1], and has the advantage that it doesn't require to pre-allocate the top pdps like here [2].
[1] http://mid.gmane.org/[email protected]
[2] http://mid.gmane.org/[email protected]v2: Combine lri writes (and save 8 commands). (Mika)
Cc Dave Gordon <[email protected]>Cc Mika Kuoppala <[email protected]>Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/intel_lrc.c | 43 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+)drivers/gpu/drm/i915/intel_lrc.c +29
-0@@ -1116,13 +1116,56 @@ static int gen9_init_render_ring(struct intel_engine_cs *ring)return init_workarounds_ring (ring); } +static int intel_logical_ring_emit_pdps (struct intel_engine_cs *ring, +struct intel_context *ctx) +{ +struct i915_hw_ppgtt *ppgtt = ctx->ppgtt; +struct intel_ringbuffer *ringbuf = ctx->engine[ring->id].ringbuf; +const int num_lri_cmds =GEN8_LEGACY_PDPES *2 ; +int i, ret; + + ret =intel_logical_ring_begin (ringbuf, ctx, num_lri_cmds *2 +2 ); +if (ret) +return ret; + +intel_logical_ring_emit (ringbuf,MI_LOAD_REGISTER_IMM (num_lri_cmds)); +for (i =GEN8_LEGACY_PDPES -1 ; i >=0 ; i--) { +const dma_addr_t pd_daddr =i915_page_dir_dma_addr (ppgtt, i); + +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_UDW (ring, i)); +intel_logical_ring_emit (ringbuf,upper_32_bits (pd_daddr)); +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_LDW (ring, i)); +intel_logical_ring_emit (ringbuf,lower_32_bits (pd_daddr)); + } + +intel_logical_ring_emit (ringbuf,MI_NOOP ); +intel_logical_ring_advance (ringbuf); + +return 0 ; +} +static int gen8_emit_bb_start (struct intel_ringbuffer *ringbuf,struct intel_context *ctx,u64 offset,unsigned dispatch_flags) { +struct intel_engine_cs *ring = ringbuf->ring;bool ppgtt = !(dispatch_flags &I915_DISPATCH_SECURE );int ret;
Quoted 9 lines
drivers/gpu/drm/i915/intel_lrc.c static int gen9_init_render_ring(struct intel_engine_cs *ring) · line 1160
Quoted 6 lines
drivers/gpu/drm/i915/intel_lrc.c +6
-0+/* Don't rely in hw updating PDPs, specially in lite-restore. +* Ideally, we should set Force PD Restore in ctx descriptor, +* but we can't. Force Restore would be a second option, but +* it is unsafe in case of lite-restore (because the ctx is +* not idle). */ + if (ctx->ppgtt &&Is this superfluous? Can the
ctx->ppgttever be null with execlists?
It's for execlists with aliasing ppgtt. In that case ctx->ppgtt is null (and we shouldn't need to update the pdps).
-Michel
Quoted 17 lines
-Mika
Quoted 14 lines
drivers/gpu/drm/i915/intel_lrc.c +8
-1+ (intel_ring_flag(ring) & ctx->ppgtt->pd_dirty_rings)) { + ret =intel_logical_ring_emit_pdps (ring, ctx); +if (ret) +return ret; + + ctx->ppgtt->pd_dirty_rings &= ~intel_ring_flag (ring); + } + ret =intel_logical_ring_begin (ringbuf, ctx,4 );if (ret)return ret; --2.4.0
Quoted 4 lines
Quoted 4 lines
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 1/1 drm/i915/lrc: Update PDPx registers with lri commands
In reply to Michel ThierryA safer way to update the PDPx registers is sending lri commands, added in the ring before the batchbuffer start. Otherwise, the ctx must be idle before trying to change anything (but the ring-tail) in the ctx image. An example where the ctx won't be idle is lite-restore.
This patch depends on 5b7e4c9ce ("drm/i915/gtt: Mark TLBS dirty for gen8+").
v2: Combine lri writes (and save 8 commands). (Mika)
v3: Rebase after ring/req changes, and removed references to deprecated patches.
--- drivers/gpu/drm/i915/intel_lrc.c | 42 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 42 insertions(+)
drivers/gpu/drm/i915/intel_lrc.c +42 -0
@@ -1379,6 +1379,34 @@ static int gen9_init_render_ring(struct intel_engine_cs *ring)return init_workarounds_ring (ring); } +static int intel_logical_ring_emit_pdps (struct drm_i915_gem_request *req) +{ +struct i915_hw_ppgtt *ppgtt = req->ctx->ppgtt; +struct intel_engine_cs *ring = req->ring; +struct intel_ringbuffer *ringbuf = req->ringbuf; +const int num_lri_cmds =GEN8_LEGACY_PDPES *2 ; +int i, ret; + + ret =intel_logical_ring_begin (req, num_lri_cmds *2 +2 ); +if (ret) +return ret; + +intel_logical_ring_emit (ringbuf,MI_LOAD_REGISTER_IMM (num_lri_cmds)); +for (i =GEN8_LEGACY_PDPES -1 ; i >=0 ; i--) { +const dma_addr_t pd_daddr =i915_page_dir_dma_addr (ppgtt, i); + +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_UDW (ring, i)); +intel_logical_ring_emit (ringbuf,upper_32_bits (pd_daddr)); +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_LDW (ring, i)); +intel_logical_ring_emit (ringbuf,lower_32_bits (pd_daddr)); + } + +intel_logical_ring_emit (ringbuf,MI_NOOP ); +intel_logical_ring_advance (ringbuf); + +return 0 ; +} +static int gen8_emit_bb_start (struct drm_i915_gem_request *req,u64 offset,unsigned dispatch_flags) {
@@ -1386,6 +1414,20 @@ static int gen8_emit_bb_start(struct drm_i915_gem_request *req,bool ppgtt = !(dispatch_flags &I915_DISPATCH_SECURE );int ret; +/* Don't rely in hw updating PDPs, specially in lite-restore. +* Ideally, we should set Force PD Restore in ctx descriptor, +* but we can't. Force Restore would be a second option, but +* it is unsafe in case of lite-restore (because the ctx is +* not idle). */ +if (req->ctx->ppgtt && + (intel_ring_flag (req->ring) & req->ctx->ppgtt->pd_dirty_rings)) { + ret =intel_logical_ring_emit_pdps (req); +if (ret) +return ret; + + req->ctx->ppgtt->pd_dirty_rings &= ~intel_ring_flag (req->ring); + } + ret =intel_logical_ring_begin (req,4 );if (ret)return ret;
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3] drm/i915/lrc: Update PDPx registers with lri commands
Quoted 5 lines · Michel Thierry
A safer way to update the PDPx registers is sending lri commands, added in the ring before the batchbuffer start. Otherwise, the ctx must be idle before trying to change anything (but the ring-tail) in the ctx image. An example where the ctx won't be idle is lite-restore.
Quoted 8 lines
This patch depends on
5b7e4c9ce("drm/i915/gtt: Mark TLBS dirty for gen8+").v2: Combine lri writes (and save 8 commands). (Mika)
v3: Rebase after ring/req changes, and removed references to deprecated patches.Cc Dave Gordon <[email protected]>Cc Mika Kuoppala <[email protected]>Signed-off-by Michel Thierry <[email protected]>
Quoted 66 lines
--- drivers/gpu/drm/i915/intel_lrc.c | 42 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 42 insertions(+)drivers/gpu/drm/i915/intel_lrc.c +42
-0@@ -1379,6 +1379,34 @@ static int gen9_init_render_ring(struct intel_engine_cs *ring)return init_workarounds_ring (ring); } +static int intel_logical_ring_emit_pdps (struct drm_i915_gem_request *req) +{ +struct i915_hw_ppgtt *ppgtt = req->ctx->ppgtt; +struct intel_engine_cs *ring = req->ring; +struct intel_ringbuffer *ringbuf = req->ringbuf; +const int num_lri_cmds =GEN8_LEGACY_PDPES *2 ; +int i, ret; + + ret =intel_logical_ring_begin (req, num_lri_cmds *2 +2 ); +if (ret) +return ret; + +intel_logical_ring_emit (ringbuf,MI_LOAD_REGISTER_IMM (num_lri_cmds)); +for (i =GEN8_LEGACY_PDPES -1 ; i >=0 ; i--) { +const dma_addr_t pd_daddr =i915_page_dir_dma_addr (ppgtt, i); + +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_UDW (ring, i)); +intel_logical_ring_emit (ringbuf,upper_32_bits (pd_daddr)); +intel_logical_ring_emit (ringbuf,GEN8_RING_PDP_LDW (ring, i)); +intel_logical_ring_emit (ringbuf,lower_32_bits (pd_daddr)); + } + +intel_logical_ring_emit (ringbuf,MI_NOOP ); +intel_logical_ring_advance (ringbuf); + +return 0 ; +} +static int gen8_emit_bb_start (struct drm_i915_gem_request *req,u64 offset,unsigned dispatch_flags) {@@ -1386,6 +1414,20 @@ static int gen8_emit_bb_start(struct drm_i915_gem_request *req,bool ppgtt = !(dispatch_flags &I915_DISPATCH_SECURE );int ret; +/* Don't rely in hw updating PDPs, specially in lite-restore. +* Ideally, we should set Force PD Restore in ctx descriptor, +* but we can't. Force Restore would be a second option, but +* it is unsafe in case of lite-restore (because the ctx is +* not idle). */ +if (req->ctx->ppgtt && + (intel_ring_flag (req->ring) & req->ctx->ppgtt->pd_dirty_rings)) { + ret =intel_logical_ring_emit_pdps (req); +if (ret) +return ret; + + req->ctx->ppgtt->pd_dirty_rings &= ~intel_ring_flag (req->ring); + } + ret =intel_logical_ring_begin (req,4 );if (ret)return ret;Signature · 1 line
2.4.5
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 2/18 drm/i915/gtt: Switch gen8_free_page_tables params
In reply to Michel ThierryAfter Mika's ppgtt cleanup series, all the other free functions have drm_device as the first parameter, except this one.
No functional changes.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +4 -2
@@ -766,7 +766,8 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,kunmap_px (ppgtt, pt_vaddr); } -static void gen8_free_page_tables (struct i915_page_directory *pd,struct drm_device *dev) +static void gen8_free_page_tables (struct drm_device *dev, +struct i915_page_directory *pd) {int i;
@@ -792,7 +793,8 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm)if (WARN_ON (!ppgtt->pdp.page_directory[i]))continue ; -gen8_free_page_tables (ppgtt->pdp.page_directory[i], ppgtt->base.dev); +gen8_free_page_tables (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]);free_pd (ppgtt->base.dev, ppgtt->pdp.page_directory[i]); }
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 02/18] drm/i915/gtt: Switch gen8_free_page_tables params
Quoted 6 lines · Michel Thierry
After Mika's ppgtt cleanup series, all the other free functions have
drm_deviceas the first parameter, except this one.No functional changes.
Signed-off-by Michel Thierry <[email protected]>
Quoted 35 lines
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)drivers/gpu/drm/i915/i915_gem_gtt.c +4
-2@@ -766,7 +766,8 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,kunmap_px (ppgtt, pt_vaddr); } -static void gen8_free_page_tables (struct i915_page_directory *pd,struct drm_device *dev) +static void gen8_free_page_tables (struct drm_device *dev, +struct i915_page_directory *pd) {int i;@@ -792,7 +793,8 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm)if (WARN_ON (!ppgtt->pdp.page_directory[i]))continue ; -gen8_free_page_tables (ppgtt->pdp.page_directory[i], ppgtt->base.dev); +gen8_free_page_tables (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]);free_pd (ppgtt->base.dev, ppgtt->pdp.page_directory[i]); }Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 02/18] drm/i915/gtt: Switch gen8_free_page_tables params
Quoted 10 lines · Mika Kuoppala
Quoted 6 lines · Michel Thierry
After Mika's ppgtt cleanup series, all the other free functions have
drm_deviceas the first parameter, except this one.No functional changes.
Signed-off-by Michel Thierry <[email protected]>Reviewed-by Mika Kuoppala <[email protected]>
First two patches merged, thanks.
-Daniel
Quoted 40 lines
Quoted 36 lines
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)drivers/gpu/drm/i915/i915_gem_gtt.c +4
-2@@ -766,7 +766,8 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,kunmap_px (ppgtt, pt_vaddr); } -static void gen8_free_page_tables (struct i915_page_directory *pd,struct drm_device *dev) +static void gen8_free_page_tables (struct drm_device *dev, +struct i915_page_directory *pd) {int i;@@ -792,7 +793,8 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm)if (WARN_ON (!ppgtt->pdp.page_directory[i]))continue ; -gen8_free_page_tables (ppgtt->pdp.page_directory[i], ppgtt->base.dev); +gen8_free_page_tables (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]);free_pd (ppgtt->base.dev, ppgtt->pdp.page_directory[i]); }Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 3/18 drm/i915: Remove unnecessary gen8_clamp_pd
In reply to Michel Thierrygen8_clamp_pd clamps to the next page directory boundary, but the macro gen8_for_each_pde already has a check to stop at the page directory boundary.
Furthermore, i915_pte_count also restricts to the next page table boundary.
v2: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 2 +- drivers/gpu/drm/i915/i915_gem_gtt.h | 11 ----------- 2 files changed, 1 insertion(+), 12 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
@@ -1023,7 +1023,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,gen8_for_each_pdpe (pd, &ppgtt->pdp, start, length, temp, pdpe) {gen8_pde_t *const page_directory =kmap_px (pd);struct i915_page_table *pt; -uint64_t pd_len =gen8_clamp_pd (start, length); +uint64_t pd_len = length;uint64_t pd_start = start;uint32_t pde;
drivers/gpu/drm/i915/i915_gem_gtt.h +0 -11
@@ -438,17 +438,6 @@ static inline uint32_t gen6_pde_index(uint32_t addr) temp = min(temp, length), \ start += temp, length -= temp) -/* Clamp length to the next page_directory boundary */ -static inline uint64_t gen8_clamp_pd (uint64_t start,uint64_t length) -{ -uint64_t next_pd =ALIGN (start +1 ,1 <<GEN8_PDPE_SHIFT ); - -if (next_pd > (start + length)) -return length; - -return next_pd - start; -} -static inline uint32_t gen8_pte_index (uint64_t address) {return i915_pte_index (address,GEN8_PDE_SHIFT );
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 4/18 drm/i915/gen8: Make pdp allocation more dynamic
In reply to Michel ThierryThis transitional patch doesn't do much for the existing code. However, it should make upcoming patches to use the full 48b address space a bit easier. The patch also introduces the PML4, ie. the new top level structure of the page tables.
v2: Renamed pdp_free to be similar to pd/pt (unmap_and_free_pdp).
v3: To facilitate testing, 48b mode will be available on Broadwell and GEN9+, when i915.enable_ppgtt = 3.
v4: Rebase after s/page_tables/page_table/, added extra information about 4-level page table formats and use IS_ENABLED macro.
v5: Check CONFIG_X86_64 instead of CONFIG_64BIT.
v6: Rebase after Mika's ppgtt cleanup / scratch merge patch series, and follow his nomenclature in pdp functions (there is no alloc_pdp yet).
--- drivers/gpu/drm/i915/i915_drv.h | 7 +- drivers/gpu/drm/i915/i915_gem_gtt.c | 125 ++++++++++++++++++++++++++++-------- drivers/gpu/drm/i915/i915_gem_gtt.h | 41 +++++++++--- 3 files changed, 135 insertions(+), 38 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +6 -1
@@ -2436,7 +2436,12 @@ struct drm_i915_cmd_table {#define HAS_HW_CONTEXTS (dev) (INTEL_INFO(dev)->gen >= 6)#define HAS_LOGICAL_RING_CONTEXTS (dev) (INTEL_INFO(dev)->gen >= 8)#define USES_PPGTT (dev) (i915.enable_ppgtt) -#define USES_FULL_PPGTT (dev) (i915.enable_ppgtt == 2) +#define USES_FULL_PPGTT (dev) (i915.enable_ppgtt >= 2) +#ifdef CONFIG_X86_64 +# define USES_FULL_48BIT_PPGTT (dev) (i915.enable_ppgtt == 3) +#else +# define USES_FULL_48BIT_PPGTT (dev) false +#endif #define HAS_OVERLAY (dev) (INTEL_INFO(dev)->has_overlay)#define OVERLAY_NEEDS_PHYSICAL (dev) (INTEL_INFO(dev)->overlay_needs_physical)
drivers/gpu/drm/i915/i915_gem_gtt.c +97 -28
271 lines not on this page · raw message
drivers/gpu/drm/i915/i915_gem_gtt.h +32 -9
78 lines not on this page · raw message
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 5/18 drm/i915/gen8: Abstract PDP usage
In reply to Michel ThierryUp until now, ppgtt->pdp has always been the root of our page tables. Legacy 32b addresses acted like it had 1 PDP with 4 PDPEs.
In preparation for 4 level page tables, we need to stop use ppgtt->pdp directly unless we know it's what we want. The future structure will use ppgtt->pml4 for the top level, and the pdp is just one of the entries being pointed to by a pml4e.
v2: Updated after dynamic page allocation changes.
v3: Rebase after s/page_tables/page_table/.
v4: Rebase after changes in "Dynamic page table allocations" patch.
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 138 +++++++++++++++++++++++------------- 1 file changed, 89 insertions(+), 49 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +89 -49
@@ -495,6 +495,25 @@ static void gen8_initialize_pd(struct i915_address_space *vm,fill_px (vm->dev, pd, scratch_pde); } +/* It's likely we'll map more than one page table at a time. This function will +* save us unnecessary kmap calls, but do no more functionally than multiple +* calls to pde_encode. The ppgtt is only needed to reuse the kunmap macro. */ +static void gen8_map_pagetable_range (struct i915_hw_ppgtt *ppgtt, +struct i915_page_directory *pd, +uint64_t start, +uint64_t length) +{ +gen8_pde_t *const page_directory =kmap_px (pd); +struct i915_page_table *pt; +uint64_t temp, pde; + +gen8_for_each_pde (pt, pd, start, length, temp, pde) + page_directory[pde] =gen8_pde_encode (px_dma (pt), +I915_CACHE_LLC ); + +kunmap_px (ppgtt, page_directory); +} +static int __pdp_init (struct drm_device *dev,struct i915_page_directory_pointer *pdp) {
@@ -721,6 +740,7 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ gen8_pte_t *pt_vaddr, scratch_pte;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;
@@ -735,10 +755,10 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm,struct i915_page_directory *pd;struct i915_page_table *pt; -if (WARN_ON (!ppgtt->pdp.page_directory[pdpe])) +if (WARN_ON (!pdp->page_directory[pdpe]))continue ; - pd = ppgtt->pdp.page_directory[pdpe]; + pd = pdp->page_directory[pdpe];if (WARN_ON (!pd->page_table[pde]))continue ;
@@ -776,6 +796,7 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ gen8_pte_t *pt_vaddr;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;
@@ -786,7 +807,7 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,for_each_sg_page (pages->sgl, &sg_iter, pages->nents,0 ) {if (pt_vaddr ==NULL ) { -struct i915_page_directory *pd = ppgtt->pdp.page_directory[pdpe]; +struct i915_page_directory *pd = pdp->page_directory[pdpe];struct i915_page_table *pt = pd->page_table[pde]; pt_vaddr =kmap_px (pt); }
@@ -832,23 +853,28 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm) container_of(vm, struct i915_hw_ppgtt, base);int i; -for_each_set_bit (i, ppgtt->pdp.used_pdpes, -I915_PDPES_PER_PDP (ppgtt->base.dev)) { -if (WARN_ON (!ppgtt->pdp.page_directory[i])) -continue ; +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { +for_each_set_bit (i, ppgtt->pdp.used_pdpes, +I915_PDPES_PER_PDP (ppgtt->base.dev)) { +if (WARN_ON (!ppgtt->pdp.page_directory[i])) +continue ; -gen8_free_page_tables (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); -free_pd (ppgtt->base.dev, ppgtt->pdp.page_directory[i]); +gen8_free_page_tables (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]); +free_pd (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]); + } +free_pdp (ppgtt->base.dev, &ppgtt->pdp); + }else { +WARN_ON (1 );/* to be implemented later */ } -free_pdp (ppgtt->base.dev, &ppgtt->pdp);cleanup_scratch (vm); }/** * gen8_ppgtt_alloc_pagetabs() - Allocate page tables for VA range. -* @ppgtt: Master ppgtt structure. +* @vm: Master vm structure. * @pd: Page directory for this address range. * @start: Starting virtual address to begin allocations. * @length Size of the allocations.
@@ -864,13 +890,15 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm)* * Return: 0 if success; negative error code otherwise. */ -static int gen8_ppgtt_alloc_pagetabs (struct i915_hw_ppgtt *ppgtt, +static int gen8_ppgtt_alloc_pagetabs (struct i915_address_space *vm,struct i915_page_directory *pd,uint64_t start,uint64_t length,unsigned long *new_pts) { -struct drm_device *dev = ppgtt->base.dev; +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct drm_device *dev = vm->dev;struct i915_page_table *pt;uint64_t temp;uint32_t pde;
@@ -887,7 +915,7 @@ static int gen8_ppgtt_alloc_pagetabs(struct i915_hw_ppgtt *ppgtt,if (IS_ERR (pt)) goto unwind_out; -gen8_initialize_pt (&ppgtt->base, pt); +gen8_initialize_pt (vm, pt); pd->page_table[pde] = pt;__set_bit (pde, new_pts); }
@@ -903,7 +931,7 @@ unwind_out:/** * gen8_ppgtt_alloc_page_directories() - Allocate page directories for VA range. -* @ppgtt: Master ppgtt structure. +* @vm: Master vm structure. * @pdp: Page directory pointer for this address range. * @start: Starting virtual address to begin allocations. * @length Size of the allocations.
@@ -924,17 +952,18 @@ unwind_out:* * Return: 0 if success; negative error code otherwise. */ -static int gen8_ppgtt_alloc_page_directories (struct i915_hw_ppgtt *ppgtt, -struct i915_page_directory_pointer *pdp, -uint64_t start, -uint64_t length, -unsigned long *new_pds) +static int +gen8_ppgtt_alloc_page_directories (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +uint64_t start, +uint64_t length, +unsigned long *new_pds) { -struct drm_device *dev = ppgtt->base.dev; +struct drm_device *dev = vm->dev;struct i915_page_directory *pd;uint64_t temp;uint32_t pdpe; -uint32_t pdpes =I915_PDPES_PER_PDP (ppgtt->base.dev); +uint32_t pdpes =I915_PDPES_PER_PDP (vm->dev);WARN_ON (!bitmap_empty (new_pds, pdpes));
@@ -946,7 +975,7 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,if (IS_ERR (pd)) goto unwind_out; -gen8_initialize_pd (&ppgtt->base, pd); +gen8_initialize_pd (vm, pd); pdp->page_directory[pdpe] = pd;__set_bit (pdpe, new_pds); }
@@ -1020,13 +1049,15 @@ static void mark_tlbs_dirty(struct i915_hw_ppgtt *ppgtt) ppgtt->pd_dirty_rings =INTEL_INFO (ppgtt->base.dev)->ring_mask; } -static int gen8_alloc_va_range (struct i915_address_space *vm, -uint64_t start, -uint64_t length) +static int gen8_alloc_va_range_3lvl (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +uint64_t start, +uint64_t length) {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);unsigned long *new_page_dirs, **new_page_tables; +struct drm_device *dev = vm->dev;struct i915_page_directory *pd;const uint64_t orig_start = start;const uint64_t orig_length = length;
@@ -1049,16 +1080,15 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,return ret;/* Do the allocations first so we can easily bail out */ - ret =gen8_ppgtt_alloc_page_directories (ppgtt, &ppgtt->pdp, start, length, - new_page_dirs); + ret =gen8_ppgtt_alloc_page_directories (vm, pdp, start, length, + new_page_dirs);if (ret) {free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);return ret; } -/* For every page directory referenced, allocate page tables */ -gen8_for_each_pdpe (pd, &ppgtt->pdp, start, length, temp, pdpe) { - ret =gen8_ppgtt_alloc_pagetabs (ppgtt, pd, start, length, +gen8_for_each_pdpe (pd, pdp, start, length, temp, pdpe) { + ret =gen8_ppgtt_alloc_pagetabs (vm, pd, start, length, new_page_tables[pdpe]);if (ret) goto err_out;
@@ -1067,10 +1097,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm, start = orig_start; length = orig_length; -/* Allocations have completed successfully, so set the bitmaps, and do -* the mappings. */ -gen8_for_each_pdpe (pd, &ppgtt->pdp, start, length, temp, pdpe) { -gen8_pde_t *const page_directory =kmap_px (pd); +gen8_for_each_pdpe (pd, pdp, start, length, temp, pdpe) {struct i915_page_table *pt;uint64_t pd_len = length;uint64_t pd_start = start;
@@ -1092,18 +1119,10 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,/* Our pde is now pointing to the pagetable, pt */ __set_bit (pde, pd->used_pdes); - -/* Map the PDE to the page table */ - page_directory[pde] =gen8_pde_encode (px_dma (pt), -I915_CACHE_LLC ); - -/* NB: We haven't yet mapped ptes to pages. At this -* point we're still relying on insert_entries() */ } -kunmap_px (ppgtt, page_directory); - -__set_bit (pdpe, ppgtt->pdp.used_pdpes); +__set_bit (pdpe, pdp->used_pdpes); +gen8_map_pagetable_range (ppgtt, pd, start, length); }free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);
@@ -1113,17 +1132,38 @@ static int gen8_alloc_va_range(struct i915_address_space *vm, err_out:while (pdpe--) {for_each_set_bit (temp, new_page_tables[pdpe],I915_PDES ) -free_pt (vm->dev, ppgtt->pdp.page_directory[pdpe]->page_table[temp]); +free_pt (dev, pdp->page_directory[pdpe]->page_table[temp]); }for_each_set_bit (pdpe, new_page_dirs, pdpes) -free_pd (vm->dev, ppgtt->pdp.page_directory[pdpe]); +free_pd (dev, pdp->page_directory[pdpe]);free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);mark_tlbs_dirty (ppgtt);return ret; } +static int gen8_alloc_va_range_4lvl (struct i915_address_space *vm, +struct i915_pml4 *pml4, +uint64_t start, +uint64_t length) +{ +WARN_ON (1 );/* to be implemented later */ +return 0 ; +} + +static int gen8_alloc_va_range (struct i915_address_space *vm, +uint64_t start,uint64_t length) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); + +if (!USES_FULL_48BIT_PPGTT (vm->dev)) +return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); +else +return gen8_alloc_va_range_4lvl (vm, &ppgtt->pml4, start, length); +} +/* With some architectures and 32bit legacy mode, hardware pre-loads the * top level pdps but the tlb invalidation only invalidates the lower levels. * This might lead to hw fetching with stale pdp entries if top level
@@ -1158,7 +1198,7 @@ static int gen8_preallocate_top_level_pdps(struct i915_hw_ppgtt *ppgtt)/* Allocate for all pdps regardless of how the ppgtt * was defined. */ - ret =gen8_ppgtt_alloc_page_directories (ppgtt, &ppgtt->pdp, + ret =gen8_ppgtt_alloc_page_directories (&ppgtt->base, &ppgtt->pdp,0 ,1ULL <<32 , new_page_dirs);
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 6/18 drm/i915/gen8: Add dynamic page trace events
In reply to Michel ThierryThe dynamic page allocation patch series added it for GEN6, this patch adds them for GEN8.
v2: Consolidate pagetable/page_directory events
v3: Multiple rebases.
v4: Rebase after s/page_tables/page_table/.
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 9 ++++++++- drivers/gpu/drm/i915/i915_trace.h | 16 ++++++++++++++++ 2 files changed, 24 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +8 -1
@@ -507,9 +507,14 @@ static void gen8_map_pagetable_range(struct i915_hw_ppgtt *ppgtt,struct i915_page_table *pt;uint64_t temp, pde; -gen8_for_each_pde (pt, pd, start, length, temp, pde) +gen8_for_each_pde (pt, pd, start, length, temp, pde) { page_directory[pde] =gen8_pde_encode (px_dma (pt),I915_CACHE_LLC ); +trace_i915_page_table_entry_map (&ppgtt->base, pde, pt, +gen8_pte_index (start), +gen8_pte_count (start, length), +GEN8_PTES ); + }kunmap_px (ppgtt, page_directory); }
@@ -918,6 +923,7 @@ static int gen8_ppgtt_alloc_pagetabs(struct i915_address_space *vm,gen8_initialize_pt (vm, pt); pd->page_table[pde] = pt;__set_bit (pde, new_pts); +trace_i915_page_table_entry_alloc (vm, pde, start,GEN8_PDE_SHIFT ); }return 0 ;
@@ -978,6 +984,7 @@ gen8_ppgtt_alloc_page_directories(struct i915_address_space *vm,gen8_initialize_pd (vm, pd); pdp->page_directory[pdpe] = pd;__set_bit (pdpe, new_pds); +trace_i915_page_directory_entry_alloc (vm, pdpe, start,GEN8_PDPE_SHIFT ); }return 0 ;
drivers/gpu/drm/i915/i915_trace.h +16 -0
@@ -213,6 +213,22 @@ DEFINE_EVENT(i915_page_table_entry, i915_page_table_entry_alloc,TP_ARGS (vm, pde, start, pde_shift) ); +DEFINE_EVENT_PRINT (i915_page_table_entry, i915_page_directory_entry_alloc, + TP_PROTO(struct i915_address_space *vm, u32 pdpe, u64 start, u64 pdpe_shift), +TP_ARGS (vm, pdpe, start, pdpe_shift), + +TP_printk ("vm=%p, pdpe=%d (0x%llx-0x%llx)" , + __entry->vm, __entry->pde, __entry->start, __entry->end) +); + +DEFINE_EVENT_PRINT (i915_page_table_entry, i915_page_directory_pointer_entry_alloc, + TP_PROTO(struct i915_address_space *vm, u32 pml4e, u64 start, u64 pml4e_shift), +TP_ARGS (vm, pml4e, start, pml4e_shift), + +TP_printk ("vm=%p, pml4e=%d (0x%llx-0x%llx)" , + __entry->vm, __entry->pde, __entry->start, __entry->end) +); +/* Avoid extra math because we only support two sizes. The format is defined by * bitmap_scnprintf. Each 32 bits is 8 HEX digits followed by comma */ #define TRACE_PT_SIZE (bits) \
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 7/18 drm/i915/gen8: implement alloc/free for 4lvl
In reply to Michel ThierryPML4 has no special attributes, and there will always be a PML4. So simply initialize it at creation, and destroy it at the end.
The code for 4lvl is able to call into the existing 3lvl page table code to handle all of the lower levels.
v2: Return something at the end of gen8_alloc_va_range_4lvl to keep the compiler happy. And define ret only in one place.
