adding support for non-cacheable memory addressing

This commit is contained in:
Blaise Tine
2021-06-06 13:35:55 -07:00
parent 84391c1a06
commit 3071fb7a29
23 changed files with 1605 additions and 205 deletions

View File

@@ -3,22 +3,26 @@ all:
$(MAKE) -C demo
$(MAKE) -C dogfood
$(MAKE) -C stress
$(MAKE) -C io_addr
run:
$(MAKE) -C basic run-vlsim
$(MAKE) -C demo run-vlsim
$(MAKE) -C dogfood run-vlsim
$(MAKE) -C stress run-vlsim
$(MAKE) -C io_addr run-vlsim
clean:
$(MAKE) -C basic clean
$(MAKE) -C demo clean
$(MAKE) -C dogfood clean
$(MAKE) -C stress clean
$(MAKE) -C io_addr clean
clean-all:
$(MAKE) -C basic clean-all
$(MAKE) -C demo clean-all
$(MAKE) -C dogfood clean-all
$(MAKE) -C stress clean-all
$(MAKE) -C io_addr clean-all

View File

@@ -0,0 +1,67 @@
RISCV_TOOLCHAIN_PATH ?= /opt/riscv-gnu-toolchain
VORTEX_RT_PATH ?= $(wildcard ../../../runtime)
OPTS ?= -n1
VX_CC = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-gcc
VX_CXX = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-g++
VX_DP = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-objdump
VX_CP = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-objcopy
VX_CFLAGS += -march=rv32imf -mabi=ilp32f -O3 -Wstack-usage=1024 -ffreestanding -nostartfiles -fdata-sections -ffunction-sections
VX_CFLAGS += -I$(VORTEX_RT_PATH)/include -I$(VORTEX_RT_PATH)/../hw
VX_LDFLAGS += -Wl,-Bstatic,-T,$(VORTEX_RT_PATH)/linker/vx_link.ld -Wl,--gc-sections $(VORTEX_RT_PATH)/libvortexrt.a
VX_SRCS = kernel.c
#CXXFLAGS += -std=c++11 -O2 -Wall -Wextra -pedantic -Wfatal-errors
CXXFLAGS += -std=c++11 -O0 -g -Wall -Wextra -pedantic -Wfatal-errors
CXXFLAGS += -I../../include -I$(VORTEX_RT_PATH)/../hw
PROJECT = io_addr
SRCS = main.cpp
all: $(PROJECT) kernel.bin kernel.dump
kernel.dump: kernel.elf
$(VX_DP) -D kernel.elf > kernel.dump
kernel.bin: kernel.elf
$(VX_CP) -O binary kernel.elf kernel.bin
kernel.elf: $(VX_SRCS)
$(VX_CC) $(VX_CFLAGS) $(VX_SRCS) $(VX_LDFLAGS) -o kernel.elf
$(PROJECT): $(SRCS)
$(CXX) $(CXXFLAGS) $^ $(LDFLAGS) -L../../stub -lvortex -o $@
run-fpga: $(PROJECT)
LD_LIBRARY_PATH=../../opae:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
run-asesim: $(PROJECT)
ASE_LOG=0 LD_LIBRARY_PATH=../../opae/ase:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
run-vlsim: $(PROJECT)
LD_LIBRARY_PATH=../../opae/vlsim:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
run-rtlsim: $(PROJECT)
LD_LIBRARY_PATH=../../rtlsim:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
run-simx: $(PROJECT)
LD_LIBRARY_PATH=../../simx:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
.depend: $(SRCS)
$(CXX) $(CXXFLAGS) -MM $^ > .depend;
clean:
rm -rf $(PROJECT) *.o .depend
clean-all: clean
rm -rf *.elf *.bin *.dump
ifneq ($(MAKECMDGOALS),clean)
-include .depend
endif

View File

@@ -0,0 +1,12 @@
#ifndef _COMMON_H_
#define _COMMON_H_
#define KERNEL_ARG_DEV_MEM_ADDR 0x7ffff000
struct kernel_arg_t {
uint32_t num_points;
uint32_t src_ptr;
uint32_t dst_ptr;
};
#endif

BIN
driver/tests/io_addr/kernel.bin Executable file

Binary file not shown.

View File

@@ -0,0 +1,19 @@
#include <stdint.h>
#include <vx_intrinsics.h>
#include <vx_spawn.h>
#include "common.h"
void kernel_body(int task_id, void* arg) {
struct kernel_arg_t* _arg = (struct kernel_arg_t*)(arg);
uint32_t* src_ptr = (uint32_t*)_arg->src_ptr;
uint32_t* dst_ptr = (uint32_t*)_arg->dst_ptr;
int32_t* addr_ptr = (int32_t*)(src_ptr[task_id]);
dst_ptr[task_id] = *addr_ptr;
}
void main() {
struct kernel_arg_t* arg = (struct kernel_arg_t*)KERNEL_ARG_DEV_MEM_ADDR;
vx_spawn_tasks(arg->num_points, kernel_body, arg);
}

