radeon: preallocate memory for command stream parsing
Subsystems (2)
PATCH 1/1 radeon: preallocate memory for command stream parsing
Command stream parsing is the most common operation and can happen hundred of times per second, we don't want to allocate/free memory each time this ioctl is call. This rework the ioctl to avoid doing so by allocating temporary memory along the ib pool.
--- drivers/gpu/drm/radeon/r100.c | 27 +++--- drivers/gpu/drm/radeon/r300.c | 6 +- drivers/gpu/drm/radeon/radeon.h | 110 ++++++++++---------- drivers/gpu/drm/radeon/radeon_cs.c | 186 +++++++++++++++------------------- drivers/gpu/drm/radeon/radeon_ring.c | 23 ++++ 5 files changed, 177 insertions(+), 175 deletions(-)
drivers/gpu/drm/radeon/r100.c +13 -14
@@ -699,7 +699,7 @@ void r100_cs_dump_packet(struct radeon_cs_parser *p,unsigned idx; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx;for (i =0 ; i <= (pkt->count +1 ); i++, idx++) {DRM_INFO ("ib[%d]=0x%08X\n" , idx, ib[idx]);
@@ -718,14 +718,15 @@ int r100_cs_packet_parse(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx) { -struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; -uint32_t header = ib_chunk->kdata[idx]; +struct radeon_cs_chunk *ib_chunk = &p->ibc; +uint32_t header;if (idx >= ib_chunk->length_dw) {DRM_ERROR ("Can not parse packet at %d after CS end %d !\n" , idx, ib_chunk->length_dw);return -EINVAL ; } + header = ib_chunk->kdata[idx]; pkt->idx = idx; pkt->type =CP_PACKET_GET_TYPE (header); pkt->count =CP_PACKET_GET_COUNT (header);
@@ -767,18 +768,16 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, struct radeon_cs_reloc **cs_reloc) {struct radeon_cs_chunk *ib_chunk; -struct radeon_cs_chunk *relocs_chunk;struct radeon_cs_packet p3reloc;unsigned idx;int r; -if (p->chunk_relocs_idx ==-1 ) { +if (p->relocs ==NULL ) {DRM_ERROR ("No relocation chunk !\n" );return -EINVAL ; } *cs_reloc =NULL ; - ib_chunk = &p->chunks[p->chunk_ib_idx]; - relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + ib_chunk = &p->ibc; r =r100_cs_packet_parse (p, &p3reloc, p->idx);if (r) {return r;
@@ -791,14 +790,14 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p,return -EINVAL ; } idx = ib_chunk->kdata[p3reloc.idx +1 ]; -if (idx >= relocs_chunk->length_dw) { +if ((idx /4 ) >= p->nrelocs) {DRM_ERROR ("Relocs at %d after relocations chunk end %d !\n" , - idx, relocs_chunk->length_dw); + idx, p->nrelocs *4 );r100_cs_dump_packet (p, &p3reloc);return -EINVAL ; }/* FIXME: we assume reloc size is 4 dwords */ - *cs_reloc = p->relocs_ptr[(idx /4 )]; + *cs_reloc = p->relocs_ptr[idx /4 ];return 0 ; }
@@ -816,7 +815,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx +1 ; reg = pkt->reg; onereg = false;
@@ -897,7 +896,7 @@ int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p,struct radeon_cs_chunk *ib_chunk;unsigned idx; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx +1 ;if ((ib_chunk->kdata[idx+2 ] +1 ) >radeon_object_size (robj)) { DRM_ERROR("[drm] Buffer too small for PACKET3 INDX_BUFFER "
@@ -920,7 +919,7 @@ static int r100_packet3_check(struct radeon_cs_parser *p,int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx +1 ;switch (pkt->opcode) {case PACKET3_3D_LOAD_VBPNTR :