Updated gen8_ppgtt_unmap_pages and gen8_ppgtt_free to handle 4lvl.
v3: Use i915_dma_unmap_single instead of pci API. Fix a couple of incorrect checks when unmapping pdp and pd pages (Akash).
v4: Call __pdp_fini also for 32b PPGTT. Clean up alloc_pdp param list.
v5: Prevent (harmless) out of range access in gen8_for_each_pml4e.
v6: Simplify alloc_vma_range_4lvl and gen8_ppgtt_init_common error paths. (Akash)
v7: Rebase, s/gen8_ppgtt_free_*/gen8_ppgtt_cleanup_*/.
v8: Change location of pml4_init/fini. It will make next patches cleaner.
v9: Rebase after Mika's ppgtt cleanup / scratch merge patch series, while trying to reuse as much as possible for pdp alloc. pml4_init/fini replaced by setup/cleanup_px macros.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 161 ++++++++++++++++++++++++++++++------ drivers/gpu/drm/i915/i915_gem_gtt.h | 12 ++- 2 files changed, 145 insertions(+), 28 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +135 -26
@@ -550,12 +550,44 @@ static void __pdp_fini(struct i915_page_directory_pointer *pdp) pdp->page_directory =NULL ; } +static struct +i915_page_directory_pointer *alloc_pdp (struct drm_device *dev) +{ +struct i915_page_directory_pointer *pdp; +int ret = -ENOMEM ; + +WARN_ON (!USES_FULL_48BIT_PPGTT (dev)); + + pdp =kzalloc (sizeof (*pdp),GFP_KERNEL ); +if (!pdp) +return ERR_PTR (-ENOMEM ); + + ret =__pdp_init (dev, pdp); +if (ret) + goto fail_bitmap; + + ret =setup_px (dev, pdp); +if (ret) + goto fail_page_m; + +return pdp; + +fail_page_m: +__pdp_fini (pdp); +fail_bitmap: +kfree (pdp); + +return ERR_PTR (ret); +} +static void free_pdp (struct drm_device *dev,struct i915_page_directory_pointer *pdp) {__pdp_fini (pdp); -if (USES_FULL_48BIT_PPGTT (dev)) +if (USES_FULL_48BIT_PPGTT (dev)) { +cleanup_px (dev, pdp);kfree (pdp); + } }#define SCRATCH_PAGE_MAGIC 0xffff00ffffff00ffULL
@@ -852,28 +884,46 @@ static void gen8_free_page_tables(struct drm_device *dev, } } -static void gen8_ppgtt_cleanup (struct i915_address_space *vm) +static void gen8_ppgtt_cleanup_3lvl (struct drm_device *dev, +struct i915_page_directory_pointer *pdp) { -struct i915_hw_ppgtt *ppgtt = - container_of(vm, struct i915_hw_ppgtt, base);int i; -if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { -for_each_set_bit (i, ppgtt->pdp.used_pdpes, -I915_PDPES_PER_PDP (ppgtt->base.dev)) { -if (WARN_ON (!ppgtt->pdp.page_directory[i])) -continue ; +for_each_set_bit (i, pdp->used_pdpes,I915_PDPES_PER_PDP (dev)) { +if (WARN_ON (!pdp->page_directory[i])) +continue ; -gen8_free_page_tables (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); -free_pd (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); - } -free_pdp (ppgtt->base.dev, &ppgtt->pdp); - }else { -WARN_ON (1 );/* to be implemented later */ +gen8_free_page_tables (dev, pdp->page_directory[i]); +free_pd (dev, pdp->page_directory[i]); } +free_pdp (dev, pdp); +} + +static void gen8_ppgtt_cleanup_4lvl (struct i915_hw_ppgtt *ppgtt) +{ +int i; + +for_each_set_bit (i, ppgtt->pml4.used_pml4es,GEN8_PML4ES_PER_PML4 ) { +if (WARN_ON (!ppgtt->pml4.pdps[i])) +continue ; + +gen8_ppgtt_cleanup_3lvl (ppgtt->base.dev, ppgtt->pml4.pdps[i]); + } + +cleanup_px (ppgtt->base.dev, &ppgtt->pml4); +} + +static void gen8_ppgtt_cleanup (struct i915_address_space *vm) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); + +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) +gen8_ppgtt_cleanup_3lvl (ppgtt->base.dev, &ppgtt->pdp); +else +gen8_ppgtt_cleanup_4lvl (ppgtt); +cleanup_scratch (vm); }
@@ -1155,8 +1205,62 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm,uint64_t start, uint64_t length) { -WARN_ON (1 );/* to be implemented later */ +DECLARE_BITMAP (new_pdps ,GEN8_PML4ES_PER_PML4 ); +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp; +const uint64_t orig_start = start; +const uint64_t orig_length = length; +uint64_t temp, pml4e; +int ret =0 ; + +/* Do the pml4 allocations first, so we don't need to track the newly +* allocated tables below the pdp */ +bitmap_zero (new_pdps,GEN8_PML4ES_PER_PML4 ); + +/* The pagedirectory and pagetable allocations are done in the shared 3 +* and 4 level code. Just allocate the pdps. +*/ +gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) { +if (!pdp) { +WARN_ON (test_bit (pml4e, pml4->used_pml4es)); + pdp =alloc_pdp (vm->dev); +if (IS_ERR (pdp)) + goto err_out; + + pml4->pdps[pml4e] = pdp; +__set_bit (pml4e, new_pdps); +trace_i915_page_directory_pointer_entry_alloc (&ppgtt->base, pml4e, + pml4e <<GEN8_PML4E_SHIFT , +GEN8_PML4E_SHIFT ); + } + } + +WARN (bitmap_weight (new_pdps,GEN8_PML4ES_PER_PML4 ) >2 , +"The allocation has spanned more than 512GB. " +"It is highly likely this is incorrect." ); + + start = orig_start; + length = orig_length; + +gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) { +WARN_ON (!pdp); + + ret =gen8_alloc_va_range_3lvl (vm, pdp, start, length); +if (ret) + goto err_out; + } + +bitmap_or (pml4->used_pml4es, new_pdps, pml4->used_pml4es, +GEN8_PML4ES_PER_PML4 ); +return 0 ; + +err_out: +for_each_set_bit (pml4e, new_pdps,GEN8_PML4ES_PER_PML4 ) +gen8_ppgtt_cleanup_3lvl (vm->dev, pml4->pdps[pml4e]); + +return ret; } static int gen8_alloc_va_range(struct i915_address_space *vm,
@@ -1165,10 +1269,10 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -if (!USES_FULL_48BIT_PPGTT (vm->dev)) -return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); -else +if (USES_FULL_48BIT_PPGTT (vm->dev))return gen8_alloc_va_range_4lvl (vm, &ppgtt->pml4, start, length); +else +return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); }/* With some architectures and 32bit legacy mode, hardware pre-loads the
@@ -1243,13 +1347,21 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)if (ret)return ret; -if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { - ret =__pdp_init (false, &ppgtt->pdp); +if (USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { + ret =setup_px (ppgtt->base.dev, &ppgtt->pml4); +if (ret) + goto clear_scratch; + ppgtt->base.total =1ULL <<48 ; + }else { + ret =__pdp_init (false, &ppgtt->pdp);if (ret) goto clear_scratch; ppgtt->base.total =1ULL <<32 ; +trace_i915_page_directory_pointer_entry_alloc (&ppgtt->base, +0 ,0 , +GEN8_PML4E_SHIFT );if (hw_wont_flush_pdp_tlbs (ppgtt)) {/* Avoid the tlb flush bug by preallocating * whole top level pdp structure so it stays
@@ -1261,9 +1373,6 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)if (ret) goto clear_pdp; } - }else { - ppgtt->base.total =1ULL <<48 ; -return -EPERM ;/* Not yet implemented */ }return 0 ;
drivers/gpu/drm/i915/i915_gem_gtt.h +10 -2
@@ -95,6 +95,7 @@ typedef uint64_t gen8_pde_t;*/ #define GEN8_PML4ES_PER_PML4 512#define GEN8_PML4E_SHIFT 39 +#define GEN8_PML4E_MASK (GEN8_PML4ES_PER_PML4 - 1)#define GEN8_PDPE_SHIFT 30/* NB: GEN8_PDPE_MASK is untrue for 32b platforms, but it has no impact on 32b page * tables */
@@ -458,6 +459,14 @@ static inline uint32_t gen6_pde_index(uint32_t addr) temp = min(temp, length), \ start += temp, length -= temp) +#define gen8_for_each_pml4e (pdp, pml4, start, length, temp, iter) \ + for (iter = gen8_pml4e_index(start); \ + pdp = (pml4)->pdps[iter], length > 0 && iter < GEN8_PML4ES_PER_PML4; \ + iter++, \ + temp = ALIGN(start+1, 1ULL << GEN8_PML4E_SHIFT) - start, \ + temp = min(temp, length), \ + start += temp, length -= temp) +#define gen8_for_each_pdpe (pd, pdp, start, length, temp, iter) \ gen8_for_each_pdpe_e(pd, pdp, start, length, temp, iter, I915_PDPES_PER_PDP(dev))
@@ -478,8 +487,7 @@ static inline uint32_t gen8_pdpe_index(uint64_t address)static inline uint32_t gen8_pml4e_index (uint64_t address) { -WARN_ON (1 );/* For 64B */ -return 0 ; +return (address >>GEN8_PML4E_SHIFT ) &GEN8_PML4E_MASK ; }static inline size_t gen8_pte_count (uint64_t address,uint64_t length)
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 8/18 drm/i915/gen8: Add 4 level switching infrastructure and lrc support
In reply to Michel ThierryIn 64b (48bit canonical) PPGTT addressing, the PDP0 register contains the base address to PML4, while the other PDP registers are ignored.
In LRC, the addressing mode must be specified in every context descriptor.
v2: PML4 update in legacy context switch is left for historic reasons, the preferred mode of operation is with lrc context based submission.
v3: s/gen8_map_page_directory/gen8_setup_page_directory and s/gen8_map_page_directory_pointer/gen8_setup_page_directory_pointer. Also, clflush will be needed for bxt. (Akash)
v4: Squashed lrc-specific code and use a macro to set PML4 register.
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series. PDP update in bb_start is only for legacy 32b mode.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 54 ++++++++++++++++++++++++++++++---- drivers/gpu/drm/i915/i915_gem_gtt.h | 2 ++ drivers/gpu/drm/i915/i915_reg.h | 5 +++- drivers/gpu/drm/i915/intel_lrc.c | 58 +++++++++++++++++++++++++++---------- 4 files changed, 96 insertions(+), 23 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +48 -6
@@ -211,6 +211,9 @@ static gen8_pde_t gen8_pde_encode(const dma_addr_t addr,return pde; } +#define gen8_pdpe_encode gen8_pde_encode +#define gen8_pml4e_encode gen8_pde_encode +static gen6_pte_t snb_pte_encode (dma_addr_t addr,enum i915_cache_level level,bool valid,u32 unused)
@@ -590,6 +593,35 @@ static void free_pdp(struct drm_device *dev, } } +static void +gen8_setup_page_directory (struct i915_hw_ppgtt *ppgtt, +struct i915_page_directory_pointer *pdp, +struct i915_page_directory *pd, +int index) +{ +gen8_ppgtt_pdpe_t *page_directorypo; + +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) +return ; + + page_directorypo =kmap_px (pdp); + page_directorypo[index] =gen8_pdpe_encode (px_dma (pd),I915_CACHE_LLC ); +kunmap_px (ppgtt, page_directorypo); +} + +static void +gen8_setup_page_directory_pointer (struct i915_hw_ppgtt *ppgtt, +struct i915_pml4 *pml4, +struct i915_page_directory_pointer *pdp, +int index) +{ +gen8_ppgtt_pml4e_t *pagemap =kmap_px (pml4); + +WARN_ON (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)); + pagemap[index] =gen8_pml4e_encode (px_dma (pdp),I915_CACHE_LLC ); +kunmap_px (ppgtt, pagemap); +} +#define SCRATCH_PAGE_MAGIC 0xffff00ffffff00ffULLstatic int alloc_scratch_page (struct i915_address_space *vm)
@@ -754,8 +786,8 @@ static int gen8_write_pdp(struct intel_engine_cs *ring,return 0 ; } -static int gen8_mm_switch (struct i915_hw_ppgtt *ppgtt, -struct intel_engine_cs *ring) +static int gen8_legacy_mm_switch (struct i915_hw_ppgtt *ppgtt, +struct intel_engine_cs *ring) {int i, ret;
@@ -770,6 +802,12 @@ static int gen8_mm_switch(struct i915_hw_ppgtt *ppgtt,return 0 ; } +static int gen8_48b_mm_switch (struct i915_hw_ppgtt *ppgtt, +struct intel_engine_cs *ring) +{ +return gen8_write_pdp (ring,0 ,px_dma (&ppgtt->pml4)); +} + static void gen8_ppgtt_clear_range(struct i915_address_space *vm, uint64_t start, uint64_t length,
@@ -1180,6 +1218,7 @@ static int gen8_alloc_va_range_3lvl(struct i915_address_space *vm,__set_bit (pdpe, pdp->used_pdpes);gen8_map_pagetable_range (ppgtt, pd, start, length); +gen8_setup_page_directory (ppgtt, pdp, pd, pdpe); }free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);
@@ -1249,6 +1288,8 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm, ret =gen8_alloc_va_range_3lvl (vm, pdp, start, length);if (ret) goto err_out; + +gen8_setup_page_directory_pointer (ppgtt, pml4, pdp, pml4e); } bitmap_or(pml4->used_pml4es, new_pdps, pml4->used_pml4es,
@@ -1284,7 +1325,7 @@ static bool hw_wont_flush_pdp_tlbs(struct i915_hw_ppgtt *ppgtt) {struct drm_device *dev = ppgtt->base.dev; -if (GEN8_CTX_ADDRESSING_MODE !=LEGACY_32B_CONTEXT ) +if (GEN8_CTX_ADDRESSING_MODE (dev) !=LEGACY_32B_CONTEXT )return false;if (IS_GEN8 (dev) ||IS_GEN9 (dev))
@@ -1341,8 +1382,6 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) ppgtt->base.unbind_vma = ppgtt_unbind_vma; ppgtt->base.bind_vma = ppgtt_bind_vma; - ppgtt->switch_mm = gen8_mm_switch; - ret =setup_scratch (&ppgtt->base);if (ret)return ret;
@@ -1353,12 +1392,14 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) goto clear_scratch; ppgtt->base.total =1ULL <<48 ; + ppgtt->switch_mm = gen8_48b_mm_switch; }else { ret =__pdp_init (false, &ppgtt->pdp);if (ret) goto clear_scratch; ppgtt->base.total =1ULL <<32 ; + ppgtt->switch_mm = gen8_legacy_mm_switch;trace_i915_page_directory_pointer_entry_alloc (&ppgtt->base,0 ,0 ,GEN8_PML4E_SHIFT );
@@ -1565,8 +1606,9 @@ static void gen8_ppgtt_enable(struct drm_device *dev)int j;for_each_ring (ring, dev_priv, j) { +u32 four_level =USES_FULL_48BIT_PPGTT (dev) ?GEN8_GFX_PPGTT_48B :0 ;I915_WRITE (RING_MODE_GEN7 (ring), -_MASKED_BIT_ENABLE (GFX_PPGTT_ENABLE )); +_MASKED_BIT_ENABLE (GFX_PPGTT_ENABLE | four_level)); } }
drivers/gpu/drm/i915/i915_gem_gtt.h +2 -0
@@ -39,6 +39,8 @@ struct drm_i915_file_private;typedef uint32_t gen6_pte_t ;typedef uint64_t gen8_pte_t ;typedef uint64_t gen8_pde_t ; +typedef uint64_t gen8_ppgtt_pdpe_t ; +typedef uint64_t gen8_ppgtt_pml4e_t ;#define gtt_total_entries (gtt) ((gtt).base.total >> PAGE_SHIFT)
drivers/gpu/drm/i915/i915_reg.h +4 -1
@@ -1642,6 +1642,7 @@ enum skl_disp_power_wells {#define GFX_REPLAY_MODE (1<<11)#define GFX_PSMI_GRANULARITY (1<<10)#define GFX_PPGTT_ENABLE (1<<9) +#define GEN8_GFX_PPGTT_48B (1<<7)#define VLV_DISPLAY_BASE 0x180000#define VLV_MIPI_BASE VLV_DISPLAY_BASE
@@ -2792,7 +2793,9 @@ enum { };#define GEN8_CTX_ADDRESSING_MODE_SHIFT 3 -#define GEN8_CTX_ADDRESSING_MODE LEGACY_32B_CONTEXT +#define GEN8_CTX_ADDRESSING_MODE (dev) (USES_FULL_48BIT_PPGTT(dev) ?\ + LEGACY_64B_CONTEXT :\ + LEGACY_32B_CONTEXT)/* * Overlay regs
drivers/gpu/drm/i915/intel_lrc.c +42 -16
90 lines not on this page · raw message
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 9/18 drm/i915/gen8: Generalize PTE writing for GEN8 PPGTT
In reply to Michel ThierryThe insert_entries function was the function used to write PTEs. For the PPGTT it was "hardcoded" to only understand two level page tables, which was the case for GEN7. We can reuse this for 4 level page tables, and remove the concept of insert_entries, which was never viable past 2 level page tables anyway, but it requires a bit of rework to make the function a bit more generic.
This patch begins the generalization work, and it will be heavily used upon when the 48b code is complete. The patch series attempts to make each function which touches a part of code specific to the page table level and here is no exception.
v2: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 51 ++++++++++++++++++++++++++++--------- 1 file changed, 39 insertions(+), 12 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +39 -12
@@ -808,23 +808,21 @@ static int gen8_48b_mm_switch(struct i915_hw_ppgtt *ppgtt,return gen8_write_pdp (ring,0 ,px_dma (&ppgtt->pml4)); } -static void gen8_ppgtt_clear_range (struct i915_address_space *vm, -uint64_t start, -uint64_t length, -bool use_scratch) +static void gen8_ppgtt_clear_pte_range (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +uint64_t start, +uint64_t length, +gen8_pte_t scratch_pte) {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ -gen8_pte_t *pt_vaddr, scratch_pte; +gen8_pte_t *pt_vaddr;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ;unsigned num_entries = length >>PAGE_SHIFT ;unsigned last_pte, i; - scratch_pte =gen8_pte_encode (px_dma (vm->scratch_page), -I915_CACHE_LLC , use_scratch);while (num_entries) {struct i915_page_directory *pd;
@@ -864,14 +862,30 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm, } } -static void gen8_ppgtt_insert_entries (struct i915_address_space *vm, -struct sg_table *pages, -uint64_t start, -enum i915_cache_level cache_level,u32 unused) +static void gen8_ppgtt_clear_range (struct i915_address_space *vm, +uint64_t start, +uint64_t length, +bool use_scratch) {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ + +gen8_pte_t scratch_pte =gen8_pte_encode (px_dma (vm->scratch_page), +I915_CACHE_LLC , use_scratch); + +gen8_ppgtt_clear_pte_range (vm, pdp, start, length, scratch_pte); +} + +static void +gen8_ppgtt_insert_pte_entries (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +struct sg_table *pages, +uint64_t start, +enum i915_cache_level cache_level) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;
@@ -905,6 +919,19 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,kunmap_px (ppgtt, pt_vaddr); } +static void gen8_ppgtt_insert_entries (struct i915_address_space *vm, +struct sg_table *pages, +uint64_t start, +enum i915_cache_level cache_level, +u32 unused) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ + +gen8_ppgtt_insert_pte_entries (vm, pdp, pages, start, cache_level); +} +static void gen8_free_page_tables (struct drm_device *dev,struct i915_page_directory *pd) {
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 10/18 drm/i915/gen8: Pass sg_iter through pte inserts
In reply to Michel ThierryAs a step towards implementing 4 levels, while not discarding the existing pte insert functions, we need to pass the sg_iter through. The current function understands to the page directory granularity. An object's pages may span the page directory, and so using the iter directly as we write the PTEs allows the iterator to stay coherent through a VMA insert operation spanning multiple page table levels.
v2: Rebase after s/page_tables/page_table/.
v3: Rebase after Mika's ppgtt cleanup / scratch merge patch series; updated commit message (s/map/insert).
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +6 -5
@@ -880,7 +880,7 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm,static void gen8_ppgtt_insert_pte_entries (struct i915_address_space *vm,struct i915_page_directory_pointer *pdp, -struct sg_table *pages, +struct sg_page_iter *sg_iter,uint64_t start,enum i915_cache_level cache_level) {
@@ -890,11 +890,10 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; -struct sg_page_iter sg_iter; pt_vaddr =NULL ; -for_each_sg_page (pages->sgl, &sg_iter, pages->nents,0 ) { +while (__sg_page_iter_next (sg_iter)) {if (pt_vaddr ==NULL ) {struct i915_page_directory *pd = pdp->page_directory[pdpe];struct i915_page_table *pt = pd->page_table[pde];
@@ -902,7 +901,7 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm, } pt_vaddr[pte] = -gen8_pte_encode (sg_page_iter_dma_address (&sg_iter), +gen8_pte_encode (sg_page_iter_dma_address (sg_iter), cache_level, true);if (++pte ==GEN8_PTES ) {kunmap_px (ppgtt, pt_vaddr);
@@ -928,8 +927,10 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ +struct sg_page_iter sg_iter; -gen8_ppgtt_insert_pte_entries (vm, pdp, pages, start, cache_level); +__sg_page_iter_start (&sg_iter, pages->sgl,sg_nents (pages->sgl),0 ); +gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, start, cache_level); } static void gen8_free_page_tables(struct drm_device *dev,
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 11/18 drm/i915/gen8: Add 4 level support in insert_entries and clear_range
In reply to Michel ThierryWhen 48b is enabled, gen8_ppgtt_insert_entries needs to read the Page Map Level 4 (PML4), before it selects which Page Directory Pointer (PDP) it will write to.
Similarly, gen8_ppgtt_clear_range needs to get the correct PDP/PD range.
This patch was inspired by Ben's "Depend exclusively on map and unmap_vma".
v2: Rebase after s/page_tables/page_table/.
v3: Remove unnecessary pdpe loop in gen8_ppgtt_clear_range_4lvl and use clamp_pdp in gen8_ppgtt_insert_entries (Akash).
v4: Merge gen8_ppgtt_clear_range_4lvl into gen8_ppgtt_clear_range to maintain symmetry with gen8_ppgtt_insert_entries (Akash).
v5: Do not mix pages and bytes in insert_entries (Akash).
v6: Prevent overflow in sg_nents << PAGE_SHIFT, when inserting 4GB at once.
v7: Rebase after Mika's ppgtt cleanup / scratch merge patch series. Use gen8_px_index functions, and remove unnecessary number of pages parameter in insert_pte_entries.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 51 ++++++++++++++++++++++++++++--------- drivers/gpu/drm/i915/i915_gem_gtt.h | 11 ++++++++ 2 files changed, 50 insertions(+), 12 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +39 -12
@@ -817,9 +817,9 @@ static void gen8_ppgtt_clear_pte_range(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr; -unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ; -unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ; -unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; +unsigned pdpe =gen8_pdpe_index (start); +unsigned pde =gen8_pde_index (start); +unsigned pte =gen8_pte_index (start);unsigned num_entries = length >>PAGE_SHIFT ;unsigned last_pte, i;
@@ -869,12 +869,24 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ -gen8_pte_t scratch_pte =gen8_pte_encode (px_dma (vm->scratch_page),I915_CACHE_LLC , use_scratch); -gen8_ppgtt_clear_pte_range (vm, pdp, start, length, scratch_pte); +if (!USES_FULL_48BIT_PPGTT (vm->dev)) { +gen8_ppgtt_clear_pte_range (vm, &ppgtt->pdp, start, length, + scratch_pte); + }else { +uint64_t templ4, pml4e; +struct i915_page_directory_pointer *pdp; + +gen8_for_each_pml4e (pdp, &ppgtt->pml4, start, length, templ4, pml4e) { +uint64_t pdp_len =gen8_clamp_pdp (start, length); +uint64_t pdp_start = start; + +gen8_ppgtt_clear_pte_range (vm, pdp, pdp_start, pdp_len, + scratch_pte); + } + } } static void
@@ -887,9 +899,9 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr; -unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ; -unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ; -unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; +unsigned pdpe =gen8_pdpe_index (start); +unsigned pde =gen8_pde_index (start); +unsigned pte =gen8_pte_index (start); pt_vaddr =NULL ;
@@ -907,7 +919,8 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,kunmap_px (ppgtt, pt_vaddr); pt_vaddr =NULL ;if (++pde ==I915_PDES ) { - pdpe++; +if (++pdpe ==I915_PDPES_PER_PDP (vm->dev)) +break ; pde =0 ; } pte =0 ;
@@ -926,11 +939,25 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ struct sg_page_iter sg_iter;__sg_page_iter_start (&sg_iter, pages->sgl,sg_nents (pages->sgl),0 ); -gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, start, cache_level); + +if (!USES_FULL_48BIT_PPGTT (vm->dev)) { +gen8_ppgtt_insert_pte_entries (vm, &ppgtt->pdp, &sg_iter, start, + cache_level); + }else { +struct i915_page_directory_pointer *pdp; +uint64_t templ4, pml4e; +uint64_t length = (uint64_t )sg_nents (pages->sgl) <<PAGE_SHIFT ; + +gen8_for_each_pml4e (pdp, &ppgtt->pml4, start, length, templ4, pml4e) { +uint64_t pdp_start = start; + +gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, + pdp_start, cache_level); + } + } } static void gen8_free_page_tables(struct drm_device *dev,
drivers/gpu/drm/i915/i915_gem_gtt.h +11 -0
@@ -472,6 +472,17 @@ static inline uint32_t gen6_pde_index(uint32_t addr)#define gen8_for_each_pdpe (pd, pdp, start, length, temp, iter) \ gen8_for_each_pdpe_e(pd, pdp, start, length, temp, iter, I915_PDPES_PER_PDP(dev)) +/* Clamp length to the next page_directory pointer boundary */ +static inline uint64_t gen8_clamp_pdp (uint64_t start,uint64_t length) +{ +uint64_t next_pdp =ALIGN (start +1 ,1ULL <<GEN8_PML4E_SHIFT ); + +if (next_pdp > (start + length)) +return length; + +return next_pdp - start; +} +static inline uint32_t gen8_pte_index (uint64_t address) {return i915_pte_index (address,GEN8_PDE_SHIFT );
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 12/18 drm/i915/gen8: Initialize PDPs
In reply to Michel ThierrySimilar to PDs, while setting up a page directory pointer, make all entries of the pdp point to the scratch pdp before mapping (and make all its entries point to the scratch page); this is to be safe in case of out of bound access or proactive prefetch.
Although the ggtt is always 32-bit, the scratch_pdp will be initialized/destroyed at the same time as the other scratch pages, to keep it consistent.
v2: Handle scratch_pdp allocation failure correctly, and keep initialize_px functions together (Akash)
v3: Rebase after Mika's ppgtt cleanup / scratch merge patch series. Rely on the added macros to initialize the pdps.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 53 ++++++++++++++++++++++++++++++++++--- drivers/gpu/drm/i915/i915_gem_gtt.h | 1 + 2 files changed, 51 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +50 -3
@@ -593,6 +593,27 @@ static void free_pdp(struct drm_device *dev, } } +static void gen8_initialize_pdp (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp) +{ +gen8_ppgtt_pdpe_t scratch_pdpe; + + scratch_pdpe =gen8_pdpe_encode (px_dma (vm->scratch_pd),I915_CACHE_LLC ); + +fill_px (vm->dev, pdp, scratch_pdpe); +} + +static void gen8_initialize_pml4 (struct i915_address_space *vm, +struct i915_pml4 *pml4) +{ +gen8_ppgtt_pml4e_t scratch_pml4e; + + scratch_pml4e =gen8_pml4e_encode (px_dma (vm->scratch_pdp), +I915_CACHE_LLC ); + +fill_px (vm->dev, pml4, scratch_pml4e); +} + static void gen8_setup_page_directory(struct i915_hw_ppgtt *ppgtt, struct i915_page_directory_pointer *pdp,
@@ -693,12 +714,30 @@ static int setup_scratch_ggtt(struct i915_address_space *vm)WARN_ON (px_dma (vm->scratch_pt) ==0 );gen8_initialize_pd (vm, vm->scratch_pd); + +/* although scratch_pdp is only needed for 48-bit ppgtt, +* keep it with the other scratch pages for consistency. +*/ +if (USES_FULL_48BIT_PPGTT (dev)) { +WARN_ON (vm->scratch_pdp); + + vm->scratch_pdp =alloc_pdp (vm->dev); +if (IS_ERR (vm->scratch_pdp)) { + ret =PTR_ERR (vm->scratch_pdp); + goto err_pdp; + } + +WARN_ON (px_dma (vm->scratch_pd) ==0 ); +gen8_initialize_pdp (vm, vm->scratch_pdp); + } }else {gen6_initialize_pt (vm, vm->scratch_pt); }return 0 ; +err_pdp: +free_pd (vm->dev, vm->scratch_pd); err_pd:free_pt (vm->dev, vm->scratch_pt);return ret;
@@ -714,6 +753,7 @@ static int setup_scratch(struct i915_address_space *vm) vm->scratch_page = ggtt_vm->scratch_page; vm->scratch_pt = ggtt_vm->scratch_pt; vm->scratch_pd = ggtt_vm->scratch_pd; + vm->scratch_pdp = ggtt_vm->scratch_pdp;return 0 ; }
@@ -748,8 +788,12 @@ static void cleanup_scratch_ggtt(struct i915_address_space *vm)free_pt (vm->dev, vm->scratch_pt); -if (INTEL_INFO (vm->dev)->gen >=8 ) +if (INTEL_INFO (vm->dev)->gen >=8 ) {free_pd (vm->dev, vm->scratch_pd); + +if (USES_FULL_48BIT_PPGTT (dev)) +free_pdp (vm->dev, vm->scratch_pdp); + } }static void cleanup_scratch (struct i915_address_space *vm)
@@ -760,6 +804,7 @@ static void cleanup_scratch(struct i915_address_space *vm) vm->scratch_page =NULL ; vm->scratch_pt =NULL ; vm->scratch_pd =NULL ; + vm->scratch_pdp =NULL ; }/* Broadwell Page Directory Pointer Descriptors */
@@ -1316,12 +1361,12 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm,* and 4 level code. Just allocate the pdps. */ gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) { -if (!pdp) { -WARN_ON (test_bit (pml4e, pml4->used_pml4es)); +if (!test_bit (pml4e, pml4->used_pml4es)) { pdp =alloc_pdp (vm->dev);if (IS_ERR (pdp)) goto err_out; +gen8_initialize_pdp (vm, pdp); pml4->pdps[pml4e] = pdp;__set_bit (pml4e, new_pdps); trace_i915_page_directory_pointer_entry_alloc(&ppgtt->base, pml4e,
@@ -1446,6 +1491,8 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)if (ret) goto clear_scratch; +gen8_initialize_pml4 (&ppgtt->base, &ppgtt->pml4); + ppgtt->base.total =1ULL <<48 ; ppgtt->switch_mm = gen8_48b_mm_switch; }else {
drivers/gpu/drm/i915/i915_gem_gtt.h +1 -0
@@ -272,6 +272,7 @@ struct i915_address_space {struct i915_page_scratch *scratch_page;struct i915_page_table *scratch_pt;struct i915_page_directory *scratch_pd; +struct i915_page_directory_pointer *scratch_pdp;/* GEN8+ & 48b PPGTT */ /** * List of objects currently involved in rendering.
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 13/18 drm/i915: Expand error state's address width to 64b
In reply to Michel Thierry--- drivers/gpu/drm/i915/i915_drv.h | 4 ++-- drivers/gpu/drm/i915/i915_gpu_error.c | 17 +++++++++-------- 2 files changed, 11 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +2 -2
@@ -507,7 +507,7 @@ struct drm_i915_error_state {struct drm_i915_error_object {int page_count; -u32 gtt_offset; +u64 gtt_offset;u32 *pages[0 ]; } *ringbuffer, *batchbuffer, *wa_batchbuffer, *ctx, *hws_page;
@@ -533,7 +533,7 @@ struct drm_i915_error_state {u32 size;u32 name;u32 rseqno[I915_NUM_RINGS ], wseqno; -u32 gtt_offset; +u64 gtt_offset;u32 read_domains;u32 write_domain; s32 fence_reg:I915_MAX_NUM_FENCE_BITS;
drivers/gpu/drm/i915/i915_gpu_error.c +9 -8
@@ -197,7 +197,7 @@ static void print_error_buffers(struct drm_i915_error_state_buf *m,err_printf (m," %s [%d]:\n" , name, count);while (count--) { - err_printf(m, " %08x %8u %02x %02x [ ", + err_printf(m, " %016llx %8u %02x %02x [ ", err->gtt_offset, err->size, err->read_domains,
@@ -426,7 +426,7 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,err_printf (m," (submitted by %s [%d])" , error->ring[i].comm, error->ring[i].pid); -err_printf (m," --- gtt_offset = 0x%08x\n" , +err_printf (m," --- gtt_offset = 0x%016llx\n" , obj->gtt_offset);print_error_obj (m, obj); }
@@ -434,7 +434,8 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m, obj = error->ring[i].wa_batchbuffer;if (obj) {err_printf (m,"%s (w/a) --- gtt_offset = 0x%08x\n" , - dev_priv->ring[i].name, obj->gtt_offset); + dev_priv->ring[i].name, +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); }
@@ -453,14 +454,14 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,if ((obj = error->ring[i].ringbuffer)) {err_printf (m,"%s --- ringbuffer = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); }if ((obj = error->ring[i].hws_page)) {err_printf (m,"%s --- HW Status = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset)); offset =0 ;for (elt =0 ; elt <PAGE_SIZE /16 ; elt +=4 ) { err_printf(m, "[%04x] %08x %08x %08x %08x\n",
@@ -476,13 +477,13 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,if ((obj = error->ring[i].ctx)) {err_printf (m,"%s --- HW Context = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); } }if ((obj = error->semaphore_obj)) { -err_printf (m,"Semaphore page = 0x%08x\n" , obj->gtt_offset); +err_printf (m,"Semaphore page = 0x%016llx\n" , obj->gtt_offset);for (elt =0 ; elt <PAGE_SIZE /16 ; elt +=4 ) { err_printf(m, "[%04x] %08x %08x %08x %08x\n", elt * 4,
@@ -590,7 +591,7 @@ i915_error_object_create(struct drm_i915_private *dev_priv,int num_pages;bool use_ggtt;int i =0 ; -u32 reloc_offset; +u64 reloc_offset;if (src ==NULL || src->pages ==NULL )return NULL ;
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 14/18 drm/i915/gen8: Add ppgtt info and debug_dump
In reply to Michel Thierryv2: Clean up patch after rebases.
v3: gen8_dump_ppgtt for 32b and 48b PPGTT.
v4: Use used_pml4es/pdpes (Akash).