View File

@@ -0,0 +1,547 @@
kernel.elf: file format elf32-littleriscv
Disassembly of section .init:
80000000 <_start>:
80000000: 00000597 auipc a1,0x0
80000004: 0ac58593 addi a1,a1,172 # 800000ac <vx_set_sp>
80000008: fc102573 csrr a0,0xfc1
8000000c: 00b5106b 0xb5106b
80000010: 09c000ef jal ra,800000ac <vx_set_sp>
80000014: 00100513 li a0,1
80000018: 0005006b 0x5006b
8000001c: 00002517 auipc a0,0x2
80000020: ae050513 addi a0,a0,-1312 # 80001afc <g_wspawn_args>
80000024: 00002617 auipc a2,0x2
80000028: b5860613 addi a2,a2,-1192 # 80001b7c <__BSS_END__>
8000002c: 40a60633 sub a2,a2,a0
80000030: 00000593 li a1,0
80000034: 3fc000ef jal ra,80000430 <memset>
80000038: 00000517 auipc a0,0x0
8000003c: 30050513 addi a0,a0,768 # 80000338 <__libc_fini_array>
80000040: 2b0000ef jal ra,800002f0 <atexit>
80000044: 350000ef jal ra,80000394 <__libc_init_array>
80000048: 008000ef jal ra,80000050 <main>
8000004c: 2b80006f j 80000304 <exit>
Disassembly of section .text:
80000050 <main>:
80000050: 7ffff7b7 lui a5,0x7ffff
80000054: 0007a503 lw a0,0(a5) # 7ffff000 <__stack_size+0x7fffec00>
80000058: 800005b7 lui a1,0x80000
8000005c: 7ffff637 lui a2,0x7ffff
80000060: 08058593 addi a1,a1,128 # 80000080 <__stack_top+0x81000080>
80000064: 1440006f j 800001a8 <vx_spawn_tasks>
80000068 <register_fini>:
80000068: 00000793 li a5,0
8000006c: 00078863 beqz a5,8000007c <register_fini+0x14>
80000070: 80000537 lui a0,0x80000
80000074: 33850513 addi a0,a0,824 # 80000338 <__stack_top+0x81000338>
80000078: 2780006f j 800002f0 <atexit>
8000007c: 00008067 ret
80000080 <kernel_body>:
80000080: 0045a783 lw a5,4(a1)
80000084: 00251513 slli a0,a0,0x2
80000088: 00a787b3 add a5,a5,a0
8000008c: 0007a703 lw a4,0(a5)
80000090: 0085a783 lw a5,8(a1)
80000094: 00072703 lw a4,0(a4)
80000098: 00a78533 add a0,a5,a0
8000009c: 00e52023 sw a4,0(a0)
800000a0: 00008067 ret
800000a4 <_exit>:
800000a4: 00000513 li a0,0
800000a8: 0005006b 0x5006b
800000ac <vx_set_sp>:
800000ac: fc002573 csrr a0,0xfc0
800000b0: 0005006b 0x5006b
800000b4: 00002197 auipc gp,0x2
800000b8: e1c18193 addi gp,gp,-484 # 80001ed0 <__global_pointer>
800000bc: 7f000117 auipc sp,0x7f000
800000c0: f4410113 addi sp,sp,-188 # ff000000 <__stack_top>
800000c4: 40000593 li a1,1024
800000c8: cc102673 csrr a2,0xcc1
800000cc: 02c585b3 mul a1,a1,a2
800000d0: 40b10133 sub sp,sp,a1
800000d4: cc3026f3 csrr a3,0xcc3
800000d8: 00068663 beqz a3,800000e4 <RETURN>
800000dc: 00000513 li a0,0
800000e0: 0005006b 0x5006b
800000e4 <RETURN>:
800000e4: 00008067 ret
800000e8 <spawn_tasks_callback>:
800000e8: fe010113 addi sp,sp,-32
800000ec: 00112e23 sw ra,28(sp)
800000f0: 00812c23 sw s0,24(sp)
800000f4: 00912a23 sw s1,20(sp)
800000f8: 01212823 sw s2,16(sp)
800000fc: 01312623 sw s3,12(sp)
80000100: fc0027f3 csrr a5,0xfc0
80000104: 0007806b 0x7806b
80000108: cc5027f3 csrr a5,0xcc5
8000010c: cc3029f3 csrr s3,0xcc3
80000110: cc002773 csrr a4,0xcc0
80000114: fc002673 csrr a2,0xfc0
80000118: 00279693 slli a3,a5,0x2
8000011c: 800027b7 lui a5,0x80002