@@ -1027,7 +1026,7 @@ int r100_cs_parse(struct radeon_cs_parser *p)if (r) {return r; } - }while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + }while (p->idx < p->ibc.length_dw);return 0 ; }
drivers/gpu/drm/radeon/r300.c +3 -3
@@ -1024,7 +1024,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p,int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; track = (struct r300_cs_track *)p->track;switch (reg) {case RADEON_DST_PITCH_OFFSET :
@@ -1361,7 +1361,7 @@ static int r300_packet3_check(struct radeon_cs_parser *p,int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx +1 ; track = (struct r300_cs_track *)p->track;switch (pkt->opcode) {
@@ -1520,7 +1520,7 @@ int r300_cs_parse(struct radeon_cs_parser *p)if (r) {return r; } - }while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + }while (p->idx < p->ibc.length_dw);return 0 ; }
drivers/gpu/drm/radeon/radeon.h +56 -54
134 lines not on this page · raw message
drivers/gpu/drm/radeon/radeon_cs.c +82 -104
264 lines not on this page · raw message
drivers/gpu/drm/radeon/radeon_ring.c +23 -0
@@ -177,6 +177,20 @@ int radeon_ib_pool_init(struct radeon_device *rdev)/* Allocate 1M object buffer */ INIT_LIST_HEAD (&rdev->ib_pool.scheduled_ibs); + rdev->ib_pool.ib =drm_calloc_large (RADEON_IB_POOL_SIZE ,64 *1024 ); +if (rdev->ib_pool.ib ==NULL ) { +return -ENOMEM ; + } + rdev->ib_pool.relocs =drm_calloc_large (RADEON_IB_POOL_SIZE *1024 , +sizeof (struct radeon_cs_reloc )); +if (rdev->ib_pool.relocs ==NULL ) { +return -ENOMEM ; + } + rdev->ib_pool.relocs_ptr =drm_calloc_large (RADEON_IB_POOL_SIZE *1024 , +sizeof (void *)); +if (rdev->ib_pool.relocs ==NULL ) { +return -ENOMEM ; + } r =radeon_object_create (rdev,NULL ,RADEON_IB_POOL_SIZE *64 *1024 , true,RADEON_GEM_DOMAIN_GTT , false, &rdev->ib_pool.robj);
@@ -202,6 +216,9 @@ int radeon_ib_pool_init(struct radeon_device *rdev) rdev->ib_pool.ibs[i].ptr = ptr + offset; rdev->ib_pool.ibs[i].idx = i; rdev->ib_pool.ibs[i].length_dw =0 ; + rdev->ib_pool.ibs[i].ib = &rdev->ib_pool.ib[i *16 *1024 ]; + rdev->ib_pool.ibs[i].relocs = &rdev->ib_pool.relocs[i *1024 ]; + rdev->ib_pool.ibs[i].relocs_ptr = &rdev->ib_pool.relocs_ptr[i *1024 ];INIT_LIST_HEAD (&rdev->ib_pool.ibs[i].list); }bitmap_zero (rdev->ib_pool.alloc_bm,RADEON_IB_POOL_SIZE );
@@ -225,6 +242,12 @@ void radeon_ib_pool_fini(struct radeon_device *rdev)radeon_object_unref (&rdev->ib_pool.robj); rdev->ib_pool.robj =NULL ; } +if (rdev->ib_pool.ib) +drm_free_large (rdev->ib_pool.ib); +if (rdev->ib_pool.relocs) +drm_free_large (rdev->ib_pool.relocs); +if (rdev->ib_pool.relocs_ptr) +drm_free_large (rdev->ib_pool.relocs_ptr);mutex_unlock (&rdev->ib_pool.mutex); }
Signature · 1 line
1.6.2.2
Hi Jerome,
Quoted 7 lines · Jerome Glisse
Command stream parsing is the most common operation and can happen hundred of times per second, we don't want to allocate/free memory each time this ioctl is call. This rework the ioctl to avoid doing so by allocating temporary memory along the ib pool.
Signed-off-by Jerome Glisse <[email protected]>
So how much does this help (i.e. where are the numbers)? I am bit surprised "hundred of times per second" is an issue for our slab allocators. Hmm?
