X-Git-Url: https://git.librecmc.org/?a=blobdiff_plain;f=arch%2Friscv%2Fcpu%2Fstart.S;h=ecf0482635b07fc410cac94dd4c0d4c375a93e61;hb=84dc9d26908798c7e9ee5469965c16653593fde5;hp=7cd7755190ce6a7569a900ff05d9d71c9802df36;hpb=14573fb78fef99266f7691ac9af2ddb388f9491f;p=oweals%2Fu-boot.git

diff --git a/arch/riscv/cpu/start.S b/arch/riscv/cpu/start.S
index 7cd7755190..ecf0482635 100644
--- a/arch/riscv/cpu/start.S
+++ b/arch/riscv/cpu/start.S
@@ -14,75 +14,208 @@
 #include <common.h>
 #include <elf.h>
 #include <asm/encoding.h>
+#include <generated/asm-offsets.h>
 
 #ifdef CONFIG_32BIT
-#define LREG 			lw
-#define SREG 			sw
-#define REGBYTES 		4
+#define LREG			lw
+#define SREG			sw
+#define REGBYTES		4
 #define RELOC_TYPE		R_RISCV_32
 #define SYM_INDEX		0x8
 #define SYM_SIZE		0x10
 #else
-#define LREG 			ld
-#define SREG 			sd
-#define REGBYTES 		8
+#define LREG			ld
+#define SREG			sd
+#define REGBYTES		8
 #define RELOC_TYPE		R_RISCV_64
 #define SYM_INDEX		0x20
 #define SYM_SIZE		0x18
 #endif
 
-.section      .text
+.section .data
+secondary_harts_relocation_error:
+	.ascii "Relocation of secondary harts has failed, error %d\n"
+
+.section .text
 .globl _start
 _start:
-	j handle_reset
+#if CONFIG_IS_ENABLED(RISCV_MMODE)
+	csrr	a0, CSR_MHARTID
+#endif
 
-nmi_vector:
-	j nmi_vector
+	/* save hart id and dtb pointer */
+	mv	tp, a0
+	mv	s1, a1
 
-trap_vector:
-	j trap_entry
+	la	t0, trap_entry
+	csrw	MODE_PREFIX(tvec), t0
 
-.global trap_entry
-handle_reset:
-	li t0, CONFIG_SYS_SDRAM_BASE
-	SREG a2, 0(t0)
-	la t0, trap_entry
-	csrw mtvec, t0
-	csrwi mstatus, 0
-	csrwi mie, 0
+	/* mask all interrupts */
+	csrw	MODE_PREFIX(ie), zero
 
-/*
- * Do CPU critical regs init only at reboot,
- * not when booting from ram
- */
-#ifdef CONFIG_INIT_CRITICAL
-	jal cpu_init_crit	/* Do CPU critical regs init */
+#ifdef CONFIG_SMP
+	/* check if hart is within range */
+	/* tp: hart id */
+	li	t0, CONFIG_NR_CPUS
+	bge	tp, t0, hart_out_of_bounds_loop
+
+	/* set xSIE bit to receive IPIs */
+#if CONFIG_IS_ENABLED(RISCV_MMODE)
+	li	t0, MIE_MSIE
+#else
+	li	t0, SIE_SSIE
+#endif
+	csrs	MODE_PREFIX(ie), t0
 #endif
 
 /*
  * Set stackpointer in internal/ex RAM to call board_init_f
  */
 call_board_init_f:
-	li  t0, -16
-	li  t1, CONFIG_SYS_INIT_SP_ADDR
-	and sp, t1, t0	/* force 16 byte alignment */
-
-#ifdef CONFIG_DEBUG_UART
-	jal	debug_uart_init
+	li	t0, -16
+#if defined(CONFIG_SPL_BUILD) && defined(CONFIG_SPL_STACK)
+	li	t1, CONFIG_SPL_STACK
+#else
+	li	t1, CONFIG_SYS_INIT_SP_ADDR
 #endif
+	and	sp, t1, t0		/* force 16 byte alignment */
 
 call_board_init_f_0:
 	mv	a0, sp
 	jal	board_init_f_alloc_reserve
+
+	/*
+	 * Set global data pointer here for all harts, uninitialized at this
+	 * point.
+	 */
+	mv	gp, a0
+
+	/* setup stack */
+#ifdef CONFIG_SMP
+	/* tp: hart id */
+	slli	t0, tp, CONFIG_STACK_SIZE_SHIFT
+	sub	sp, a0, t0
+#else
 	mv	sp, a0
+#endif
+
+#ifndef CONFIG_XIP
+	/*
+	 * Pick hart to initialize global data and run U-Boot. The other harts
+	 * wait for initialization to complete.
+	 */
+	la	t0, hart_lottery
+	li	s2, 1
+	amoswap.w s2, t1, 0(t0)
+	bnez	s2, wait_for_gd_init
+#else
+	bnez	tp, secondary_hart_loop
+#endif
+
+#ifdef CONFIG_OF_PRIOR_STAGE
+	la	t0, prior_stage_fdt_address
+	SREG	s1, 0(t0)
+#endif
+
 	jal	board_init_f_init_reserve
 