80000120: afc78793 addi a5,a5,-1284 # 80001afc <__stack_top+0x81001afc>
80000124: 00d787b3 add a5,a5,a3
80000128: 0007a483 lw s1,0(a5)
8000012c: 0104a403 lw s0,16(s1)
80000130: 00c4a683 lw a3,12(s1)
80000134: 0089a933 slt s2,s3,s0
80000138: 00040793 mv a5,s0
8000013c: 00d90933 add s2,s2,a3
80000140: 03368433 mul s0,a3,s3
80000144: 00f9d463 bge s3,a5,8000014c <spawn_tasks_callback+0x64>
80000148: 00098793 mv a5,s3
8000014c: 00f40433 add s0,s0,a5
80000150: 0084a683 lw a3,8(s1)
80000154: 02c40433 mul s0,s0,a2
80000158: 02e907b3 mul a5,s2,a4
8000015c: 00d40433 add s0,s0,a3
80000160: 00f40433 add s0,s0,a5
80000164: 00890933 add s2,s2,s0
80000168: 01245e63 bge s0,s2,80000184 <spawn_tasks_callback+0x9c>
8000016c: 0004a783 lw a5,0(s1)
80000170: 0044a583 lw a1,4(s1)
80000174: 00040513 mv a0,s0
80000178: 00140413 addi s0,s0,1
8000017c: 000780e7 jalr a5
80000180: fe8916e3 bne s2,s0,8000016c <spawn_tasks_callback+0x84>
80000184: 0019b993 seqz s3,s3
80000188: 0009806b 0x9806b
8000018c: 01c12083 lw ra,28(sp)
80000190: 01812403 lw s0,24(sp)
80000194: 01412483 lw s1,20(sp)
80000198: 01012903 lw s2,16(sp)
8000019c: 00c12983 lw s3,12(sp)
800001a0: 02010113 addi sp,sp,32
800001a4: 00008067 ret
800001a8 <vx_spawn_tasks>:
800001a8: fc010113 addi sp,sp,-64
800001ac: 02112e23 sw ra,60(sp)
800001b0: 02812c23 sw s0,56(sp)
800001b4: 02912a23 sw s1,52(sp)
800001b8: 03212823 sw s2,48(sp)
800001bc: 03312623 sw s3,44(sp)
800001c0: fc2026f3 csrr a3,0xfc2
800001c4: fc102873 csrr a6,0xfc1
800001c8: fc002473 csrr s0,0xfc0
800001cc: cc5027f3 csrr a5,0xcc5
800001d0: 01f00713 li a4,31
800001d4: 0cf74463 blt a4,a5,8000029c <vx_spawn_tasks+0xf4>
800001d8: 030408b3 mul a7,s0,a6
800001dc: 00100713 li a4,1
800001e0: 00a8d463 bge a7,a0,800001e8 <vx_spawn_tasks+0x40>
800001e4: 03154733 div a4,a0,a7
800001e8: 0ce6c863 blt a3,a4,800002b8 <vx_spawn_tasks+0x110>
800001ec: 0ae7d863 bge a5,a4,8000029c <vx_spawn_tasks+0xf4>
800001f0: fff68693 addi a3,a3,-1
800001f4: 02e54333 div t1,a0,a4
800001f8: 00030893 mv a7,t1
800001fc: 00f69663 bne a3,a5,80000208 <vx_spawn_tasks+0x60>
80000200: 02e56533 rem a0,a0,a4
80000204: 006508b3 add a7,a0,t1
80000208: 0288c4b3 div s1,a7,s0
8000020c: 0288e933 rem s2,a7,s0
80000210: 0b04ca63 blt s1,a6,800002c4 <vx_spawn_tasks+0x11c>
80000214: 00100693 li a3,1
80000218: 0304c733 div a4,s1,a6
8000021c: 00070663 beqz a4,80000228 <vx_spawn_tasks+0x80>
80000220: 00070693 mv a3,a4
80000224: 0304e733 rem a4,s1,a6
80000228: 800029b7 lui s3,0x80002
8000022c: afc98993 addi s3,s3,-1284 # 80001afc <__stack_top+0x81001afc>
80000230: 00e12e23 sw a4,28(sp)
80000234: 00c10713 addi a4,sp,12
80000238: 00b12623 sw a1,12(sp)
8000023c: 00c12823 sw a2,16(sp)
80000240: 00d12c23 sw a3,24(sp)
80000244: 02f30333 mul t1,t1,a5
80000248: 00279793 slli a5,a5,0x2
8000024c: 00f987b3 add a5,s3,a5
80000250: 00e7a023 sw a4,0(a5)
80000254: 00612a23 sw t1,20(sp)
80000258: 06904c63 bgtz s1,800002d0 <vx_spawn_tasks+0x128>
8000025c: 04090063 beqz s2,8000029c <vx_spawn_tasks+0xf4>
80000260: 02848433 mul s0,s1,s0
80000264: 00812a23 sw s0,20(sp)
80000268: 0009006b 0x9006b