Quoted 616 lines
---
drivers/gpu/drm/radeon/r100.c | 27 +++---drivers/gpu/drm/radeon/r300.c| 6 +-drivers/gpu/drm/radeon/radeon.h| 110 ++++++++++----------drivers/gpu/drm/radeon/radeon_cs.c| 186 +++++++++++++++-------------------drivers/gpu/drm/radeon/radeon_ring.c| 23 ++++5 files changed, 177 insertions(+), 175 deletions(-)drivers/gpu/drm/radeon/r100.c +11
-12@@ -699,7 +699,7 @@ void r100_cs_dump_packet(struct radeon_cs_parser *p, unsigned idx; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx; for (i = 0; i <= (pkt->count + 1); i++, idx++) { DRM_INFO("ib[%d]=0x%08X\n", idx, ib[idx]);@@ -718,14 +718,15 @@ int r100_cs_packet_parse(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx) { - struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; - uint32_t header = ib_chunk->kdata[idx]; + struct radeon_cs_chunk *ib_chunk = &p->ibc; + uint32_t header; if (idx >= ib_chunk->length_dw) { DRM_ERROR("Can not parse packet at %d after CS end %d !\n", idx, ib_chunk->length_dw); return -EINVAL; } + header = ib_chunk->kdata[idx]; pkt->idx = idx; pkt->type = CP_PACKET_GET_TYPE(header); pkt->count = CP_PACKET_GET_COUNT(header);@@ -767,18 +768,16 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, struct radeon_cs_reloc **cs_reloc) { struct radeon_cs_chunk *ib_chunk; - struct radeon_cs_chunk *relocs_chunk; struct radeon_cs_packet p3reloc; unsigned idx; int r; - if (p->chunk_relocs_idx == -1) { + if (p->relocs == NULL) { DRM_ERROR("No relocation chunk !\n"); return -EINVAL; } *cs_reloc = NULL; - ib_chunk = &p->chunks[p->chunk_ib_idx]; - relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + ib_chunk = &p->ibc; r = r100_cs_packet_parse(p, &p3reloc, p->idx); if (r) { return r;@@ -791,14 +790,14 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, return -EINVAL; } idx = ib_chunk->kdata[p3reloc.idx + 1]; - if (idx >= relocs_chunk->length_dw) { + if ((idx / 4) >= p->nrelocs) { DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", - idx, relocs_chunk->length_dw); + idx, p->nrelocs * 4); r100_cs_dump_packet(p, &p3reloc); return -EINVAL; } /* FIXME: we assume reloc size is 4 dwords */ - *cs_reloc = p->relocs_ptr[(idx / 4)]; + *cs_reloc = p->relocs_ptr[idx / 4]; return 0; }@@ -816,7 +815,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx + 1; reg = pkt->reg; onereg = false;@@ -897,7 +896,7 @@ int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, struct radeon_cs_chunk *ib_chunk; unsigned idx; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx + 1; if ((ib_chunk->kdata[idx+2] + 1) > radeon_object_size(robj)) {DRM_ERROR("[drm] Buffer too small for PACKET3
INDX_BUFFER"drivers/gpu/drm/radeon/r100.c +1
-1@@ -920,7 +919,7 @@ static int r100_packet3_check(struct radeon_cs_parser *p, int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx + 1; switch (pkt->opcode) {case
PACKET3_3D_LOAD_VBPNTR:drivers/gpu/drm/radeon/r100.c +1
-1@@ -1027,7 +1026,7 @@ int r100_cs_parse(struct radeon_cs_parser *p) if (r) { return r; } - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->ibc.length_dw); return 0; }drivers/gpu/drm/radeon/r300.c +1
-1@@ -1024,7 +1024,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; track = (struct r300_cs_track*)p->track; switch(reg) {case
RADEON_DST_PITCH_OFFSET:drivers/gpu/drm/radeon/r300.c +2
-2@@ -1361,7 +1361,7 @@ static int r300_packet3_check(struct radeon_cs_parser *p, int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib_chunk = &p->ibc; idx = pkt->idx + 1; track = (struct r300_cs_track*)p->track; switch(pkt->opcode) {@@ -1520,7 +1520,7 @@ int r300_cs_parse(struct radeon_cs_parser *p) if (r) { return r; } - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->ibc.length_dw); return 0; }drivers/gpu/drm/radeon/radeon.h +56
