On 2017-11-28 15:52:10, Scott D Phillips wrote:
> Memtrace aubs are similar to classic aubs, with the major
> difference being how command submission is serialized (as register
> writes instead of a high-level submit message). Some internal
> tools generate or consume only memtrace aubs.
> ---
>  src/intel/tools/aubinator.c | 117 
> +++++++++++++++++++++++++++++++-------------
>  1 file changed, 82 insertions(+), 35 deletions(-)
> 
> diff --git a/src/intel/tools/aubinator.c b/src/intel/tools/aubinator.c
> index 671296c0c2..60869b33c8 100644
> --- a/src/intel/tools/aubinator.c
> +++ b/src/intel/tools/aubinator.c
> @@ -883,6 +883,80 @@ handle_trace_header(uint32_t *p)
>     aubinator_init(aub_pci_id, app_name);
>  }
>  
> +static void
> +handle_memtrace_version(uint32_t *p)
> +{
> +   int header_length = p[0] & 0xffff;
> +   char app_name[64];
> +   int app_name_len = MIN2(4 * (header_length + 1 - 5), ARRAY_SIZE(app_name) 
> - 1);
> +   int pci_id_len = 0;
> +   int aub_pci_id = 0;
> +
> +   strncpy(app_name, (char *)&p[5], app_name_len);
> +   app_name[app_name_len] = 0;
> +   sscanf(app_name, "PCI-ID=%i %n", &aub_pci_id, &pci_id_len);
> +   if (pci_id == 0)
> +      pci_id = aub_pci_id;
> +   aubinator_init(aub_pci_id, app_name + pci_id_len);
> +}
> +
> +static void
> +handle_memtrace_reg_write(uint32_t *p)
> +{
> +   uint32_t offset = p[1];
> +   uint32_t value = p[5];
> +   int engine;
> +   static int render_elsp_writes = 0;
> +   static int blitter_elsp_writes = 0;
> +
> +   if (offset == 0x2230) {
> +      render_elsp_writes++;
> +      engine = GEN_ENGINE_RENDER;
> +   } else if (offset == 0x22230) {
> +      blitter_elsp_writes++;
> +      engine = GEN_ENGINE_BLITTER;
> +   } else {
> +      return;
> +   }
> +
> +   if (render_elsp_writes > 3)
> +      render_elsp_writes = 0;
> +   else if (blitter_elsp_writes > 3)
> +      blitter_elsp_writes = 0;
> +   else
> +      return;
> +
> +   uint8_t *pphwsp = (uint8_t*)gtt + (value & 0xfffff000);
> +   const uint32_t pphwsp_size = 4096;
> +   uint32_t *context = (uint32_t*)(pphwsp + pphwsp_size);
> +   uint32_t ring_buffer_head = context[5];
> +   uint32_t ring_buffer_tail = context[7];
> +   uint32_t ring_buffer_start = context[9];
> +   uint32_t *commands = (uint32_t*)((uint8_t*)gtt + ring_buffer_start + 
> ring_buffer_head);
> +   parse_commands(spec, commands, ring_buffer_tail - ring_buffer_head, 
> engine);
> +}
> +
> +static void
> +handle_memtrace_mem_write(uint32_t *p)
> +{
> +   uint64_t address = *(uint64_t*)&p[1];
> +   uint32_t address_space = p[3] >> 28;
> +   uint32_t size = p[4];
> +   uint32_t *data = p + 5;
> +
> +   if (address_space != 1)
> +      return;
> +
> +   if (gtt_size < address + size) {
> +      fprintf(stderr, "overflow gtt space: %s\n", strerror(errno));
> +      exit(EXIT_FAILURE);
> +   }
> +
> +   memcpy((char *) gtt + address, data, size);
> +   if (gtt_end < address + size)
> +      gtt_end = address + size;
> +}
> +
>  struct aub_file {
>     FILE *stream;
>  
> @@ -952,35 +1026,14 @@ aub_file_stdin(void)
>  
>  /* Newer version AUB opcode */
>  #define OPCODE_NEW_AUB      0x2e
> -#define SUBOPCODE_VERSION   0x00
> +#define SUBOPCODE_REG_POLL  0x02
>  #define SUBOPCODE_REG_WRITE 0x03
>  #define SUBOPCODE_MEM_POLL  0x05
>  #define SUBOPCODE_MEM_WRITE 0x06
> +#define SUBOPCODE_VERSION   0x0e
>  
>  #define MAKE_GEN(major, minor) ( ((major) << 8) | (minor) )
>  
> -struct {
> -   const char *name;
> -   uint32_t gen;
> -} device_map[] = {
> -   { "bwr", MAKE_GEN(4, 0) },
> -   { "cln", MAKE_GEN(4, 0) },
> -   { "blc", MAKE_GEN(4, 0) },
> -   { "ctg", MAKE_GEN(4, 0) },
> -   { "el", MAKE_GEN(4, 0) },
> -   { "il", MAKE_GEN(4, 0) },
> -   { "sbr", MAKE_GEN(6, 0) },
> -   { "ivb", MAKE_GEN(7, 0) },
> -   { "lrb2", MAKE_GEN(0, 0) },
> -   { "hsw", MAKE_GEN(7, 5) },
> -   { "vlv", MAKE_GEN(7, 0) },
> -   { "bdw", MAKE_GEN(8, 0) },
> -   { "skl", MAKE_GEN(9, 0) },
> -   { "chv", MAKE_GEN(8, 0) },
> -   { "bxt", MAKE_GEN(9, 0) },
> -   { "cnl", MAKE_GEN(10, 0) },
> -};
> -
>  enum {
>     AUB_ITEM_DECODE_OK,
>     AUB_ITEM_DECODE_FAILED,
> @@ -990,7 +1043,7 @@ enum {
>  static int
>  aub_file_decode_batch(struct aub_file *file)
>  {
> -   uint32_t *p, h, device, data_type, *new_cursor;
> +   uint32_t *p, h, *new_cursor;
>     int header_length, bias;
>  
>     if (file->end - file->cursor < 1)
> @@ -1034,25 +1087,19 @@ aub_file_decode_batch(struct aub_file *file)
>     case MAKE_HEADER(TYPE_AUB, OPCODE_AUB, SUBOPCODE_BMP):
>        break;
>     case MAKE_HEADER(TYPE_AUB, OPCODE_NEW_AUB, SUBOPCODE_VERSION):
> -      fprintf(outfile, "version block: dw1 %08x\n", p[1]);
> -      device = (p[1] >> 8) & 0xff;
> -      fprintf(outfile, "  device %s\n", device_map[device].name);