8000026c: cc5027f3 csrr a5,0xcc5
80000270: cc202573 csrr a0,0xcc2
80000274: 00279793 slli a5,a5,0x2
80000278: 00f989b3 add s3,s3,a5
8000027c: 0009a783 lw a5,0(s3)
80000280: 0087a683 lw a3,8(a5)
80000284: 0007a703 lw a4,0(a5)
80000288: 0047a583 lw a1,4(a5)
8000028c: 00d50533 add a0,a0,a3
80000290: 000700e7 jalr a4
80000294: 00100793 li a5,1
80000298: 0007806b 0x7806b
8000029c: 03c12083 lw ra,60(sp)
800002a0: 03812403 lw s0,56(sp)
800002a4: 03412483 lw s1,52(sp)
800002a8: 03012903 lw s2,48(sp)
800002ac: 02c12983 lw s3,44(sp)
800002b0: 04010113 addi sp,sp,64
800002b4: 00008067 ret
800002b8: 00068713 mv a4,a3
800002bc: f2e7cae3 blt a5,a4,800001f0 <vx_spawn_tasks+0x48>
800002c0: fddff06f j 8000029c <vx_spawn_tasks+0xf4>
800002c4: 00000713 li a4,0
800002c8: 00100693 li a3,1
800002cc: f5dff06f j 80000228 <vx_spawn_tasks+0x80>
800002d0: 00048713 mv a4,s1
800002d4: 00985463 bge a6,s1,800002dc <vx_spawn_tasks+0x134>
800002d8: 00080713 mv a4,a6
800002dc: 800007b7 lui a5,0x80000
800002e0: 0e878793 addi a5,a5,232 # 800000e8 <__stack_top+0x810000e8>
800002e4: 00f7106b 0xf7106b
800002e8: e01ff0ef jal ra,800000e8 <spawn_tasks_callback>
800002ec: f71ff06f j 8000025c <vx_spawn_tasks+0xb4>
800002f0 <atexit>:
800002f0: 00050593 mv a1,a0
800002f4: 00000693 li a3,0
800002f8: 00000613 li a2,0
800002fc: 00000513 li a0,0
80000300: 20c0006f j 8000050c <__register_exitproc>
80000304 <exit>:
80000304: ff010113 addi sp,sp,-16
80000308: 00000593 li a1,0
8000030c: 00812423 sw s0,8(sp)
80000310: 00112623 sw ra,12(sp)
80000314: 00050413 mv s0,a0
80000318: 290000ef jal ra,800005a8 <__call_exitprocs>
8000031c: 800027b7 lui a5,0x80002
80000320: af87a503 lw a0,-1288(a5) # 80001af8 <__stack_top+0x81001af8>
80000324: 03c52783 lw a5,60(a0)
80000328: 00078463 beqz a5,80000330 <exit+0x2c>
8000032c: 000780e7 jalr a5
80000330: 00040513 mv a0,s0
80000334: d71ff0ef jal ra,800000a4 <_exit>
80000338 <__libc_fini_array>:
80000338: ff010113 addi sp,sp,-16
8000033c: 00812423 sw s0,8(sp)
80000340: 800017b7 lui a5,0x80001
80000344: 80001437 lui s0,0x80001
80000348: 6d040413 addi s0,s0,1744 # 800016d0 <__stack_top+0x810016d0>
8000034c: 6d078793 addi a5,a5,1744 # 800016d0 <__stack_top+0x810016d0>
80000350: 408787b3 sub a5,a5,s0
80000354: 00912223 sw s1,4(sp)
80000358: 00112623 sw ra,12(sp)
8000035c: 4027d493 srai s1,a5,0x2
80000360: 02048063 beqz s1,80000380 <__libc_fini_array+0x48>
80000364: ffc78793 addi a5,a5,-4
80000368: 00878433 add s0,a5,s0
8000036c: 00042783 lw a5,0(s0)
80000370: fff48493 addi s1,s1,-1
80000374: ffc40413 addi s0,s0,-4
80000378: 000780e7 jalr a5
8000037c: fe0498e3 bnez s1,8000036c <__libc_fini_array+0x34>
80000380: 00c12083 lw ra,12(sp)
80000384: 00812403 lw s0,8(sp)
80000388: 00412483 lw s1,4(sp)
8000038c: 01010113 addi sp,sp,16
80000390: 00008067 ret
80000394 <__libc_init_array>:
80000394: ff010113 addi sp,sp,-16
80000398: 00812423 sw s0,8(sp)
8000039c: 01212023 sw s2,0(sp)
800003a0: 80001437 lui s0,0x80001
800003a4: 80001937 lui s2,0x80001
800003a8: 6cc40793 addi a5,s0,1740 # 800016cc <__stack_top+0x810016cc>