-54@@ -352,6 +352,56 @@ void radeon_irq_kms_fini(struct radeon_device *rdev); /* +* CS. +*/ +struct radeon_ib ; + +struct radeon_cs_reloc { + struct drm_gem_object *gobj; + struct radeon_object *robj; + struct radeon_object_list lobj; + uint32_t handle; + uint32_t flags; +}; + +struct radeon_cs_chunk { + uint32_t chunk_id; + uint32_t length_dw; + uint32_t *kdata; +}; + +struct radeon_cs_parser { + struct radeon_device *rdev; + struct drm_file *filp; + struct radeon_cs_chunk ibc; + /* IB */ + unsigned idx; + /* relocations */ + unsigned nrelocs; + struct radeon_cs_reloc *relocs; + struct radeon_cs_reloc **relocs_ptr; + struct list_head validated; + struct radeon_ib *ib; + void *track; +}; + +struct radeon_cs_packet { + unsigned idx; + unsigned type; + unsigned reg; + unsigned opcode; + int count; + unsigned one_reg_wr; +}; + +typedef int (*radeon_packet0_check_t )(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg); +typedef int (*radeon_packet3_check_t )(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt); + + +/* * CP & ring. */ struct radeon_ib {@@ -360,12 +410,18 @@ struct radeon_ib { uint64_t gpu_addr; struct radeon_fence *fence; volatile uint32_t *ptr; + uint32_t *ib; + struct radeon_cs_reloc *relocs; + struct radeon_cs_reloc **relocs_ptr; uint32_t length_dw; }; struct radeon_ib_pool { struct mutex mutex; struct radeon_object *robj; + uint32_t *ib; + struct radeon_cs_reloc *relocs; + struct radeon_cs_reloc **relocs_ptr; struct list_head scheduled_ibs; struct radeon_ib ibs[RADEON_IB_POOL_SIZE]; bool ready;@@ -405,60 +461,6 @@ void radeon_ring_fini(struct radeon_device *rdev); /* -* CS. -*/ -struct radeon_cs_reloc { - struct drm_gem_object *gobj; - struct radeon_object *robj; - struct radeon_object_list lobj; - uint32_t handle; - uint32_t flags; -}; - -struct radeon_cs_chunk { - uint32_t chunk_id; - uint32_t length_dw; - uint32_t *kdata; -}; - -struct radeon_cs_parser { - struct radeon_device *rdev; - struct drm_file *filp; - /* chunks */ - unsigned nchunks; - struct radeon_cs_chunk *chunks; - uint64_t *chunks_array; - /* IB */ - unsigned idx; - /* relocations */ - unsigned nrelocs; - struct radeon_cs_reloc *relocs; - struct radeon_cs_reloc **relocs_ptr; - struct list_head validated; - /* indices of various chunks */ - int chunk_ib_idx; - int chunk_relocs_idx; - struct radeon_ib *ib; - void *track; -}; - -struct radeon_cs_packet { - unsigned idx; - unsigned type; - unsigned reg; - unsigned opcode; - int count; - unsigned one_reg_wr; -}; - -typedef int (*radeon_packet0_check_t )(struct radeon_cs_parser *p, - struct radeon_cs_packet *pkt, - unsigned idx, unsigned reg); -typedef int (*radeon_packet3_check_t )(struct radeon_cs_parser *p, - struct radeon_cs_packet *pkt); - - -/* * AGP */ int radeon_agp_init(struct radeon_device *rdev);drivers/gpu/drm/radeon/radeon_cs.c +82
-104@@ -29,61 +29,52 @@ #include "radeon_reg.h" #include "radeon.h" -void r100_cs_dump_packet (struct radeon_cs_parser *p, - struct radeon_cs_packet *pkt); - -int radeon_cs_parser_relocs (struct radeon_cs_parser *p) +static int radeon_cs_parser_reloc (struct radeon_cs_parser *p, + struct drm_radeon_cs_chunk *chunk) { - struct drm_device *ddev = p->rdev->ddev; - struct radeon_cs_chunk *chunk; - unsigned i, j; + struct drm_radeon_cs_reloc __user *reloc_ptr; + struct drm_radeon_cs_reloc reloc; + unsigned i, j, c; bool duplicate; - if (p->chunk_relocs_idx == -1) { - return 0; - } - chunk = &p->chunks[p->chunk_relocs_idx]; - /* FIXME: we assume that each relocs use 4 dwords */ p->nrelocs = chunk->length_dw / 4; - p->relocs_ptr = kcalloc(p->nrelocs, sizeof(void *), GFP_KERNEL); - if (p->relocs_ptr == NULL) { - return -ENOMEM; - } - p->relocs = kcalloc(p->nrelocs, sizeof(struct