-	mv  a0, zero	/* a0 <-- boot_flags = 0 */
-	la t5, board_init_f
-	jr t5		/* jump to board_init_f() */
+	/* save the boot hart id to global_data */
+	SREG	tp, GD_BOOT_HART(gp)
+
+#ifndef CONFIG_XIP
+	la	t0, available_harts_lock
+	fence	rw, w
+	amoswap.w zero, zero, 0(t0)
+
+wait_for_gd_init:
+	la	t0, available_harts_lock
+	li	t1, 1
+1:	amoswap.w t1, t1, 0(t0)
+	fence	r, rw
+	bnez	t1, 1b
+
+	/* register available harts in the available_harts mask */
+	li	t1, 1
+	sll	t1, t1, tp
+	LREG	t2, GD_AVAILABLE_HARTS(gp)
+	or	t2, t2, t1
+	SREG	t2, GD_AVAILABLE_HARTS(gp)
+
+	fence	rw, w
+	amoswap.w zero, zero, 0(t0)
+
+	/*
+	 * Continue on hart lottery winner, others branch to
+	 * secondary_hart_loop.
+	 */
+	bnez	s2, secondary_hart_loop
+#endif
+
+	/* Enable cache */
+	jal	icache_enable
+	jal	dcache_enable
+
+#ifdef CONFIG_DEBUG_UART
+	jal	debug_uart_init
+#endif
+
+	mv	a0, zero		/* a0 <-- boot_flags = 0 */
+	la	t5, board_init_f
+	jalr	t5			/* jump to board_init_f() */
+
+#ifdef CONFIG_SPL_BUILD
+spl_clear_bss:
+	la	t0, __bss_start
+	la	t1, __bss_end
+	beq	t0, t1, spl_stack_gd_setup
+
+spl_clear_bss_loop:
+	SREG	zero, 0(t0)
+	addi	t0, t0, REGBYTES
+	blt	t0, t1, spl_clear_bss_loop
+
+spl_stack_gd_setup:
+	jal	spl_relocate_stack_gd
+
+	/* skip setup if we did not relocate */
+	beqz	a0, spl_call_board_init_r
+	mv	s0, a0
+
+	/* setup stack on main hart */
+#ifdef CONFIG_SMP
+	/* tp: hart id */
+	slli	t0, tp, CONFIG_STACK_SIZE_SHIFT
+	sub	sp, s0, t0
+#else
+	mv	sp, s0
+#endif
+
+	/* set new stack and global data pointer on secondary harts */
+spl_secondary_hart_stack_gd_setup:
+	la	a0, secondary_hart_relocate
+	mv	a1, s0
+	mv	a2, s0
+	mv	a3, zero
+	jal	smp_call_function
+
+	/* hang if relocation of secondary harts has failed */
+	beqz	a0, 1f
+	mv	a1, a0
+	la	a0, secondary_harts_relocation_error
+	jal	printf
+	jal	hang
+
+	/* set new global data pointer on main hart */
+1:	mv	gp, s0
+
+spl_call_board_init_r:
+	mv	a0, zero
+	mv	a1, zero
+	jal	board_init_r
+#endif
 
 /*
- * void relocate_code (addr_sp, gd, addr_moni)
+ * void relocate_code(addr_sp, gd, addr_moni)
  *
  * This "function" does not return, instead it continues in RAM
  * after relocating the monitor code.
@@ -90,203 +223,187 @@ call_board_init_f_0:
  */
 .globl relocate_code
 relocate_code:
-	mv  s2, a0	/* save addr_sp */
-	mv  s3, a1	/* save addr of gd */
-	mv  s4, a2	/* save addr of destination */
+	mv	s2, a0			/* save addr_sp */
+	mv	s3, a1			/* save addr of gd */
+	mv	s4, a2			/* save addr of destination */
 