800003ac: 6cc90913 addi s2,s2,1740 # 800016cc <__stack_top+0x810016cc>
800003b0: 40f90933 sub s2,s2,a5
800003b4: 00112623 sw ra,12(sp)
800003b8: 00912223 sw s1,4(sp)
800003bc: 40295913 srai s2,s2,0x2
800003c0: 02090063 beqz s2,800003e0 <__libc_init_array+0x4c>
800003c4: 6cc40413 addi s0,s0,1740
800003c8: 00000493 li s1,0
800003cc: 00042783 lw a5,0(s0)
800003d0: 00148493 addi s1,s1,1
800003d4: 00440413 addi s0,s0,4
800003d8: 000780e7 jalr a5
800003dc: fe9918e3 bne s2,s1,800003cc <__libc_init_array+0x38>
800003e0: 80001437 lui s0,0x80001
800003e4: 80001937 lui s2,0x80001
800003e8: 6cc40793 addi a5,s0,1740 # 800016cc <__stack_top+0x810016cc>
800003ec: 6d090913 addi s2,s2,1744 # 800016d0 <__stack_top+0x810016d0>
800003f0: 40f90933 sub s2,s2,a5
800003f4: 40295913 srai s2,s2,0x2
800003f8: 02090063 beqz s2,80000418 <__libc_init_array+0x84>
800003fc: 6cc40413 addi s0,s0,1740
80000400: 00000493 li s1,0
80000404: 00042783 lw a5,0(s0)
80000408: 00148493 addi s1,s1,1
8000040c: 00440413 addi s0,s0,4
80000410: 000780e7 jalr a5
80000414: fe9918e3 bne s2,s1,80000404 <__libc_init_array+0x70>
80000418: 00c12083 lw ra,12(sp)
8000041c: 00812403 lw s0,8(sp)
80000420: 00412483 lw s1,4(sp)
80000424: 00012903 lw s2,0(sp)
80000428: 01010113 addi sp,sp,16
8000042c: 00008067 ret
80000430 <memset>:
80000430: 00f00313 li t1,15
80000434: 00050713 mv a4,a0
80000438: 02c37e63 bgeu t1,a2,80000474 <memset+0x44>
8000043c: 00f77793 andi a5,a4,15
80000440: 0a079063 bnez a5,800004e0 <memset+0xb0>
80000444: 08059263 bnez a1,800004c8 <memset+0x98>
80000448: ff067693 andi a3,a2,-16
8000044c: 00f67613 andi a2,a2,15
80000450: 00e686b3 add a3,a3,a4
80000454: 00b72023 sw a1,0(a4)
80000458: 00b72223 sw a1,4(a4)
8000045c: 00b72423 sw a1,8(a4)
80000460: 00b72623 sw a1,12(a4)
80000464: 01070713 addi a4,a4,16
80000468: fed766e3 bltu a4,a3,80000454 <memset+0x24>
8000046c: 00061463 bnez a2,80000474 <memset+0x44>
80000470: 00008067 ret
80000474: 40c306b3 sub a3,t1,a2
80000478: 00269693 slli a3,a3,0x2
8000047c: 00000297 auipc t0,0x0
80000480: 005686b3 add a3,a3,t0
80000484: 00c68067 jr 12(a3)
80000488: 00b70723 sb a1,14(a4)
8000048c: 00b706a3 sb a1,13(a4)
80000490: 00b70623 sb a1,12(a4)
80000494: 00b705a3 sb a1,11(a4)
80000498: 00b70523 sb a1,10(a4)
8000049c: 00b704a3 sb a1,9(a4)
800004a0: 00b70423 sb a1,8(a4)
800004a4: 00b703a3 sb a1,7(a4)
800004a8: 00b70323 sb a1,6(a4)
800004ac: 00b702a3 sb a1,5(a4)
800004b0: 00b70223 sb a1,4(a4)
800004b4: 00b701a3 sb a1,3(a4)
800004b8: 00b70123 sb a1,2(a4)
800004bc: 00b700a3 sb a1,1(a4)
800004c0: 00b70023 sb a1,0(a4)
800004c4: 00008067 ret
800004c8: 0ff5f593 andi a1,a1,255
800004cc: 00859693 slli a3,a1,0x8
800004d0: 00d5e5b3 or a1,a1,a3
800004d4: 01059693 slli a3,a1,0x10
800004d8: 00d5e5b3 or a1,a1,a3
800004dc: f6dff06f j 80000448 <memset+0x18>
800004e0: 00279693 slli a3,a5,0x2
800004e4: 00000297 auipc t0,0x0
800004e8: 005686b3 add a3,a3,t0
800004ec: 00008293 mv t0,ra
800004f0: fa0680e7 jalr -96(a3)
800004f4: 00028093 mv ra,t0
800004f8: ff078793 addi a5,a5,-16