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
--- drivers/gpu/drm/i915/i915_debugfs.c | 18 ++++---- drivers/gpu/drm/i915/i915_gem_gtt.c | 92 +++++++++++++++++++++++++++++++++++++ 2 files changed, 102 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +10 -8
@@ -2227,7 +2227,6 @@ static void gen6_ppgtt_info(struct seq_file *m, struct drm_device *dev) {struct drm_i915_private *dev_priv = dev->dev_private;struct intel_engine_cs *ring; -struct drm_file *file;int i;if (INTEL_INFO (dev)->gen ==6 )
@@ -2250,13 +2249,6 @@ static void gen6_ppgtt_info(struct seq_file *m, struct drm_device *dev) ppgtt->debug_dump (ppgtt, m); } -list_for_each_entry_reverse (file, &dev->filelist, lhead) { -struct drm_i915_file_private *file_priv = file->driver_priv; - -seq_printf (m,"proc: %s\n" , -get_pid_task (file->pid,PIDTYPE_PID )->comm); -idr_for_each (&file_priv->context_idr, per_file_ctx, m); - }seq_printf (m,"ECOCHK: 0x%08x\n" ,I915_READ (GAM_ECOCHK )); }
@@ -2265,6 +2257,7 @@ static int i915_ppgtt_info(struct seq_file *m, void *data)struct drm_info_node *node = m->private;struct drm_device *dev = node->minor->dev;struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_file *file;int ret =mutex_lock_interruptible (&dev->struct_mutex);if (ret)
@@ -2276,6 +2269,15 @@ static int i915_ppgtt_info(struct seq_file *m, void *data)else if (INTEL_INFO (dev)->gen >=6 )gen6_ppgtt_info (m, dev); +list_for_each_entry_reverse (file, &dev->filelist, lhead) { +struct drm_i915_file_private *file_priv = file->driver_priv; + +seq_printf (m,"\nproc: %s\n" , +get_pid_task (file->pid,PIDTYPE_PID )->comm); +idr_for_each (&file_priv->context_idr, per_file_ctx, + (void *)(unsigned long )m); + } +intel_runtime_pm_put (dev_priv);mutex_unlock (&dev->struct_mutex);
drivers/gpu/drm/i915/i915_gem_gtt.c +92 -0
104 lines not on this page · raw message
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 15/18 drm/i915: object size needs to be u64
In reply to Michel ThierryIn a 48b world, users can try to allocate buffers bigger than 4GB; in these cases it is important that size is a 64b variable.
Also added a warning for illegal bind with size = 0.
--- drivers/gpu/drm/i915/i915_gem.c | 5 +++-- drivers/gpu/drm/i915/i915_gem_gtt.c | 3 +++ 2 files changed, 6 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +3 -2
@@ -3670,7 +3670,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, {struct drm_device *dev = obj->base.dev;struct drm_i915_private *dev_priv = dev->dev_private; -u32 size, fence_size, fence_alignment, unfenced_alignment; +u32 fence_alignment, unfenced_alignment; +u64 size, fence_size;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =
@@ -3729,7 +3730,7 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,* attempt to find space. */ if (size > end) { - DRM_DEBUG("Attempting to bind an object (view type=%u) larger than the aperture: size=%u > %s aperture=%llu\n", + DRM_DEBUG("Attempting to bind an object (view type=%u) larger than the aperture: size=%llu > %s aperture=%llu\n", ggtt_view ? ggtt_view->type : 0, size, flags & PIN_MAPPABLE ? "mappable" : "total",
drivers/gpu/drm/i915/i915_gem_gtt.c +3 -0
@@ -3433,6 +3433,9 @@ int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,if (WARN_ON (flags ==0 ))return -EINVAL ; +if (WARN_ON (vma->node.size ==0 )) +return -EINVAL ; + bind_flags =0 ;if (flags &PIN_GLOBAL ) bind_flags |=GLOBAL_BIND ;
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 16/18 drm/i915: Check against correct user_size limit in 48b ppgtt mode
In reply to Michel ThierryGTT is only 32b and its max value is 4GB. In order to allow objects bigger than 4GB in 48b PPGTT, i915_gem_userptr_ioctl needs to check against max 48b range (1ULL << 48).
Whenever possible, read the PPGTT's total instead of the GTT one, this will be accurate in 32 and 48 bit modes.
v2: Use the default ctx to infer the ppgtt max size (Akash).
--- drivers/gpu/drm/i915/i915_gem_userptr.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_userptr.c +10 -2
@@ -789,8 +789,10 @@ int i915_gem_userptr_ioctl(struct drm_device *dev, void *data, struct drm_file *file) { struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_i915_file_private *file_priv = file->driver_priv;struct drm_i915_gem_userptr *args = data;struct drm_i915_gem_object *obj; +struct intel_context *ctx;int ret;u32 handle;
@@ -801,8 +803,14 @@ i915_gem_userptr_ioctl(struct drm_device *dev, void *data, struct drm_file *fileif (offset_in_page (args->user_ptr | args->user_size))return -EINVAL ; -if (args->user_size > dev_priv->gtt.base.total) -return -E2BIG ; + ctx =i915_gem_context_get (file_priv,DEFAULT_CONTEXT_HANDLE ); +if (ctx->ppgtt) { +if (args->user_size > ctx->ppgtt->base.total) +return -E2BIG ; + }else { +if (args->user_size > dev_priv->gtt.base.total) +return -E2BIG ; + }if (!access_ok (args->flags &I915_USERPTR_READ_ONLY ?VERIFY_READ :VERIFY_WRITE , (char __user *)(unsigned long )args->user_ptr, args->user_size))
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 16/18] drm/i915: Check against correct user_size limit in 48b ppgtt mode
Quoted 6 lines · Michel Thierry
GTT is only 32b and its max value is 4GB. In order to allow objects bigger than 4GB in 48b PPGTT,
i915_gem_userptr_ioctlneeds to check against max 48b range (1ULL << 48).Whenever possible, read the PPGTT's total instead of the GTT one, this will be accurate in 32 and 48 bit modes.
Just kill the limit. It is only there for early detection of an error when it is used for execbuffer - however, we may be using the bo for other purposes where the limit doesn't apply. Or the check may be invalid (such as now).
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 17/18 drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
In reply to Michel ThierryThere are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/ DRM_MM_CREATE_TOP flags
User must pass I915_EXEC_SUPPORTS_48BADDRESS flag to indicate it can be allocated above the 32b address range.
The flag is ignored in 32b PPGTT.
v2: Changed flag logic from neeeds_32b, to supports_48b.
--- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 19 ++++++++++++++-- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 36 +++++++++++++++++++++--------- include/uapi/drm/i915_drm.h | 4 +++- 4 files changed, 47 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +1 -0
@@ -2714,6 +2714,7 @@ void i915_gem_vma_destroy(struct i915_vma *vma);#define PIN_OFFSET_BIAS (1<<3)#define PIN_USER (1<<4)#define PIN_UPDATE (1<<5) +#define PIN_FULL_RANGE (1<<6)#define PIN_OFFSET_MASK (~4095) int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,
drivers/gpu/drm/i915/i915_gem.c +17 -2
@@ -3672,6 +3672,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,struct drm_i915_private *dev_priv = dev->dev_private;u32 fence_alignment, unfenced_alignment;u64 size, fence_size; +u32 search_flag =DRM_MM_SEARCH_DEFAULT ; +u32 alloc_flag =DRM_MM_CREATE_DEFAULT ;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =
@@ -3713,6 +3715,19 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, obj->tiling_mode, false); size = flags &PIN_MAPPABLE ? fence_size : obj->base.size; + +/* Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset, +* limit address to 4GB-1 for objects requiring this wa; for +* others, set alloc flag to TOP. +*/ +if (USES_FULL_48BIT_PPGTT (dev)) { +if (flags &PIN_FULL_RANGE ) { + search_flag =DRM_MM_SEARCH_BELOW ; + alloc_flag =DRM_MM_CREATE_TOP ; + }else { + end = ((4ULL <<GEN8_PDPE_SHIFT ) -1 ); + } + } }if (alignment ==0 )
@@ -3755,8 +3770,8 @@ search_free: size, alignment, obj->cache_level, start, end, -DRM_MM_SEARCH_DEFAULT , - DRM_MM_CREATE_DEFAULT); + search_flag, + alloc_flag);if (ret) { ret = i915_gem_evict_something(dev, vm, size, alignment, obj->cache_level,
drivers/gpu/drm/i915/i915_gem_execbuffer.c +26 -10
@@ -577,7 +577,8 @@ static bool only_mappable_for_reloc(unsigned int flags)static int i915_gem_execbuffer_reserve_vma (struct i915_vma *vma,struct intel_engine_cs *ring, -bool *need_reloc) +bool *need_reloc, +bool support_48baddr) {struct drm_i915_gem_object *obj = vma->obj;struct drm_i915_gem_exec_object2 *entry = vma->exec_entry;
@@ -588,6 +589,9 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,if (entry->flags &EXEC_OBJECT_NEEDS_GTT ) flags |=PIN_GLOBAL ; +if (support_48baddr) + flags |=PIN_FULL_RANGE ; +if (!drm_mm_node_allocated (&vma->node)) {if (entry->flags & __EXEC_OBJECT_NEEDS_MAP) flags |=PIN_GLOBAL |PIN_MAPPABLE ;
@@ -650,7 +654,7 @@ need_reloc_mappable(struct i915_vma *vma) }static bool -eb_vma_misplaced (struct i915_vma *vma) +eb_vma_misplaced (struct i915_vma *vma,bool support_48baddr) {struct drm_i915_gem_exec_object2 *entry = vma->exec_entry;struct drm_i915_gem_object *obj = vma->obj;
@@ -670,13 +674,18 @@ eb_vma_misplaced(struct i915_vma *vma)if (entry->flags & __EXEC_OBJECT_NEEDS_MAP && !obj->map_and_fenceable)return !only_mappable_for_reloc (entry->flags); +if (!support_48baddr && + vma->node.start >= (1ULL <<32 )) +return true; +return false; }static int i915_gem_execbuffer_reserve (struct intel_engine_cs *ring,struct list_head *vmas, -bool *need_relocs) +bool *need_relocs, +bool support_48baddr) {struct drm_i915_gem_object *obj;struct i915_vma *vma;
@@ -737,10 +746,11 @@ i915_gem_execbuffer_reserve(struct intel_engine_cs *ring,if (!drm_mm_node_allocated (&vma->node))continue ; -if (eb_vma_misplaced (vma)) +if (eb_vma_misplaced (vma, support_48baddr)) ret =i915_vma_unbind (vma);else - ret =i915_gem_execbuffer_reserve_vma (vma, ring, need_relocs); + ret =i915_gem_execbuffer_reserve_vma (vma, ring, need_relocs, + support_48baddr);if (ret) goto err; }
@@ -750,7 +760,9 @@ i915_gem_execbuffer_reserve(struct intel_engine_cs *ring,if (drm_mm_node_allocated (&vma->node))continue ; - ret =i915_gem_execbuffer_reserve_vma (vma, ring, need_relocs); + ret =i915_gem_execbuffer_reserve_vma (vma, ring, + need_relocs, + support_48baddr);if (ret) goto err; }
@@ -780,7 +792,7 @@ i915_gem_execbuffer_relocate_slow(struct drm_device *dev,struct drm_i915_gem_relocation_entry *reloc;struct i915_address_space *vm;struct i915_vma *vma; -bool need_relocs; +bool need_relocs, support_48baddr;int *reloc_offset;int i, total, ret;unsigned count = args->buffer_count;
@@ -861,7 +873,9 @@ i915_gem_execbuffer_relocate_slow(struct drm_device *dev, goto err; need_relocs = (args->flags &I915_EXEC_NO_RELOC ) ==0 ; - ret =i915_gem_execbuffer_reserve (ring, &eb->vmas, &need_relocs); + support_48baddr = (args->flags &I915_EXEC_SUPPORT_48BADDRESS ); + ret =i915_gem_execbuffer_reserve (ring, &eb->vmas, &need_relocs, + support_48baddr);if (ret) goto err;
@@ -1411,7 +1425,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,u64 exec_start = args->batch_start_offset;u32 dispatch_flags;int ret; -bool need_relocs; +bool need_relocs, support_48baddr;if (!i915_gem_check_execbuffer (args))return -EINVAL ;
@@ -1519,7 +1533,9 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,/* Move the objects en-masse into the GTT, evicting if necessary. */ need_relocs = (args->flags &I915_EXEC_NO_RELOC ) ==0 ; - ret =i915_gem_execbuffer_reserve (ring, &eb->vmas, &need_relocs); + support_48baddr = (args->flags &I915_EXEC_SUPPORT_48BADDRESS ); + ret =i915_gem_execbuffer_reserve (ring, &eb->vmas, &need_relocs, + support_48baddr);if (ret) goto err;
include/uapi/drm/i915_drm.h +3 -1
@@ -760,7 +760,9 @@ struct drm_i915_gem_execbuffer2 {#define I915_EXEC_BSD_RING1 (1<<13)#define I915_EXEC_BSD_RING2 (2<<13) -#define __I915_EXEC_UNKNOWN_FLAGS -(1<<15) +#define I915_EXEC_SUPPORT_48BADDRESS (1<<15) + +#define __I915_EXEC_UNKNOWN_FLAGS -(1<<16)#define I915_EXEC_CONTEXT_ID_MASK (0xffffffff)#define i915_execbuffer2_set_context_id (eb2, context) \
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 7 lines · Michel Thierry
There are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsUser must pass
I915_EXEC_SUPPORTS_48BADDRESSflag to indicate it can be allocated above the 32b address range.
This should be a per-object flag not per-execbuffer.
Quoted 17 lines
The flag is ignored in 32b PPGTT.
v2: Changed flag logic from
neeeds_32b, tosupports_48b.Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 19 ++++++++++++++-- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 36 +++++++++++++++++++++--------- include/uapi/drm/i915_drm.h | 4 +++- 4 files changed, 47 insertions(+), 13 deletions(-)drivers/gpu/drm/i915/i915_drv.h +0
-0
drivers/gpu/drm/i915/i915_drv.h void i915_gem_vma_destroy(struct i915_vma *vma); · line 2717
Quoted 4 lines
drivers/gpu/drm/i915/i915_drv.h +1
-0@@ -2714,6 +2714,7 @@ void i915_gem_vma_destroy(struct i915_vma *vma);#define PIN_OFFSET_BIAS (1<<3)#define PIN_USER (1<<4)#define PIN_UPDATE (1<<5) +#define PIN_FULL_RANGE (1<<6)
Halfway through our flags. We should get around to putting a reminder that 1<<11 is the last flag.
Quoted 21 lines
drivers/gpu/drm/i915/i915_drv.h +0
-0#define PIN_OFFSET_MASK (~4095) int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,drivers/gpu/drm/i915/i915_gem.c +3
-0@@ -3672,6 +3672,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,struct drm_i915_private *dev_priv = dev->dev_private;u32 fence_alignment, unfenced_alignment;u64 size, fence_size; +u32 search_flag =DRM_MM_SEARCH_DEFAULT ; +u32 alloc_flag =DRM_MM_CREATE_DEFAULT ;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =@@ -3713,6 +3715,19 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, obj->tiling_mode, false); size = flags &PIN_MAPPABLE ? fence_size : obj->base.size; +
drivers/gpu/drm/i915/i915_gem.c i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, · line 3728
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem.c +10
-0+/* Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset, +* limit address to 4GB-1 for objects requiring this wa; for +* others, set alloc flag to TOP. +*/ +if (USES_FULL_48BIT_PPGTT (dev)) { +if (flags &PIN_FULL_RANGE ) { + search_flag =DRM_MM_SEARCH_BELOW ; + alloc_flag =DRM_MM_CREATE_TOP ; + }else { + end = ((4ULL <<GEN8_PDPE_SHIFT ) -1 );
Looking at this, I think this is better as two flags. I have used SEARCH_BELOW in the past to try and keep objects out of the mappable aperture. Having that as separate flag is quite useful in its own right.
Then the second flag is PIN_BELOW_4G which we can set by default (and cleared when the user specifies EXEC_OBJECT_48BIT). Not so sure, but I think with the right combination of flags you can avoid having device and w/a specific logic here. (This should be mechanism, push the policy out to the boundary, preferrably into userspace.)
All the i915_gem_execbuffer changes are wrong due to the flag not being on the object.
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 10 lines · Chris Wilson
Quoted 7 lines · Michel Thierry
There are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsUser must pass
I915_EXEC_SUPPORTS_48BADDRESSflag to indicate it can be allocated above the 32b address range.This should be a per-object flag not per-execbuffer.
We need both. This one to opt into the large address space, the per-object one to apply the w/a. Also libdrm/mesa patches for this are still missing. -Daniel
Quoted 78 lines
Quoted 18 lines
The flag is ignored in 32b PPGTT.
v2: Changed flag logic from
neeeds_32b, tosupports_48b.Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 19 ++++++++++++++-- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 36 +++++++++++++++++++++--------- include/uapi/drm/i915_drm.h | 4 +++- 4 files changed, 47 insertions(+), 13 deletions(-)drivers/gpu/drm/i915/i915_drv.h +0
-0drivers/gpu/drm/i915/i915_drv.h void i915_gem_vma_destroy(struct i915_vma *vma); · line 2717
Quoted 4 lines
drivers/gpu/drm/i915/i915_drv.h +1
-0@@ -2714,6 +2714,7 @@ void i915_gem_vma_destroy(struct i915_vma *vma);#define PIN_OFFSET_BIAS (1<<3)#define PIN_USER (1<<4)#define PIN_UPDATE (1<<5) +#define PIN_FULL_RANGE (1<<6)Halfway through our flags. We should get around to putting a reminder that 1<<11 is the last flag.
Quoted 21 lines
drivers/gpu/drm/i915/i915_drv.h +0
-0#define PIN_OFFSET_MASK (~4095) int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,drivers/gpu/drm/i915/i915_gem.c +3
-0@@ -3672,6 +3672,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,struct drm_i915_private *dev_priv = dev->dev_private;u32 fence_alignment, unfenced_alignment;u64 size, fence_size; +u32 search_flag =DRM_MM_SEARCH_DEFAULT ; +u32 alloc_flag =DRM_MM_CREATE_DEFAULT ;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =@@ -3713,6 +3715,19 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, obj->tiling_mode, false); size = flags &PIN_MAPPABLE ? fence_size : obj->base.size; +drivers/gpu/drm/i915/i915_gem.c i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, · line 3728
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem.c +10
-0+/* Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset, +* limit address to 4GB-1 for objects requiring this wa; for +* others, set alloc flag to TOP. +*/ +if (USES_FULL_48BIT_PPGTT (dev)) { +if (flags &PIN_FULL_RANGE ) { + search_flag =DRM_MM_SEARCH_BELOW ; + alloc_flag =DRM_MM_CREATE_TOP ; + }else { + end = ((4ULL <<GEN8_PDPE_SHIFT ) -1 );Looking at this, I think this is better as two flags. I have used
SEARCH_BELOWin the past to try and keep objects out of the mappable aperture. Having that as separate flag is quite useful in its own right.Then the second flag is
PIN_BELOW_4Gwhich we can set by default (and cleared when the user specifiesEXEC_OBJECT_48BIT). Not so sure, but I think with the right combination of flags you can avoid having device and w/a specific logic here. (This should be mechanism, push the policy out to the boundary, preferrably into userspace.)All the
i915_gem_execbufferchanges are wrong due to the flag not being on the object.
-ChrisSignature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 7 lines · Daniel Vetter
Quoted 6 lines · Chris Wilson
Quoted 5 lines · Michel Thierry
There are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflags
Quoted 7 lines
Quoted 4 lines
Quoted 2 lines
User must pass
I915_EXEC_SUPPORTS_48BADDRESSflag to indicate it can be allocated above the 32b address range.This should be a per-object flag not per-execbuffer.
We need both. This one to opt into the large address space, the per-object one to apply the w/a. Also libdrm/mesa patches for this are still missing.
Do we need the opt in on the context? The 48bit vm is lazily constructed, if no object asks to use the high range, it will never be populated. Or is there a cost with preparing a 48bit vm? -Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 11 lines · Chris Wilson
Quoted 10 lines · Daniel Vetter
Quoted 9 lines · Chris Wilson
Quoted 7 lines · Michel Thierry
There are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsUser must pass
I915_EXEC_SUPPORTS_48BADDRESSflag to indicate it can be allocated above the 32b address range.
Quoted 8 lines
Quoted 4 lines
Quoted 1 line
This should be a per-object flag not per-execbuffer.
We need both. This one to opt into the large address space, the per-object one to apply the w/a. Also libdrm/mesa patches for this are still missing.
Do we need the opt in on the context? The 48bit vm is lazily constructed, if no object asks to use the high range, it will never be populated. Or is there a cost with preparing a 48bit vm?
If we restrict to 4G we'll evict objects if we run out, and will stay correct even when processing fairly large workloads. With just lazily eating into 48b that won't be the case. A bit far-fetched, but if we go to the trouble of implementing this might as well do it right. -Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 17 lines · Daniel Vetter
Quoted 16 lines · Chris Wilson
Quoted 14 lines · Daniel Vetter
Quoted 10 lines · Chris Wilson
Quoted 7 lines · Michel Thierry
There are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsUser must pass
I915_EXEC_SUPPORTS_48BADDRESSflag to indicate it can be allocated above the 32b address range.This should be a per-object flag not per-execbuffer.
We need both. This one to opt into the large address space, the per-object one to apply the w/a. Also libdrm/mesa patches for this are still missing.
Quoted 8 lines
Quoted 3 lines
Do we need the opt in on the context? The 48bit vm is lazily constructed, if no object asks to use the high range, it will never be populated. Or is there a cost with preparing a 48bit vm?
If we restrict to 4G we'll evict objects if we run out, and will stay correct even when processing fairly large workloads. With just lazily eating into 48b that won't be the case. A bit far-fetched, but if we go to the trouble of implementing this might as well do it right.
i915_evict_something runs between the range requested for pinning. If we run out of 4G space and the desired pin does not opt into 48bit, we will evict from the lower 4G.
I obviously missed your concern. Care to elaborate? -Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 27 lines · Chris Wilson
Quoted 17 lines · Daniel Vetter
Quoted 16 lines · Chris Wilson
Quoted 14 lines · Daniel Vetter
Quoted 10 lines · Chris Wilson
Quoted 7 lines · Michel Thierry
There are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsUser must pass
I915_EXEC_SUPPORTS_48BADDRESSflag to indicate it can be allocated above the 32b address range.This should be a per-object flag not per-execbuffer.
We need both. This one to opt into the large address space, the per-object one to apply the w/a. Also libdrm/mesa patches for this are still missing.
Quoted 8 lines
Quoted 3 lines
Do we need the opt in on the context? The 48bit vm is lazily constructed, if no object asks to use the high range, it will never be populated. Or is there a cost with preparing a 48bit vm?
If we restrict to 4G we'll evict objects if we run out, and will stay correct even when processing fairly large workloads. With just lazily eating into 48b that won't be the case. A bit far-fetched, but if we go to the trouble of implementing this might as well do it right.
Quoted 5 lines
i915_evict_somethingruns between the range requested for pinning. If we run out of 4G space and the desired pin does not opt into 48bit, we will evict from the lower 4G.I obviously missed your concern. Care to elaborate?
Current situation: You always get an address below 4G for all objects, even if you use more than 4G of textures - the evict code will make space.
New situation with 48b address space enabled but existing userspace and a total BO set bigger than 4G: The kernel will eventually hand out ppgtt addresses > 4G, which means if we get such an address potentially even for an object where this wa needs to apply. This would be a regression. But if we make 48b strictly opt-in the kernel will restrict _all_ objects to below 4G, creating no regression.
Ofc new userspace on 48b would set both the execbuf opt-in (or context flag, we have those now) plus the per-obj "I need this below 4G" flag for the objects that need this wa.
-Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 37 lines · Daniel Vetter
Quoted 27 lines · Chris Wilson
Quoted 17 lines · Daniel Vetter
Quoted 16 lines · Chris Wilson
Quoted 14 lines · Daniel Vetter
Quoted 10 lines · Chris Wilson
Quoted 7 lines · Michel Thierry
There are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsUser must pass
I915_EXEC_SUPPORTS_48BADDRESSflag to indicate it can be allocated above the 32b address range.This should be a per-object flag not per-execbuffer.
We need both. This one to opt into the large address space, the per-object one to apply the w/a. Also libdrm/mesa patches for this are still missing.
Quoted 8 lines
Quoted 3 lines
Do we need the opt in on the context? The 48bit vm is lazily constructed, if no object asks to use the high range, it will never be populated. Or is there a cost with preparing a 48bit vm?
If we restrict to 4G we'll evict objects if we run out, and will stay correct even when processing fairly large workloads. With just lazily eating into 48b that won't be the case. A bit far-fetched, but if we go to the trouble of implementing this might as well do it right.
Quoted 5 lines
i915_evict_somethingruns between the range requested for pinning. If we run out of 4G space and the desired pin does not opt into 48bit, we will evict from the lower 4G.I obviously missed your concern. Care to elaborate?
Current situation: You always get an address below 4G for all objects, even if you use more than 4G of textures - the evict code will make space.
Quoted 6 lines
New situation with 48b address space enabled but existing userspace and a total BO set bigger than 4G: The kernel will eventually hand out ppgtt addresses > 4G, which means if we get such an address potentially even for an object where this wa needs to apply. This would be a regression. But if we make 48b strictly opt-in the kernel will restrict _all_ objects to below 4G, creating no regression.
How? The pin code requires PIN_48BIT to be set to hand out higher addresses. That is only set by execbuffer if execobject->flags is also set.
Quoted 3 lines
Ofc new userspace on 48b would set both the execbuf opt-in (or context flag, we have those now) plus the per-obj "I need this below 4G" flag for the objects that need this wa.
I don't see why we need another flag beyond the per-object flag. If you are thinking validation, we have to validate per-object flags anyway. -Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 45 lines · Chris Wilson
Quoted 37 lines · Daniel Vetter
Quoted 27 lines · Chris Wilson
Quoted 17 lines · Daniel Vetter
Quoted 16 lines · Chris Wilson
Quoted 14 lines · Daniel Vetter
Quoted 10 lines · Chris Wilson
Quoted 7 lines · Michel Thierry
There are some allocations that must be only referenced by 32bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsUser must pass
I915_EXEC_SUPPORTS_48BADDRESSflag to indicate it can be allocated above the 32b address range.This should be a per-object flag not per-execbuffer.
We need both. This one to opt into the large address space, the per-object one to apply the w/a. Also libdrm/mesa patches for this are still missing.
Quoted 8 lines
Quoted 3 lines
Do we need the opt in on the context? The 48bit vm is lazily constructed, if no object asks to use the high range, it will never be populated. Or is there a cost with preparing a 48bit vm?
If we restrict to 4G we'll evict objects if we run out, and will stay correct even when processing fairly large workloads. With just lazily eating into 48b that won't be the case. A bit far-fetched, but if we go to the trouble of implementing this might as well do it right.
Quoted 5 lines
i915_evict_somethingruns between the range requested for pinning. If we run out of 4G space and the desired pin does not opt into 48bit, we will evict from the lower 4G.I obviously missed your concern. Care to elaborate?
Current situation: You always get an address below 4G for all objects, even if you use more than 4G of textures - the evict code will make space.
Quoted 6 lines
New situation with 48b address space enabled but existing userspace and a total BO set bigger than 4G: The kernel will eventually hand out ppgtt addresses > 4G, which means if we get such an address potentially even for an object where this wa needs to apply. This would be a regression. But if we make 48b strictly opt-in the kernel will restrict _all_ objects to below 4G, creating no regression.
Quoted 2 lines
How? The pin code requires
PIN_48BITto be set to hand out higher addresses. That is only set by execbuffer ifexecobject->flagsis also set.
I've been dense, somehow I thought we need the execbuf opt-in with the object opt-out. But opt-in at the object level is indeed all we need. -Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 17/18] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 2 lines · Daniel Vetter
I've been dense, somehow I thought we need the execbuf opt-in with the object opt-out. But opt-in at the object level is indeed all we need.
To be fair and recap our discussion on irc, the other side of the coin is that at some point we want to use 48bit by default (gen9, gen10, whenever it is robust!) Daniel's argument is that with an high level enable bit + opt-out, there is less work in userspace to dtrt.
Imo, having changed userspace to opt-in when possible with gen8, having userspace opt-in for all objects is then trivial (plus it is then easier for userspace to disable it again). Having a flag at the execbuf level would be nice, but given the changes we need in userspace today (to support either the opt-in/opt-out model), I think a second flag is of no practical value.
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 1/1 drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
In reply to Michel ThierryThere are some allocations that must be only referenced by 32-bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/ DRM_MM_CREATE_TOP flags
In specific, any resource used with flat/heapless (0x00000000-0xfffff000) General State Heap (GSH) or Intructions State Heap (ISH) must be in a 32-bit range, because the General State Offset and Instruction State Offset are limited to 32-bits.
Objects must have EXEC_OBJECT_SUPPORTS_48BADDRESS flag to indicate if they can be allocated above the 32-bit address range.
v2: Changed flag logic from neeeds_32b, to supports_48b.
v3: Moved 48-bit support flag back to exec_object. (Chris, Daniel)
--- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 19 +++++++++++++++++-- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 7 +++++++ include/uapi/drm/i915_drm.h | 3 ++- 4 files changed, 27 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +1 -0
@@ -2739,6 +2739,7 @@ void i915_gem_vma_destroy(struct i915_vma *vma);#define PIN_OFFSET_BIAS (1<<3)#define PIN_USER (1<<4)#define PIN_UPDATE (1<<5) +#define PIN_FULL_RANGE (1<<6)#define PIN_OFFSET_MASK (~4095) int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,
drivers/gpu/drm/i915/i915_gem.c +17 -2
@@ -3669,6 +3669,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,struct drm_i915_private *dev_priv = dev->dev_private;u32 fence_alignment, unfenced_alignment;u64 size, fence_size; +u32 search_flag =DRM_MM_SEARCH_DEFAULT ; +u32 alloc_flag =DRM_MM_CREATE_DEFAULT ;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =
@@ -3710,6 +3712,19 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, obj->tiling_mode, false); size = flags &PIN_MAPPABLE ? fence_size : obj->base.size; + +/* Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset, +* limit address to 4GB-1 for objects requiring this wa; for +* others, set alloc flag to TOP. +*/ +if (USES_FULL_48BIT_PPGTT (dev)) { +if (flags &PIN_FULL_RANGE ) { + search_flag =DRM_MM_SEARCH_BELOW ; + alloc_flag =DRM_MM_CREATE_TOP ; + }else { + end = ((4ULL <<GEN8_PDPE_SHIFT ) -1 ); + } + } }if (alignment ==0 )
@@ -3752,8 +3767,8 @@ search_free: size, alignment, obj->cache_level, start, end, -DRM_MM_SEARCH_DEFAULT , - DRM_MM_CREATE_DEFAULT); + search_flag, + alloc_flag);if (ret) { ret = i915_gem_evict_something(dev, vm, size, alignment, obj->cache_level,
drivers/gpu/drm/i915/i915_gem_execbuffer.c +7 -0
@@ -588,6 +588,9 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,if (entry->flags &EXEC_OBJECT_NEEDS_GTT ) flags |=PIN_GLOBAL ; +if (entry->flags &EXEC_OBJECT_SUPPORTS_48BBADDRESS ) + flags |=PIN_FULL_RANGE ; +if (!drm_mm_node_allocated (&vma->node)) {if (entry->flags & __EXEC_OBJECT_NEEDS_MAP) flags |=PIN_GLOBAL |PIN_MAPPABLE ;
@@ -670,6 +673,10 @@ eb_vma_misplaced(struct i915_vma *vma)if (entry->flags & __EXEC_OBJECT_NEEDS_MAP && !obj->map_and_fenceable)return !only_mappable_for_reloc (entry->flags); +if (!(entry->flags &EXEC_OBJECT_SUPPORTS_48BBADDRESS ) && + vma->node.start >= (1ULL <<32 )) +return true; +return false; }
include/uapi/drm/i915_drm.h +2 -1
@@ -685,7 +685,8 @@ struct drm_i915_gem_exec_object2 {#define EXEC_OBJECT_NEEDS_FENCE (1<<0)#define EXEC_OBJECT_NEEDS_GTT (1<<1)#define EXEC_OBJECT_WRITE (1<<2) -#define __EXEC_OBJECT_UNKNOWN_FLAGS -(EXEC_OBJECT_WRITE<<1) +#define EXEC_OBJECT_SUPPORTS_48BBADDRESS (1<<3) +#define __EXEC_OBJECT_UNKNOWN_FLAGS -(EXEC_OBJECT_SUPPORTS_48BBADDRESS<<1)__u64 flags;__u64 rsvd1;
Signature · 6 lines
2.4.3 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 52 lines · Michel Thierry
There are some allocations that must be only referenced by 32-bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsIn specific, any resource used with flat/heapless (0x00000000-0xfffff000) General State Heap (GSH) or Intructions State Heap (ISH) must be in a 32-bit range, because the General State Offset and Instruction State Offset are limited to 32-bits.
Objects must have
EXEC_OBJECT_SUPPORTS_48BADDRESSflag to indicate if they can be allocated above the 32-bit address range.v2: Changed flag logic from
neeeds_32b, tosupports_48b.
v3: Moved 48-bit support flag back toexec_object. (Chris, Daniel)Cc Chris Wilson <[email protected]>Cc Daniel Vetter <[email protected]>Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 19 +++++++++++++++++-- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 7 +++++++ include/uapi/drm/i915_drm.h | 3 ++- 4 files changed, 27 insertions(+), 3 deletions(-)drivers/gpu/drm/i915/i915_drv.h +1
-0@@ -2739,6 +2739,7 @@ void i915_gem_vma_destroy(struct i915_vma *vma);#define PIN_OFFSET_BIAS (1<<3)#define PIN_USER (1<<4)#define PIN_UPDATE (1<<5) +#define PIN_FULL_RANGE (1<<6)#define PIN_OFFSET_MASK (~4095) int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,drivers/gpu/drm/i915/i915_gem.c +2
-0@@ -3669,6 +3669,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,struct drm_i915_private *dev_priv = dev->dev_private;u32 fence_alignment, unfenced_alignment;u64 size, fence_size; +u32 search_flag =DRM_MM_SEARCH_DEFAULT ; +u32 alloc_flag =DRM_MM_CREATE_DEFAULT ;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =
drivers/gpu/drm/i915/i915_gem.c i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, · line 3721
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem.c +7
-0@@ -3710,6 +3712,19 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, obj->tiling_mode, false); size = flags &PIN_MAPPABLE ? fence_size : obj->base.size; + +/* Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset, +* limit address to 4GB-1 for objects requiring this wa; for +* others, set alloc flag to TOP. +*/ +if (USES_FULL_48BIT_PPGTT (dev)) { +if (flags &PIN_FULL_RANGE ) {
I wanted this as a separate PIN_FLAG. (a) it is generally useful (for example, I think anything that has PIN_GLOBAL but not PIN_MAPPABLE is a candidate for this flag), but (b) there are bugs in the drm_mm implementation for searching below...
drivers/gpu/drm/i915/i915_gem.c i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, · line 3724
Quoted 3 lines
drivers/gpu/drm/i915/i915_gem.c +3
-0+ search_flag =DRM_MM_SEARCH_BELOW ; + alloc_flag =DRM_MM_CREATE_TOP ; + }else {
This would be better internally as a PIN_ZONE_4G flag.
drivers/gpu/drm/i915/i915_gem.c i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, · line 3725
Quoted 1 line
drivers/gpu/drm/i915/i915_gem.c +1
-0+ end = ((4ULL <<GEN8_PDPE_SHIFT ) -1 );
end should not be -1 here (or if that is actually required by the hardware -4096).