 /*
  *Set up the stack
  */
 stack_setup:
-	mv sp, s2
-	la t0, _start
-	sub t6, s4, t0	/* t6 <- relocation offset */
-	beq t0, s4, clear_bss	/* skip relocation */
+#ifdef CONFIG_SMP
+	/* tp: hart id */
+	slli	t0, tp, CONFIG_STACK_SIZE_SHIFT
+	sub	sp, s2, t0
+#else
+	mv	sp, s2
+#endif
 
-	mv t1, s4	/* t1 <- scratch for copy_loop */
-	la t3, __bss_start
-	sub t3, t3, t0	/* t3 <- __bss_start_ofs */
-	add t2, t0, t3	/* t2 <- source end address */
+	la	t0, _start
+	sub	t6, s4, t0		/* t6 <- relocation offset */
+	beq	t0, s4, clear_bss	/* skip relocation */
+
+	mv	t1, s4			/* t1 <- scratch for copy_loop */
+	la	t3, __bss_start
+	sub	t3, t3, t0		/* t3 <- __bss_start_ofs */
+	add	t2, t0, t3		/* t2 <- source end address */
 
 copy_loop:
-	LREG t5, 0(t0)
-	addi t0, t0, REGBYTES
-	SREG t5, 0(t1)
-	addi t1, t1, REGBYTES
-	blt t0, t2, copy_loop
+	LREG	t5, 0(t0)
+	addi	t0, t0, REGBYTES
+	SREG	t5, 0(t1)
+	addi	t1, t1, REGBYTES
+	blt	t0, t2, copy_loop
 
 /*
  * Update dynamic relocations after board_init_f
  */
 fix_rela_dyn:
-	la  t1, __rel_dyn_start
-	la  t2, __rel_dyn_end
-	beq t1, t2, clear_bss
-	add t1, t1, t6			/* t1 <- rela_dyn_start in RAM */
-	add t2, t2, t6			/* t2 <- rela_dyn_end in RAM */
+	la	t1, __rel_dyn_start
+	la	t2, __rel_dyn_end
+	beq	t1, t2, clear_bss
+	add	t1, t1, t6		/* t1 <- rela_dyn_start in RAM */
+	add	t2, t2, t6		/* t2 <- rela_dyn_end in RAM */
 
 /*
  * skip first reserved entry: address, type, addend
  */
-	bne t1, t2, 7f
+	j	10f
 
 6:
-	LREG  t5, -(REGBYTES*2)(t1)	/* t5 <-- relocation info:type */
-	li  t3, R_RISCV_RELATIVE	/* reloc type R_RISCV_RELATIVE */
-	bne t5, t3, 8f			/* skip non-RISCV_RELOC entries */
-	LREG t3, -(REGBYTES*3)(t1)
-	LREG t5, -(REGBYTES)(t1)	/* t5 <-- addend */
-	add t5, t5, t6			/* t5 <-- location to fix up in RAM */
-	add t3, t3, t6			/* t3 <-- location to fix up in RAM */
-	SREG t5, 0(t3)
-7:
-	addi t1, t1, (REGBYTES*3)
-	ble t1, t2, 6b
+	LREG	t5, -(REGBYTES*2)(t1)	/* t5 <-- relocation info:type */
+	li	t3, R_RISCV_RELATIVE	/* reloc type R_RISCV_RELATIVE */
+	bne	t5, t3, 8f		/* skip non-RISCV_RELOC entries */
+	LREG	t3, -(REGBYTES*3)(t1)
+	LREG	t5, -(REGBYTES)(t1)	/* t5 <-- addend */
+	add	t5, t5, t6		/* t5 <-- location to fix up in RAM */
+	add	t3, t3, t6		/* t3 <-- location to fix up in RAM */
+	SREG	t5, 0(t3)
+	j	10f
 
 8:
-	la  t4, __dyn_sym_start
-	add t4, t4, t6
+	la	t4, __dyn_sym_start
+	add	t4, t4, t6
 