I guess we lose this device print by moving to the new 'VERSION'
header? It doesn't seem like too big of a loss.

Series Reviewed-by: Jordan Justen <jordan.l.jus...@intel.com>

> +      handle_memtrace_version(p);
>        break;
>     case MAKE_HEADER(TYPE_AUB, OPCODE_NEW_AUB, SUBOPCODE_REG_WRITE):
> -      fprintf(outfile, "register write block: (dwords %d)\n", h & 0xffff);
> -      fprintf(outfile, "  reg 0x%x, data 0x%x\n", p[1], p[5]);
> +      handle_memtrace_reg_write(p);
>        break;
>     case MAKE_HEADER(TYPE_AUB, OPCODE_NEW_AUB, SUBOPCODE_MEM_WRITE):
> -      fprintf(outfile, "memory write block (dwords %d):\n", h & 0xffff);
> -      fprintf(outfile, "  address 0x%"PRIx64"\n", *(uint64_t *) &p[1]);
> -      data_type = (p[3] >> 20) & 0xff;
> -      if (data_type != 0)
> -         fprintf(outfile, "  data type 0x%x\n", data_type);
> -      fprintf(outfile, "  address space 0x%x\n", (p[3] >> 28) & 0xf);
> +      handle_memtrace_mem_write(p);
>        break;
>     case MAKE_HEADER(TYPE_AUB, OPCODE_NEW_AUB, SUBOPCODE_MEM_POLL):
>        fprintf(outfile, "memory poll block (dwords %d):\n", h & 0xffff);
>        break;
> +   case MAKE_HEADER(TYPE_AUB, OPCODE_NEW_AUB, SUBOPCODE_REG_POLL):
> +      break;
>     default:
>        fprintf(outfile, "unknown block type=0x%x, opcode=0x%x, "
>               "subopcode=0x%x (%08x)\n", TYPE(h), OPCODE(h), SUBOPCODE(h), h);
> -- 
> 2.14.3
> 
> _______________________________________________
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Reply via email to