Quoted 21 lines
drivers/gpu/drm/i915/i915_gem.c +4
-2+ } + } }if (alignment ==0 )@@ -3752,8 +3767,8 @@ search_free: size, alignment, obj->cache_level, start, end, -DRM_MM_SEARCH_DEFAULT , - DRM_MM_CREATE_DEFAULT); + search_flag, + alloc_flag);if (ret) { ret = i915_gem_evict_something(dev, vm, size, alignment, obj->cache_level,drivers/gpu/drm/i915/i915_gem_execbuffer.c +0
-0
drivers/gpu/drm/i915/i915_gem_execbuffer.c i915_gem_execbuffer_reserve_vma(struct i915_vma *vma, · line 592
Quoted 5 lines
drivers/gpu/drm/i915/i915_gem_execbuffer.c +2
-0@@ -588,6 +588,9 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,if (entry->flags &EXEC_OBJECT_NEEDS_GTT ) flags |=PIN_GLOBAL ; +if (entry->flags &EXEC_OBJECT_SUPPORTS_48BBADDRESS ) + flags |=PIN_FULL_RANGE ;
flags |= PIN_ZONE_4G;
if (entry->flags & EXEC_OBJECT_SUPPORTS_48BBADDRESS)
flags &= ~PIN_ZONE_4G;
drivers/gpu/drm/i915/i915_gem_execbuffer.c i915_gem_execbuffer_reserve_vma(struct i915_vma *vma, · line 596
Quoted 3 lines
drivers/gpu/drm/i915/i915_gem_execbuffer.c +0
-0if (!drm_mm_node_allocated (&vma->node)) {if (entry->flags & __EXEC_OBJECT_NEEDS_MAP) flags |=PIN_GLOBAL |PIN_MAPPABLE ;
if ((flags & PIN_MAPPABLE) == 0)
flags |= PIN_HIGH;
drivers/gpu/drm/i915/i915_gem_execbuffer.c eb_vma_misplaced(struct i915_vma *vma) · line 677
Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_execbuffer.c +2
-0@@ -670,6 +673,10 @@ eb_vma_misplaced(struct i915_vma *vma)if (entry->flags & __EXEC_OBJECT_NEEDS_MAP && !obj->map_and_fenceable)return !only_mappable_for_reloc (entry->flags); +if (!(entry->flags &EXEC_OBJECT_SUPPORTS_48BBADDRESS ) && + vma->node.start >= (1ULL <<32 ))
vma->node.start + vma->node.size > 1<<32
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 18/18 drm/i915/gen8: Flip the 48b switch
In reply to Michel ThierryUse 48b addresses if hw supports it and i915.enable_ppgtt=3.
Note, aliasing PPGTT remains 32b only.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 2 +- drivers/gpu/drm/i915/i915_params.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
@@ -110,7 +110,7 @@ static int sanitize_enable_ppgtt(struct drm_device *dev, int enable_ppgtt) has_full_ppgtt =INTEL_INFO (dev)->gen >=7 ; has_full_64bit_ppgtt =IS_ENABLED (CONFIG_X86_64 ) && (IS_BROADWELL (dev) || -INTEL_INFO (dev)->gen >=9 ) && false;/* FIXME: 64b */ +INTEL_INFO (dev)->gen >=9 );if (intel_vgpu_active (dev)) has_full_ppgtt = false;/* emulation is too hard */
drivers/gpu/drm/i915/i915_params.c +1 -1
@@ -116,7 +116,7 @@ MODULE_PARM_DESC(enable_hangcheck,module_param_named_unsafe (enable_ppgtt, i915.enable_ppgtt, int,0400 );MODULE_PARM_DESC (enable_ppgtt,"Override PPGTT usage. " -"(-1=auto [default], 0=disabled, 1=aliasing, 2=full)" ); +"(-1=auto [default], 0=disabled, 1=aliasing, 2=full, 3=full_64b)" );module_param_named (enable_execlists, i915.enable_execlists, int,0400 ); MODULE_PARM_DESC(enable_execlists,
Signature · 6 lines
2.4.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 1/1 tests/gem_ppgtt: Check Wa32bitOffsets workarounds
In reply to Michel ThierryTest I915_EXEC_SUPPORTS_48BADDRESS flag to use 32b+ segment. Driver will try to use lower PDPs of each PPGTT for the objects requiring Wa32bitGeneralStateOffset or Wa32bitInstructionBaseOffset.
v2: Add flink cases, (suggested by Daniel Vetter).
--- tests/gem_ppgtt.c | 152 ++++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 147 insertions(+), 5 deletions(-)
tests/gem_ppgtt.c +147 -5
@@ -47,8 +47,18 @@#define STRIDE (WIDTH*4)#define HEIGHT 512#define SIZE (HEIGHT*STRIDE) +#define K (1024l) +#define M (1024l * K) +#define G (1024l * M) +#define LOCAL_EXEC_SUPPORT_48BADDRESS (1<<15) -static bool uses_full_ppgtt (int fd) +/* +* 0 - No PPGTT +* 1 - Aliasing PPGTT +* 2 - Full PPGTT (32b) +* 3 - Full PPGTT (48b) +*/ +static bool __uses_full_ppgtt (int fd,int min) {struct drm_i915_getparam gp;int val =0 ;
@@ -61,7 +71,17 @@ static bool uses_full_ppgtt(int fd)return 0 ; errno =0 ; -return val >1 ; +return val >= min; +} + +static bool uses_full_ppgtt (int fd) +{ +return __uses_full_ppgtt (fd,2 ); +} + +static bool uses_48b_full_ppgtt (int fd) +{ +return __uses_full_ppgtt (fd,3 ); } static drm_intel_bo *create_bo(drm_intel_bufmgr *bufmgr,
@@ -216,7 +236,7 @@ static void surfaces_check(dri_bo **bo, int count, uint32_t expected) } } -static uint64_t exec_and_get_offset (int fd,uint32_t batch) +static uint64_t exec_and_get_offset (int fd,uint32_t batch,bool needs_32b_addr) {struct drm_i915_gem_execbuffer2 execbuf;struct drm_i915_gem_exec_object2 exec[1 ];
@@ -230,6 +250,7 @@ static uint64_t exec_and_get_offset(int fd, uint32_t batch)memset (&execbuf,0 ,sizeof (execbuf)); execbuf.buffers_ptr = (uintptr_t )exec; execbuf.buffer_count =1 ; + execbuf.flags = (needs_32b_addr) ?0 :LOCAL_EXEC_SUPPORT_48BADDRESS ;gem_execbuf (fd, &execbuf);igt_assert_neq (exec[0 ].offset,-1 );
@@ -252,7 +273,7 @@ static void flink_and_close(void) fd2 =drm_open_any (); flinked_bo =gem_open (fd2, name); - offset =exec_and_get_offset (fd2, flinked_bo); + offset =exec_and_get_offset (fd2, flinked_bo,0 );gem_sync (fd2, flinked_bo);gem_close (fd2, flinked_bo);
@@ -260,7 +281,7 @@ static void flink_and_close(void)* same size should get the same offset */ new_bo =gem_create (fd2,4096 ); - offset_new =exec_and_get_offset (fd2, new_bo); + offset_new =exec_and_get_offset (fd2, new_bo,0 );gem_close (fd2, new_bo);igt_assert_eq (offset, offset_new);
@@ -270,6 +291,124 @@ static void flink_and_close(void)close (fd2); } +static bool is_32b (uint64_t offset) +{ +return (offset < (1ULL <<32 )); +} + +static void reusebo_and_compare_offsets (uint32_t fd, +uint64_t buf_size) +{ +uint32_t bo; +uint64_t offset, offset2; + + bo =gem_create (fd, buf_size); +/* support 48b addresses */ + offset =exec_and_get_offset (fd, bo,0 ); +gem_sync (fd, bo); + +/* require 32b address */ + offset2 =exec_and_get_offset (fd, bo,1 ); +gem_sync (fd, bo); + +igt_assert (is_32b (offset2)); +igt_assert_neq (offset, offset2); +gem_close (fd, bo); +} + +static void flinkbo_and_compare_offsets (uint32_t fd,uint32_t fd2, +uint64_t buf_size) +{ +uint32_t bo, flinked_bo, name; +uint64_t offset, offset2; + + bo =gem_create (fd, buf_size); + name =gem_flink (fd, bo); + +/* support 48b addresses */ + offset =exec_and_get_offset (fd, bo,0 ); +gem_sync (fd, bo); + +/* require 32b address */ + flinked_bo =gem_open (fd2, name); + offset2 =exec_and_get_offset (fd2, flinked_bo,1 ); +gem_sync (fd2, flinked_bo); + +igt_assert (is_32b (offset2)); +igt_assert_neq (offset, offset2); +gem_close (fd2, flinked_bo); +gem_close (fd, bo); +} + +static void createbo_and_compare_offsets (uint32_t fd,uint32_t fd2, +uint64_t buf_size, +bool needs_32b,bool needs_32b2) +{ +uint32_t bo, bo2; +uint64_t offset, offset2; + + bo =gem_create (fd, buf_size); + offset =exec_and_get_offset (fd, bo, needs_32b); +gem_sync (fd, bo); + + bo2 =gem_create (fd2, buf_size); + offset2 =exec_and_get_offset (fd2, bo2, needs_32b2); +gem_sync (fd2, bo2); + +if (needs_32b == needs_32b2) +igt_assert_eq (offset, offset2); +else +igt_assert_neq (offset, offset2); + + +/* lower PDPs of each PPGTT are reserved for the objects +* requiring this workaround +*/ +if (needs_32b) +igt_assert (is_32b (offset)); + +if (needs_32b2) +igt_assert (is_32b (offset2)); + +gem_close (fd, bo); +gem_close (fd2, bo2); +} + +static void wa_32b_offset_test (void ) +{ +uint32_t fd, fd2, gb; + + fd =drm_open_any (); +igt_require (uses_48b_full_ppgtt (fd)); + +intel_require_memory (1 ,4 *G ,CHECK_RAM ); + + fd2 =drm_open_any (); + +for (gb =1 ; gb <4 ; gb++) { +/* same bo test */ +reusebo_and_compare_offsets (fd, gb*G ); + + +/* allow full addr range */ +createbo_and_compare_offsets (fd, fd2, gb*G ,0 ,0 ); + +/* limit 32b addr range */ +createbo_and_compare_offsets (fd, fd2, gb*G ,1 ,1 ); + +/* mixed */ +createbo_and_compare_offsets (fd, fd2, gb*G ,0 ,1 ); +createbo_and_compare_offsets (fd, fd2, gb*G ,1 ,0 ); + +/* flink obj */ +flinkbo_and_compare_offsets (fd, fd2, gb*G ); + } + +close (fd); +close (fd2); +} + +#define N_CHILD 8int main (int argc,char **argv) {
@@ -302,5 +441,8 @@ int main(int argc, char **argv) igt_subtest("flink-and-close-vma-leak")flink_and_close (); + igt_subtest("wa-32b-offset-test") +wa_32b_offset_test (); +igt_exit (); }
Signature · 6 lines
2.3.6 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
v3 cover 48-bit PPGTT
In reply to Michel ThierryThese are the rebased patches, after Mika's final ppgtt clean-up series landed (it relies in the macros added). New functions also follow these changes.
In order expand the GPU address space, a 4th level translation is added, the Page Map Level 4 (PML4). This PML4 has 256 PML4 Entries (PML4E), PML4[0-255], each pointing to a PDP. All the existing "dynamic alloc ppgtt" functions are used, only adding the 4th level changes. I also updated some remaining variables that were 32b only.
There are 2 hardware workarounds needed to allow correct operation with 48b addresses (Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset). This new patchset version includes the comments and suggestions from Chris Wilson. A flag (EXEC_OBJECT_SUPPORTS_48B_ADDRESS) will indicate if a given object can be allocated outside the first 4 PDPs; if not, the end range is forced to 4GB. Also, more objects now use the DRM_MM_CREATE_TOP flag. To maintain compatibility, in libdrm I added a new drm_intel_bo_emit_reloc_48bit function that will flag these objects, while the existing drm_intel_bo_emit_reloc clears it.
Finally, this feature is only available in BDW and Gen9, requires LRC submission mode (execlists) and it can be detected by i915.enable_ppgtt=3.
Also note that this expanded address space is only available for full PPGTT, aliasing PPGTT and Global GTT remain 32-bit.
Michel Thierry (17):
drm/i915: Remove unnecessary gen8_clamp_pd
drm/i915/gen8: Make pdp allocation more dynamic drm/i915/gen8: Abstract PDP usage
drm/i915/gen8: Add dynamic page trace events drm/i915/gen8: implement alloc/free for 4lvl drm/i915/gen8: Add 4 level switching infrastructure and lrc support drm/i915/gen8: Generalize PTE writing for GEN8 PPGTT drm/i915/gen8: Pass sg_iter through pte inserts drm/i915/gen8: Add 4 level support in insert_entries and clear_range drm/i915/gen8: Initialize PDPs
drm/i915: Expand error state's address width to 64b drm/i915/gen8: Add ppgtt info and debug_dump
drm/i915: object size needs to be u64
drm/i915: batch_obj vm offset must be u64 drm/i915/userptr: Kill user_size limit check
drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset drm/i915/gen8: Flip the 48b switch
drivers/gpu/drm/i915/i915_debugfs.c | 18 +- drivers/gpu/drm/i915/i915_drv.h | 17 +- drivers/gpu/drm/i915/i915_gem.c | 22 +- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 10 + drivers/gpu/drm/i915/i915_gem_gtt.c | 649 ++++++++++++++++++++++++----- drivers/gpu/drm/i915/i915_gem_gtt.h | 66 ++- drivers/gpu/drm/i915/i915_gem_userptr.c | 4 - drivers/gpu/drm/i915/i915_gpu_error.c | 17 +- drivers/gpu/drm/i915/i915_params.c | 2 +- drivers/gpu/drm/i915/i915_reg.h | 1 + drivers/gpu/drm/i915/i915_trace.h | 16 + drivers/gpu/drm/i915/intel_lrc.c | 65 ++- include/uapi/drm/i915_drm.h | 3 +- 13 files changed, 725 insertions(+), 165 deletions(-)
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 1/17 drm/i915: Remove unnecessary gen8_clamp_pd
gen8_clamp_pd clamps to the next page directory boundary, but the macro gen8_for_each_pde already has a check to stop at the page directory boundary.
Furthermore, i915_pte_count also restricts to the next page table boundary.
v2: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 2 +- drivers/gpu/drm/i915/i915_gem_gtt.h | 11 ----------- 2 files changed, 1 insertion(+), 12 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
@@ -955,7 +955,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,gen8_for_each_pdpe (pd, &ppgtt->pdp, start, length, temp, pdpe) {gen8_pde_t *const page_directory =kmap_px (pd);struct i915_page_table *pt; -uint64_t pd_len =gen8_clamp_pd (start, length); +uint64_t pd_len = length;uint64_t pd_start = start;uint32_t pde;
drivers/gpu/drm/i915/i915_gem_gtt.h +0 -11
@@ -444,17 +444,6 @@ static inline uint32_t gen6_pde_index(uint32_t addr) temp = min(temp, length), \ start += temp, length -= temp) -/* Clamp length to the next page_directory boundary */ -static inline uint64_t gen8_clamp_pd (uint64_t start,uint64_t length) -{ -uint64_t next_pd =ALIGN (start +1 ,1 <<GEN8_PDPE_SHIFT ); - -if (next_pd > (start + length)) -return length; - -return next_pd - start; -} -static inline uint32_t gen8_pte_index (uint64_t address) {return i915_pte_index (address,GEN8_PDE_SHIFT );
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 2/17 drm/i915/gen8: Make pdp allocation more dynamic
In reply to Michel ThierryThis transitional patch doesn't do much for the existing code. However, it should make upcoming patches to use the full 48b address space a bit easier. The patch also introduces the PML4, ie. the new top level structure of the page tables.
v2: Renamed pdp_free to be similar to pd/pt (unmap_and_free_pdp).
v3: To facilitate testing, 48b mode will be available on Broadwell and GEN9+, when i915.enable_ppgtt = 3.
v4: Rebase after s/page_tables/page_table/, added extra information about 4-level page table formats and use IS_ENABLED macro.
v5: Check CONFIG_X86_64 instead of CONFIG_64BIT.
v6: Rebase after Mika's ppgtt cleanup / scratch merge patch series, and follow his nomenclature in pdp functions (there is no alloc_pdp yet).
v7: Rebase after merged version of Mika's ppgtt cleanup patch series.
v8: Rebase after final merged version of Mika's ppgtt/scratch patches.
--- drivers/gpu/drm/i915/i915_drv.h | 7 ++- drivers/gpu/drm/i915/i915_gem_gtt.c | 116 ++++++++++++++++++++++++++++-------- drivers/gpu/drm/i915/i915_gem_gtt.h | 41 ++++++++++--- 3 files changed, 128 insertions(+), 36 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +6 -1
@@ -2490,7 +2490,12 @@ struct drm_i915_cmd_table {#define HAS_HW_CONTEXTS (dev) (INTEL_INFO(dev)->gen >= 6)#define HAS_LOGICAL_RING_CONTEXTS (dev) (INTEL_INFO(dev)->gen >= 8)#define USES_PPGTT (dev) (i915.enable_ppgtt) -#define USES_FULL_PPGTT (dev) (i915.enable_ppgtt == 2) +#define USES_FULL_PPGTT (dev) (i915.enable_ppgtt >= 2) +#ifdef CONFIG_X86_64 +# define USES_FULL_48BIT_PPGTT (dev) (i915.enable_ppgtt == 3) +#else +# define USES_FULL_48BIT_PPGTT (dev) false +#endif #define HAS_OVERLAY (dev) (INTEL_INFO(dev)->has_overlay)#define OVERLAY_NEEDS_PHYSICAL (dev) (INTEL_INFO(dev)->overlay_needs_physical)
drivers/gpu/drm/i915/i915_gem_gtt.c +90 -26
247 lines not on this page · raw message
drivers/gpu/drm/i915/i915_gem_gtt.h +32 -9
78 lines not on this page · raw message
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 02/17] drm/i915/gen8: Make pdp allocation more dynamic
Quoted 5 lines · Michel Thierry
This transitional patch doesn't do much for the existing code. However, it should make upcoming patches to use the full 48b address space a bit easier. The patch also introduces the PML4, ie. the new top level structure of the page tables.
Would be better to move the introduction of PML4 to a separate patch & keep this patch only for the dynamic allocation of pdp changes.
Quoted 383 lines
v2: Renamed
pdp_freeto be similar to pd/pt (unmap_and_free_pdp).
v3: To facilitate testing, 48b mode will be available on Broadwell and GEN9+, when i915.enable_ppgtt = 3.
v4: Rebase after s/page_tables/page_table/, added extra information about 4-level page table formats and useIS_ENABLEDmacro.
v5: CheckCONFIG_X86_64instead ofCONFIG_64BIT.
v6: Rebase after Mika's ppgtt cleanup / scratch merge patch series, and follow his nomenclature in pdp functions (there is noalloc_pdpyet).
v7: Rebase after merged version of Mika's ppgtt cleanup patch series.
v8: Rebase after final merged version of Mika's ppgtt/scratch patches.Signed-off-by Ben Widawsky <[email protected]>Signed-off-by Michel Thierry <[email protected]> (v2+)--- drivers/gpu/drm/i915/i915_drv.h | 7 ++- drivers/gpu/drm/i915/i915_gem_gtt.c | 116 ++++++++++++++++++++++++++++-------- drivers/gpu/drm/i915/i915_gem_gtt.h | 41 ++++++++++--- 3 files changed, 128 insertions(+), 36 deletions(-)drivers/gpu/drm/i915/i915_drv.h +6
-1@@ -2490,7 +2490,12 @@ struct drm_i915_cmd_table {#define HAS_HW_CONTEXTS (dev) (INTEL_INFO(dev)->gen >= 6)#define HAS_LOGICAL_RING_CONTEXTS (dev) (INTEL_INFO(dev)->gen >= 8)#define USES_PPGTT (dev) (i915.enable_ppgtt) -#define USES_FULL_PPGTT (dev) (i915.enable_ppgtt == 2) +#define USES_FULL_PPGTT (dev) (i915.enable_ppgtt >= 2) +#ifdef CONFIG_X86_64 +# define USES_FULL_48BIT_PPGTT (dev) (i915.enable_ppgtt == 3) +#else +# define USES_FULL_48BIT_PPGTT (dev) false +#endif #define HAS_OVERLAY (dev) (INTEL_INFO(dev)->has_overlay)#define OVERLAY_NEEDS_PHYSICAL (dev) (INTEL_INFO(dev)->overlay_needs_physical)drivers/gpu/drm/i915/i915_gem_gtt.c +90
-26@@ -104,9 +104,13 @@ static int sanitize_enable_ppgtt(struct drm_device *dev, int enable_ppgtt) {bool has_aliasing_ppgtt;bool has_full_ppgtt; +bool has_full_64bit_ppgtt; has_aliasing_ppgtt =INTEL_INFO (dev)->gen >=6 ; has_full_ppgtt =INTEL_INFO (dev)->gen >=7 ; + has_full_64bit_ppgtt =IS_ENABLED (CONFIG_X86_64 ) && + (IS_BROADWELL (dev) || +INTEL_INFO (dev)->gen >=9 ) && false;/* FIXME: 64b */ if (intel_vgpu_active (dev)) has_full_ppgtt = false;/* emulation is too hard */ @@ -125,6 +129,9 @@ static int sanitize_enable_ppgtt(struct drm_device *dev, int enable_ppgtt)if (enable_ppgtt ==2 && has_full_ppgtt)return 2 ; +if (enable_ppgtt ==3 && has_full_64bit_ppgtt) +return 3 ; +#ifdef CONFIG_INTEL_IOMMU /* Disable ppgtt on SNB if VT-d is on. */ if (INTEL_INFO (dev)->gen ==6 && intel_iommu_gfx_mapped) {@@ -522,6 +529,45 @@ static void gen8_initialize_pd(struct i915_address_space *vm,fill_px (vm->dev, pd, scratch_pde); } +static int __pdp_init (struct drm_device *dev, +struct i915_page_directory_pointer *pdp) +{ +size_t pdpes =I915_PDPES_PER_PDP (dev); + + pdp->used_pdpes =kcalloc (BITS_TO_LONGS (pdpes), +sizeof (unsigned long ), +GFP_KERNEL ); +if (!pdp->used_pdpes) +return -ENOMEM ; + + pdp->page_directory =kcalloc (pdpes,sizeof (*pdp->page_directory), +GFP_KERNEL ); +if (!pdp->page_directory) { +kfree (pdp->used_pdpes); +/* the PDP might be the statically allocated top level. Keep it +* as clean as possible */ + pdp->used_pdpes =NULL ; +return -ENOMEM ; + } + +return 0 ; +} + +static void __pdp_fini (struct i915_page_directory_pointer *pdp) +{ +kfree (pdp->used_pdpes); +kfree (pdp->page_directory); + pdp->page_directory =NULL ; +} + +static void free_pdp (struct drm_device *dev, +struct i915_page_directory_pointer *pdp) +{ +__pdp_fini (pdp); +if (USES_FULL_48BIT_PPGTT (dev)) +kfree (pdp); +} +/* Broadwell Page Directory Pointer Descriptors */ static int gen8_write_pdp(struct drm_i915_gem_request *req, unsigned entry,@@ -634,9 +680,6 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, pt_vaddr =NULL ;for_each_sg_page (pages->sgl, &sg_iter, pages->nents,0 ) { -if (WARN_ON (pdpe >=GEN8_LEGACY_PDPES )) -break ; -if (pt_vaddr ==NULL ) {struct i915_page_directory *pd = ppgtt->pdp.page_directory[pdpe];struct i915_page_table *pt = pd->page_table[pde];@@ -720,7 +763,8 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm) container_of(vm, struct i915_hw_ppgtt, base);int i; -for_each_set_bit (i, ppgtt->pdp.used_pdpes,GEN8_LEGACY_PDPES ) { +for_each_set_bit (i, ppgtt->pdp.used_pdpes, +I915_PDPES_PER_PDP (ppgtt->base.dev)) {if (WARN_ON (!ppgtt->pdp.page_directory[i]))continue ;@@ -729,6 +773,7 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm)free_pd (ppgtt->base.dev, ppgtt->pdp.page_directory[i]); } +free_pdp (ppgtt->base.dev, &ppgtt->pdp);gen8_free_scratch (vm); }@@ -820,8 +865,9 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,struct i915_page_directory *pd;uint64_t temp;uint32_t pdpe; +uint32_t pdpes =I915_PDPES_PER_PDP (ppgtt->base.dev); -WARN_ON (!bitmap_empty (new_pds,GEN8_LEGACY_PDPES )); +WARN_ON (!bitmap_empty (new_pds, pdpes));gen8_for_each_pdpe (pd, pdp, start, length, temp, pdpe) {if (pd)@@ -839,18 +885,19 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,return 0 ; unwind_out: -for_each_set_bit (pdpe, new_pds,GEN8_LEGACY_PDPES ) +for_each_set_bit (pdpe, new_pds, pdpes)free_pd (dev, pdp->page_directory[pdpe]);return -ENOMEM ; }static void -free_gen8_temp_bitmaps (unsigned long *new_pds,unsigned long **new_pts) +free_gen8_temp_bitmaps (unsigned long *new_pds,unsigned long **new_pts, +uint32_t pdpes) {int i; -for (i =0 ; i <GEN8_LEGACY_PDPES ; i++) +for (i =0 ; i < pdpes; i++)kfree (new_pts[i]);kfree (new_pts);kfree (new_pds);@@ -861,23 +908,24 @@ free_gen8_temp_bitmaps(unsigned long *new_pds, unsigned long **new_pts)*/ static int __must_check alloc_gen8_temp_bitmaps (unsigned long **new_pds, -unsigned long ***new_pts) +unsigned long ***new_pts, +uint32_t pdpes) {int i;unsigned long *pds;unsigned long **pts; - pds =kcalloc (BITS_TO_LONGS (GEN8_LEGACY_PDPES ),sizeof (unsigned long ),GFP_KERNEL ); + pds =kcalloc (BITS_TO_LONGS (pdpes),sizeof (unsigned long ),GFP_KERNEL );if (!pds)return -ENOMEM ; - pts =kcalloc (GEN8_LEGACY_PDPES ,sizeof (unsigned long *),GFP_KERNEL ); + pts =kcalloc (pdpes,sizeof (unsigned long *),GFP_KERNEL );if (!pts) {kfree (pds);return -ENOMEM ; } -for (i =0 ; i <GEN8_LEGACY_PDPES ; i++) { +for (i =0 ; i < pdpes; i++) { pts[i] =kcalloc (BITS_TO_LONGS (I915_PDES ),sizeof (unsigned long ),GFP_KERNEL );if (!pts[i])@@ -890,7 +938,7 @@ int __must_check alloc_gen8_temp_bitmaps(unsigned long **new_pds,return 0 ; err_out: -free_gen8_temp_bitmaps (pds, pts); +free_gen8_temp_bitmaps (pds, pts, pdpes);return -ENOMEM ; }@@ -916,6 +964,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,const uint64_t orig_length = length;uint64_t temp;uint32_t pdpe; +uint32_t pdpes =I915_PDPES_PER_PDP (dev);int ret;/* Wrap is never okay since we can only represent 48b, and we don't @@ -927,7 +976,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,if (WARN_ON (start + length > ppgtt->base.total))return -ENODEV ; - ret =alloc_gen8_temp_bitmaps (&new_page_dirs, &new_page_tables); + ret =alloc_gen8_temp_bitmaps (&new_page_dirs, &new_page_tables, pdpes);if (ret)return ret;@@ -935,7 +984,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm, ret =gen8_ppgtt_alloc_page_directories (ppgtt, &ppgtt->pdp, start, length, new_page_dirs);if (ret) { -free_gen8_temp_bitmaps (new_page_dirs, new_page_tables); +free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);return ret; }@@ -989,7 +1038,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,__set_bit (pdpe, ppgtt->pdp.used_pdpes); } -free_gen8_temp_bitmaps (new_page_dirs, new_page_tables); +free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);mark_tlbs_dirty (ppgtt);return 0 ;@@ -999,10 +1048,10 @@ err_out:free_pt (vm->dev, ppgtt->pdp.page_directory[pdpe]->page_table[temp]); } -for_each_set_bit (pdpe, new_page_dirs,GEN8_LEGACY_PDPES ) +for_each_set_bit (pdpe, new_page_dirs, pdpes)free_pd (vm->dev, ppgtt->pdp.page_directory[pdpe]); -free_gen8_temp_bitmaps (new_page_dirs, new_page_tables); +free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);mark_tlbs_dirty (ppgtt);return ret; }@@ -1023,14 +1072,6 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)return ret; ppgtt->base.start =0 ; - ppgtt->base.total =1ULL <<32 ; -if (IS_ENABLED (CONFIG_X86_32 )) -/* While we have a proliferation of size_t variables -* we cannot represent the full ppgtt size on 32bit, -* so limit it to the same size as the GGTT (currently -* 2GiB). -*/ - ppgtt->base.total =to_i915 (ppgtt->base.dev)->gtt.base.total; ppgtt->base.cleanup = gen8_ppgtt_cleanup; ppgtt->base.allocate_va_range = gen8_alloc_va_range; ppgtt->base.insert_entries = gen8_ppgtt_insert_entries;@@ -1040,7 +1081,30 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) ppgtt->switch_mm = gen8_mm_switch; +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { + ret =__pdp_init (false, &ppgtt->pdp); + +if (ret) + goto free_scratch; + + ppgtt->base.total =1ULL <<32 ; +if (IS_ENABLED (CONFIG_X86_32 )) +/* While we have a proliferation of size_t variables +* we cannot represent the full ppgtt size on 32bit, +* so limit it to the same size as the GGTT (currently +* 2GiB). +*/ + ppgtt->base.total =to_i915 (ppgtt->base.dev)->gtt.base.total; + }else { + ppgtt->base.total =1ULL <<48 ; +return -EPERM ;/* Not yet implemented */ + } +return 0 ; + +free_scratch: +gen8_free_scratch (&ppgtt->base); +return ret; }static void gen6_dump_ppgtt (struct i915_hw_ppgtt *ppgtt,struct seq_file *m)drivers/gpu/drm/i915/i915_gem_gtt.h +29
-9@@ -88,9 +88,17 @@ typedef uint64_t gen8_pde_t;* PDPE | PDE | PTE | offset * The difference as compared to normal x86 3 level page table is the PDPEs are * programmed via register. +* +* GEN8 48b legacy style address is defined as a 4 level page table: +* 47:39 | 38:30 | 29:21 | 20:12 | 11:0 +* PML4E | PDPE | PDE | PTE | offset */ +#define GEN8_PML4ES_PER_PML4 512 +#define GEN8_PML4E_SHIFT 39#define GEN8_PDPE_SHIFT 30 -#define GEN8_PDPE_MASK 0x3 +/* NB: GEN8_PDPE_MASK is untrue for 32b platforms, but it has no impact on 32b page +* tables */ +#define GEN8_PDPE_MASK 0x1ff#define GEN8_PDE_SHIFT 21#define GEN8_PDE_MASK 0x1ff#define GEN8_PTE_SHIFT 12@@ -98,6 +106,9 @@ typedef uint64_t gen8_pde_t;#define GEN8_LEGACY_PDPES 4#define GEN8_PTES I915_PTES(sizeof(gen8_pte_t)) +#define I915_PDPES_PER_PDP (dev) (USES_FULL_48BIT_PPGTT(dev) ?\ + GEN8_PML4ES_PER_PML4 : GEN8_LEGACY_PDPES) +#define PPAT_UNCACHED_INDEX (_PAGE_PWT | _PAGE_PCD)#define PPAT_CACHED_PDE_INDEX 0/* WB LLC */ #define PPAT_CACHED_INDEX _PAGE_PAT/* WB LLCeLLC */ @@ -241,9 +252,17 @@ struct i915_page_directory { };struct i915_page_directory_pointer { -/* struct page *page; */ - DECLARE_BITMAP(used_pdpes, GEN8_LEGACY_PDPES); -struct i915_page_directory *page_directory[GEN8_LEGACY_PDPES ]; +struct i915_page_dma base; + +unsigned long *used_pdpes; +struct i915_page_directory **page_directory; +}; + +struct i915_pml4 { +struct i915_page_dma base; + + DECLARE_BITMAP(used_pml4es, GEN8_PML4ES_PER_PML4); +struct i915_page_directory_pointer *pdps[GEN8_PML4ES_PER_PML4 ]; };struct i915_address_space {@@ -341,8 +360,9 @@ struct i915_hw_ppgtt {struct drm_mm_node node;unsigned long pd_dirty_rings;union { -struct i915_page_directory_pointer pdp; -struct i915_page_directory pd; +struct i915_pml4 pml4;/* GEN8+ & 48b PPGTT */ +struct i915_page_directory_pointer pdp;/* GEN8+ */ +struct i915_page_directory pd;/* GEN6-7 */ };struct drm_i915_file_private *file_priv;@@ -436,14 +456,17 @@ static inline uint32_t gen6_pde_index(uint32_t addr) temp = min(temp, length), \ start += temp, length -= temp) -#define gen8_for_each_pdpe (pd, pdp, start, length, temp, iter) \ - for (iter = gen8_pdpe_index(start); \ - pd = (pdp)->page_directory[iter], length > 0 && iter < GEN8_LEGACY_PDPES; \ +#define gen8_for_each_pdpe_e (pd, pdp, start, length, temp, iter, b) \ + for (iter = gen8_pdpe_index(start); \ + pd = (pdp)->page_directory[iter], length > 0 && (iter < b); \ iter++, \
drivers/gpu/drm/i915/i915_gem_gtt.h static inline uint32_t gen6_pde_index(uint32_t addr) · line 472
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.h +3
-0temp = ALIGN(start+1, 1 << GEN8_PDPE_SHIFT) - start, \ temp = min(temp, length), \ start += temp, length -= temp) +#define gen8_for_each_pdpe (pd, pdp, start, length, temp, iter) \ + gen8_for_each_pdpe_e(pd, pdp, start, length, temp, iter, I915_PDPES_PER_PDP(dev)) +static inline uint32_t gen8_pte_index (uint64_t address) {return i915_pte_index (address,GEN8_PDE_SHIFT );
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 02/17] drm/i915/gen8: Make pdp allocation more dynamic
Quoted 10 lines · Akash
Quoted 6 lines · Michel Thierry
This transitional patch doesn't do much for the existing code. However, it should make upcoming patches to use the full 48b address space a bit easier. The patch also introduces the PML4, ie. the new top level structure
of the page tables.