radeon_cs_reloc), GFP_KERNEL); - if (p->relocs == NULL) { - return -ENOMEM; + if (p->nrelocs > 1024) { + DRM_ERROR("Too many relocations %d\n", p->nrelocs); + return -EINVAL; } - for (i = 0; i < p->nrelocs; i++) { - struct drm_radeon_cs_reloc *r; - + for (i = 0, c = 0; i < p->nrelocs; i++) { + reloc_ptr = (void __user*)(unsigned long)(chunk->chunk_data + i * 16); + if (DRM_COPY_FROM_USER(&reloc, reloc_ptr, 16)) { + return -EFAULT; + } duplicate = false; - r = (struct drm_radeon_cs_reloc *)&chunk->kdata[i*4]; - for (j = 0; j < p->nrelocs; j++) { - if (r->handle == p->relocs[j].handle) { + for (j = 0; j < c; j++) { + if (reloc.handle == p->relocs[j].handle) { p->relocs_ptr[i] = &p->relocs[j]; duplicate = true; break; } } if (!duplicate) { - p->relocs[i].gobj = drm_gem_object_lookup(ddev, + p->relocs[i].gobj = drm_gem_object_lookup(p->rdev->ddev, p->filp, - r->handle); + reloc.handle); if (p->relocs[i].gobj == NULL) { DRM_ERROR("gem object lookup failed 0x%x\n", - r->handle); + reloc.handle); return -EINVAL; } p->relocs_ptr[i] = &p->relocs[i]; p->relocs[i].robj = p->relocs[i].gobj->driver_private; p->relocs[i].lobj.robj = p->relocs[i].robj; - p->relocs[i].lobj.rdomain = r->read_domains; - p->relocs[i].lobj.wdomain = r->write_domain; - p->relocs[i].handle = r->handle; - p->relocs[i].flags = r->flags; + p->relocs[i].lobj.rdomain = reloc.read_domains; + p->relocs[i].lobj.wdomain = reloc.write_domain; + p->relocs[i].handle = reloc.handle; + p->relocs[i].flags = reloc.flags; INIT_LIST_HEAD(&p->relocs[i].lobj.list); radeon_object_list_add_object(&p->relocs[i].lobj, &p->validated); + c++; } } return radeon_object_list_validate(&p->validated, p->ib->fence);@@ -92,65 +83,53 @@ int radeon_cs_parser_relocs(struct radeon_cs_parser *p) int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) { struct drm_radeon_cs *cs = data; - uint64_t *chunk_array_ptr; - unsigned size, i; + unsigned i; + int r; + bool found_ib = false; - if (!cs->num_chunks) { - return 0; - } - /* get chunks */ - INIT_LIST_HEAD(&p->validated); - p->idx = 0; - p->chunk_ib_idx = -1; - p->chunk_relocs_idx = -1; - p->chunks_array = kcalloc(cs->num_chunks, sizeof(uint64_t), GFP_KERNEL); - if (p->chunks_array == NULL) { - return -ENOMEM; - } - chunk_array_ptr = (uint64_t *)(unsigned long)(cs->chunks); - if (DRM_COPY_FROM_USER(p->chunks_array, chunk_array_ptr, - sizeof(uint64_t)*cs->num_chunks)) { - return -EFAULT; - } - p->nchunks = cs->num_chunks; - p->chunks = kcalloc(p->nchunks, sizeof(struct radeon_cs_chunk), GFP_KERNEL); - if (p->chunks == NULL) { - return -ENOMEM; - } - for (i = 0; i < p->nchunks; i++) { - struct drm_radeon_cs_chunk __user **chunk_ptr = NULL; - struct drm_radeon_cs_chunk user_chunk; + for (i = 0; i < cs->num_chunks; i++) { + struct drm_radeon_cs_chunk __user *chunk_ptr = NULL; + struct drm_radeon_cs_chunk chunk; uint32_t __user *cdata; + unsigned tmp; + u64 ptr; - chunk_ptr = (void __user*)(unsigned long)p->chunks_array[i]; - if (DRM_COPY_FROM_USER(&user_chunk, chunk_ptr, - sizeof(struct drm_radeon_cs_chunk))) { + chunk_ptr = (void __user*)(unsigned long)(cs->chunks + i * 8); + if (DRM_COPY_FROM_USER(&ptr, chunk_ptr, sizeof(u64))) { return -EFAULT; } - p->chunks[i].chunk_id = user_chunk.chunk_id; - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_RELOCS) { - p->chunk_relocs_idx = i; - } - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_IB) { - p->chunk_ib_idx = i; - } - p->chunks[i].length_dw = user_chunk.length_dw; - cdata = (uint32_t *)(unsigned long)user_chunk.chunk_data; - - p->chunks[i].kdata = NULL; - size = p->chunks[i].length_dw * sizeof(uint32_t); - p->chunks[i].kdata = kzalloc(size, GFP_KERNEL); - if (p->chunks[i].kdata == NULL) { - return -ENOMEM; - } - if (DRM_COPY_FROM_USER(p->chunks[i].kdata, cdata, size)) { + chunk_ptr = (void __user*)(unsigned long)ptr; + if (DRM_COPY_FROM_USER(&chunk, chunk_ptr, sizeof(struct drm_radeon_cs_chunk))) { return -EFAULT; } - } - if (p->chunks[p->chunk_ib_idx].length_dw > (16 * 1024)) { - DRM_ERROR("cs IB too big: %d\n", - p->chunks[p->chunk_ib_idx].length_dw); - return -EINVAL; + switch (chunk.chunk_id) { + case RADEON_CHUNK_ID_RELOCS: + r = radeon_cs_parser_reloc(p, &chunk); + if (r) { + return r; + } + break; + case RADEON_CHUNK_ID_IB: + if (found_ib) { + DRM_ERROR("Multiple IB chunks not supported\n"); + return -EINVAL; + } + found_ib = true; + if (chunk.length_dw > 16 * 1024) { + DRM_ERROR("cs IB too big: %d\n", chunk.length_dw); + return -EINVAL; + } + p->ibc.chunk_id = chunk.chunk_id; + p->ibc.length_dw = chunk.length_dw; + tmp = p->ibc.length_dw * sizeof(u32); + cdata = (uint32_t *)(unsigned long)chunk.chunk_data; + if (DRM_COPY_FROM_USER(p->ibc.kdata, cdata, tmp)) { + return -EFAULT; + } + break; + default: + break; + } } return 0; }@@ -179,13 +158,6 @@ static void radeon_cs_parser_fini(struct radeon_cs_parser *parser, int error) mutex_unlock(&parser->rdev->ddev->struct_mutex); } } - kfree(parser->relocs); - kfree(parser->relocs_ptr); - for (i = 0; i < parser->nchunks; i++) { - kfree(parser->chunks[i].kdata); - } - kfree(parser->chunks); - kfree(parser->chunks_array); radeon_ib_free(parser->rdev, &parser->ib); }@@ -193,7 +165,6 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) { struct radeon_device *rdev = dev->dev_private; struct radeon_cs_parser parser; - struct radeon_cs_chunk *ib_chunk; int r; mutex_lock(&rdev->cs_mutex);@@ -202,36 +173,43 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) return -EINVAL; } /* initialize parser */ - memset(&parser, 0, sizeof(struct radeon_cs_parser)); + INIT_LIST_HEAD(&parser.validated); + parser.idx = 0; + parser.ibc.kdata = NULL; + parser.ibc.length_dw = 0; + parser.relocs = NULL; + parser.relocs_ptr = NULL; + parser.ib = NULL; parser.filp = filp; parser.rdev = rdev; - r = radeon_cs_parser_init(&parser, data); + r = radeon_ib_get(rdev, &parser.ib); if (r) { - DRM_ERROR("Failed to initialize parser !\n"); + DRM_ERROR("Failed to get ib !\n"); radeon_cs_parser_fini(&parser, r); mutex_unlock(&rdev->cs_mutex); return r; } - r = radeon_ib_get(rdev, &parser.ib); + parser.ibc.kdata = parser.ib->ib; + parser.relocs = parser.ib->relocs; + parser.relocs_ptr = parser.ib->relocs_ptr; + r = radeon_cs_parser_init(&parser, data); if (r) { - DRM_ERROR("Failed to get ib !\n"); + DRM_ERROR("Failed to initialize parser !\n"); radeon_cs_parser_fini(&parser, r); mutex_unlock(&rdev->cs_mutex); return r; } - r = radeon_cs_parser_relocs(&parser); - if (r) { - DRM_ERROR("Failed to parse relocation !\n"); + if (parser.ibc.kdata == NULL) { + DRM_INFO("No IB chunk in cs, nothing to do\n"); radeon_cs_parser_fini(&parser, r); mutex_unlock(&rdev->cs_mutex); - return r; + return 0; } /* Copy the packet into the IB, the parser will read from the * input memory (cached) and write to the IB (which can be * uncached). */ - ib_chunk = &parser.chunks[parser.chunk_ib_idx]; - parser.ib->length_dw = ib_chunk->length_dw; - memcpy((void *)parser.ib->ptr, ib_chunk->kdata, ib_chunk->length_dw*4); + parser.ib->length_dw = parser.ibc.length_dw; + memcpy((void *)parser.ib->ptr, parser.ibc.kdata, parser.ibc.length_dw * 4); r = radeon_cs_parse(&parser); if (r) { DRM_ERROR("Invalid command stream !\n");drivers/gpu/drm/radeon/radeon_ring.c +23