 9:
-	LREG  t5, -(REGBYTES*2)(t1)	/* t5 <-- relocation info:type */
-	srli t0, t5, SYM_INDEX		/* t0 <--- sym table index */
-	andi t5, t5, 0xFF		/* t5 <--- relocation type */
-	li  t3, RELOC_TYPE
-	bne t5, t3, 10f 		/* skip non-addned entries */
-
-	LREG t3, -(REGBYTES*3)(t1)
-	li t5, SYM_SIZE
-	mul t0, t0, t5
-	add s1, t4, t0
-	LREG t5, REGBYTES(s1)
-	add t5, t5, t6			/* t5 <-- location to fix up in RAM */
-	add t3, t3, t6			/* t3 <-- location to fix up in RAM */
-	SREG t5, 0(t3)
+	LREG	t5, -(REGBYTES*2)(t1)	/* t5 <-- relocation info:type */
+	srli	t0, t5, SYM_INDEX	/* t0 <--- sym table index */
+	andi	t5, t5, 0xFF		/* t5 <--- relocation type */
+	li	t3, RELOC_TYPE
+	bne	t5, t3, 10f		/* skip non-addned entries */
+
+	LREG	t3, -(REGBYTES*3)(t1)
+	li	t5, SYM_SIZE
+	mul	t0, t0, t5
+	add	s5, t4, t0
+	LREG	t0, -(REGBYTES)(t1)	/* t0 <-- addend */
+	LREG	t5, REGBYTES(s5)
+	add	t5, t5, t0
+	add	t5, t5, t6		/* t5 <-- location to fix up in RAM */
+	add	t3, t3, t6		/* t3 <-- location to fix up in RAM */
+	SREG	t5, 0(t3)
 10:
-	addi t1, t1, (REGBYTES*3)
-	ble t1, t2, 9b
+	addi	t1, t1, (REGBYTES*3)
+	ble	t1, t2, 6b
 
 /*
  * trap update
 */
-	la t0, trap_entry
-	add t0, t0, t6
-	csrw mtvec, t0
+	la	t0, trap_entry
+	add	t0, t0, t6
+	csrw	MODE_PREFIX(tvec), t0
 
 clear_bss:
-	la t0, __bss_start		/* t0 <- rel __bss_start in FLASH */
-	add t0, t0, t6			/* t0 <- rel __bss_start in RAM */
-	la t1, __bss_end		/* t1 <- rel __bss_end in FLASH */
-	add t1, t1, t6			/* t1 <- rel __bss_end in RAM */
-	li t2, 0x00000000		/* clear */
-	beq t0, t1, call_board_init_r
+	la	t0, __bss_start		/* t0 <- rel __bss_start in FLASH */
+	add	t0, t0, t6		/* t0 <- rel __bss_start in RAM */
+	la	t1, __bss_end		/* t1 <- rel __bss_end in FLASH */
+	add	t1, t1, t6		/* t1 <- rel __bss_end in RAM */
+	beq	t0, t1, relocate_secondary_harts
 
 clbss_l:
-	SREG t2, 0(t0)			/* clear loop... */
-	addi t0, t0, REGBYTES
-	bne t0, t1, clbss_l
+	SREG	zero, 0(t0)		/* clear loop... */
+	addi	t0, t0, REGBYTES
+	blt	t0, t1, clbss_l
+
+relocate_secondary_harts:
+#ifdef CONFIG_SMP
+	/* send relocation IPI */
+	la	t0, secondary_hart_relocate
+	add	a0, t0, t6
+
+	/* store relocation offset */
+	mv	s5, t6
+
+	mv	a1, s2
+	mv	a2, s3
+	mv	a3, zero
+	jal	smp_call_function
+
+	/* hang if relocation of secondary harts has failed */
+	beqz	a0, 1f
+	mv	a1, a0
+	la	a0, secondary_harts_relocation_error
+	jal	printf
+	jal	hang
+
+	/* restore relocation offset */
+1:	mv	t6, s5
+#endif
 
 /*
  * We are done. Do not return, instead branch to second part of board
  * initialization, now running from RAM.
  */
 call_board_init_r:
-	la t0, board_init_r
-	mv t4, t0			/* offset of board_init_r() */
-	add t4, t4, t6			/* real address of board_init_r() */
+	jal	invalidate_icache_all
+	jal	flush_dcache_all
+	la	t0, board_init_r        /* offset of board_init_r() */
+	add	t4, t0, t6		/* real address of board_init_r() */
 /*
  * setup parameters for board_init_r
  */
-	mv a0, s3			/* gd_t */
-	mv a1, s4			/* dest_addr */
+	mv	a0, s3			/* gd_t */
+	mv	a1, s4			/* dest_addr */
 
 /*
  * jump to it ...
  */
-	jr t4				/* jump to board_init_r() */
+	jr	t4			/* jump to board_init_r() */
 