Quoted 3 lines
Would be better to move the introduction of PML4 to a separate patch & keep this patch only for the dynamic allocation of pdp changes.
I'll move the PML4 declaration to a subsequent patch.
Thanks
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 3/17 drm/i915/gen8: Abstract PDP usage
In reply to Michel ThierryUp until now, ppgtt->pdp has always been the root of our page tables. Legacy 32b addresses acted like it had 1 PDP with 4 PDPEs.
In preparation for 4 level page tables, we need to stop use ppgtt->pdp directly unless we know it's what we want. The future structure will use ppgtt->pml4 for the top level, and the pdp is just one of the entries being pointed to by a pml4e.
v2: Updated after dynamic page allocation changes.
v3: Rebase after s/page_tables/page_table/.
v4: Rebase after changes in "Dynamic page table allocations" patch.
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
v6: Rebase after final merged version of Mika's ppgtt/scratch patches.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 136 +++++++++++++++++++++++------------- 1 file changed, 88 insertions(+), 48 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +88 -48
@@ -529,6 +529,25 @@ static void gen8_initialize_pd(struct i915_address_space *vm,fill_px (vm->dev, pd, scratch_pde); } +/* It's likely we'll map more than one page table at a time. This function will +* save us unnecessary kmap calls, but do no more functionally than multiple +* calls to pde_encode. The ppgtt is only needed to reuse the kunmap macro. */ +static void gen8_map_pagetable_range (struct i915_hw_ppgtt *ppgtt, +struct i915_page_directory *pd, +uint64_t start, +uint64_t length) +{ +gen8_pde_t *const page_directory =kmap_px (pd); +struct i915_page_table *pt; +uint64_t temp, pde; + +gen8_for_each_pde (pt, pd, start, length, temp, pde) + page_directory[pde] =gen8_pde_encode (px_dma (pt), +I915_CACHE_LLC ); + +kunmap_px (ppgtt, page_directory); +} +static int __pdp_init (struct drm_device *dev,struct i915_page_directory_pointer *pdp) {
@@ -616,6 +635,7 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ gen8_pte_t *pt_vaddr, scratch_pte;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;
@@ -630,10 +650,10 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm,struct i915_page_directory *pd;struct i915_page_table *pt; -if (WARN_ON (!ppgtt->pdp.page_directory[pdpe])) +if (WARN_ON (!pdp->page_directory[pdpe]))break ; - pd = ppgtt->pdp.page_directory[pdpe]; + pd = pdp->page_directory[pdpe];if (WARN_ON (!pd->page_table[pde]))break ;
@@ -671,6 +691,7 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ gen8_pte_t *pt_vaddr;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;
@@ -681,7 +702,7 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,for_each_sg_page (pages->sgl, &sg_iter, pages->nents,0 ) {if (pt_vaddr ==NULL ) { -struct i915_page_directory *pd = ppgtt->pdp.page_directory[pdpe]; +struct i915_page_directory *pd = pdp->page_directory[pdpe];struct i915_page_table *pt = pd->page_table[pde]; pt_vaddr =kmap_px (pt); }
@@ -763,23 +784,28 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm) container_of(vm, struct i915_hw_ppgtt, base);int i; -for_each_set_bit (i, ppgtt->pdp.used_pdpes, -I915_PDPES_PER_PDP (ppgtt->base.dev)) { -if (WARN_ON (!ppgtt->pdp.page_directory[i])) -continue ; +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { +for_each_set_bit (i, ppgtt->pdp.used_pdpes, +I915_PDPES_PER_PDP (ppgtt->base.dev)) { +if (WARN_ON (!ppgtt->pdp.page_directory[i])) +continue ; -gen8_free_page_tables (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); -free_pd (ppgtt->base.dev, ppgtt->pdp.page_directory[i]); +gen8_free_page_tables (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]); +free_pd (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]); + } +free_pdp (ppgtt->base.dev, &ppgtt->pdp); + }else { +WARN_ON (1 );/* to be implemented later */ } -free_pdp (ppgtt->base.dev, &ppgtt->pdp);gen8_free_scratch (vm); }/** * gen8_ppgtt_alloc_pagetabs() - Allocate page tables for VA range. -* @ppgtt: Master ppgtt structure. +* @vm: Master vm structure. * @pd: Page directory for this address range. * @start: Starting virtual address to begin allocations. * @length Size of the allocations.
@@ -795,13 +821,15 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm)* * Return: 0 if success; negative error code otherwise. */ -static int gen8_ppgtt_alloc_pagetabs (struct i915_hw_ppgtt *ppgtt, +static int gen8_ppgtt_alloc_pagetabs (struct i915_address_space *vm,struct i915_page_directory *pd,uint64_t start,uint64_t length,unsigned long *new_pts) { -struct drm_device *dev = ppgtt->base.dev; +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct drm_device *dev = vm->dev;struct i915_page_table *pt;uint64_t temp;uint32_t pde;
@@ -818,7 +846,7 @@ static int gen8_ppgtt_alloc_pagetabs(struct i915_hw_ppgtt *ppgtt,if (IS_ERR (pt)) goto unwind_out; -gen8_initialize_pt (&ppgtt->base, pt); +gen8_initialize_pt (vm, pt); pd->page_table[pde] = pt;__set_bit (pde, new_pts); }
@@ -834,7 +862,7 @@ unwind_out:/** * gen8_ppgtt_alloc_page_directories() - Allocate page directories for VA range. -* @ppgtt: Master ppgtt structure. +* @vm: Master vm structure. * @pdp: Page directory pointer for this address range. * @start: Starting virtual address to begin allocations. * @length Size of the allocations.
@@ -855,17 +883,18 @@ unwind_out:* * Return: 0 if success; negative error code otherwise. */ -static int gen8_ppgtt_alloc_page_directories (struct i915_hw_ppgtt *ppgtt, -struct i915_page_directory_pointer *pdp, -uint64_t start, -uint64_t length, -unsigned long *new_pds) +static int +gen8_ppgtt_alloc_page_directories (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +uint64_t start, +uint64_t length, +unsigned long *new_pds) { -struct drm_device *dev = ppgtt->base.dev; +struct drm_device *dev = vm->dev;struct i915_page_directory *pd;uint64_t temp;uint32_t pdpe; -uint32_t pdpes =I915_PDPES_PER_PDP (ppgtt->base.dev); +uint32_t pdpes =I915_PDPES_PER_PDP (vm->dev);WARN_ON (!bitmap_empty (new_pds, pdpes));
@@ -877,7 +906,7 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,if (IS_ERR (pd)) goto unwind_out; -gen8_initialize_pd (&ppgtt->base, pd); +gen8_initialize_pd (vm, pd); pdp->page_directory[pdpe] = pd;__set_bit (pdpe, new_pds); }
@@ -952,13 +981,15 @@ static void mark_tlbs_dirty(struct i915_hw_ppgtt *ppgtt) ppgtt->pd_dirty_rings =INTEL_INFO (ppgtt->base.dev)->ring_mask; } -static int gen8_alloc_va_range (struct i915_address_space *vm, -uint64_t start, -uint64_t length) +static int gen8_alloc_va_range_3lvl (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +uint64_t start, +uint64_t length) {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);unsigned long *new_page_dirs, **new_page_tables; +struct drm_device *dev = vm->dev;struct i915_page_directory *pd;const uint64_t orig_start = start;const uint64_t orig_length = length;
@@ -981,16 +1012,15 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,return ret;/* Do the allocations first so we can easily bail out */ - ret =gen8_ppgtt_alloc_page_directories (ppgtt, &ppgtt->pdp, start, length, - new_page_dirs); + ret =gen8_ppgtt_alloc_page_directories (vm, pdp, start, length, + new_page_dirs);if (ret) {free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);return ret; } -/* For every page directory referenced, allocate page tables */ -gen8_for_each_pdpe (pd, &ppgtt->pdp, start, length, temp, pdpe) { - ret =gen8_ppgtt_alloc_pagetabs (ppgtt, pd, start, length, +gen8_for_each_pdpe (pd, pdp, start, length, temp, pdpe) { + ret =gen8_ppgtt_alloc_pagetabs (vm, pd, start, length, new_page_tables[pdpe]);if (ret) goto err_out;
@@ -999,10 +1029,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm, start = orig_start; length = orig_length; -/* Allocations have completed successfully, so set the bitmaps, and do -* the mappings. */ -gen8_for_each_pdpe (pd, &ppgtt->pdp, start, length, temp, pdpe) { -gen8_pde_t *const page_directory =kmap_px (pd); +gen8_for_each_pdpe (pd, pdp, start, length, temp, pdpe) {struct i915_page_table *pt;uint64_t pd_len = length;uint64_t pd_start = start;
@@ -1024,18 +1051,10 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,/* Our pde is now pointing to the pagetable, pt */ __set_bit (pde, pd->used_pdes); - -/* Map the PDE to the page table */ - page_directory[pde] =gen8_pde_encode (px_dma (pt), -I915_CACHE_LLC ); - -/* NB: We haven't yet mapped ptes to pages. At this -* point we're still relying on insert_entries() */ } -kunmap_px (ppgtt, page_directory); - -__set_bit (pdpe, ppgtt->pdp.used_pdpes); +__set_bit (pdpe, pdp->used_pdpes); +gen8_map_pagetable_range (ppgtt, pd, start, length); }free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);
@@ -1045,17 +1064,38 @@ static int gen8_alloc_va_range(struct i915_address_space *vm, err_out:while (pdpe--) {for_each_set_bit (temp, new_page_tables[pdpe],I915_PDES ) -free_pt (vm->dev, ppgtt->pdp.page_directory[pdpe]->page_table[temp]); +free_pt (dev, pdp->page_directory[pdpe]->page_table[temp]); }for_each_set_bit (pdpe, new_page_dirs, pdpes) -free_pd (vm->dev, ppgtt->pdp.page_directory[pdpe]); +free_pd (dev, pdp->page_directory[pdpe]);free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);mark_tlbs_dirty (ppgtt);return ret; } +static int gen8_alloc_va_range_4lvl (struct i915_address_space *vm, +struct i915_pml4 *pml4, +uint64_t start, +uint64_t length) +{ +WARN_ON (1 );/* to be implemented later */ +return 0 ; +} + +static int gen8_alloc_va_range (struct i915_address_space *vm, +uint64_t start,uint64_t length) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); + +if (!USES_FULL_48BIT_PPGTT (vm->dev)) +return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); +else +return gen8_alloc_va_range_4lvl (vm, &ppgtt->pml4, start, length); +} +/* * GEN8 legacy ppgtt programming is accomplished through a max 4 PDP registers * with a net effect resembling a 2-level page table in normal x86 terms. Each
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 03/17] drm/i915/gen8: Abstract PDP usage
Quoted 131 lines · Michel Thierry
Up until now,
ppgtt->pdphas always been the root of our page tables. Legacy 32b addresses acted like it had 1 PDP with 4 PDPEs.In preparation for 4 level page tables, we need to stop use
ppgtt->pdpdirectly unless we know it's what we want. The future structure will useppgtt->pml4for the top level, and the pdp is just one of the entries being pointed to by a pml4e.v2: Updated after dynamic page allocation changes.
v3: Rebase after s/page_tables/page_table/.
v4: Rebase after changes in "Dynamic page table allocations" patch.
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
v6: Rebase after final merged version of Mika's ppgtt/scratch patches.Signed-off-by Ben Widawsky <[email protected]>Signed-off-by Michel Thierry <[email protected]> (v2+)--- drivers/gpu/drm/i915/i915_gem_gtt.c | 136 +++++++++++++++++++++++------------- 1 file changed, 88 insertions(+), 48 deletions(-)drivers/gpu/drm/i915/i915_gem_gtt.c +38
-13@@ -529,6 +529,25 @@ static void gen8_initialize_pd(struct i915_address_space *vm,fill_px (vm->dev, pd, scratch_pde); } +/* It's likely we'll map more than one page table at a time. This function will +* save us unnecessary kmap calls, but do no more functionally than multiple +* calls to pde_encode. The ppgtt is only needed to reuse the kunmap macro. */ +static void gen8_map_pagetable_range (struct i915_hw_ppgtt *ppgtt, +struct i915_page_directory *pd, +uint64_t start, +uint64_t length) +{ +gen8_pde_t *const page_directory =kmap_px (pd); +struct i915_page_table *pt; +uint64_t temp, pde; + +gen8_for_each_pde (pt, pd, start, length, temp, pde) + page_directory[pde] =gen8_pde_encode (px_dma (pt), +I915_CACHE_LLC ); + +kunmap_px (ppgtt, page_directory); +} +static int __pdp_init (struct drm_device *dev,struct i915_page_directory_pointer *pdp) {@@ -616,6 +635,7 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ gen8_pte_t *pt_vaddr, scratch_pte;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;@@ -630,10 +650,10 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm,struct i915_page_directory *pd;struct i915_page_table *pt; -if (WARN_ON (!ppgtt->pdp.page_directory[pdpe])) +if (WARN_ON (!pdp->page_directory[pdpe]))break ; - pd = ppgtt->pdp.page_directory[pdpe]; + pd = pdp->page_directory[pdpe];if (WARN_ON (!pd->page_table[pde]))break ;@@ -671,6 +691,7 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ gen8_pte_t *pt_vaddr;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;@@ -681,7 +702,7 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,for_each_sg_page (pages->sgl, &sg_iter, pages->nents,0 ) {if (pt_vaddr ==NULL ) { -struct i915_page_directory *pd = ppgtt->pdp.page_directory[pdpe]; +struct i915_page_directory *pd = pdp->page_directory[pdpe];struct i915_page_table *pt = pd->page_table[pde]; pt_vaddr =kmap_px (pt); }@@ -763,23 +784,28 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm) container_of(vm, struct i915_hw_ppgtt, base);int i; -for_each_set_bit (i, ppgtt->pdp.used_pdpes, -I915_PDPES_PER_PDP (ppgtt->base.dev)) { -if (WARN_ON (!ppgtt->pdp.page_directory[i])) -continue ; +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { +for_each_set_bit (i, ppgtt->pdp.used_pdpes, +I915_PDPES_PER_PDP (ppgtt->base.dev)) { +if (WARN_ON (!ppgtt->pdp.page_directory[i])) +continue ; -gen8_free_page_tables (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); -free_pd (ppgtt->base.dev, ppgtt->pdp.page_directory[i]); +gen8_free_page_tables (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]); +free_pd (ppgtt->base.dev, + ppgtt->pdp.page_directory[i]); + } +free_pdp (ppgtt->base.dev, &ppgtt->pdp); + }else { +WARN_ON (1 );/* to be implemented later */ } -free_pdp (ppgtt->base.dev, &ppgtt->pdp);gen8_free_scratch (vm); }/** * gen8_ppgtt_alloc_pagetabs() - Allocate page tables for VA range. -* @ppgtt: Master ppgtt structure. +* @vm: Master vm structure. * @pd: Page directory for this address range. * @start: Starting virtual address to begin allocations. * @length Size of the allocations. @@ -795,13 +821,15 @@ static void gen8_ppgtt_cleanup(struct i915_address_space *vm)* * Return: 0 if success; negative error code otherwise. */ -static int gen8_ppgtt_alloc_pagetabs(struct i915_hw_ppgtt *ppgtt,
drivers/gpu/drm/i915/i915_gem_gtt.c static void gen8_ppgtt_cleanup(struct i915_address_space *vm) · line 832
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +4
-1+static int gen8_ppgtt_alloc_pagetabs (struct i915_address_space *vm,struct i915_page_directory *pd,uint64_t start,uint64_t length,unsigned long *new_pts) { - struct drm_device *dev = ppgtt->base.dev; +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct drm_device *dev = vm->dev;
The 'ppgtt' pointer can be completely dispensed with, by the use of 'vm' pointer.
Quoted 116 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +21
-26struct i915_page_table *pt;uint64_t temp;uint32_t pde;@@ -818,7 +846,7 @@ static int gen8_ppgtt_alloc_pagetabs(struct i915_hw_ppgtt *ppgtt,if (IS_ERR (pt)) goto unwind_out; -gen8_initialize_pt (&ppgtt->base, pt); +gen8_initialize_pt (vm, pt); pd->page_table[pde] = pt;__set_bit (pde, new_pts); }@@ -834,7 +862,7 @@ unwind_out:/** * gen8_ppgtt_alloc_page_directories() - Allocate page directories for VA range. -* @ppgtt: Master ppgtt structure. +* @vm: Master vm structure. * @pdp: Page directory pointer for this address range. * @start: Starting virtual address to begin allocations. * @length Size of the allocations. @@ -855,17 +883,18 @@ unwind_out:* * Return: 0 if success; negative error code otherwise. */ -static int gen8_ppgtt_alloc_page_directories (struct i915_hw_ppgtt *ppgtt, -struct i915_page_directory_pointer *pdp, -uint64_t start, -uint64_t length, -unsigned long *new_pds) +static int +gen8_ppgtt_alloc_page_directories (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +uint64_t start, +uint64_t length, +unsigned long *new_pds) { -struct drm_device *dev = ppgtt->base.dev; +struct drm_device *dev = vm->dev;struct i915_page_directory *pd;uint64_t temp;uint32_t pdpe; -uint32_t pdpes =I915_PDPES_PER_PDP (ppgtt->base.dev); +uint32_t pdpes =I915_PDPES_PER_PDP (vm->dev);WARN_ON (!bitmap_empty (new_pds, pdpes));@@ -877,7 +906,7 @@ static int gen8_ppgtt_alloc_page_directories(struct i915_hw_ppgtt *ppgtt,if (IS_ERR (pd)) goto unwind_out; -gen8_initialize_pd (&ppgtt->base, pd); +gen8_initialize_pd (vm, pd); pdp->page_directory[pdpe] = pd;__set_bit (pdpe, new_pds); }@@ -952,13 +981,15 @@ static void mark_tlbs_dirty(struct i915_hw_ppgtt *ppgtt) ppgtt->pd_dirty_rings =INTEL_INFO (ppgtt->base.dev)->ring_mask; } -static int gen8_alloc_va_range (struct i915_address_space *vm, -uint64_t start, -uint64_t length) +static int gen8_alloc_va_range_3lvl (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +uint64_t start, +uint64_t length) {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);unsigned long *new_page_dirs, **new_page_tables; +struct drm_device *dev = vm->dev;struct i915_page_directory *pd;const uint64_t orig_start = start;const uint64_t orig_length = length;@@ -981,16 +1012,15 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,return ret;/* Do the allocations first so we can easily bail out */ - ret =gen8_ppgtt_alloc_page_directories (ppgtt, &ppgtt->pdp, start, length, - new_page_dirs); + ret =gen8_ppgtt_alloc_page_directories (vm, pdp, start, length, + new_page_dirs);if (ret) {free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);return ret; } -/* For every page directory referenced, allocate page tables */ -gen8_for_each_pdpe (pd, &ppgtt->pdp, start, length, temp, pdpe) { - ret =gen8_ppgtt_alloc_pagetabs (ppgtt, pd, start, length, +gen8_for_each_pdpe (pd, pdp, start, length, temp, pdpe) { + ret =gen8_ppgtt_alloc_pagetabs (vm, pd, start, length, new_page_tables[pdpe]);if (ret) goto err_out;@@ -999,10 +1029,7 @@ static int gen8_alloc_va_range(struct i915_address_space *vm, start = orig_start; length = orig_length; -/* Allocations have completed successfully, so set the bitmaps, and do -* the mappings. */ -gen8_for_each_pdpe (pd, &ppgtt->pdp, start, length, temp, pdpe) { -gen8_pde_t *const page_directory =kmap_px (pd); +gen8_for_each_pdpe (pd, pdp, start, length, temp, pdpe) {struct i915_page_table *pt;uint64_t pd_len = length;uint64_t pd_start = start;@@ -1024,18 +1051,10 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,/* Our pde is now pointing to the pagetable, pt */ __set_bit (pde, pd->used_pdes); - -/* Map the PDE to the page table */ - page_directory[pde] =gen8_pde_encode (px_dma (pt), -I915_CACHE_LLC );
drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_alloc_va_range(struct i915_address_space *vm, · line 1057
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2
-6- -/* NB: We haven't yet mapped ptes to pages. At this -* point we're still relying on insert_entries() */ } -kunmap_px (ppgtt, page_directory); - -__set_bit (pdpe, ppgtt->pdp.used_pdpes); +__set_bit (pdpe, pdp->used_pdpes); +gen8_map_pagetable_range (ppgtt, pd, start, length);
No apparent benefit in use of "gen8_map_pagetable_range", considering that the Page Directory page is mapped at the start of outer loop (PDPEs) only and the inner loop (PDEs) maps the PDEs to the page tables, in an inline manner. The 'gen8_map_pagetable_range', will repeat the same inner loop of PDEs.
Quoted 35 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +16
-2}free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);@@ -1045,17 +1064,38 @@ static int gen8_alloc_va_range(struct i915_address_space *vm, err_out:while (pdpe--) {for_each_set_bit (temp, new_page_tables[pdpe],I915_PDES ) -free_pt (vm->dev, ppgtt->pdp.page_directory[pdpe]->page_table[temp]); +free_pt (dev, pdp->page_directory[pdpe]->page_table[temp]); }for_each_set_bit (pdpe, new_page_dirs, pdpes) -free_pd (vm->dev, ppgtt->pdp.page_directory[pdpe]); +free_pd (dev, pdp->page_directory[pdpe]);free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);mark_tlbs_dirty (ppgtt);return ret; } +static int gen8_alloc_va_range_4lvl (struct i915_address_space *vm, +struct i915_pml4 *pml4, +uint64_t start, +uint64_t length) +{ +WARN_ON (1 );/* to be implemented later */ +return 0 ; +} + +static int gen8_alloc_va_range (struct i915_address_space *vm, +uint64_t start,uint64_t length) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base);
drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_alloc_va_range(struct i915_address_space *vm, · line 1101
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +7
-0+ +if (!USES_FULL_48BIT_PPGTT (vm->dev)) +return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); +else +return gen8_alloc_va_range_4lvl (vm, &ppgtt->pml4, start, length); +} +/* * GEN8 legacy ppgtt programming is accomplished through a max 4 PDP registers * with a net effect resembling a 2-level page table in normal x86 terms. Each
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 03/17] drm/i915/gen8: Abstract PDP usage
Quoted 14 lines · Akash
Quoted 11 lines · Michel Thierry
drivers/gpu/drm/i915/i915_gem_gtt.c +1
-1@@ -795,13 +821,15 @@ static void gen8_ppgtt_cleanup(structi915_address_space *vm) * * Return: 0 if success; negative error code otherwise. */ -static int gen8_ppgtt_alloc_pagetabs (struct i915_hw_ppgtt *ppgtt, +static int gen8_ppgtt_alloc_pagetabs (struct i915_address_space *vm,struct i915_page_directory *pd,uint64_t start,uint64_t length,unsigned long *new_pts)
Quoted 9 lines
drivers/gpu/drm/i915/i915_gem_gtt.c static void gen8_ppgtt_cleanup(struct i915_address_space *vm) · line 832
Quoted 5 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +3
-1{ -struct drm_device *dev = ppgtt->base.dev; +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct drm_device *dev = vm->dev;The 'ppgtt' pointer can be completely dispensed with, by the use of 'vm' pointer.
Leftovers from old rebases, I'll clean these functions up.
Quoted 17 lines
Quoted 17 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +0
-0struct i915_page_table *pt;uint64_t temp;uint32_t pde;drivers/gpu/drm/i915/i915_gem_gtt.c +0
-7@@ -1024,18 +1051,10 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,/* Our pde is now pointing to the pagetable, pt */ __set_bit (pde, pd->used_pdes); - -/* Map the PDE to the page table */ - page_directory[pde] =gen8_pde_encode (px_dma (pt), -I915_CACHE_LLC ); - -/* NB: We haven't yet mapped ptes to pages. At this -* point we're still relying on insert_entries() */ }
Quoted 12 lines
drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_alloc_va_range(struct i915_address_space *vm, · line 1057
Quoted 5 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2
-3-kunmap_px (ppgtt, page_directory); - -__set_bit (pdpe, ppgtt->pdp.used_pdpes); +__set_bit (pdpe, pdp->used_pdpes); +gen8_map_pagetable_range (ppgtt, pd, start, length);No apparent benefit in use of "
gen8_map_pagetable_range", considering that the Page Directory page is mapped at the start of outer loop (PDPEs) only and the inner loop (PDEs) maps the PDEs to the page tables, in an inline manner.
The 'gen8_map_pagetable_range', will repeat the same inner loop of PDEs.
And I'll discard these changes.
-Michel
Quoted 3 lines
Quoted 3 lines
(quoted diff) +0
-0}free_gen8_temp_bitmaps(new_page_dirs, new_page_tables, pdpes);
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 4/17 drm/i915/gen8: Add dynamic page trace events
In reply to Michel ThierryThe dynamic page allocation patch series added it for GEN6, this patch adds them for GEN8.
v2: Consolidate pagetable/page_directory events
v3: Multiple rebases.
v4: Rebase after s/page_tables/page_table/.
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 9 ++++++++- drivers/gpu/drm/i915/i915_trace.h | 16 ++++++++++++++++ 2 files changed, 24 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +8 -1
@@ -541,9 +541,14 @@ static void gen8_map_pagetable_range(struct i915_hw_ppgtt *ppgtt,struct i915_page_table *pt;uint64_t temp, pde; -gen8_for_each_pde (pt, pd, start, length, temp, pde) +gen8_for_each_pde (pt, pd, start, length, temp, pde) { page_directory[pde] =gen8_pde_encode (px_dma (pt),I915_CACHE_LLC ); +trace_i915_page_table_entry_map (&ppgtt->base, pde, pt, +gen8_pte_index (start), +gen8_pte_count (start, length), +GEN8_PTES ); + }kunmap_px (ppgtt, page_directory); }
@@ -849,6 +854,7 @@ static int gen8_ppgtt_alloc_pagetabs(struct i915_address_space *vm,gen8_initialize_pt (vm, pt); pd->page_table[pde] = pt;__set_bit (pde, new_pts); +trace_i915_page_table_entry_alloc (vm, pde, start,GEN8_PDE_SHIFT ); }return 0 ;
@@ -909,6 +915,7 @@ gen8_ppgtt_alloc_page_directories(struct i915_address_space *vm,gen8_initialize_pd (vm, pd); pdp->page_directory[pdpe] = pd;__set_bit (pdpe, new_pds); +trace_i915_page_directory_entry_alloc (vm, pdpe, start,GEN8_PDPE_SHIFT ); }return 0 ;
drivers/gpu/drm/i915/i915_trace.h +16 -0
@@ -213,6 +213,22 @@ DEFINE_EVENT(i915_page_table_entry, i915_page_table_entry_alloc,TP_ARGS (vm, pde, start, pde_shift) ); +DEFINE_EVENT_PRINT (i915_page_table_entry, i915_page_directory_entry_alloc, + TP_PROTO(struct i915_address_space *vm, u32 pdpe, u64 start, u64 pdpe_shift), +TP_ARGS (vm, pdpe, start, pdpe_shift), + +TP_printk ("vm=%p, pdpe=%d (0x%llx-0x%llx)" , + __entry->vm, __entry->pde, __entry->start, __entry->end) +); + +DEFINE_EVENT_PRINT (i915_page_table_entry, i915_page_directory_pointer_entry_alloc, + TP_PROTO(struct i915_address_space *vm, u32 pml4e, u64 start, u64 pml4e_shift), +TP_ARGS (vm, pml4e, start, pml4e_shift), + +TP_printk ("vm=%p, pml4e=%d (0x%llx-0x%llx)" , + __entry->vm, __entry->pde, __entry->start, __entry->end) +); +/* Avoid extra math because we only support two sizes. The format is defined by * bitmap_scnprintf. Each 32 bits is 8 HEX digits followed by comma */ #define TRACE_PT_SIZE (bits) \
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 5/17 drm/i915/gen8: implement alloc/free for 4lvl
In reply to Michel ThierryPML4 has no special attributes, and there will always be a PML4. So simply initialize it at creation, and destroy it at the end.
The code for 4lvl is able to call into the existing 3lvl page table code to handle all of the lower levels.
v2: Return something at the end of gen8_alloc_va_range_4lvl to keep the compiler happy. And define ret only in one place.