-1@@ -177,6 +177,20 @@ int radeon_ib_pool_init(struct radeon_device *rdev) /* Allocate 1M object buffer */ INIT_LIST_HEAD(&rdev->ib_pool.scheduled_ibs); + rdev->ib_pool.ib = drm_calloc_large(RADEON_IB_POOL_SIZE, 64 * 1024); + if (rdev->ib_pool.ib == NULL) { + return -ENOMEM; + } + rdev->ib_pool.relocs = drm_calloc_large(RADEON_IB_POOL_SIZE * 1024, + sizeof(struct radeon_cs_reloc)); + if (rdev->ib_pool.relocs == NULL) { + return -ENOMEM; + } + rdev->ib_pool.relocs_ptr = drm_calloc_large(RADEON_IB_POOL_SIZE * 1024, + sizeof(void *)); + if (rdev->ib_pool.relocs == NULL) { + return -ENOMEM; + } r = radeon_object_create(rdev, NULL, RADEON_IB_POOL_SIZE*64*1024, true, RADEON_GEM_DOMAIN_GTT, false, &rdev->ib_pool.robj);@@ -202,6 +216,9 @@ int radeon_ib_pool_init(struct radeon_device *rdev) rdev->ib_pool.ibs[i].ptr = ptr + offset; rdev->ib_pool.ibs[i].idx = i; rdev->ib_pool.ibs[i].length_dw = 0; + rdev->ib_pool.ibs[i].ib = &rdev->ib_pool.ib[i * 16 * 1024]; + rdev->ib_pool.ibs[i].relocs = &rdev->ib_pool.relocs[i * 1024]; + rdev->ib_pool.ibs[i].relocs_ptr = &rdev->ib_pool.relocs_ptr[i * 1024]; INIT_LIST_HEAD(&rdev->ib_pool.ibs[i].list); } bitmap_zero(rdev->ib_pool.alloc_bm, RADEON_IB_POOL_SIZE);@@ -225,6 +242,12 @@ void radeon_ib_pool_fini(struct radeon_device *rdev) radeon_object_unref(&rdev->ib_pool.robj); rdev->ib_pool.robj = NULL; } + if (rdev->ib_pool.ib) + drm_free_large(rdev->ib_pool.ib); + if (rdev->ib_pool.relocs) + drm_free_large(rdev->ib_pool.relocs); + if (rdev->ib_pool.relocs_ptr) + drm_free_large(rdev->ib_pool.relocs_ptr); mutex_unlock(&rdev->ib_pool.mutex); } --1.6.2.2
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Quoted 9 lines · Pekka Enberg
Hi Jerome,
Quoted 6 lines · Jerome Glisse
Command stream parsing is the most common operation and can happen hundred of times per second, we don't want to allocate/free memory each time this ioctl is call. This rework the ioctl to avoid doing so by allocating temporary memory along the ib pool.
Quoted 6 lines
Quoted 1 line
Signed-off-by Jerome Glisse <[email protected]>So how much does this help (i.e. where are the numbers)? I am bit surprised "hundred of times per second" is an issue for our slab allocators. Hmm?
I didn't have real number but the vmap path was really slower, quake3 fps goes from ~20 to ~40 on average when going from vmap to preallocated. When using kmalloc i don't thing there was so much performance hit. But i think the biggest hit was that in previous code i asked for zeroed memory so i am pretty sure kernel spend a bit of time clearing page. I reworked the code to avoid needing cleared memory and so avoid memset, this is likely why we get a performance boost.
Cheers,
Jerome
Hi Jerome,
Quoted 12 lines · Pekka Enberg
Quoted 12 lines · Jerome Glisse
Quoted 7 lines
Command stream parsing is the most common operation and can happen hundred of times per second, we don't want to allocate/free memory each time this ioctl is call. This rework the ioctl to avoid doing so by allocating temporary memory along the ib pool.
Signed-off-by Jerome Glisse <[email protected]>So how much does this help (i.e. where are the numbers)? I am bit surprised "hundred of times per second" is an issue for our slab allocators. Hmm?
Quoted 8 lines · Jerome Glisse
I didn't have real number but the vmap path was really slower, quake3 fps goes from ~20 to ~40 on average when going from vmap to preallocated. When using kmalloc i don't thing there was so much performance hit. But i think the biggest hit was that in previous code i asked for zeroed memory so i am pretty sure kernel spend a bit of time clearing page. I reworked the code to avoid needing cleared memory and so avoid memset, this is likely why we get a performance boost.
OK. If kmalloc() (without memset) really was too slow for your case, I'd be interested in looking at it in more detail. I'm not completely convinced the memory pool is needed here but I'm not a DRM expert so I'm not NAK'ing this either...