-/*
- * trap entry
- */
-trap_entry:
-	addi sp, sp, -32*REGBYTES
-	SREG x1, 1*REGBYTES(sp)
-	SREG x2, 2*REGBYTES(sp)
-	SREG x3, 3*REGBYTES(sp)
-	SREG x4, 4*REGBYTES(sp)
-	SREG x5, 5*REGBYTES(sp)
-	SREG x6, 6*REGBYTES(sp)
-	SREG x7, 7*REGBYTES(sp)
-	SREG x8, 8*REGBYTES(sp)
-	SREG x9, 9*REGBYTES(sp)
-	SREG x10, 10*REGBYTES(sp)
-	SREG x11, 11*REGBYTES(sp)
-	SREG x12, 12*REGBYTES(sp)
-	SREG x13, 13*REGBYTES(sp)
-	SREG x14, 14*REGBYTES(sp)
-	SREG x15, 15*REGBYTES(sp)
-	SREG x16, 16*REGBYTES(sp)
-	SREG x17, 17*REGBYTES(sp)
-	SREG x18, 18*REGBYTES(sp)
-	SREG x19, 19*REGBYTES(sp)
-	SREG x20, 20*REGBYTES(sp)
-	SREG x21, 21*REGBYTES(sp)
-	SREG x22, 22*REGBYTES(sp)
-	SREG x23, 23*REGBYTES(sp)
-	SREG x24, 24*REGBYTES(sp)
-	SREG x25, 25*REGBYTES(sp)
-	SREG x26, 26*REGBYTES(sp)
-	SREG x27, 27*REGBYTES(sp)
-	SREG x28, 28*REGBYTES(sp)
-	SREG x29, 29*REGBYTES(sp)
-	SREG x30, 30*REGBYTES(sp)
-	SREG x31, 31*REGBYTES(sp)
-	csrr a0, mcause
-	csrr a1, mepc
-	mv a2, sp
-	jal handle_trap
-	csrw mepc, a0
+#ifdef CONFIG_SMP
+hart_out_of_bounds_loop:
+	/* Harts in this loop are out of bounds, increase CONFIG_NR_CPUS. */
+	wfi
+	j	hart_out_of_bounds_loop
 
-/*
- * Remain in M-mode after mret
- */
-	li t0, MSTATUS_MPP
-	csrs mstatus, t0
-	LREG x1, 1*REGBYTES(sp)
-	LREG x2, 2*REGBYTES(sp)
-	LREG x3, 3*REGBYTES(sp)
-	LREG x4, 4*REGBYTES(sp)
-	LREG x5, 5*REGBYTES(sp)
-	LREG x6, 6*REGBYTES(sp)
-	LREG x7, 7*REGBYTES(sp)
-	LREG x8, 8*REGBYTES(sp)
-	LREG x9, 9*REGBYTES(sp)
-	LREG x10, 10*REGBYTES(sp)
-	LREG x11, 11*REGBYTES(sp)
-	LREG x12, 12*REGBYTES(sp)
-	LREG x13, 13*REGBYTES(sp)
-	LREG x14, 14*REGBYTES(sp)
-	LREG x15, 15*REGBYTES(sp)
-	LREG x16, 16*REGBYTES(sp)
-	LREG x17, 17*REGBYTES(sp)
-	LREG x18, 18*REGBYTES(sp)
-	LREG x19, 19*REGBYTES(sp)
-	LREG x20, 20*REGBYTES(sp)
-	LREG x21, 21*REGBYTES(sp)
-	LREG x22, 22*REGBYTES(sp)
-	LREG x23, 23*REGBYTES(sp)
-	LREG x24, 24*REGBYTES(sp)
-	LREG x25, 25*REGBYTES(sp)
-	LREG x26, 26*REGBYTES(sp)
-	LREG x27, 27*REGBYTES(sp)
-	LREG x28, 28*REGBYTES(sp)
-	LREG x29, 29*REGBYTES(sp)
-	LREG x30, 30*REGBYTES(sp)
-	LREG x31, 31*REGBYTES(sp)
-	addi sp, sp, 32*REGBYTES
-	mret
-
-#ifdef CONFIG_INIT_CRITICAL
-cpu_init_crit:
-    ret
+/* SMP relocation entry */
+secondary_hart_relocate:
+	/* a1: new sp */
+	/* a2: new gd */
+	/* tp: hart id */
+
+	/* setup stack */
+	slli	t0, tp, CONFIG_STACK_SIZE_SHIFT
+	sub	sp, a1, t0
+
+	/* update global data pointer */
+	mv	gp, a2
+#endif
+
+secondary_hart_loop:
+	wfi
+
+#ifdef CONFIG_SMP
+	csrr	t0, MODE_PREFIX(ip)
+#if CONFIG_IS_ENABLED(RISCV_MMODE)
+	andi	t0, t0, MIE_MSIE
+#else
+	andi	t0, t0, SIE_SSIE
+#endif
+	beqz	t0, secondary_hart_loop
+
+	mv	a0, tp
+	jal	handle_ipi
 #endif
+
+	j	secondary_hart_loop