Hi Bin,
On Tue, 2018-12-11 at 01:34 -0800, Bin Meng wrote:
This updates supports_extension() implementation to use the desc string from the cpu driver whenever possible, which avoids the reading of misa CSR for S-mode U-Boot.
Signed-off-by: Bin Meng bmeng.cn@gmail.com
Changes in v3:
- new patch to update supports_extension() to use desc from cpu
driver
Changes in v2: None
arch/riscv/cpu/cpu.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+)
diff --git a/arch/riscv/cpu/cpu.c b/arch/riscv/cpu/cpu.c index d3c59da..704ae70 100644 --- a/arch/riscv/cpu/cpu.c +++ b/arch/riscv/cpu/cpu.c @@ -5,8 +5,10 @@
#include <common.h> #include <cpu.h> +#include <dm.h> #include <log.h> #include <asm/csr.h> +#include <dm/uclass-internal.h>
/*
- prior_stage_fdt_address must be stored in the data section since
it is used @@ -16,7 +18,21 @@ phys_addr_t prior_stage_fdt_address __attribute__((section(".data")));
static inline bool supports_extension(char ext) { +#ifdef CONFIG_CPU
- struct udevice *dev;
- char desc[32];
- uclass_find_first_device(UCLASS_CPU, &dev);
- if (!cpu_get_desc(dev, desc, sizeof(desc))) {
/* skip the first 4 characters (rv32|rv64) */if (strchr(desc + 4, ext))return true;- }
- return false;
+#else return csr_read(misa) & (1 << (ext - 'a'));
This should not be used in supervisor mode, so that we don't need a patched BBL. I think it is better to use #elif defined(CONFIG_RISCV_MMODE) here.
There should also be a warning message to the user if supports_extension() does not have a way to determine the available extensions (supervisor-mode and no CONFIG_CPU or CONFIG_CPU, but no device found). This would help to prevent difficult-to-debug errors caused by part of the initialization missing.
Thanks, Lukas
+#endif }
static int riscv_cpu_probe(void)