800004fc: 40f70733 sub a4,a4,a5
80000500: 00f60633 add a2,a2,a5
80000504: f6c378e3 bgeu t1,a2,80000474 <memset+0x44>
80000508: f3dff06f j 80000444 <memset+0x14>
8000050c <__register_exitproc>:
8000050c: 800027b7 lui a5,0x80002
80000510: af87a703 lw a4,-1288(a5) # 80001af8 <__stack_top+0x81001af8>
80000514: 14872783 lw a5,328(a4)
80000518: 04078c63 beqz a5,80000570 <__register_exitproc+0x64>
8000051c: 0047a703 lw a4,4(a5)
80000520: 01f00813 li a6,31
80000524: 06e84e63 blt a6,a4,800005a0 <__register_exitproc+0x94>
80000528: 00271813 slli a6,a4,0x2
8000052c: 02050663 beqz a0,80000558 <__register_exitproc+0x4c>
80000530: 01078333 add t1,a5,a6
80000534: 08c32423 sw a2,136(t1)
80000538: 1887a883 lw a7,392(a5)
8000053c: 00100613 li a2,1
80000540: 00e61633 sll a2,a2,a4
80000544: 00c8e8b3 or a7,a7,a2
80000548: 1917a423 sw a7,392(a5)
8000054c: 10d32423 sw a3,264(t1)
80000550: 00200693 li a3,2
80000554: 02d50463 beq a0,a3,8000057c <__register_exitproc+0x70>
80000558: 00170713 addi a4,a4,1
8000055c: 00e7a223 sw a4,4(a5)
80000560: 010787b3 add a5,a5,a6
80000564: 00b7a423 sw a1,8(a5)
80000568: 00000513 li a0,0
8000056c: 00008067 ret
80000570: 14c70793 addi a5,a4,332
80000574: 14f72423 sw a5,328(a4)
80000578: fa5ff06f j 8000051c <__register_exitproc+0x10>
8000057c: 18c7a683 lw a3,396(a5)
80000580: 00170713 addi a4,a4,1
80000584: 00e7a223 sw a4,4(a5)
80000588: 00c6e633 or a2,a3,a2
8000058c: 18c7a623 sw a2,396(a5)
80000590: 010787b3 add a5,a5,a6
80000594: 00b7a423 sw a1,8(a5)
80000598: 00000513 li a0,0
8000059c: 00008067 ret
800005a0: fff00513 li a0,-1
800005a4: 00008067 ret
800005a8 <__call_exitprocs>:
800005a8: fd010113 addi sp,sp,-48
800005ac: 800027b7 lui a5,0x80002
800005b0: 01412c23 sw s4,24(sp)
800005b4: af87aa03 lw s4,-1288(a5) # 80001af8 <__stack_top+0x81001af8>
800005b8: 03212023 sw s2,32(sp)
800005bc: 02112623 sw ra,44(sp)
800005c0: 148a2903 lw s2,328(s4)
800005c4: 02812423 sw s0,40(sp)
800005c8: 02912223 sw s1,36(sp)
800005cc: 01312e23 sw s3,28(sp)
800005d0: 01512a23 sw s5,20(sp)
800005d4: 01612823 sw s6,16(sp)
800005d8: 01712623 sw s7,12(sp)
800005dc: 01812423 sw s8,8(sp)
800005e0: 04090063 beqz s2,80000620 <__call_exitprocs+0x78>
800005e4: 00050b13 mv s6,a0
800005e8: 00058b93 mv s7,a1
800005ec: 00100a93 li s5,1
800005f0: fff00993 li s3,-1
800005f4: 00492483 lw s1,4(s2)
800005f8: fff48413 addi s0,s1,-1
800005fc: 02044263 bltz s0,80000620 <__call_exitprocs+0x78>
80000600: 00249493 slli s1,s1,0x2
80000604: 009904b3 add s1,s2,s1
80000608: 040b8463 beqz s7,80000650 <__call_exitprocs+0xa8>
8000060c: 1044a783 lw a5,260(s1)
80000610: 05778063 beq a5,s7,80000650 <__call_exitprocs+0xa8>
80000614: fff40413 addi s0,s0,-1
80000618: ffc48493 addi s1,s1,-4
8000061c: ff3416e3 bne s0,s3,80000608 <__call_exitprocs+0x60>
80000620: 02c12083 lw ra,44(sp)
80000624: 02812403 lw s0,40(sp)
80000628: 02412483 lw s1,36(sp)
8000062c: 02012903 lw s2,32(sp)
80000630: 01c12983 lw s3,28(sp)
80000634: 01812a03 lw s4,24(sp)
80000638: 01412a83 lw s5,20(sp)
8000063c: 01012b03 lw s6,16(sp)