Updated gen8_ppgtt_unmap_pages and gen8_ppgtt_free to handle 4lvl.
v3: Use i915_dma_unmap_single instead of pci API. Fix a couple of incorrect checks when unmapping pdp and pd pages (Akash).
v4: Call __pdp_fini also for 32b PPGTT. Clean up alloc_pdp param list.
v5: Prevent (harmless) out of range access in gen8_for_each_pml4e.
v6: Simplify alloc_vma_range_4lvl and gen8_ppgtt_init_common error paths. (Akash)
v7: Rebase, s/gen8_ppgtt_free_*/gen8_ppgtt_cleanup_*/.
v8: Change location of pml4_init/fini. It will make next patches cleaner.
v9: Rebase after Mika's ppgtt cleanup / scratch merge patch series, while trying to reuse as much as possible for pdp alloc. pml4_init/fini replaced by setup/cleanup_px macros.
v10: Rebase after Mika's merged ppgtt cleanup patch series.
v11: Rebase after final merged version of Mika's ppgtt/scratch patches.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 162 ++++++++++++++++++++++++++++++------ drivers/gpu/drm/i915/i915_gem_gtt.h | 12 ++- 2 files changed, 146 insertions(+), 28 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +136 -26
@@ -584,12 +584,44 @@ static void __pdp_fini(struct i915_page_directory_pointer *pdp) pdp->page_directory =NULL ; } +static struct +i915_page_directory_pointer *alloc_pdp (struct drm_device *dev) +{ +struct i915_page_directory_pointer *pdp; +int ret = -ENOMEM ; + +WARN_ON (!USES_FULL_48BIT_PPGTT (dev)); + + pdp =kzalloc (sizeof (*pdp),GFP_KERNEL ); +if (!pdp) +return ERR_PTR (-ENOMEM ); + + ret =__pdp_init (dev, pdp); +if (ret) + goto fail_bitmap; + + ret =setup_px (dev, pdp); +if (ret) + goto fail_page_m; + +return pdp; + +fail_page_m: +__pdp_fini (pdp); +fail_bitmap: +kfree (pdp); + +return ERR_PTR (ret); +} +static void free_pdp (struct drm_device *dev,struct i915_page_directory_pointer *pdp) {__pdp_fini (pdp); -if (USES_FULL_48BIT_PPGTT (dev)) +if (USES_FULL_48BIT_PPGTT (dev)) { +cleanup_px (dev, pdp);kfree (pdp); + } }/* Broadwell Page Directory Pointer Descriptors */
@@ -783,28 +815,46 @@ static void gen8_free_scratch(struct i915_address_space *vm)free_scratch_page (dev, vm->scratch_page); } -static void gen8_ppgtt_cleanup (struct i915_address_space *vm) +static void gen8_ppgtt_cleanup_3lvl (struct drm_device *dev, +struct i915_page_directory_pointer *pdp) { -struct i915_hw_ppgtt *ppgtt = - container_of(vm, struct i915_hw_ppgtt, base);int i; -if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { -for_each_set_bit (i, ppgtt->pdp.used_pdpes, -I915_PDPES_PER_PDP (ppgtt->base.dev)) { -if (WARN_ON (!ppgtt->pdp.page_directory[i])) -continue ; +for_each_set_bit (i, pdp->used_pdpes,I915_PDPES_PER_PDP (dev)) { +if (WARN_ON (!pdp->page_directory[i])) +continue ; -gen8_free_page_tables (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); -free_pd (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); - } -free_pdp (ppgtt->base.dev, &ppgtt->pdp); - }else { -WARN_ON (1 );/* to be implemented later */ +gen8_free_page_tables (dev, pdp->page_directory[i]); +free_pd (dev, pdp->page_directory[i]); } +free_pdp (dev, pdp); +} + +static void gen8_ppgtt_cleanup_4lvl (struct i915_hw_ppgtt *ppgtt) +{ +int i; + +for_each_set_bit (i, ppgtt->pml4.used_pml4es,GEN8_PML4ES_PER_PML4 ) { +if (WARN_ON (!ppgtt->pml4.pdps[i])) +continue ; + +gen8_ppgtt_cleanup_3lvl (ppgtt->base.dev, ppgtt->pml4.pdps[i]); + } + +cleanup_px (ppgtt->base.dev, &ppgtt->pml4); +} + +static void gen8_ppgtt_cleanup (struct i915_address_space *vm) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); + +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) +gen8_ppgtt_cleanup_3lvl (ppgtt->base.dev, &ppgtt->pdp); +else +gen8_ppgtt_cleanup_4lvl (ppgtt); +gen8_free_scratch (vm); }
@@ -1087,8 +1137,62 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm,uint64_t start, uint64_t length) { -WARN_ON (1 );/* to be implemented later */ +DECLARE_BITMAP (new_pdps ,GEN8_PML4ES_PER_PML4 ); +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp; +const uint64_t orig_start = start; +const uint64_t orig_length = length; +uint64_t temp, pml4e; +int ret =0 ; + +/* Do the pml4 allocations first, so we don't need to track the newly +* allocated tables below the pdp */ +bitmap_zero (new_pdps,GEN8_PML4ES_PER_PML4 ); + +/* The pagedirectory and pagetable allocations are done in the shared 3 +* and 4 level code. Just allocate the pdps. +*/ +gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) { +if (!pdp) { +WARN_ON (test_bit (pml4e, pml4->used_pml4es)); + pdp =alloc_pdp (vm->dev); +if (IS_ERR (pdp)) + goto err_out; + + pml4->pdps[pml4e] = pdp; +__set_bit (pml4e, new_pdps); +trace_i915_page_directory_pointer_entry_alloc (&ppgtt->base, pml4e, + pml4e <<GEN8_PML4E_SHIFT , +GEN8_PML4E_SHIFT ); + } + } + +WARN (bitmap_weight (new_pdps,GEN8_PML4ES_PER_PML4 ) >2 , +"The allocation has spanned more than 512GB. " +"It is highly likely this is incorrect." ); + + start = orig_start; + length = orig_length; + +gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) { +WARN_ON (!pdp); + + ret =gen8_alloc_va_range_3lvl (vm, pdp, start, length); +if (ret) + goto err_out; + } + +bitmap_or (pml4->used_pml4es, new_pdps, pml4->used_pml4es, +GEN8_PML4ES_PER_PML4 ); +return 0 ; + +err_out: +for_each_set_bit (pml4e, new_pdps,GEN8_PML4ES_PER_PML4 ) +gen8_ppgtt_cleanup_3lvl (vm->dev, pml4->pdps[pml4e]); + +return ret; } static int gen8_alloc_va_range(struct i915_address_space *vm,
@@ -1097,10 +1201,10 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -if (!USES_FULL_48BIT_PPGTT (vm->dev)) -return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); -else +if (USES_FULL_48BIT_PPGTT (vm->dev))return gen8_alloc_va_range_4lvl (vm, &ppgtt->pml4, start, length); +else +return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); }/*
@@ -1128,9 +1232,14 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) ppgtt->switch_mm = gen8_mm_switch; -if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { - ret =__pdp_init (false, &ppgtt->pdp); +if (USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { + ret =setup_px (ppgtt->base.dev, &ppgtt->pml4); +if (ret) + goto free_scratch; + ppgtt->base.total =1ULL <<48 ; + }else { + ret =__pdp_init (false, &ppgtt->pdp);if (ret) goto free_scratch;
@@ -1142,9 +1251,10 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)* 2GiB). */ ppgtt->base.total =to_i915 (ppgtt->base.dev)->gtt.base.total; - }else { - ppgtt->base.total =1ULL <<48 ; -return -EPERM ;/* Not yet implemented */ + +trace_i915_page_directory_pointer_entry_alloc (&ppgtt->base, +0 ,0 , +GEN8_PML4E_SHIFT ); }return 0 ;
drivers/gpu/drm/i915/i915_gem_gtt.h +10 -2
@@ -95,6 +95,7 @@ typedef uint64_t gen8_pde_t;*/ #define GEN8_PML4ES_PER_PML4 512#define GEN8_PML4E_SHIFT 39 +#define GEN8_PML4E_MASK (GEN8_PML4ES_PER_PML4 - 1)#define GEN8_PDPE_SHIFT 30/* NB: GEN8_PDPE_MASK is untrue for 32b platforms, but it has no impact on 32b page * tables */
@@ -464,6 +465,14 @@ static inline uint32_t gen6_pde_index(uint32_t addr) temp = min(temp, length), \ start += temp, length -= temp) +#define gen8_for_each_pml4e (pdp, pml4, start, length, temp, iter) \ + for (iter = gen8_pml4e_index(start); \ + pdp = (pml4)->pdps[iter], length > 0 && iter < GEN8_PML4ES_PER_PML4; \ + iter++, \ + temp = ALIGN(start+1, 1ULL << GEN8_PML4E_SHIFT) - start, \ + temp = min(temp, length), \ + start += temp, length -= temp) +#define gen8_for_each_pdpe (pd, pdp, start, length, temp, iter) \ gen8_for_each_pdpe_e(pd, pdp, start, length, temp, iter, I915_PDPES_PER_PDP(dev))
@@ -484,8 +493,7 @@ static inline uint32_t gen8_pdpe_index(uint64_t address)static inline uint32_t gen8_pml4e_index (uint64_t address) { -WARN_ON (1 );/* For 64B */ -return 0 ; +return (address >>GEN8_PML4E_SHIFT ) &GEN8_PML4E_MASK ; }static inline size_t gen8_pte_count (uint64_t address,uint64_t length)
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 05/17] drm/i915/gen8: implement alloc/free for 4lvl
Quoted 167 lines · Michel Thierry
PML4 has no special attributes, and there will always be a PML4. So simply initialize it at creation, and destroy it at the end.
The code for 4lvl is able to call into the existing 3lvl page table code to handle all of the lower levels.
v2: Return something at the end of
gen8_alloc_va_range_4lvlto keep the compiler happy. And define ret only in one place.
Updatedgen8_ppgtt_unmap_pagesandgen8_ppgtt_freeto handle 4lvl.
v3: Usei915_dma_unmap_singleinstead of pci API. Fix a couple of incorrect checks when unmapping pdp and pd pages (Akash).
v4: Call__pdp_finialso for 32b PPGTT. Clean upalloc_pdpparam list.
v5: Prevent (harmless) out of range access ingen8_for_each_pml4e.
v6: Simplifyalloc_vma_range_4lvlandgen8_ppgtt_init_commonerror paths. (Akash)
v7: Rebase, s/gen8_ppgtt_free_*/gen8_ppgtt_cleanup_*/.
v8: Change location of pml4_init/fini. It will make next patches cleaner.
v9: Rebase after Mika's ppgtt cleanup / scratch merge patch series, while trying to reuse as much as possible for pdp alloc. pml4_init/fini replaced by setup/cleanup_px macros.
v10: Rebase after Mika's merged ppgtt cleanup patch series.
v11: Rebase after final merged version of Mika's ppgtt/scratch patches.Cc Akash Goel <[email protected]>Signed-off-by Ben Widawsky <[email protected]>Signed-off-by Michel Thierry <[email protected]> (v2+)--- drivers/gpu/drm/i915/i915_gem_gtt.c | 162 ++++++++++++++++++++++++++++++------ drivers/gpu/drm/i915/i915_gem_gtt.h | 12 ++- 2 files changed, 146 insertions(+), 28 deletions(-)drivers/gpu/drm/i915/i915_gem_gtt.c +84
-18@@ -584,12 +584,44 @@ static void __pdp_fini(struct i915_page_directory_pointer *pdp) pdp->page_directory =NULL ; } +static struct +i915_page_directory_pointer *alloc_pdp (struct drm_device *dev) +{ +struct i915_page_directory_pointer *pdp; +int ret = -ENOMEM ; + +WARN_ON (!USES_FULL_48BIT_PPGTT (dev)); + + pdp =kzalloc (sizeof (*pdp),GFP_KERNEL ); +if (!pdp) +return ERR_PTR (-ENOMEM ); + + ret =__pdp_init (dev, pdp); +if (ret) + goto fail_bitmap; + + ret =setup_px (dev, pdp); +if (ret) + goto fail_page_m; + +return pdp; + +fail_page_m: +__pdp_fini (pdp); +fail_bitmap: +kfree (pdp); + +return ERR_PTR (ret); +} +static void free_pdp (struct drm_device *dev,struct i915_page_directory_pointer *pdp) {__pdp_fini (pdp); -if (USES_FULL_48BIT_PPGTT (dev)) +if (USES_FULL_48BIT_PPGTT (dev)) { +cleanup_px (dev, pdp);kfree (pdp); + } }/* Broadwell Page Directory Pointer Descriptors */ @@ -783,28 +815,46 @@ static void gen8_free_scratch(struct i915_address_space *vm)free_scratch_page (dev, vm->scratch_page); } -static void gen8_ppgtt_cleanup (struct i915_address_space *vm) +static void gen8_ppgtt_cleanup_3lvl (struct drm_device *dev, +struct i915_page_directory_pointer *pdp) { -struct i915_hw_ppgtt *ppgtt = - container_of(vm, struct i915_hw_ppgtt, base);int i; -if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { -for_each_set_bit (i, ppgtt->pdp.used_pdpes, -I915_PDPES_PER_PDP (ppgtt->base.dev)) { -if (WARN_ON (!ppgtt->pdp.page_directory[i])) -continue ; +for_each_set_bit (i, pdp->used_pdpes,I915_PDPES_PER_PDP (dev)) { +if (WARN_ON (!pdp->page_directory[i])) +continue ; -gen8_free_page_tables (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); -free_pd (ppgtt->base.dev, - ppgtt->pdp.page_directory[i]); - } -free_pdp (ppgtt->base.dev, &ppgtt->pdp); - }else { -WARN_ON (1 );/* to be implemented later */ +gen8_free_page_tables (dev, pdp->page_directory[i]); +free_pd (dev, pdp->page_directory[i]); } +free_pdp (dev, pdp); +} + +static void gen8_ppgtt_cleanup_4lvl (struct i915_hw_ppgtt *ppgtt) +{ +int i; + +for_each_set_bit (i, ppgtt->pml4.used_pml4es,GEN8_PML4ES_PER_PML4 ) { +if (WARN_ON (!ppgtt->pml4.pdps[i])) +continue ; + +gen8_ppgtt_cleanup_3lvl (ppgtt->base.dev, ppgtt->pml4.pdps[i]); + } + +cleanup_px (ppgtt->base.dev, &ppgtt->pml4); +} + +static void gen8_ppgtt_cleanup (struct i915_address_space *vm) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); + +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) +gen8_ppgtt_cleanup_3lvl (ppgtt->base.dev, &ppgtt->pdp); +else +gen8_ppgtt_cleanup_4lvl (ppgtt); +gen8_free_scratch (vm); }@@ -1087,8 +1137,62 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm,uint64_t start, uint64_t length) { -WARN_ON (1 );/* to be implemented later */ +DECLARE_BITMAP (new_pdps ,GEN8_PML4ES_PER_PML4 ); +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp; +const uint64_t orig_start = start; +const uint64_t orig_length = length; +uint64_t temp, pml4e; +int ret =0 ; + +/* Do the pml4 allocations first, so we don't need to track the newly +* allocated tables below the pdp */ +bitmap_zero (new_pdps,GEN8_PML4ES_PER_PML4 ); + +/* The pagedirectory and pagetable allocations are done in the shared 3 +* and 4 level code. Just allocate the pdps. +*/ +gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) {
drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm, · line 1166
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +10
-0+if (!pdp) { +WARN_ON (test_bit (pml4e, pml4->used_pml4es)); + pdp =alloc_pdp (vm->dev); +if (IS_ERR (pdp)) + goto err_out; + + pml4->pdps[pml4e] = pdp; +__set_bit (pml4e, new_pdps); + trace_i915_page_directory_pointer_entry_alloc(&ppgtt->base, pml4e, + pml4e << GEN8_PML4E_SHIFT,
The ‘start’ variable should be used here in place of ‘pml4e << GEN8_PML4E_SHIFT’ ?
Quoted 106 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +42
-8+ GEN8_PML4E_SHIFT); + } + } + +WARN (bitmap_weight (new_pdps,GEN8_PML4ES_PER_PML4 ) >2 , +"The allocation has spanned more than 512GB. " +"It is highly likely this is incorrect." ); + + start = orig_start; + length = orig_length; + +gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) { +WARN_ON (!pdp); + + ret =gen8_alloc_va_range_3lvl (vm, pdp, start, length); +if (ret) + goto err_out; + } + +bitmap_or (pml4->used_pml4es, new_pdps, pml4->used_pml4es, +GEN8_PML4ES_PER_PML4 ); +return 0 ; + +err_out: +for_each_set_bit (pml4e, new_pdps,GEN8_PML4ES_PER_PML4 ) +gen8_ppgtt_cleanup_3lvl (vm->dev, pml4->pdps[pml4e]); + +return ret; } static int gen8_alloc_va_range(struct i915_address_space *vm,@@ -1097,10 +1201,10 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -if (!USES_FULL_48BIT_PPGTT (vm->dev)) -return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); -else +if (USES_FULL_48BIT_PPGTT (vm->dev))return gen8_alloc_va_range_4lvl (vm, &ppgtt->pml4, start, length); +else +return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); }/* @@ -1128,9 +1232,14 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) ppgtt->switch_mm = gen8_mm_switch; -if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { - ret =__pdp_init (false, &ppgtt->pdp); +if (USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { + ret =setup_px (ppgtt->base.dev, &ppgtt->pml4); +if (ret) + goto free_scratch; + ppgtt->base.total =1ULL <<48 ; + }else { + ret =__pdp_init (false, &ppgtt->pdp);if (ret) goto free_scratch;@@ -1142,9 +1251,10 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)* 2GiB). */ ppgtt->base.total =to_i915 (ppgtt->base.dev)->gtt.base.total; - }else { - ppgtt->base.total =1ULL <<48 ; -return -EPERM ;/* Not yet implemented */ + +trace_i915_page_directory_pointer_entry_alloc (&ppgtt->base, +0 ,0 , +GEN8_PML4E_SHIFT ); }return 0 ;drivers/gpu/drm/i915/i915_gem_gtt.h +9
-0@@ -95,6 +95,7 @@ typedef uint64_t gen8_pde_t;*/ #define GEN8_PML4ES_PER_PML4 512#define GEN8_PML4E_SHIFT 39 +#define GEN8_PML4E_MASK (GEN8_PML4ES_PER_PML4 - 1)#define GEN8_PDPE_SHIFT 30/* NB: GEN8_PDPE_MASK is untrue for 32b platforms, but it has no impact on 32b page * tables */ @@ -464,6 +465,14 @@ static inline uint32_t gen6_pde_index(uint32_t addr) temp = min(temp, length), \ start += temp, length -= temp) +#define gen8_for_each_pml4e (pdp, pml4, start, length, temp, iter) \ + for (iter = gen8_pml4e_index(start); \ + pdp = (pml4)->pdps[iter], length > 0 && iter < GEN8_PML4ES_PER_PML4; \ + iter++, \ + temp = ALIGN(start+1, 1ULL << GEN8_PML4E_SHIFT) - start, \ + temp = min(temp, length), \ + start += temp, length -= temp) +#define gen8_for_each_pdpe (pd, pdp, start, length, temp, iter) \ gen8_for_each_pdpe_e(pd, pdp, start, length, temp, iter, I915_PDPES_PER_PDP(dev))
drivers/gpu/drm/i915/i915_gem_gtt.h static inline uint32_t gen8_pdpe_index(uint64_t address) · line 499
Quoted 9 lines
drivers/gpu/drm/i915/i915_gem_gtt.h +1
-2@@ -484,8 +493,7 @@ static inline uint32_t gen8_pdpe_index(uint64_t address)static inline uint32_t gen8_pml4e_index (uint64_t address) { -WARN_ON (1 );/* For 64B */ -return 0 ; +return (address >>GEN8_PML4E_SHIFT ) &GEN8_PML4E_MASK ; }static inline size_t gen8_pte_count (uint64_t address,uint64_t length)
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 05/17] drm/i915/gen8: implement alloc/free for 4lvl
Quoted 33 lines · Akash
Quoted 30 lines · Michel Thierry
drivers/gpu/drm/i915/i915_gem_gtt.c +10
-1@@ -1087,8 +1137,62 @@ static int gen8_alloc_va_range_4lvl(structi915_address_space *vm, uint64_t start, uint64_t length) { - WARN_ON(1); /* to be implemented later */ +DECLARE_BITMAP (new_pdps ,GEN8_PML4ES_PER_PML4 ); +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp; +const uint64_t orig_start = start; +const uint64_t orig_length = length; +uint64_t temp, pml4e; +int ret =0 ; + +/* Do the pml4 allocations first, so we don't need to track the newly
drivers/gpu/drm/i915/i915_gem_gtt.c +4
-0+* allocated tables below the pdp */ +bitmap_zero (new_pdps,GEN8_PML4ES_PER_PML4 ); + +/* The pagedirectory and pagetable allocations are done in the shared 3
drivers/gpu/drm/i915/i915_gem_gtt.c +8
-0+* and 4 level code. Just allocate the pdps. +*/ +gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) { +if (!pdp) { +WARN_ON (test_bit (pml4e, pml4->used_pml4es)); + pdp =alloc_pdp (vm->dev); +if (IS_ERR (pdp)) + goto err_out;
Quoted 8 lines
Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +4
-0+ + pml4->pdps[pml4e] = pdp; +__set_bit (pml4e, new_pdps); +trace_i915_page_directory_pointer_entry_alloc(&ppgtt->base, pml4e,
+ pml4e <<GEN8_PML4E_SHIFT,The ‘start’ variable should be used here in place of ‘pml4e << GEN8_PML4E_SHIFT’ ?
Correct, should be ‘start’.
Thanks
Quoted 4 lines
drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm, · line 1170
Quoted 4 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +4
-0+ GEN8_PML4E_SHIFT); + } + } +
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 6/17 drm/i915/gen8: Add 4 level switching infrastructure and lrc support
In reply to Michel ThierryIn 64b (48bit canonical) PPGTT addressing, the PDP0 register contains the base address to PML4, while the other PDP registers are ignored.
In LRC, the addressing mode must be specified in every context descriptor.
v2: PML4 update in legacy context switch is left for historic reasons, the preferred mode of operation is with lrc context based submission.
v3: s/gen8_map_page_directory/gen8_setup_page_directory and s/gen8_map_page_directory_pointer/gen8_setup_page_directory_pointer. Also, clflush will be needed for bxt. (Akash)
v4: Squashed lrc-specific code and use a macro to set PML4 register.
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series. PDP update in bb_start is only for legacy 32b mode.
v6: Rebase after final merged version of Mika's ppgtt/scratch patches.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 52 ++++++++++++++++++++++++++--- drivers/gpu/drm/i915/i915_gem_gtt.h | 2 ++ drivers/gpu/drm/i915/i915_reg.h | 1 + drivers/gpu/drm/i915/intel_lrc.c | 65 +++++++++++++++++++++++++++---------- 4 files changed, 97 insertions(+), 23 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +47 -5
@@ -211,6 +211,9 @@ static gen8_pde_t gen8_pde_encode(const dma_addr_t addr,return pde; } +#define gen8_pdpe_encode gen8_pde_encode +#define gen8_pml4e_encode gen8_pde_encode +static gen6_pte_t snb_pte_encode (dma_addr_t addr,enum i915_cache_level level,bool valid,u32 unused)
@@ -624,6 +627,35 @@ static void free_pdp(struct drm_device *dev, } } +static void +gen8_setup_page_directory (struct i915_hw_ppgtt *ppgtt, +struct i915_page_directory_pointer *pdp, +struct i915_page_directory *pd, +int index) +{ +gen8_ppgtt_pdpe_t *page_directorypo; + +if (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) +return ; + + page_directorypo =kmap_px (pdp); + page_directorypo[index] =gen8_pdpe_encode (px_dma (pd),I915_CACHE_LLC ); +kunmap_px (ppgtt, page_directorypo); +} + +static void +gen8_setup_page_directory_pointer (struct i915_hw_ppgtt *ppgtt, +struct i915_pml4 *pml4, +struct i915_page_directory_pointer *pdp, +int index) +{ +gen8_ppgtt_pml4e_t *pagemap =kmap_px (pml4); + +WARN_ON (!USES_FULL_48BIT_PPGTT (ppgtt->base.dev)); + pagemap[index] =gen8_pml4e_encode (px_dma (pdp),I915_CACHE_LLC ); +kunmap_px (ppgtt, pagemap); +} +/* Broadwell Page Directory Pointer Descriptors */ static int gen8_write_pdp(struct drm_i915_gem_request *req, unsigned entry,
@@ -649,8 +681,8 @@ static int gen8_write_pdp(struct drm_i915_gem_request *req,return 0 ; } -static int gen8_mm_switch (struct i915_hw_ppgtt *ppgtt, -struct drm_i915_gem_request *req) +static int gen8_legacy_mm_switch (struct i915_hw_ppgtt *ppgtt, +struct drm_i915_gem_request *req) {int i, ret;
@@ -665,6 +697,12 @@ static int gen8_mm_switch(struct i915_hw_ppgtt *ppgtt,return 0 ; } +static int gen8_48b_mm_switch (struct i915_hw_ppgtt *ppgtt, +struct drm_i915_gem_request *req) +{ +return gen8_write_pdp (req,0 ,px_dma (&ppgtt->pml4)); +} + static void gen8_ppgtt_clear_range(struct i915_address_space *vm, uint64_t start, uint64_t length,
@@ -1112,6 +1150,7 @@ static int gen8_alloc_va_range_3lvl(struct i915_address_space *vm,__set_bit (pdpe, pdp->used_pdpes);gen8_map_pagetable_range (ppgtt, pd, start, length); +gen8_setup_page_directory (ppgtt, pdp, pd, pdpe); }free_gen8_temp_bitmaps (new_page_dirs, new_page_tables, pdpes);
@@ -1181,6 +1220,8 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm, ret =gen8_alloc_va_range_3lvl (vm, pdp, start, length);if (ret) goto err_out; + +gen8_setup_page_directory_pointer (ppgtt, pml4, pdp, pml4e); } bitmap_or(pml4->used_pml4es, new_pdps, pml4->used_pml4es,
@@ -1230,14 +1271,13 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) ppgtt->base.unbind_vma = ppgtt_unbind_vma; ppgtt->base.bind_vma = ppgtt_bind_vma; - ppgtt->switch_mm = gen8_mm_switch; -if (USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { ret =setup_px (ppgtt->base.dev, &ppgtt->pml4);if (ret) goto free_scratch; ppgtt->base.total =1ULL <<48 ; + ppgtt->switch_mm = gen8_48b_mm_switch; }else { ret =__pdp_init (false, &ppgtt->pdp);if (ret)
@@ -1252,6 +1292,7 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)*/ ppgtt->base.total =to_i915 (ppgtt->base.dev)->gtt.base.total; + ppgtt->switch_mm = gen8_legacy_mm_switch;trace_i915_page_directory_pointer_entry_alloc (&ppgtt->base,0 ,0 ,GEN8_PML4E_SHIFT );
@@ -1449,8 +1490,9 @@ static void gen8_ppgtt_enable(struct drm_device *dev)int j;for_each_ring (ring, dev_priv, j) { +u32 four_level =USES_FULL_48BIT_PPGTT (dev) ?GEN8_GFX_PPGTT_48B :0 ;I915_WRITE (RING_MODE_GEN7 (ring), -_MASKED_BIT_ENABLE (GFX_PPGTT_ENABLE )); +_MASKED_BIT_ENABLE (GFX_PPGTT_ENABLE | four_level)); } }
drivers/gpu/drm/i915/i915_gem_gtt.h +2 -0
@@ -39,6 +39,8 @@ struct drm_i915_file_private;typedef uint32_t gen6_pte_t ;typedef uint64_t gen8_pte_t ;typedef uint64_t gen8_pde_t ; +typedef uint64_t gen8_ppgtt_pdpe_t ; +typedef uint64_t gen8_ppgtt_pml4e_t ;#define gtt_total_entries (gtt) ((gtt).base.total >> PAGE_SHIFT)
drivers/gpu/drm/i915/i915_reg.h +1 -0
@@ -1664,6 +1664,7 @@ enum skl_disp_power_wells {#define GFX_REPLAY_MODE (1<<11)#define GFX_PSMI_GRANULARITY (1<<10)#define GFX_PPGTT_ENABLE (1<<9) +#define GEN8_GFX_PPGTT_48B (1<<7)#define VLV_DISPLAY_BASE 0x180000#define VLV_MIPI_BASE VLV_DISPLAY_BASE
drivers/gpu/drm/i915/intel_lrc.c +47 -18
102 lines not on this page · raw message
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 7/17 drm/i915/gen8: Generalize PTE writing for GEN8 PPGTT
In reply to Michel ThierryThe insert_entries function was the function used to write PTEs. For the PPGTT it was "hardcoded" to only understand two level page tables, which was the case for GEN7. We can reuse this for 4 level page tables, and remove the concept of insert_entries, which was never viable past 2 level page tables anyway, but it requires a bit of rework to make the function a bit more generic.
This patch begins the generalization work, and it will be heavily used upon when the 48b code is complete. The patch series attempts to make each function which touches a part of code specific to the page table level and here is no exception.
v2: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
v3: Rebase after final merged version of Mika's ppgtt/scratch patches.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 52 +++++++++++++++++++++++++++---------- 1 file changed, 39 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +39 -13
@@ -703,24 +703,21 @@ static int gen8_48b_mm_switch(struct i915_hw_ppgtt *ppgtt,return gen8_write_pdp (req,0 ,px_dma (&ppgtt->pml4)); } -static void gen8_ppgtt_clear_range (struct i915_address_space *vm, -uint64_t start, -uint64_t length, -bool use_scratch) +static void gen8_ppgtt_clear_pte_range (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +uint64_t start, +uint64_t length, +gen8_pte_t scratch_pte) {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ -gen8_pte_t *pt_vaddr, scratch_pte; +gen8_pte_t *pt_vaddr;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ;unsigned num_entries = length >>PAGE_SHIFT ;unsigned last_pte, i; - scratch_pte =gen8_pte_encode (px_dma (ppgtt->base.scratch_page), -I915_CACHE_LLC , use_scratch); -while (num_entries) {struct i915_page_directory *pd;struct i915_page_table *pt;
@@ -759,14 +756,30 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm, } } -static void gen8_ppgtt_insert_entries (struct i915_address_space *vm, -struct sg_table *pages, -uint64_t start, -enum i915_cache_level cache_level,u32 unused) +static void gen8_ppgtt_clear_range (struct i915_address_space *vm, +uint64_t start, +uint64_t length, +bool use_scratch) {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ + +gen8_pte_t scratch_pte =gen8_pte_encode (px_dma (vm->scratch_page), +I915_CACHE_LLC , use_scratch); + +gen8_ppgtt_clear_pte_range (vm, pdp, start, length, scratch_pte); +} + +static void +gen8_ppgtt_insert_pte_entries (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp, +struct sg_table *pages, +uint64_t start, +enum i915_cache_level cache_level) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr;unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;
@@ -800,6 +813,19 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,kunmap_px (ppgtt, pt_vaddr); } +static void gen8_ppgtt_insert_entries (struct i915_address_space *vm, +struct sg_table *pages, +uint64_t start, +enum i915_cache_level cache_level, +u32 unused) +{ +struct i915_hw_ppgtt *ppgtt = + container_of(vm, struct i915_hw_ppgtt, base); +struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ + +gen8_ppgtt_insert_pte_entries (vm, pdp, pages, start, cache_level); +} +static void gen8_free_page_tables (struct drm_device *dev,struct i915_page_directory *pd) {
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 8/17 drm/i915/gen8: Pass sg_iter through pte inserts
In reply to Michel ThierryAs a step towards implementing 4 levels, while not discarding the existing pte insert functions, we need to pass the sg_iter through. The current function understands to the page directory granularity. An object's pages may span the page directory, and so using the iter directly as we write the PTEs allows the iterator to stay coherent through a VMA insert operation spanning multiple page table levels.
v2: Rebase after s/page_tables/page_table/.
v3: Rebase after Mika's ppgtt cleanup / scratch merge patch series; updated commit message (s/map/insert).
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +6 -5
@@ -774,7 +774,7 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm,static void gen8_ppgtt_insert_pte_entries (struct i915_address_space *vm,struct i915_page_directory_pointer *pdp, -struct sg_table *pages, +struct sg_page_iter *sg_iter,uint64_t start,enum i915_cache_level cache_level) {
@@ -784,11 +784,10 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ;unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ;unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; -struct sg_page_iter sg_iter; pt_vaddr =NULL ; -for_each_sg_page (pages->sgl, &sg_iter, pages->nents,0 ) { +while (__sg_page_iter_next (sg_iter)) {if (pt_vaddr ==NULL ) {struct i915_page_directory *pd = pdp->page_directory[pdpe];struct i915_page_table *pt = pd->page_table[pde];
@@ -796,7 +795,7 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm, } pt_vaddr[pte] = -gen8_pte_encode (sg_page_iter_dma_address (&sg_iter), +gen8_pte_encode (sg_page_iter_dma_address (sg_iter), cache_level, true);if (++pte ==GEN8_PTES ) {kunmap_px (ppgtt, pt_vaddr);
@@ -822,8 +821,10 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ +struct sg_page_iter sg_iter; -gen8_ppgtt_insert_pte_entries (vm, pdp, pages, start, cache_level); +__sg_page_iter_start (&sg_iter, pages->sgl,sg_nents (pages->sgl),0 ); +gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, start, cache_level); } static void gen8_free_page_tables(struct drm_device *dev,
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 9/17 drm/i915/gen8: Add 4 level support in insert_entries and clear_range
In reply to Michel ThierryWhen 48b is enabled, gen8_ppgtt_insert_entries needs to read the Page Map Level 4 (PML4), before it selects which Page Directory Pointer (PDP) it will write to.
Similarly, gen8_ppgtt_clear_range needs to get the correct PDP/PD range.
