-----Original Message----- From: U-Boot u-boot-bounces@lists.denx.de On Behalf Of Auer, Lukas Sent: Sunday, March 10, 2019 11:42 PM To: anup@brainfault.org Cc: rickchen36@gmail.com; baruch@tkos.co.il; sr@denx.de; cmchen@andestech.com; greentime@andestech.com; schwab@suse.de; palmer@sifive.com; agraf@suse.de; u-boot@lists.denx.de; kito@andestech.com Subject: Re: [U-Boot] [PATCH 5/7] riscv: add support for multi-hart systems
On Sun, 2019-03-10 at 20:24 +0530, Anup Patel wrote:
On Sun, Mar 10, 2019 at 7:28 PM Auer, Lukas lukas.auer@aisec.fraunhofer.de wrote:
Hi Rick,
On Thu, 2019-03-07 at 17:30 +0800, Rick Chen wrote:
Hi Lukas
From: Lukas Auer [mailto:lukas.auer@aisec.fraunhofer.de] Sent: Tuesday, February 12, 2019 6:14 AM To: u-boot@lists.denx.de Cc: Atish Patra; Anup Patel; Bin Meng; Andreas Schwab; Palmer Dabbelt; Alexander Graf; Lukas Auer; Anup Patel; Rick Jian-Zhi Chen(陳建志); Baruch Siach; Stefan Roese Subject: [PATCH 5/7] riscv: add support for multi-hart systems
On RISC-V, all harts boot independently. To be able to run on a multi-hart system, U-Boot must be extended with the functionality to manage all harts in the system. A new config option, CONFIG_MAIN_HART, is used to select the hart U-Boot runs on. All other harts are halted. U-Boot can delegate functions to them using smp_call_function().
Every hart has a valid pointer to the global data structure and a 8KiB stack by default. The stack size is set with CONFIG_STACK_SIZE_SHIFT.
Signed-off-by: Lukas Auer lukas.auer@aisec.fraunhofer.de
arch/riscv/Kconfig | 12 +++++ arch/riscv/cpu/start.S | 102 ++++++++++++++++++++++++++++++++++- arch/riscv/include/asm/csr.h | 1 + 3 files changed, 114 insertions(+), 1 deletion(-)
diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig index 3a51339c4d..af8d0f8d67 100644 --- a/arch/riscv/Kconfig +++ b/arch/riscv/Kconfig @@ -140,4 +140,16 @@ config SBI_IPI default y if RISCV_SMODE depends on SMP
+config MAIN_HART
int "Main hart in system"default 0helpSome SoCs include harts of various sizes, some ofwhich might not
be suitable for running U-Boot. CONFIG_MAIN_HART isused to select
the hart U-Boot runs on.+config STACK_SIZE_SHIFT
intdefault 13endmenu diff --git a/arch/riscv/cpu/start.S b/arch/riscv/cpu/start.S index a30f6f7194..ce7230df37 100644 --- a/arch/riscv/cpu/start.S +++ b/arch/riscv/cpu/start.S @@ -13,6 +13,7 @@ #include <config.h> #include <common.h> #include <elf.h> +#include <asm/csr.h> #include <asm/encoding.h> #include <generated/asm-offsets.h>
@@ -45,6 +46,23 @@ _start: /* mask all interrupts */ csrw MODE_PREFIX(ie), zero
+#ifdef CONFIG_SMP
/* check if hart is within range *//* s0: hart id */li t0, CONFIG_NR_CPUSbge s0, t0, hart_out_of_bounds_loop+#endif
+#ifdef CONFIG_SMP
/* set xSIE bit to receive IPIs */ #ifdef+CONFIG_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
*/ @@ -56,7 +74,25 @@ call_board_init_f: 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
/* s0: hart id */slli t0, s0, CONFIG_STACK_SIZE_SHIFTsub sp, a0, t0+#else mv sp, a0 +#endif
/* Continue on main hart, others branch tosecondary_hart_loop */
li t0, CONFIG_MAIN_HARTbne s0, t0, secondary_hart_loop la t0, prior_stage_fdt_address SREG s1, 0(t0)@@ -95,7 +131,14 @@ relocate_code: *Set up the stack */ stack_setup: +#ifdef CONFIG_SMP
/* s0: hart id */slli t0, s0, CONFIG_STACK_SIZE_SHIFTsub sp, s2, t0+#else mv sp, s2 +#endif
la t0, _start sub t6, s4, t0 /* t6 <- relocationoffset */ beq t0, s4, clear_bss /* skip relocation */ @@ -175,13 +218,30 @@ clear_bss: 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, call_board_init_r
beq t0, t1, relocate_secondary_hartsclbss_l: SREG zero, 0(t0) /* clear loop... */ addi t0, t0, REGBYTES bne t0, t1, clbss_l
+relocate_secondary_harts: +#ifdef CONFIG_SMP
/* send relocation IPI */la t0, secondary_hart_relocateadd a0, t0, t6/* store relocation offset */mv s5, t6mv a1, s2mv a2, s3jal smp_call_function/* restore relocation offset */mv t6, s5+#endif
/*
- We are done. Do not return, instead branch to second part
of board
- initialization, now running from RAM.
@@ -202,3 +262,43 @@ call_board_init_r:
- jump to it ...
*/ jr t4 /* jump to board_init_r() */
+#ifdef CONFIG_SMP +hart_out_of_bounds_loop:
/* Harts in this loop are out of bounds, increaseCONFIG_NR_CPUS. */
wfij hart_out_of_bounds_loop+#endif
+#ifdef CONFIG_SMP +/* SMP relocation entry */ +secondary_hart_relocate:
/* a1: new sp *//* a2: new gd *//* s0: hart id *//* setup stack */slli t0, s0, CONFIG_STACK_SIZE_SHIFTsub sp, a1, t0/* update global data pointer */mv gp, a2+#endif
+secondary_hart_loop:
wfi+#ifdef CONFIG_SMP
csrr t0, MODE_PREFIX(ip)+#ifdef CONFIG_RISCV_MMODE
andi t0, t0, MIE_MSIE+#else
andi t0, t0, SIE_SSIE+#endif
beqz t0, secondary_hart_loopmv a0, s0jal handle_ipiI found that s0 maybe corrupted after execute handle_ipi. Because smp_function will be treated as a return function by compiler, so compiler will generate codes to execute restore after smp_function().
But actually it is a no-return function. So there maybe no chance to execute restore. And s0 will be corrupted somehow.
The usage of s0 in v2 flow seem the same as v1. So I reply mail in v1 patch.
Thanks for reporting this issue! What compiler version are you using? I never saw this issue with GCC 8.2.0, because optimization removes the ret following the call to smp_function(), it is called using jr. The registers are therefore restored before jumping to smp_function().
This system is probably too fragile. A simple solution would be to store the hart ID somewhere else, for example register tp. What do you think?
I think you can also use scratch/mscratch (if it is not used anywhere else).
You are right, sscratch and mscratch are also not currently used in U-Boot. Is there an advantage to use the scratch CSRs instead of tp?
No specific advantage as such. This was another possibility.
I suggest you go ahead with TP because mscratch will be used by M-mode U-Boot when linked with OpenSBI as library.
Regards, Anup