80000640: 00c12b83 lw s7,12(sp)
80000644: 00812c03 lw s8,8(sp)
80000648: 03010113 addi sp,sp,48
8000064c: 00008067 ret
80000650: 00492783 lw a5,4(s2)
80000654: 0044a683 lw a3,4(s1)
80000658: fff78793 addi a5,a5,-1
8000065c: 04878e63 beq a5,s0,800006b8 <__call_exitprocs+0x110>
80000660: 0004a223 sw zero,4(s1)
80000664: fa0688e3 beqz a3,80000614 <__call_exitprocs+0x6c>
80000668: 18892783 lw a5,392(s2)
8000066c: 008a9733 sll a4,s5,s0
80000670: 00492c03 lw s8,4(s2)
80000674: 00f777b3 and a5,a4,a5
80000678: 02079263 bnez a5,8000069c <__call_exitprocs+0xf4>
8000067c: 000680e7 jalr a3
80000680: 00492703 lw a4,4(s2)
80000684: 148a2783 lw a5,328(s4)
80000688: 01871463 bne a4,s8,80000690 <__call_exitprocs+0xe8>
8000068c: f92784e3 beq a5,s2,80000614 <__call_exitprocs+0x6c>
80000690: f80788e3 beqz a5,80000620 <__call_exitprocs+0x78>
80000694: 00078913 mv s2,a5
80000698: f5dff06f j 800005f4 <__call_exitprocs+0x4c>
8000069c: 18c92783 lw a5,396(s2)
800006a0: 0844a583 lw a1,132(s1)
800006a4: 00f77733 and a4,a4,a5
800006a8: 00071c63 bnez a4,800006c0 <__call_exitprocs+0x118>
800006ac: 000b0513 mv a0,s6
800006b0: 000680e7 jalr a3
800006b4: fcdff06f j 80000680 <__call_exitprocs+0xd8>
800006b8: 00892223 sw s0,4(s2)
800006bc: fa9ff06f j 80000664 <__call_exitprocs+0xbc>
800006c0: 00058513 mv a0,a1
800006c4: 000680e7 jalr a3
800006c8: fb9ff06f j 80000680 <__call_exitprocs+0xd8>
Disassembly of section .init_array:
800016cc <__init_array_start>:
800016cc: 0068 addi a0,sp,12
800016ce: 8000 0x8000
Disassembly of section .data:
800016d0 <impure_data>:
800016d0: 0000 unimp
800016d2: 0000 unimp
800016d4: 19bc addi a5,sp,248
800016d6: 8000 0x8000
800016d8: 1a24 addi s1,sp,312
800016da: 8000 0x8000
800016dc: 1a8c addi a1,sp,368
800016de: 8000 0x8000
...
80001778: 0001 nop
8000177a: 0000 unimp
8000177c: 0000 unimp
8000177e: 0000 unimp
80001780: 330e fld ft6,224(sp)
80001782: abcd j 80001d74 <__BSS_END__+0x1f8>
80001784: 1234 addi a3,sp,296
80001786: e66d bnez a2,80001870 <impure_data+0x1a0>
80001788: deec sw a1,124(a3)
8000178a: 0005 c.nop 1
8000178c: 0000000b 0xb
...
Disassembly of section .sdata:
80001af8 <_global_impure_ptr>:
80001af8: 16d0 addi a2,sp,868
80001afa: 8000 0x8000
Disassembly of section .bss:
80001afc <g_wspawn_args>:
...
Disassembly of section .comment:
00000000 <.comment>:
0: 3a434347 fmsub.d ft6,ft6,ft4,ft7,rmm
4: 2820 fld fs0,80(s0)
6: 29554e47 fmsub.s ft8,fa0,fs5,ft5,rmm
a: 3120 fld fs0,96(a0)
c: 2e30 fld fa2,88(a2)
e: 2e32 fld ft8,264(sp)
10: 0030 addi a2,sp,8
Disassembly of section .riscv.attributes:
00000000 <.riscv.attributes>:
0: 2941 jal 490 <__stack_size+0x90>
2: 0000 unimp
4: 7200 flw fs0,32(a2)
6: 7369 lui t1,0xffffa
8: 01007663 bgeu zero,a6,14 <__stack_usage+0x14>
c: 001f 0000 1004 0x10040000001f
12: 7205 lui tp,0xfffe1
14: 3376 fld ft6,376(sp)
16: 6932 flw fs2,12(sp)
18: 7032 flw ft0,44(sp)
1a: 5f30 lw a2,120(a4)
1c: 326d jal fffff9c6 <__stack_top+0xfff9c6>
1e: 3070 fld fa2,224(s0)
20: 665f 7032 0030 0x307032665f
26: 0108 addi a0,sp,128
28: 0b0a slli s6,s6,0x2