This patch was inspired by Ben's "Depend exclusively on map and unmap_vma".
v2: Rebase after s/page_tables/page_table/.
v3: Remove unnecessary pdpe loop in gen8_ppgtt_clear_range_4lvl and use clamp_pdp in gen8_ppgtt_insert_entries (Akash).
v4: Merge gen8_ppgtt_clear_range_4lvl into gen8_ppgtt_clear_range to maintain symmetry with gen8_ppgtt_insert_entries (Akash).
v5: Do not mix pages and bytes in insert_entries (Akash).
v6: Prevent overflow in sg_nents << PAGE_SHIFT, when inserting 4GB at once.
v7: Rebase after Mika's ppgtt cleanup / scratch merge patch series. Use gen8_px_index functions, and remove unnecessary number of pages parameter in insert_pte_entries.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 51 ++++++++++++++++++++++++++++--------- drivers/gpu/drm/i915/i915_gem_gtt.h | 11 ++++++++ 2 files changed, 50 insertions(+), 12 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +39 -12
@@ -712,9 +712,9 @@ static void gen8_ppgtt_clear_pte_range(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr; -unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ; -unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ; -unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; +unsigned pdpe =gen8_pdpe_index (start); +unsigned pde =gen8_pde_index (start); +unsigned pte =gen8_pte_index (start);unsigned num_entries = length >>PAGE_SHIFT ;unsigned last_pte, i;
@@ -763,12 +763,24 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ -gen8_pte_t scratch_pte =gen8_pte_encode (px_dma (vm->scratch_page),I915_CACHE_LLC , use_scratch); -gen8_ppgtt_clear_pte_range (vm, pdp, start, length, scratch_pte); +if (!USES_FULL_48BIT_PPGTT (vm->dev)) { +gen8_ppgtt_clear_pte_range (vm, &ppgtt->pdp, start, length, + scratch_pte); + }else { +uint64_t templ4, pml4e; +struct i915_page_directory_pointer *pdp; + +gen8_for_each_pml4e (pdp, &ppgtt->pml4, start, length, templ4, pml4e) { +uint64_t pdp_len =gen8_clamp_pdp (start, length); +uint64_t pdp_start = start; + +gen8_ppgtt_clear_pte_range (vm, pdp, pdp_start, pdp_len, + scratch_pte); + } + } } static void
@@ -781,9 +793,9 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr; -unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ; -unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ; -unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; +unsigned pdpe =gen8_pdpe_index (start); +unsigned pde =gen8_pde_index (start); +unsigned pte =gen8_pte_index (start); pt_vaddr =NULL ;
@@ -801,7 +813,8 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,kunmap_px (ppgtt, pt_vaddr); pt_vaddr =NULL ;if (++pde ==I915_PDES ) { - pdpe++; +if (++pdpe ==I915_PDPES_PER_PDP (vm->dev)) +break ; pde =0 ; } pte =0 ;
@@ -820,11 +833,25 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ struct sg_page_iter sg_iter;__sg_page_iter_start (&sg_iter, pages->sgl,sg_nents (pages->sgl),0 ); -gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, start, cache_level); + +if (!USES_FULL_48BIT_PPGTT (vm->dev)) { +gen8_ppgtt_insert_pte_entries (vm, &ppgtt->pdp, &sg_iter, start, + cache_level); + }else { +struct i915_page_directory_pointer *pdp; +uint64_t templ4, pml4e; +uint64_t length = (uint64_t )sg_nents (pages->sgl) <<PAGE_SHIFT ; + +gen8_for_each_pml4e (pdp, &ppgtt->pml4, start, length, templ4, pml4e) { +uint64_t pdp_start = start; + +gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, + pdp_start, cache_level); + } + } } static void gen8_free_page_tables(struct drm_device *dev,
drivers/gpu/drm/i915/i915_gem_gtt.h +11 -0
@@ -478,6 +478,17 @@ static inline uint32_t gen6_pde_index(uint32_t addr)#define gen8_for_each_pdpe (pd, pdp, start, length, temp, iter) \ gen8_for_each_pdpe_e(pd, pdp, start, length, temp, iter, I915_PDPES_PER_PDP(dev)) +/* Clamp length to the next page_directory pointer boundary */ +static inline uint64_t gen8_clamp_pdp (uint64_t start,uint64_t length) +{ +uint64_t next_pdp =ALIGN (start +1 ,1ULL <<GEN8_PML4E_SHIFT ); + +if (next_pdp > (start + length)) +return length; + +return next_pdp - start; +} +static inline uint32_t gen8_pte_index (uint64_t address) {return i915_pte_index (address,GEN8_PDE_SHIFT );
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 09/17] drm/i915/gen8: Add 4 level support in insert_entries and clear_range
Quoted 87 lines · Michel Thierry
When 48b is enabled,
gen8_ppgtt_insert_entriesneeds to read the Page Map Level 4 (PML4), before it selects which Page Directory Pointer (PDP) it will write to.Similarly,
gen8_ppgtt_clear_rangeneeds to get the correct PDP/PD range.This patch was inspired by Ben's "Depend exclusively on map and
unmap_vma".v2: Rebase after s/page_tables/page_table/.
v3: Remove unnecessary pdpe loop ingen8_ppgtt_clear_range_4lvland useclamp_pdpingen8_ppgtt_insert_entries(Akash).
v4: Mergegen8_ppgtt_clear_range_4lvlintogen8_ppgtt_clear_rangeto maintain symmetry withgen8_ppgtt_insert_entries(Akash).
v5: Do not mix pages and bytes ininsert_entries(Akash).
v6: Prevent overflow insg_nents<<PAGE_SHIFT, when inserting 4GB at once.
v7: Rebase after Mika's ppgtt cleanup / scratch merge patch series. Usegen8_px_indexfunctions, and remove unnecessary number of pages parameter ininsert_pte_entries.Cc Akash Goel <[email protected]>Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/i915_gem_gtt.c | 51 ++++++++++++++++++++++++++++--------- drivers/gpu/drm/i915/i915_gem_gtt.h | 11 ++++++++ 2 files changed, 50 insertions(+), 12 deletions(-)drivers/gpu/drm/i915/i915_gem_gtt.c +21
-9@@ -712,9 +712,9 @@ static void gen8_ppgtt_clear_pte_range(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr; -unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ; -unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ; -unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; +unsigned pdpe =gen8_pdpe_index (start); +unsigned pde =gen8_pde_index (start); +unsigned pte =gen8_pte_index (start);unsigned num_entries = length >>PAGE_SHIFT ;unsigned last_pte, i;@@ -763,12 +763,24 @@ static void gen8_ppgtt_clear_range(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ -gen8_pte_t scratch_pte =gen8_pte_encode (px_dma (vm->scratch_page),I915_CACHE_LLC , use_scratch); -gen8_ppgtt_clear_pte_range (vm, pdp, start, length, scratch_pte); +if (!USES_FULL_48BIT_PPGTT (vm->dev)) { +gen8_ppgtt_clear_pte_range (vm, &ppgtt->pdp, start, length, + scratch_pte); + }else { +uint64_t templ4, pml4e; +struct i915_page_directory_pointer *pdp; + +gen8_for_each_pml4e (pdp, &ppgtt->pml4, start, length, templ4, pml4e) { +uint64_t pdp_len =gen8_clamp_pdp (start, length); +uint64_t pdp_start = start; + +gen8_ppgtt_clear_pte_range (vm, pdp, pdp_start, pdp_len, + scratch_pte); + } + } } static void@@ -781,9 +793,9 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr; -unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ; -unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ; -unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; +unsigned pdpe =gen8_pdpe_index (start); +unsigned pde =gen8_pde_index (start); +unsigned pte =gen8_pte_index (start); pt_vaddr =NULL ;
drivers/gpu/drm/i915/i915_gem_gtt.c gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm, · line 817
Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2
-1@@ -801,7 +813,8 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,kunmap_px (ppgtt, pt_vaddr); pt_vaddr =NULL ;if (++pde ==I915_PDES ) { - pdpe++; +if (++pdpe ==I915_PDPES_PER_PDP (vm->dev)) +break ;
Can the same pdpe check (for Page directory pointer boundary) be added in the gen8_ppgtt_clear_pte_range function also, to make it consistent with gen8_ppgtt_insert_pte_entries and this will also obviate the need for gen8_clamp_pdp macro.
Quoted 14 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2
-2pde =0 ; } pte =0 ;@@ -820,11 +833,25 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */ struct sg_page_iter sg_iter;__sg_page_iter_start (&sg_iter, pages->sgl,sg_nents (pages->sgl),0 ); -gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, start, cache_level); + +if (!USES_FULL_48BIT_PPGTT (vm->dev)) {
drivers/gpu/drm/i915/i915_gem_gtt.c static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, · line 850
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +10
-0+gen8_ppgtt_insert_pte_entries (vm, &ppgtt->pdp, &sg_iter, start, + cache_level); + }else { +struct i915_page_directory_pointer *pdp; +uint64_t templ4, pml4e; +uint64_t length = (uint64_t )sg_nents (pages->sgl) <<PAGE_SHIFT ; + +gen8_for_each_pml4e (pdp, &ppgtt->pml4, start, length, templ4, pml4e) { +uint64_t pdp_start = start; +
Isn't the 'pdp_start' dispensable here ? ‘start’ can be used directly.
Quoted 20 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +4
-0+gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, + pdp_start, cache_level); + } + } } static void gen8_free_page_tables(struct drm_device *dev,drivers/gpu/drm/i915/i915_gem_gtt.h +4
-0@@ -478,6 +478,17 @@ static inline uint32_t gen6_pde_index(uint32_t addr)#define gen8_for_each_pdpe (pd, pdp, start, length, temp, iter) \ gen8_for_each_pdpe_e(pd, pdp, start, length, temp, iter, I915_PDPES_PER_PDP(dev)) +/* Clamp length to the next page_directory pointer boundary */ +static inline uint64_t gen8_clamp_pdp (uint64_t start,uint64_t length) +{ +uint64_t next_pdp =ALIGN (start +1 ,1ULL <<GEN8_PML4E_SHIFT );
drivers/gpu/drm/i915/i915_gem_gtt.h static inline uint32_t gen6_pde_index(uint32_t addr) · line 494
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.h +7
-0+ +if (next_pdp > (start + length)) +return length; + +return next_pdp - start; +} +static inline uint32_t gen8_pte_index (uint64_t address) {return i915_pte_index (address,GEN8_PDE_SHIFT );
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 09/17] drm/i915/gen8: Add 4 level support in insert_entries and clear_range
Quoted 16 lines · Akash
Quoted 15 lines · Michel Thierry
static void
drivers/gpu/drm/i915/i915_gem_gtt.c +3
-3@@ -781,9 +793,9 @@ gen8_ppgtt_insert_pte_entries(structi915_address_space *vm, struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base);gen8_pte_t *pt_vaddr; -unsigned pdpe = start >>GEN8_PDPE_SHIFT &GEN8_PDPE_MASK ; -unsigned pde = start >>GEN8_PDE_SHIFT &GEN8_PDE_MASK ; -unsigned pte = start >>GEN8_PTE_SHIFT &GEN8_PTE_MASK ; +unsigned pdpe =gen8_pdpe_index (start); +unsigned pde =gen8_pde_index (start); +unsigned pte =gen8_pte_index (start); pt_vaddr =NULL ;
Quoted 14 lines
drivers/gpu/drm/i915/i915_gem_gtt.c gen8_ppgtt_insert_pte_entries(struct · line 818
Quoted 8 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2
-1@@ -801,7 +813,8 @@ gen8_ppgtt_insert_pte_entries(structi915_address_space *vm,kunmap_px (ppgtt ,pt_vaddr ); pt_vaddr =NULL ;if (++pde ==I915_PDES ) { - pdpe++; +if (++pdpe ==I915_PDPES_PER_PDP (vm->dev)) +break ;Can the same pdpe check (for Page directory pointer boundary) be added in the
gen8_ppgtt_clear_pte_rangefunction also, to make it consistent withgen8_ppgtt_insert_pte_entriesand this will also obviate the need forgen8_clamp_pdpmacro.
I will change gen8_ppgtt_clear_pte_range to stop at PDP boundary as you suggests (and clamp_pdp will go away).
Quoted 23 lines
Quoted 23 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +0
-0pde =0 ; }pte = 0;drivers/gpu/drm/i915/i915_gem_gtt.c +0
-1@@ -820,11 +833,25 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, {struct i915_hw_ppgtt *ppgtt = container_of(vm, struct i915_hw_ppgtt, base); -struct i915_page_directory_pointer *pdp = &ppgtt->pdp;/* FIXME: 48b */
struct sg_page_iter sg_iter; __sg_page_iter_start(&sg_iter, pages->sgl, sg_nents(pages->sgl),0);
- gen8_ppgtt_insert_pte_entries(vm, pdp,&sg_iter, start,cache_level);drivers/gpu/drm/i915/i915_gem_gtt.c +7
-0+ +if (!USES_FULL_48BIT_PPGTT (vm->dev)) { +gen8_ppgtt_insert_pte_entries (vm, &ppgtt->pdp, &sg_iter, start, + cache_level); + }else { +struct i915_page_directory_pointer *pdp; +uint64_t templ4, pml4e;
Quoted 9 lines
Review
Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +3
-0+uint64_t length = (uint64_t )sg_nents (pages->sgl) <<PAGE_SHIFT ; + + gen8_for_each_pml4e(pdp, &ppgtt->pml4, start, length, templ4,pml4e) {
(quoted diff) +2
-0+ uint64_t pdp_start = start; +Isn't the '
pdp_start' dispensable here ? ‘start’ can be used directly.
Yes, and the same applies in gen8_ppgtt_clear_range, pdp_len and pdp_start are redundant there.
Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_gtt.c static void gen8_ppgtt_insert_entries(struct i915_address_space *vm, · line 855
Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +4
-0+gen8_ppgtt_insert_pte_entries (vm, pdp, &sg_iter, + pdp_start, cache_level); + } + } }
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 10/17 drm/i915/gen8: Initialize PDPs
In reply to Michel ThierrySimilar to PDs, while setting up a page directory pointer, make all entries of the pdp point to the scratch pdp before mapping (and make all its entries point to the scratch page); this is to be safe in case of out of bound access or proactive prefetch.
Although the ggtt is always 32-bit, the scratch_pdp will be initialized/destroyed at the same time as the other scratch pages, to keep it consistent.
v2: Handle scratch_pdp allocation failure correctly, and keep initialize_px functions together (Akash)
v3: Rebase after Mika's ppgtt cleanup / scratch merge patch series. Rely on the added macros to initialize the pdps.
v4: Rebase after final merged version of Mika's ppgtt/scratch patches.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 41 +++++++++++++++++++++++++++++++++++-- drivers/gpu/drm/i915/i915_gem_gtt.h | 1 + 2 files changed, 40 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +39 -2
@@ -627,6 +627,27 @@ static void free_pdp(struct drm_device *dev, } } +static void gen8_initialize_pdp (struct i915_address_space *vm, +struct i915_page_directory_pointer *pdp) +{ +gen8_ppgtt_pdpe_t scratch_pdpe; + + scratch_pdpe =gen8_pdpe_encode (px_dma (vm->scratch_pd),I915_CACHE_LLC ); + +fill_px (vm->dev, pdp, scratch_pdpe); +} + +static void gen8_initialize_pml4 (struct i915_address_space *vm, +struct i915_pml4 *pml4) +{ +gen8_ppgtt_pml4e_t scratch_pml4e; + + scratch_pml4e =gen8_pml4e_encode (px_dma (vm->scratch_pdp), +I915_CACHE_LLC ); + +fill_px (vm->dev, pml4, scratch_pml4e); +} + static void gen8_setup_page_directory(struct i915_hw_ppgtt *ppgtt, struct i915_page_directory_pointer *pdp,
@@ -892,8 +913,20 @@ static int gen8_init_scratch(struct i915_address_space *vm)return PTR_ERR (vm->scratch_pd); } +if (USES_FULL_48BIT_PPGTT (dev)) { + vm->scratch_pdp =alloc_pdp (dev); +if (IS_ERR (vm->scratch_pdp)) { +free_pd (dev, vm->scratch_pd); +free_pt (dev, vm->scratch_pt); +free_scratch_page (dev, vm->scratch_page); +return PTR_ERR (vm->scratch_pdp); + } + } +gen8_initialize_pt (vm, vm->scratch_pt);gen8_initialize_pd (vm, vm->scratch_pd); +if (USES_FULL_48BIT_PPGTT (dev)) +gen8_initialize_pdp (vm, vm->scratch_pdp);return 0 ; }
@@ -902,6 +935,8 @@ static void gen8_free_scratch(struct i915_address_space *vm) {struct drm_device *dev = vm->dev; +if (USES_FULL_48BIT_PPGTT (dev)) +free_pdp (dev, vm->scratch_pdp);free_pd (dev, vm->scratch_pd);free_pt (dev, vm->scratch_pt);free_scratch_page (dev, vm->scratch_page);
@@ -1247,12 +1282,12 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm,* and 4 level code. Just allocate the pdps. */ gen8_for_each_pml4e (pdp, pml4, start, length, temp, pml4e) { -if (!pdp) { -WARN_ON (test_bit (pml4e, pml4->used_pml4es)); +if (!test_bit (pml4e, pml4->used_pml4es)) { pdp =alloc_pdp (vm->dev);if (IS_ERR (pdp)) goto err_out; +gen8_initialize_pdp (vm, pdp); pml4->pdps[pml4e] = pdp;__set_bit (pml4e, new_pdps); trace_i915_page_directory_pointer_entry_alloc(&ppgtt->base, pml4e,
@@ -1330,6 +1365,8 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)if (ret) goto free_scratch; +gen8_initialize_pml4 (&ppgtt->base, &ppgtt->pml4); + ppgtt->base.total =1ULL <<48 ; ppgtt->switch_mm = gen8_48b_mm_switch; }else {
drivers/gpu/drm/i915/i915_gem_gtt.h +1 -0
@@ -278,6 +278,7 @@ struct i915_address_space {struct i915_page_scratch *scratch_page;struct i915_page_table *scratch_pt;struct i915_page_directory *scratch_pd; +struct i915_page_directory_pointer *scratch_pdp;/* GEN8+ & 48b PPGTT */ /** * List of objects currently involved in rendering.
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 11/17 drm/i915: Expand error state's address width to 64b
In reply to Michel Thierry--- drivers/gpu/drm/i915/i915_drv.h | 4 ++-- drivers/gpu/drm/i915/i915_gpu_error.c | 17 +++++++++-------- 2 files changed, 11 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +2 -2
@@ -546,7 +546,7 @@ struct drm_i915_error_state {struct drm_i915_error_object {int page_count; -u32 gtt_offset; +u64 gtt_offset;u32 *pages[0 ]; } *ringbuffer, *batchbuffer, *wa_batchbuffer, *ctx, *hws_page;
@@ -572,7 +572,7 @@ struct drm_i915_error_state {u32 size;u32 name;u32 rseqno[I915_NUM_RINGS ], wseqno; -u32 gtt_offset; +u64 gtt_offset;u32 read_domains;u32 write_domain; s32 fence_reg:I915_MAX_NUM_FENCE_BITS;
drivers/gpu/drm/i915/i915_gpu_error.c +9 -8
@@ -197,7 +197,7 @@ static void print_error_buffers(struct drm_i915_error_state_buf *m,err_printf (m," %s [%d]:\n" , name, count);while (count--) { - err_printf(m, " %08x %8u %02x %02x [ ", + err_printf(m, " %016llx %8u %02x %02x [ ", err->gtt_offset, err->size, err->read_domains,
@@ -426,7 +426,7 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,err_printf (m," (submitted by %s [%d])" , error->ring[i].comm, error->ring[i].pid); -err_printf (m," --- gtt_offset = 0x%08x\n" , +err_printf (m," --- gtt_offset = 0x%016llx\n" , obj->gtt_offset);print_error_obj (m, obj); }
@@ -434,7 +434,8 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m, obj = error->ring[i].wa_batchbuffer;if (obj) {err_printf (m,"%s (w/a) --- gtt_offset = 0x%08x\n" , - dev_priv->ring[i].name, obj->gtt_offset); + dev_priv->ring[i].name, +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); }
@@ -453,14 +454,14 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,if ((obj = error->ring[i].ringbuffer)) {err_printf (m,"%s --- ringbuffer = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); }if ((obj = error->ring[i].hws_page)) {err_printf (m,"%s --- HW Status = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset)); offset =0 ;for (elt =0 ; elt <PAGE_SIZE /16 ; elt +=4 ) { err_printf(m, "[%04x] %08x %08x %08x %08x\n",
@@ -476,13 +477,13 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,if ((obj = error->ring[i].ctx)) {err_printf (m,"%s --- HW Context = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); } }if ((obj = error->semaphore_obj)) { -err_printf (m,"Semaphore page = 0x%08x\n" , obj->gtt_offset); +err_printf (m,"Semaphore page = 0x%016llx\n" , obj->gtt_offset);for (elt =0 ; elt <PAGE_SIZE /16 ; elt +=4 ) { err_printf(m, "[%04x] %08x %08x %08x %08x\n", elt * 4,
@@ -590,7 +591,7 @@ i915_error_object_create(struct drm_i915_private *dev_priv,int num_pages;bool use_ggtt;int i =0 ; -u32 reloc_offset; +u64 reloc_offset;if (src ==NULL || src->pages ==NULL )return NULL ;
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 11/17] drm/i915: Expand error state's address width to 64b
Quoted 81 lines · Michel Thierry
Signed-off-by Ben Widawsky <[email protected]>Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/i915_drv.h | 4 ++-- drivers/gpu/drm/i915/i915_gpu_error.c | 17 +++++++++-------- 2 files changed, 11 insertions(+), 10 deletions(-)drivers/gpu/drm/i915/i915_drv.h +2
-2@@ -546,7 +546,7 @@ struct drm_i915_error_state {struct drm_i915_error_object {int page_count; -u32 gtt_offset; +u64 gtt_offset;u32 *pages[0 ]; } *ringbuffer, *batchbuffer, *wa_batchbuffer, *ctx, *hws_page;@@ -572,7 +572,7 @@ struct drm_i915_error_state {u32 size;u32 name;u32 rseqno[I915_NUM_RINGS ], wseqno; -u32 gtt_offset; +u64 gtt_offset;u32 read_domains;u32 write_domain; s32 fence_reg:I915_MAX_NUM_FENCE_BITS;drivers/gpu/drm/i915/i915_gpu_error.c +6
-5@@ -197,7 +197,7 @@ static void print_error_buffers(struct drm_i915_error_state_buf *m,err_printf (m," %s [%d]:\n" , name, count);while (count--) { - err_printf(m, " %08x %8u %02x %02x [ ", + err_printf(m, " %016llx %8u %02x %02x [ ", err->gtt_offset, err->size, err->read_domains,@@ -426,7 +426,7 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,err_printf (m," (submitted by %s [%d])" , error->ring[i].comm, error->ring[i].pid); -err_printf (m," --- gtt_offset = 0x%08x\n" , +err_printf (m," --- gtt_offset = 0x%016llx\n" , obj->gtt_offset);print_error_obj (m, obj); }@@ -434,7 +434,8 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m, obj = error->ring[i].wa_batchbuffer;if (obj) {err_printf (m,"%s (w/a) --- gtt_offset = 0x%08x\n" , - dev_priv->ring[i].name, obj->gtt_offset); + dev_priv->ring[i].name, +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); }@@ -453,14 +454,14 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,if ((obj = error->ring[i].ringbuffer)) {err_printf (m,"%s --- ringbuffer = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); }if ((obj = error->ring[i].hws_page)) {err_printf (m,"%s --- HW Status = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset)); offset =0 ;for (elt =0 ; elt <PAGE_SIZE /16 ; elt +=4 ) { err_printf(m, "[%04x] %08x %08x %08x %08x\n",@@ -476,13 +477,13 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m,if ((obj = error->ring[i].ctx)) { err_printf(m, "%s --- HW Context = 0x%08x\n",
drivers/gpu/drm/i915/i915_gpu_error.c int i915_error_state_to_str(struct drm_i915_error_state_buf *m, · line 486
Quoted 10 lines
drivers/gpu/drm/i915/i915_gpu_error.c +2
-2dev_priv->ring[i].name, - obj->gtt_offset); + lower_32_bits(obj->gtt_offset));print_error_obj (m, obj); } }if ((obj = error->semaphore_obj)) { -err_printf (m,"Semaphore page = 0x%08x\n" , obj->gtt_offset); +err_printf (m,"Semaphore page = 0x%016llx\n" , obj->gtt_offset);
Can the 'lower_32_bits' be used for the semaphore object also. Its mapped into GGTT during ring init time, so may not have an offset > 4GB.
drivers/gpu/drm/i915/i915_gpu_error.c i915_error_object_create(struct drm_i915_private *dev_priv, · line 597
Quoted 13 lines
drivers/gpu/drm/i915/i915_gpu_error.c +1
-1for (elt =0 ; elt <PAGE_SIZE /16 ; elt +=4 ) { err_printf(m, "[%04x] %08x %08x %08x %08x\n", elt * 4,@@ -590,7 +591,7 @@ i915_error_object_create(struct drm_i915_private *dev_priv,int num_pages;bool use_ggtt;int i =0 ; -u32 reloc_offset; +u64 reloc_offset;if (src ==NULL || src->pages ==NULL )return NULL ;
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 11/17] drm/i915: Expand error state's address width to 64b
Quoted 12 lines · Akash
Quoted 11 lines · Michel Thierry
drivers/gpu/drm/i915/i915_gpu_error.c +1
-1@@ -476,13 +477,13 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m, if ((obj = error->ring[i].ctx)) {err_printf (m,"%s --- HW Context = 0x%08x\n" , dev_priv->ring[i].name, - obj->gtt_offset); +lower_32_bits (obj->gtt_offset));print_error_obj (m, obj); } }
Quoted 7 lines
drivers/gpu/drm/i915/i915_gpu_error.c int i915_error_state_to_str(struct drm_i915_error_state_buf *m, · line 486
Quoted 3 lines
drivers/gpu/drm/i915/i915_gpu_error.c +1
-1if ((obj = error->semaphore_obj)) { -err_printf (m,"Semaphore page = 0x%08x\n" , obj->gtt_offset); +err_printf (m,"Semaphore page = 0x%016llx\n" , obj->gtt_offset);Can the '
lower_32_bits' be used for the semaphore object also. Its mapped into GGTT during ring init time, so may not have an offset > 4GB.
Makes sense, will change to:
if ((obj = error->semaphore_obj)) {
err_printf(m, "Semaphore page = 0x%08x\n",
lower_32_bits(obj->gtt_offset));Quoted 11 lines
Quoted 11 lines
(quoted diff) +0
-0for (elt = 0; elt < PAGE_SIZE/16; elt += 4) { err_printf(m, "[%04x] %08x %08x %08x %08x\n",elt * 4,drivers/gpu/drm/i915/i915_gpu_error.c +1
-1@@ -590,7 +591,7 @@ i915_error_object_create(struct drm_i915_private *dev_priv, int num_pages;bool use_ggtt;int i =0 ; -u32 reloc_offset; +u64 reloc_offset;
Quoted 3 lines
Quoted 3 lines
drivers/gpu/drm/i915/i915_gpu_error.c +0
-0if (src ==NULL || src->pages ==NULL )return NULL;
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 12/17 drm/i915/gen8: Add ppgtt info and debug_dump
In reply to Michel Thierryv2: Clean up patch after rebases.
v3: gen8_dump_ppgtt for 32b and 48b PPGTT.
v4: Use used_pml4es/pdpes (Akash).
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.
--- drivers/gpu/drm/i915/i915_debugfs.c | 18 ++++---- drivers/gpu/drm/i915/i915_gem_gtt.c | 92 +++++++++++++++++++++++++++++++++++++ 2 files changed, 102 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +10 -8
@@ -2223,7 +2223,6 @@ static void gen6_ppgtt_info(struct seq_file *m, struct drm_device *dev) {struct drm_i915_private *dev_priv = dev->dev_private;struct intel_engine_cs *ring; -struct drm_file *file;int i;if (INTEL_INFO (dev)->gen ==6 )
@@ -2246,13 +2245,6 @@ static void gen6_ppgtt_info(struct seq_file *m, struct drm_device *dev) ppgtt->debug_dump (ppgtt, m); } -list_for_each_entry_reverse (file, &dev->filelist, lhead) { -struct drm_i915_file_private *file_priv = file->driver_priv; - -seq_printf (m,"proc: %s\n" , -get_pid_task (file->pid,PIDTYPE_PID )->comm); -idr_for_each (&file_priv->context_idr, per_file_ctx, m); - }seq_printf (m,"ECOCHK: 0x%08x\n" ,I915_READ (GAM_ECOCHK )); }
@@ -2261,6 +2253,7 @@ static int i915_ppgtt_info(struct seq_file *m, void *data)struct drm_info_node *node = m->private;struct drm_device *dev = node->minor->dev;struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_file *file;int ret =mutex_lock_interruptible (&dev->struct_mutex);if (ret)
@@ -2272,6 +2265,15 @@ static int i915_ppgtt_info(struct seq_file *m, void *data)else if (INTEL_INFO (dev)->gen >=6 )gen6_ppgtt_info (m, dev); +list_for_each_entry_reverse (file, &dev->filelist, lhead) { +struct drm_i915_file_private *file_priv = file->driver_priv; + +seq_printf (m,"\nproc: %s\n" , +get_pid_task (file->pid,PIDTYPE_PID )->comm); +idr_for_each (&file_priv->context_idr, per_file_ctx, + (void *)(unsigned long )m); + } +intel_runtime_pm_put (dev_priv);mutex_unlock (&dev->struct_mutex);
drivers/gpu/drm/i915/i915_gem_gtt.c +92 -0
104 lines not on this page · raw message
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 12/17] drm/i915/gen8: Add ppgtt info and debug_dump
Quoted 142 lines · Michel Thierry
v2: Clean up patch after rebases.
v3:gen8_dump_ppgttfor 32b and 48b PPGTT.
v4: Use used_pml4es/pdpes (Akash).
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.Signed-off-by Ben Widawsky <[email protected]>Signed-off-by Michel Thierry <[email protected]> (v2+)--- drivers/gpu/drm/i915/i915_debugfs.c | 18 ++++---- drivers/gpu/drm/i915/i915_gem_gtt.c | 92 +++++++++++++++++++++++++++++++++++++ 2 files changed, 102 insertions(+), 8 deletions(-)drivers/gpu/drm/i915/i915_debugfs.c +10
-8@@ -2223,7 +2223,6 @@ static void gen6_ppgtt_info(struct seq_file *m, struct drm_device *dev) {struct drm_i915_private *dev_priv = dev->dev_private;struct intel_engine_cs *ring; -struct drm_file *file;int i;if (INTEL_INFO (dev)->gen ==6 )@@ -2246,13 +2245,6 @@ static void gen6_ppgtt_info(struct seq_file *m, struct drm_device *dev) ppgtt->debug_dump (ppgtt, m); } -list_for_each_entry_reverse (file, &dev->filelist, lhead) { -struct drm_i915_file_private *file_priv = file->driver_priv; - -seq_printf (m,"proc: %s\n" , -get_pid_task (file->pid,PIDTYPE_PID )->comm); -idr_for_each (&file_priv->context_idr, per_file_ctx, m); - }seq_printf (m,"ECOCHK: 0x%08x\n" ,I915_READ (GAM_ECOCHK )); }@@ -2261,6 +2253,7 @@ static int i915_ppgtt_info(struct seq_file *m, void *data)struct drm_info_node *node = m->private;struct drm_device *dev = node->minor->dev;struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_file *file;int ret =mutex_lock_interruptible (&dev->struct_mutex);if (ret)@@ -2272,6 +2265,15 @@ static int i915_ppgtt_info(struct seq_file *m, void *data)else if (INTEL_INFO (dev)->gen >=6 )gen6_ppgtt_info (m, dev); +list_for_each_entry_reverse (file, &dev->filelist, lhead) { +struct drm_i915_file_private *file_priv = file->driver_priv; + +seq_printf (m,"\nproc: %s\n" , +get_pid_task (file->pid,PIDTYPE_PID )->comm); +idr_for_each (&file_priv->context_idr, per_file_ctx, + (void *)(unsigned long )m); + } +intel_runtime_pm_put (dev_priv);mutex_unlock (&dev->struct_mutex);drivers/gpu/drm/i915/i915_gem_gtt.c +72
-0@@ -1337,6 +1337,97 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,return gen8_alloc_va_range_3lvl (vm, &ppgtt->pdp, start, length); } +static void gen8_dump_pdp (struct i915_page_directory_pointer *pdp, +uint64_t start,uint64_t length, +gen8_pte_t scratch_pte, +struct seq_file *m) +{ +struct i915_page_directory *pd; +uint64_t temp; +uint32_t pdpe; + +gen8_for_each_pdpe (pd, pdp, start, length, temp, pdpe) { +struct i915_page_table *pt; +uint64_t pd_len = length; +uint64_t pd_start = start; +uint32_t pde; + +if (!pd) +continue ; + +if (!test_bit (pdpe, pdp->used_pdpes)) +continue ; + +seq_printf (m,"\tPDPE #%d\n" , pdpe); +gen8_for_each_pde (pt, pd, pd_start, pd_len, temp, pde) { +uint32_t pte; +gen8_pte_t *pt_vaddr; + +if (!pt) +continue ; + + pt_vaddr =kmap_px (pt); +for (pte =0 ; pte <GEN8_PTES ; pte+=4 ) { +uint64_t va = + (pdpe <<GEN8_PDPE_SHIFT ) | + (pde <<GEN8_PDE_SHIFT ) | + (pte <<GEN8_PTE_SHIFT ); +int i; +bool found = false; +for (i =0 ; i <4 ; i++) +if (pt_vaddr[pte + i] != scratch_pte) + found = true; +if (!found) +continue ; + +seq_printf (m,"\t\t0x%llx [%03d,%03d,%04d]: =" , va, pdpe, pde, pte); +for (i =0 ; i <4 ; i++) { +if (pt_vaddr[pte + i] != scratch_pte) +seq_printf (m," %llx" , pt_vaddr[pte + i]); +else +seq_puts (m," SCRATCH " ); + } +seq_puts (m,"\n" ); + } +/* don't use kunmap_px, it could trigger +* an unnecessary flush. +*/ +kunmap_atomic (pt_vaddr); + } + } +} + +static void gen8_dump_ppgtt (struct i915_hw_ppgtt *ppgtt,struct seq_file *m) +{ +struct i915_address_space *vm = &ppgtt->base; +uint64_t start = ppgtt->base.start; +uint64_t length = ppgtt->base.total; +gen8_pte_t scratch_pte =gen8_pte_encode (px_dma (vm->scratch_page), +I915_CACHE_LLC , true); + +if (!USES_FULL_48BIT_PPGTT (vm->dev)) { +gen8_dump_pdp (&ppgtt->pdp, start, length, scratch_pte, m); + }else { +uint64_t templ4, pml4e;
drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_alloc_va_range(struct i915_address_space *vm, · line 1421
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +10
-0+struct i915_pml4 *pml4 = &ppgtt->pml4; +struct i915_page_directory_pointer *pdp; + +gen8_for_each_pml4e (pdp, pml4, start, length, templ4, pml4e) { +uint64_t pdp_len = length; +uint64_t pdp_start = start; + +if (!pdp) +continue ; +
I think the "if (!test_bit(pml4e, pml4->used_pml4es))" check is foolproof & should suffice. No real need of the extra check of 'if (!pdp)'.