Pekka
Pekka Enberg skrev:
Quoted 31 lines
Hi Jerome,
Quoted 14 lines · Pekka Enberg
Quoted 14 lines · Jerome Glisse
Quoted 9 lines
Command stream parsing is the most common operation and can happen hundred of times per second, we don't want to allocate/free memory each time this ioctl is call. This rework the ioctl to avoid doing so by allocating temporary memory along the ib pool.
Signed-off-by Jerome Glisse <[email protected]>So how much does this help (i.e. where are the numbers)? I am bit surprised "hundred of times per second" is an issue for our slab allocators. Hmm?
Quoted 9 lines · Jerome Glisse
I didn't have real number but the vmap path was really slower, quake3 fps goes from ~20 to ~40 on average when going from vmap to preallocated. When using kmalloc i don't thing there was so much performance hit. But i think the biggest hit was that in previous code i asked for zeroed memory so i am pretty sure kernel spend a bit of time clearing page. I reworked the code to avoid needing cleared memory and so avoid memset, this is likely why we get a performance boost.
Quoted 8 lines
OK. If
kmalloc()(without memset) really was too slow for your case, I'd be interested in looking at it in more detail. I'm not completely convinced the memory pool is needed here but I'm not a DRM expert so I'm not NAK'ing this either...Pekka
Hi!
From previous experience with other drivers kmalloc() is just fine performance-wise.
I've also never seen memsetting pages turn up on the profile. It would be interesting to see an oprofile timing of this to try and pinpoint what's happening.
However, in this case, I believe Jerome was forced to use vmalloc to guarantee that the allocation would succeed, and frequent vmallocs seem to be a performance killer.
One should also be careful about frame-rates. Tuning memory manager / command submission operation is usually a matter of how much CPU is consumed for a given framerate. If one compares framerates one must make sure that the CPU is at nearly 100% while benchmarking.
/Thomas
Quoted 7 lines
------------------------------------------------------------------------------
--
_______________________________________________
Dri-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dri-devel
Quoted 51 lines · Thomas Hellström
Pekka Enberg skrev:
Quoted 31 lines
Hi Jerome,
Quoted 14 lines · Pekka Enberg
Quoted 14 lines · Jerome Glisse
Quoted 9 lines
Command stream parsing is the most common operation and can happen hundred of times per second, we don't want to allocate/free memory each time this ioctl is call. This rework the ioctl to avoid doing so by allocating temporary memory along the ib pool.
Signed-off-by Jerome Glisse <[email protected]>So how much does this help (i.e. where are the numbers)? I am bit surprised "hundred of times per second" is an issue for our slab allocators. Hmm?
Quoted 9 lines · Jerome Glisse
I didn't have real number but the vmap path was really slower, quake3 fps goes from ~20 to ~40 on average when going from vmap to preallocated. When using kmalloc i don't thing there was so much performance hit. But i think the biggest hit was that in previous code i asked for zeroed memory so i am pretty sure kernel spend a bit of time clearing page. I reworked the code to avoid needing cleared memory and so avoid memset, this is likely why we get a performance boost.
Quoted 8 lines
OK. If
kmalloc()(without memset) really was too slow for your case, I'd be interested in looking at it in more detail. I'm not completely convinced the memory pool is needed here but I'm not a DRM expert so I'm not NAK'ing this either...Pekka
Hi!
From previous experience with other driverskmalloc()is just fine
performance-wise.
I've also never seen memsetting pages turn up on the profile. It would be interesting to see an oprofile timing of this to try and pinpoint what's happening.However, in this case, I believe Jerome was forced to use vmalloc to guarantee that the allocation would succeed, and frequent vmallocs seem to be a performance killer.
Quoted 7 lines
One should also be careful about frame-rates. Tuning memory manager / command submission operation is usually a matter of how much CPU is consumed for a given framerate. If one compares framerates one must make sure that the CPU is at nearly 100% while benchmarking.
/Thomas
To give a more correct rough estimate, at 60fps we will issue somethings like 10 to 50 cs ioctl per frame so it's several thousands of cs ioctl so several thousands of 64K allocation, and memory clearing, i believe such things would show up on profile. I am not running kernel without my patch as i am working on other stuff now, but i will lower the pool size so that we don't waste too much memory right now i think the preallocation use somethings around 8M of memory. I think i can get it down to 1M without impacting performance (even less if we are on pcie).
Cheers,
Jerome