BIN
driver/tests/io_addr/kernel.elf Executable file

Binary file not shown.

View File

@@ -0,0 +1,246 @@
#include <iostream>
#include <unistd.h>
#include <string.h>
#include <vortex.h>
#include <vector>
#include <VX_config.h>
#include "common.h"
#define NUM_ADDRS 16
#define RT_CHECK(_expr) \
do { \
int _ret = _expr; \
if (0 == _ret) \
break; \
printf("Error: '%s' returned %d!\n", #_expr, (int)_ret); \
cleanup(); \
exit(-1); \
} while (false)
///////////////////////////////////////////////////////////////////////////////
const char* kernel_file = "kernel.bin";
uint32_t count = 0;
size_t usr_test_mem;
std::vector<uint32_t> src_data;
std::vector<int32_t> ref_data;
vx_device_h device = nullptr;
vx_buffer_h staging_buf = nullptr;
static void show_usage() {
std::cout << "Vortex Driver Test." << std::endl;
std::cout << "Usage: [-k: kernel] [-n words] [-h: help]" << std::endl;
}
static void parse_args(int argc, char **argv) {
int c;
while ((c = getopt(argc, argv, "n:k:h?")) != -1) {
switch (c) {
case 'n':
count = atoi(optarg);
break;
case 'k':
kernel_file = optarg;
break;
case 'h':
case '?': {
show_usage();
exit(0);
} break;
default:
show_usage();
exit(-1);
}
}
}
void cleanup() {
if (staging_buf) {
vx_buf_release(staging_buf);
}
if (device) {
vx_dev_close(device);
}
}
void gen_input_data(uint32_t num_points) {
src_data.resize(num_points);
uint32_t u = 0, k = 0;
for (uint32_t i = 0; i < num_points; ++i) {
if (0 ==(i % 4)) {
k = (i + u) % NUM_ADDRS;
++u;
}
uint32_t j = i % NUM_ADDRS;
uint32_t v = ((j == k) ? usr_test_mem : IO_BASE_ADDR) + j * sizeof(uint32_t);
src_data[i] = v;
std::cout << std::dec << i << "," << k << ": value=0x" << std::hex << v << std::endl;
}
}
void gen_ref_data(uint32_t num_points) {
ref_data.resize(num_points);
for (uint32_t i = 0; i < num_points; ++i) {
uint32_t j = i % NUM_ADDRS;
ref_data[i] = j * j;
}
}
int run_test(const kernel_arg_t& kernel_arg,
uint32_t buf_size,
uint32_t num_points) {
// start device
std::cout << "start device" << std::endl;
RT_CHECK(vx_start(device));
// wait for completion
std::cout << "wait for completion" << std::endl;
RT_CHECK(vx_ready_wait(device, -1));
// download destination buffer
std::cout << "download destination buffer" << std::endl;
RT_CHECK(vx_copy_from_dev(staging_buf, kernel_arg.dst_ptr, buf_size, 0));
// verify result
std::cout << "verify result" << std::endl;
{
int errors = 0;
auto buf_ptr = (int32_t*)vx_host_ptr(staging_buf);
for (uint32_t i = 0; i < num_points; ++i) {
int ref = ref_data.at(i);
int cur = buf_ptr[i];
if (cur != ref) {
std::cout << "error at result #" << std::dec << i
<< std::hex << ": actual 0x" << cur << ", expected 0x" << ref << std::endl;
++errors;
}
}
if (errors != 0) {
std::cout << "Found " << std::dec << errors << " errors!" << std::endl;
std::cout << "FAILED!" << std::endl;
return 1;
}
}
return 0;
}
int main(int argc, char *argv[]) {
size_t value;
kernel_arg_t kernel_arg;
// parse command arguments
parse_args(argc, argv);
if (count == 0) {
count = 1;
}
std::srand(50);
// open device connection
std::cout << "open device connection" << std::endl;
RT_CHECK(vx_dev_open(&device));
unsigned max_cores, max_warps, max_threads;
RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_CORES, &max_cores));
RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_WARPS, &max_warps));
RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_THREADS, &max_threads));
uint32_t num_tasks = max_cores * max_warps * max_threads;
uint32_t num_points = count * num_tasks;
RT_CHECK(vx_alloc_dev_mem(device, NUM_ADDRS * sizeof(uint32_t), &usr_test_mem));
// generate input data
gen_input_data(num_points);
// generate reference data
gen_ref_data(num_points);
uint32_t src_buf_size = src_data.size() * sizeof(int32_t);
uint32_t dst_buf_size = src_data.size() * sizeof(int32_t);
std::cout << "number of points: " << num_points << std::endl;
std::cout << "buffer size: " << dst_buf_size << " bytes" << std::endl;
// upload program
std::cout << "upload program" << std::endl;
RT_CHECK(vx_upload_kernel_file(device, kernel_file));
// allocate device memory
std::cout << "allocate device memory" << std::endl;
RT_CHECK(vx_alloc_dev_mem(device, src_buf_size, &value));
kernel_arg.src_ptr = value;
RT_CHECK(vx_alloc_dev_mem(device, dst_buf_size, &value));
kernel_arg.dst_ptr = value;
kernel_arg.num_points = num_points;
std::cout << "dev_src=" << std::hex << kernel_arg.src_ptr << std::endl;
std::cout << "dev_dst=" << std::hex << kernel_arg.dst_ptr << std::endl;
// allocate shared memory
std::cout << "allocate shared memory" << std::endl;
uint32_t staging_buf_size = std::max<uint32_t>(src_buf_size,
std::max<uint32_t>(dst_buf_size,
sizeof(kernel_arg_t)));
RT_CHECK(vx_alloc_shared_mem(device, staging_buf_size, &staging_buf));
// upload kernel argument
std::cout << "upload kernel argument" << std::endl;
{
auto buf_ptr = (int*)vx_host_ptr(staging_buf);
memcpy(buf_ptr, &kernel_arg, sizeof(kernel_arg_t));
RT_CHECK(vx_copy_to_dev(staging_buf, KERNEL_ARG_DEV_MEM_ADDR, sizeof(kernel_arg_t), 0));
}
// upload test address data
{
auto buf_ptr = (int32_t*)vx_host_ptr(staging_buf);
for (uint32_t i = 0; i < NUM_ADDRS; ++i) {
buf_ptr[i] = i * i;
}
}
RT_CHECK(vx_copy_to_dev(staging_buf, 0xFF000000, NUM_ADDRS * sizeof(uint32_t), 0));
RT_CHECK(vx_copy_to_dev(staging_buf, usr_test_mem, NUM_ADDRS * sizeof(uint32_t), 0));
// upload source buffer
{
auto buf_ptr = (int32_t*)vx_host_ptr(staging_buf);
for (uint32_t i = 0; i < num_points; ++i) {
buf_ptr[i] = src_data.at(i);
}
}
std::cout << "upload source buffer" << std::endl;
RT_CHECK(vx_copy_to_dev(staging_buf, kernel_arg.src_ptr, src_buf_size, 0));
// clear destination buffer
{
auto buf_ptr = (int32_t*)vx_host_ptr(staging_buf);
for (uint32_t i = 0; i < num_points; ++i) {
buf_ptr[i] = 0xdeadbeef;
}
}
std::cout << "clear destination buffer" << std::endl;
RT_CHECK(vx_copy_to_dev(staging_buf, kernel_arg.dst_ptr, dst_buf_size, 0));
// run tests
std::cout << "run tests" << std::endl;
RT_CHECK(run_test(kernel_arg, dst_buf_size, num_points));
// cleanup
std::cout << "cleanup" << std::endl;
cleanup();
std::cout << "PASSED!" << std::endl;
return 0;
}