Same for pdpe & pde loops in gen8_dump_pdp function
Quoted 14 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +9
-0+if (!test_bit (pml4e, pml4->used_pml4es)) +continue ; + +seq_printf (m," PML4E #%llu\n" , pml4e); +gen8_dump_pdp (pdp, pdp_start, pdp_len, scratch_pte, m); + } + } +} +/* * GEN8 legacy ppgtt programming is accomplished through a max 4 PDP registers * with a net effect resembling a 2-level page table in normal x86 terms. Each
drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) · line 1456
Quoted 7 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +1
-0@@ -1359,6 +1450,7 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) ppgtt->base.clear_range = gen8_ppgtt_clear_range; ppgtt->base.unbind_vma = ppgtt_unbind_vma; ppgtt->base.bind_vma = ppgtt_bind_vma; + ppgtt->debug_dump = gen8_dump_ppgtt;if (USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) { ret =setup_px (ppgtt->base.dev, &ppgtt->pml4);
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 12/17] drm/i915/gen8: Add ppgtt info and debug_dump
Quoted 32 lines · Akash
Quoted 31 lines · Michel Thierry
v2: Clean up patch after rebases.
v3:gen8_dump_ppgttfor 32b and 48b PPGTT.
v4: Use used_pml4es/pdpes (Akash).
v5: Rebase after Mika's ppgtt cleanup / scratch merge patch series.Signed-off-by Ben Widawsky <[email protected]>Signed-off-by Michel Thierry <[email protected]> (v2+)--- drivers/gpu/drm/i915/i915_debugfs.c | 18 ++++---- drivers/gpu/drm/i915/i915_gem_gtt.c | 92 +++++++++++++++++++++++++++++++++++++ 2 files changed, 102 insertions(+), 8 deletions(-)+static void
gen8_dump_ppgtt(struct i915_hw_ppgtt *ppgtt, struct seq_file *m)drivers/gpu/drm/i915/i915_gem_gtt.c +17
-0+{ +struct i915_address_space *vm = &ppgtt->base; +uint64_t start = ppgtt->base.start; +uint64_t length = ppgtt->base.total; +gen8_pte_t scratch_pte =gen8_pte_encode (px_dma (vm->scratch_page), +I915_CACHE_LLC , true); + +if (!USES_FULL_48BIT_PPGTT (vm->dev)) { +gen8_dump_pdp (&ppgtt->pdp, start, length, scratch_pte, m); + }else { +uint64_t templ4, pml4e; +struct i915_pml4 *pml4 = &ppgtt->pml4; +struct i915_page_directory_pointer *pdp; + +gen8_for_each_pml4e (pdp, pml4, start, length, templ4, pml4e) { +uint64_t pdp_len = length; +uint64_t pdp_start = start;
Quoted 9 lines
drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_alloc_va_range(struct i915_address_space *vm, · line 1421
Quoted 4 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +4
-0+ +if (!pdp) +continue ; +I think the "if (!test_bit(pml4e,
pml4->used_pml4es))" check is foolproof & should suffice.
No real need of the extra check of 'if (!pdp)'.
Same for pdpe & pde loops ingen8_dump_pdpfunction
Right, I'll changed it to use test_bit across the board, also remove the unnecessary pdp_len/pdp_start variables:
- if (!pd)
- continue;
-
if(!test_bit(pdpe, pdp->used_pdpes))
continue;
-----------
- if (!pt)
+ if(!test_bit(pde, pd->used_pdes))
continue;
-----------
gen8_for_each_pml4e(pdp, pml4, start, length, templ4, pml4e) {- uint64_t pdp_len = length;
- uint64_t pdp_start = start;
-
- if (!pdp)
- continue;
-
if (!test_bit(pml4e, pml4->used_pml4es))
continue;
seq_printf(m, " PML4E #%llu\n", pml4e);- gen8_dump_pdp(pdp, pdp_start, pdp_len, scratch_pte, m);
+ gen8_dump_pdp(pdp, start, length, scratch_pte, m);
-----------
Quoted 21 lines
Quoted 21 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +9
-0+if (!test_bit (pml4e, pml4->used_pml4es)) +continue ; + +seq_printf (m," PML4E #%llu\n" , pml4e); +gen8_dump_pdp (pdp, pdp_start, pdp_len, scratch_pte, m); + } + } +} +/* * GEN8 legacy ppgtt programming is accomplished through a max 4 PDP registers
* with a net effect resembling a 2-level page table in normal x86
terms. Eachdrivers/gpu/drm/i915/i915_gem_gtt.c +1
-0@@ -1359,6 +1450,7 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) ppgtt->base.clear_range = gen8_ppgtt_clear_range; ppgtt->base.unbind_vma = ppgtt_unbind_vma; ppgtt->base.bind_vma = ppgtt_bind_vma; + ppgtt->debug_dump = gen8_dump_ppgtt;
Quoted 3 lines
Quoted 3 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +0
-0if (USES_FULL_48BIT_PPGTT (ppgtt->base.dev)) {ret = setup_px(ppgtt->base.dev, &ppgtt->pml4);
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 13/17 drm/i915: object size needs to be u64
In reply to Michel ThierryIn a 48b world, users can try to allocate buffers bigger than 4GB; in these cases it is important that size is a 64b variable.
Also added a warning for illegal bind with size = 0.
--- drivers/gpu/drm/i915/i915_gem.c | 5 +++-- drivers/gpu/drm/i915/i915_gem_gtt.c | 3 +++ 2 files changed, 6 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +3 -2
@@ -3716,7 +3716,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, {struct drm_device *dev = obj->base.dev;struct drm_i915_private *dev_priv = dev->dev_private; -u32 size, fence_size, fence_alignment, unfenced_alignment; +u32 fence_alignment, unfenced_alignment; +u64 size, fence_size;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =
@@ -3775,7 +3776,7 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,* attempt to find space. */ if (size > end) { - DRM_DEBUG("Attempting to bind an object (view type=%u) larger than the aperture: size=%u > %s aperture=%llu\n", + DRM_DEBUG("Attempting to bind an object (view type=%u) larger than the aperture: size=%llu > %s aperture=%llu\n", ggtt_view ? ggtt_view->type : 0, size, flags & PIN_MAPPABLE ? "mappable" : "total",
drivers/gpu/drm/i915/i915_gem_gtt.c +3 -0
@@ -3336,6 +3336,9 @@ int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,if (WARN_ON (flags ==0 ))return -EINVAL ; +if (WARN_ON (vma->node.size ==0 )) +return -EINVAL ; + bind_flags =0 ;if (flags &PIN_GLOBAL ) bind_flags |=GLOBAL_BIND ;
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 14/17 drm/i915: batch_obj vm offset must be u64
In reply to Michel ThierryOtherwise it can overflow in 48-bit mode, and cause an incorrect exec_start.
Before commit 5f19e2bff ("drm/i915: Merged the many do_execbuf() parameters into a structure"), it was already an u64, so it could be seen as a regression (or as an optimization that looked good at that time).
--- drivers/gpu/drm/i915/i915_drv.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_drv.h +1 -1
@@ -1664,7 +1664,7 @@ struct i915_execbuffer_params {struct drm_file *file;uint32_t dispatch_flags;uint32_t args_batch_start_offset; -uint32_t batch_obj_vm_offset; +uint64_t batch_obj_vm_offset;struct intel_engine_cs *ring;struct drm_i915_gem_object *batch_obj;struct intel_context *ctx;
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 14/17] drm/i915: batch_obj vm offset must be u64
Quoted 6 lines · Michel Thierry
Otherwise it can overflow in 48-bit mode, and cause an incorrect
exec_start.Before commit
5f19e2bff("drm/i915: Merged the manydo_execbuf()parameters into a structure"), it was already an u64, so it could be seen as a regression (or as an optimization that looked good at that time).
Almost certainly a merge failure. The above patch moved the variable when it was a uint32_t but by the time it got merged, another patch had updated it to uint64_t. Unfortunately, the merge conflict either didn't conflict or didn't get resolved correctly. Either way, the downgrade was certainly not intentional.
Quoted 11 lines
Cc John Harrison <[email protected]>Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/i915_drv.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/gpu/drm/i915/i915_drv.h +0
-0
drivers/gpu/drm/i915/i915_drv.h struct i915_execbuffer_params { · line 1670
Quoted 8 lines
drivers/gpu/drm/i915/i915_drv.h +1
-1@@ -1664,7 +1664,7 @@ struct i915_execbuffer_params {struct drm_file *file;uint32_t dispatch_flags;uint32_t args_batch_start_offset; -uint32_t batch_obj_vm_offset; +uint64_t batch_obj_vm_offset;struct intel_engine_cs *ring;struct drm_i915_gem_object *batch_obj;struct intel_context *ctx;
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 15/17 drm/i915/userptr: Kill user_size limit check
In reply to Michel ThierryGTT was only 32b and its max value is 4GB. In order to allow objects bigger than 4GB in 48b PPGTT, i915_gem_userptr_ioctl we could check against max 48b range (1ULL << 48).
But since the check no longer applies, just kill the limit.
v2: Use the default ctx to infer the ppgtt max size (Akash).
v3: Just kill the limit, it was only there for early detection of an error when used for execbuffer (Chris).
--- drivers/gpu/drm/i915/i915_gem_userptr.c | 4 ---- 1 file changed, 4 deletions(-)
drivers/gpu/drm/i915/i915_gem_userptr.c +0 -4
@@ -788,7 +788,6 @@ static const struct drm_i915_gem_object_ops i915_gem_userptr_ops = {int i915_gem_userptr_ioctl (struct drm_device *dev,void *data,struct drm_file *file) { -struct drm_i915_private *dev_priv = dev->dev_private;struct drm_i915_gem_userptr *args = data;struct drm_i915_gem_object *obj;int ret;
@@ -801,9 +800,6 @@ i915_gem_userptr_ioctl(struct drm_device *dev, void *data, struct drm_file *fileif (offset_in_page (args->user_ptr | args->user_size))return -EINVAL ; -if (args->user_size > dev_priv->gtt.base.total) -return -E2BIG ; -if (!access_ok (args->flags &I915_USERPTR_READ_ONLY ?VERIFY_READ :VERIFY_WRITE , (char __user *)(unsigned long )args->user_ptr, args->user_size))return -EFAULT ;
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 15/17] drm/i915/userptr: Kill user_size limit check
Quoted 6 lines · Michel Thierry
GTT was only 32b and its max value is 4GB. In order to allow objects bigger than 4GB in 48b PPGTT,
i915_gem_userptr_ioctlwe could check against max 48b range (1ULL << 48).But since the check no longer applies, just kill the limit.
Quoted 7 lines
v2: Use the default ctx to infer the ppgtt max size (Akash).
v3: Just kill the limit, it was only there for early detection of an error when used for execbuffer (Chris).Cc Akash Goel <[email protected]>Cc Chris Wilson <[email protected]>Signed-off-by Michel Thierry <[email protected]>
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 16/17 drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
In reply to Michel ThierryThere are some allocations that must be only referenced by 32-bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/ DRM_MM_CREATE_TOP flags
In specific, any resource used with flat/heapless (0x00000000-0xfffff000) General State Heap (GSH) or Intructions State Heap (ISH) must be in a 32-bit range, because the General State Offset and Instruction State Offset are limited to 32-bits.
Objects must have EXEC_OBJECT_SUPPORTS_48B_ADDRESS flag to indicate if they can be allocated above the 32-bit address range. To limit the chances of having the first 4GB already full, objects will use DRM_MM_SEARCH_BELOW + DRM_MM_CREATE_TOP flags when possible.
v2: Changed flag logic from neeeds_32b, to supports_48b.
v3: Moved 48-bit support flag back to exec_object. (Chris, Daniel)
v4: Split pin flags into PIN_ZONE_4G and PIN_HIGH; update PIN_OFFSET_MASK to use last PIN_ defined instead of hard-coded value; use correct limit check in eb_vma_misplaced. (Chris)
--- drivers/gpu/drm/i915/i915_drv.h | 4 +++- drivers/gpu/drm/i915/i915_gem.c | 17 +++++++++++++++-- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 10 ++++++++++ include/uapi/drm/i915_drm.h | 3 ++- 4 files changed, 30 insertions(+), 4 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +3 -1
@@ -2765,7 +2765,9 @@ void i915_gem_vma_destroy(struct i915_vma *vma);#define PIN_OFFSET_BIAS (1<<3)#define PIN_USER (1<<4)#define PIN_UPDATE (1<<5) -#define PIN_OFFSET_MASK (~4095) +#define PIN_ZONE_4G (1<<6) +#define PIN_HIGH (1<<7) +#define PIN_OFFSET_MASK -(PIN_HIGH<<1) int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj, struct i915_address_space *vm,
drivers/gpu/drm/i915/i915_gem.c +15 -2
@@ -3718,6 +3718,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,struct drm_i915_private *dev_priv = dev->dev_private;u32 fence_alignment, unfenced_alignment;u64 size, fence_size; +u32 search_flag =DRM_MM_SEARCH_DEFAULT ; +u32 alloc_flag =DRM_MM_CREATE_DEFAULT ;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =
@@ -3759,6 +3761,17 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, obj->tiling_mode, false); size = flags &PIN_MAPPABLE ? fence_size : obj->base.size; + +if (flags &PIN_HIGH ) { + search_flag =DRM_MM_SEARCH_BELOW ; + alloc_flag =DRM_MM_CREATE_TOP ; + } + +/* Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset, +* limit address to the first 4GBs for flagged objects. +*/ +if (flags &PIN_ZONE_4G ) + end = (1ULL <<32 ); }if (alignment ==0 )
@@ -3801,8 +3814,8 @@ search_free: size, alignment, obj->cache_level, start, end, -DRM_MM_SEARCH_DEFAULT , - DRM_MM_CREATE_DEFAULT); + search_flag, + alloc_flag);if (ret) { ret = i915_gem_evict_something(dev, vm, size, alignment, obj->cache_level,
drivers/gpu/drm/i915/i915_gem_execbuffer.c +10 -0
@@ -588,11 +588,17 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,if (entry->flags &EXEC_OBJECT_NEEDS_GTT ) flags |=PIN_GLOBAL ; + flags |=PIN_ZONE_4G ; +if (entry->flags &EXEC_OBJECT_SUPPORTS_48B_ADDRESS ) + flags &= ~PIN_ZONE_4G ; +if (!drm_mm_node_allocated (&vma->node)) {if (entry->flags & __EXEC_OBJECT_NEEDS_MAP) flags |=PIN_GLOBAL |PIN_MAPPABLE ;if (entry->flags & __EXEC_OBJECT_NEEDS_BIAS) flags |=BATCH_OFFSET_BIAS |PIN_OFFSET_BIAS ; +if ((flags &PIN_MAPPABLE ) ==0 ) + flags |=PIN_HIGH ; } ret =i915_gem_object_pin (obj, vma->vm, entry->alignment, flags);
@@ -670,6 +676,10 @@ eb_vma_misplaced(struct i915_vma *vma)if (entry->flags & __EXEC_OBJECT_NEEDS_MAP && !obj->map_and_fenceable)return !only_mappable_for_reloc (entry->flags); +if (!(entry->flags &EXEC_OBJECT_SUPPORTS_48B_ADDRESS ) && + (vma->node.start + vma->node.size) >= (1ULL <<32 )) +return true; +return false; }
include/uapi/drm/i915_drm.h +2 -1
@@ -685,7 +685,8 @@ struct drm_i915_gem_exec_object2 {#define EXEC_OBJECT_NEEDS_FENCE (1<<0)#define EXEC_OBJECT_NEEDS_GTT (1<<1)#define EXEC_OBJECT_WRITE (1<<2) -#define __EXEC_OBJECT_UNKNOWN_FLAGS -(EXEC_OBJECT_WRITE<<1) +#define EXEC_OBJECT_SUPPORTS_48B_ADDRESS (1<<3) +#define __EXEC_OBJECT_UNKNOWN_FLAGS -(EXEC_OBJECT_SUPPORTS_48B_ADDRESS<<1)__u64 flags;__u64 rsvd1;
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 16/17] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 36 lines · Michel Thierry
There are some allocations that must be only referenced by 32-bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/
DRM_MM_CREATE_TOPflagsIn specific, any resource used with flat/heapless (0x00000000-0xfffff000) General State Heap (GSH) or Intructions State Heap (ISH) must be in a 32-bit range, because the General State Offset and Instruction State Offset are limited to 32-bits.
Objects must have
EXEC_OBJECT_SUPPORTS_48B_ADDRESSflag to indicate if they can be allocated above the 32-bit address range. To limit the chances of having the first 4GB already full, objects will useDRM_MM_SEARCH_BELOW+DRM_MM_CREATE_TOPflags when possible.v2: Changed flag logic from
neeeds_32b, tosupports_48b.
v3: Moved 48-bit support flag back toexec_object. (Chris, Daniel)
v4: Split pin flags intoPIN_ZONE_4GandPIN_HIGH; updatePIN_OFFSET_MASKto use lastPIN_defined instead of hard-coded value; use correct limit check ineb_vma_misplaced. (Chris)Cc Chris Wilson <[email protected]>Cc Ben Widawsky <[email protected]>Signed-off-by Michel Thierry <[email protected]>--- drivers/gpu/drm/i915/i915_drv.h | 4 +++- drivers/gpu/drm/i915/i915_gem.c | 17 +++++++++++++++-- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 10 ++++++++++ include/uapi/drm/i915_drm.h | 3 ++- 4 files changed, 30 insertions(+), 4 deletions(-)drivers/gpu/drm/i915/i915_drv.h +0
-0
drivers/gpu/drm/i915/i915_drv.h void i915_gem_vma_destroy(struct i915_vma *vma); · line 2770
Quoted 7 lines
drivers/gpu/drm/i915/i915_drv.h +3
-1@@ -2765,7 +2765,9 @@ void i915_gem_vma_destroy(struct i915_vma *vma);#define PIN_OFFSET_BIAS (1<<3)#define PIN_USER (1<<4)#define PIN_UPDATE (1<<5) -#define PIN_OFFSET_MASK (~4095) +#define PIN_ZONE_4G (1<<6) +#define PIN_HIGH (1<<7) +#define PIN_OFFSET_MASK -(PIN_HIGH<<1)
The offset has to be 4096 aligned - it imposes an upper limit on how many low bits we can use for flags. When we exceed it, it probably past time for a params struct!
Quoted 19 lines
drivers/gpu/drm/i915/i915_drv.h +0
-0int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj, struct i915_address_space *vm,drivers/gpu/drm/i915/i915_gem.c +2
-0@@ -3718,6 +3718,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,struct drm_i915_private *dev_priv = dev->dev_private;u32 fence_alignment, unfenced_alignment;u64 size, fence_size; +u32 search_flag =DRM_MM_SEARCH_DEFAULT ; +u32 alloc_flag =DRM_MM_CREATE_DEFAULT ;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =@@ -3759,6 +3761,17 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, obj->tiling_mode, false);
drivers/gpu/drm/i915/i915_gem.c i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, · line 3772
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem.c +9
-0size = flags &PIN_MAPPABLE ? fence_size : obj->base.size; + +if (flags &PIN_HIGH ) { + search_flag =DRM_MM_SEARCH_BELOW ; + alloc_flag =DRM_MM_CREATE_TOP ; + } + +/* Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset, +* limit address to the first 4GBs for flagged objects. +*/
This note is best next to where we set PIN_ZONE_4G in execbuffer.
Quoted 15 lines
drivers/gpu/drm/i915/i915_gem.c +2
-0+if (flags &PIN_ZONE_4G ) + end = (1ULL <<32 ); }if (alignment ==0 )drivers/gpu/drm/i915/i915_gem_execbuffer.c +1
-0@@ -588,11 +588,17 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,if (entry->flags &EXEC_OBJECT_NEEDS_GTT ) flags |=PIN_GLOBAL ; + flags |=PIN_ZONE_4G ;
drivers/gpu/drm/i915/i915_gem_execbuffer.c i915_gem_execbuffer_reserve_vma(struct i915_vma *vma, · line 601
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_execbuffer.c +5
-0+if (entry->flags &EXEC_OBJECT_SUPPORTS_48B_ADDRESS ) + flags &= ~PIN_ZONE_4G ; +if (!drm_mm_node_allocated (&vma->node)) {if (entry->flags & __EXEC_OBJECT_NEEDS_MAP) flags |=PIN_GLOBAL |PIN_MAPPABLE ;if (entry->flags & __EXEC_OBJECT_NEEDS_BIAS) flags |=BATCH_OFFSET_BIAS |PIN_OFFSET_BIAS ; +if ((flags &PIN_MAPPABLE ) ==0 ) + flags |=PIN_HIGH ;
I'm still debating the right semantics to use, but I'm happy with this until I can find something better. (The biggest issue is that drm_mm is not indexed for fast top-down searching. The current search code I put into drm_mm is unfortunately broken, the idea I have in mind to fix it is to add a hole_list into drm_mm/drm_mm_node, so that we can just walk holes in up/down, recent/old order. And with that allocating top-down will not be any more expensive than the current reuse recent hole - though perhaps given a fragment drm_mm the hole stack probably requires fewer steps to find a large hole.)
Other than don't touch PIN_OFFSET_MASK and move the w/a note next to where we tweak the PIN_ZONE_4G, it lgtm, so with those changes,
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 16/17] drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
Quoted 24 lines · Chris Wilson
drivers/gpu/drm/i915/i915_gem_execbuffer.c i915_gem_execbuffer_reserve_vma(struct i915_vma *vma, · line 601
Quoted 12 lines · Michel Thierry
drivers/gpu/drm/i915/i915_gem_execbuffer.c +6
-0+ flags |=PIN_ZONE_4G ; +if (entry->flags &EXEC_OBJECT_SUPPORTS_48B_ADDRESS ) + flags &= ~PIN_ZONE_4G ; +if (!drm_mm_node_allocated (&vma->node)) {if (entry->flags & __EXEC_OBJECT_NEEDS_MAP) flags |=PIN_GLOBAL |PIN_MAPPABLE ;if (entry->flags & __EXEC_OBJECT_NEEDS_BIAS) flags |=BATCH_OFFSET_BIAS |PIN_OFFSET_BIAS ; +if ((flags &PIN_MAPPABLE ) ==0 ) + flags |=PIN_HIGH ;I'm still debating the right semantics to use, but I'm happy with this until I can find something better. (The biggest issue is that
drm_mmis not indexed for fast top-down searching. The current search code I put intodrm_mmis unfortunately broken, the idea I have in mind to fix it is to add ahole_listinto drm_mm/drm_mm_node, so that we can just walk holes in up/down, recent/old order. And with that allocating top-down will not be any more expensive than the current reuse recent hole - though perhaps given a fragmentdrm_mmthe hole stack probably requires fewer steps to find a large hole.)
Quoted 2 lines
Other than don't touch
PIN_OFFSET_MASKand move the w/a note next to where we tweak thePIN_ZONE_4G, it lgtm, so with those changes,
Also the comment should be "limit address to the first 4GBs for _unflagged_ objects". I didn't update it after changing the logic.
I'll resend with these changes.
Thanks,
Quoted 4 lines
Reviewed-by Chris Wilson <[email protected]>-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v5 1/1 drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset
In reply to Michel ThierryThere are some allocations that must be only referenced by 32-bit offsets. To limit the chances of having the first 4GB already full, objects not requiring this workaround use DRM_MM_SEARCH_BELOW/ DRM_MM_CREATE_TOP flags
In specific, any resource used with flat/heapless (0x00000000-0xfffff000) General State Heap (GSH) or Intructions State Heap (ISH) must be in a 32-bit range, because the General State Offset and Instruction State Offset are limited to 32-bits.
Objects must have EXEC_OBJECT_SUPPORTS_48B_ADDRESS flag to indicate if they can be allocated above the 32-bit address range. To limit the chances of having the first 4GB already full, objects will use DRM_MM_SEARCH_BELOW + DRM_MM_CREATE_TOP flags when possible.
v2: Changed flag logic from neeeds_32b, to supports_48b.
v3: Moved 48-bit support flag back to exec_object. (Chris, Daniel)
v4: Split pin flags into PIN_ZONE_4G and PIN_HIGH; update PIN_OFFSET_MASK to use last PIN_ defined instead of hard-coded value; use correct limit check in eb_vma_misplaced. (Chris)
v5: Don't touch PIN_OFFSET_MASK and update workaround comment (Chris)
--- drivers/gpu/drm/i915/i915_drv.h | 2 ++ drivers/gpu/drm/i915/i915_gem.c | 14 ++++++++++++-- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 13 +++++++++++++ include/uapi/drm/i915_drm.h | 3 ++- 4 files changed, 29 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +2 -0
@@ -2768,6 +2768,8 @@ void i915_gem_vma_destroy(struct i915_vma *vma);#define PIN_OFFSET_BIAS (1<<3)#define PIN_USER (1<<4)#define PIN_UPDATE (1<<5) +#define PIN_ZONE_4G (1<<6) +#define PIN_HIGH (1<<7)#define PIN_OFFSET_MASK (~4095) int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,
drivers/gpu/drm/i915/i915_gem.c +12 -2
@@ -3722,6 +3722,8 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,struct drm_i915_private *dev_priv = dev->dev_private;u32 fence_alignment, unfenced_alignment;u64 size, fence_size; +u32 search_flag =DRM_MM_SEARCH_DEFAULT ; +u32 alloc_flag =DRM_MM_CREATE_DEFAULT ;u64 start = flags &PIN_OFFSET_BIAS ? flags &PIN_OFFSET_MASK :0 ; u64 end =
@@ -3763,6 +3765,14 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj, obj->tiling_mode, false); size = flags &PIN_MAPPABLE ? fence_size : obj->base.size; + +if (flags &PIN_HIGH ) { + search_flag =DRM_MM_SEARCH_BELOW ; + alloc_flag =DRM_MM_CREATE_TOP ; + } + +if (flags &PIN_ZONE_4G ) + end = (1ULL <<32 ); }if (alignment ==0 )
@@ -3805,8 +3815,8 @@ search_free: size, alignment, obj->cache_level, start, end, -DRM_MM_SEARCH_DEFAULT , - DRM_MM_CREATE_DEFAULT); + search_flag, + alloc_flag);if (ret) { ret = i915_gem_evict_something(dev, vm, size, alignment, obj->cache_level,
drivers/gpu/drm/i915/i915_gem_execbuffer.c +13 -0
@@ -588,11 +588,20 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,if (entry->flags &EXEC_OBJECT_NEEDS_GTT ) flags |=PIN_GLOBAL ; +/* Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset, +* limit address to the first 4GBs for unflagged objects. +*/ + flags |=PIN_ZONE_4G ; +if (entry->flags &EXEC_OBJECT_SUPPORTS_48B_ADDRESS ) + flags &= ~PIN_ZONE_4G ; +if (!drm_mm_node_allocated (&vma->node)) {if (entry->flags & __EXEC_OBJECT_NEEDS_MAP) flags |=PIN_GLOBAL |PIN_MAPPABLE ;if (entry->flags & __EXEC_OBJECT_NEEDS_BIAS) flags |=BATCH_OFFSET_BIAS |PIN_OFFSET_BIAS ; +if ((flags &PIN_MAPPABLE ) ==0 ) + flags |=PIN_HIGH ; } ret =i915_gem_object_pin (obj, vma->vm, entry->alignment, flags);
@@ -670,6 +679,10 @@ eb_vma_misplaced(struct i915_vma *vma)if (entry->flags & __EXEC_OBJECT_NEEDS_MAP && !obj->map_and_fenceable)return !only_mappable_for_reloc (entry->flags); +if (!(entry->flags &EXEC_OBJECT_SUPPORTS_48B_ADDRESS ) && + (vma->node.start + vma->node.size) >= (1ULL <<32 )) +return true; +return false; }
include/uapi/drm/i915_drm.h +2 -1
@@ -685,7 +685,8 @@ struct drm_i915_gem_exec_object2 {#define EXEC_OBJECT_NEEDS_FENCE (1<<0)#define EXEC_OBJECT_NEEDS_GTT (1<<1)#define EXEC_OBJECT_WRITE (1<<2) -#define __EXEC_OBJECT_UNKNOWN_FLAGS -(EXEC_OBJECT_WRITE<<1) +#define EXEC_OBJECT_SUPPORTS_48B_ADDRESS (1<<3) +#define __EXEC_OBJECT_UNKNOWN_FLAGS -(EXEC_OBJECT_SUPPORTS_48B_ADDRESS<<1)__u64 flags;__u64 rsvd1;
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 17/17 drm/i915/gen8: Flip the 48b switch
In reply to Michel ThierryUse 48b addresses if hw supports it (i915.enable_ppgtt=3).
Note, aliasing PPGTT remains 32b only.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 4 ++-- drivers/gpu/drm/i915/i915_params.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -2
@@ -110,7 +110,7 @@ static int sanitize_enable_ppgtt(struct drm_device *dev, int enable_ppgtt) has_full_ppgtt =INTEL_INFO (dev)->gen >=7 ; has_full_64bit_ppgtt =IS_ENABLED (CONFIG_X86_64 ) && (IS_BROADWELL (dev) || -INTEL_INFO (dev)->gen >=9 ) && false;/* FIXME: 64b */ +INTEL_INFO (dev)->gen >=9 );if (intel_vgpu_active (dev)) has_full_ppgtt = false;/* emulation is too hard */
@@ -148,7 +148,7 @@ static int sanitize_enable_ppgtt(struct drm_device *dev, int enable_ppgtt) }if (INTEL_INFO (dev)->gen >=8 && i915.enable_execlists) -return 2 ; +return has_full_64bit_ppgtt ?3 :2 ;else return has_aliasing_ppgtt ?1 :0 ; }
drivers/gpu/drm/i915/i915_params.c +1 -1
@@ -110,7 +110,7 @@ MODULE_PARM_DESC(enable_hangcheck,module_param_named_unsafe (enable_ppgtt, i915.enable_ppgtt, int,0400 );MODULE_PARM_DESC (enable_ppgtt,"Override PPGTT usage. " -"(-1=auto [default], 0=disabled, 1=aliasing, 2=full)" ); +"(-1=auto [default], 0=disabled, 1=aliasing, 2=full, 3=full_64b)" );module_param_named (enable_execlists, i915.enable_execlists, int,0400 ); MODULE_PARM_DESC(enable_execlists,
Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 34 lines · Michel Thierry
These are the rebased patches, after Mika's final ppgtt clean-up series landed (it relies in the macros added). New functions also follow these changes.
In order expand the GPU address space, a 4th level translation is added, the Page Map Level 4 (PML4). This PML4 has 256 PML4 Entries (PML4E), PML4[0-255], each pointing to a PDP. All the existing "dynamic alloc ppgtt" functions are used, only adding the 4th level changes. I also updated some remaining variables that were 32b only.
There are 2 hardware workarounds needed to allow correct operation with 48b addresses (Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset). This new patchset version includes the comments and suggestions from Chris Wilson. A flag (
EXEC_OBJECT_SUPPORTS_48B_ADDRESS) will indicate if a given object can be allocated outside the first 4 PDPs; if not, the end range is forced to 4GB. Also, more objects now use theDRM_MM_CREATE_TOPflag. To maintain compatibility, in libdrm I added a newdrm_intel_bo_emit_reloc_48bitfunction that will flag these objects, while the existingdrm_intel_bo_emit_relocclears it.Finally, this feature is only available in BDW and Gen9, requires LRC submission mode (execlists) and it can be detected by i915.enable_ppgtt=3.
Also note that this expanded address space is only available for full PPGTT, aliasing PPGTT and Global GTT remain 32-bit.
Michel Thierry (17):
drm/i915: Remove unnecessarygen8_clamp_pd
drm/i915/gen8: Make pdp allocation more dynamic drm/i915/gen8: Abstract PDP usage
drm/i915/gen8: Add dynamic page trace events drm/i915/gen8: implement alloc/free for 4lvl drm/i915/gen8: Add 4 level switching infrastructure and lrc support drm/i915/gen8: Generalize PTE writing for GEN8 PPGTT drm/i915/gen8: Passsg_iterthrough pte inserts drm/i915/gen8: Add 4 level support ininsert_entriesandclear_range
Quoted 8 lines
drm/i915/gen8: Initialize PDPs
drm/i915: Expand error state's address width to 64b drm/i915/gen8: Add ppgtt info anddebug_dump
drm/i915: object size needs to be u64
drm/i915:batch_objvm offset must be u64 drm/i915/userptr: Killuser_sizelimit check
drm/i915: Wa32bitGeneralStateOffset & Wa32bitInstructionBaseOffset drm/i915/gen8: Flip the 48b switch
Please start a new thread when resending the entire patch series. Only in-reply-to parts of a series. It's harder to piece the series together this way and hence doesn't really improve things compared to just in-reply-to all the patches individually. The point of a full resend is to make restart/consolidate the review discussions.
But don't resend this one here now since that will make a discussion split guaranteed.
-Daniel
Quoted 23 lines
drivers/gpu/drm/i915/i915_debugfs.c | 18 +- drivers/gpu/drm/i915/i915_drv.h | 17 +- drivers/gpu/drm/i915/i915_gem.c | 22 +- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 10 + drivers/gpu/drm/i915/i915_gem_gtt.c | 649 ++++++++++++++++++++++++----- drivers/gpu/drm/i915/i915_gem_gtt.h | 66 ++- drivers/gpu/drm/i915/i915_gem_userptr.c | 4 - drivers/gpu/drm/i915/i915_gpu_error.c | 17 +- drivers/gpu/drm/i915/i915_params.c | 2 +- drivers/gpu/drm/i915/i915_reg.h | 1 + drivers/gpu/drm/i915/i915_trace.h | 16 + drivers/gpu/drm/i915/intel_lrc.c | 65 ++- include/uapi/drm/i915_drm.h | 3 +- 13 files changed, 725 insertions(+), 165 deletions(-)Signature · 6 lines
2.4.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx