target/rx/cpu: Use address_space_ldl() to read reset vector address
PATCH 1/1 target/rx/cpu: Use address_space_ldl() to read reset vector address
From: Philippe Mathieu-Daudé <[email protected]>
The RX code flash is not a Masked ROM but a EEPROM (electrically erasable programmable flash memory).
When implementing the flash hardware, the rom_ptr() returns NULL and the reset vector is not set.
Instead, use the address_space ld/st API to fetch the reset vector address from the code flash.
---
Based-on: <[email protected]>
Same issue might occurs in Cortex-M arm_cpu_reset()
--- target/rx/cpu.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)
target/rx/cpu.c +4 -6
@@ -26,6 +26,8 @@#include "hw/loader.h" #include "fpu/softfloat.h" +#define CPU_RESET_VECTOR 0xfffffffc +static void rx_cpu_set_pc (CPUState *cs,vaddr value) {RXCPU *cpu =RXCPU (cs);
@@ -51,17 +53,13 @@ static void rx_cpu_reset(CPUState *s)RXCPU *cpu =RXCPU (s);RXCPUClass *rcc =RXCPU_GET_CLASS (cpu);CPURXState *env = &cpu->env; -uint32_t *resetvec; rcc->parent_reset (s);memset (env,0 , offsetof(CPURXState , end_reset_fields)); - resetvec =rom_ptr (0xfffffffc ,4 ); -if (resetvec) { -/* In the case of kernel, it is ignored because it is not set. */ - env->pc =ldl_p (resetvec); - } + env->pc =address_space_ldl (cpu_get_address_space (s,0 ), +CPU_RESET_VECTOR ,MEMTXATTRS_UNSPECIFIED ,NULL );rx_cpu_unpack_psw (env,0 ,1 ); env->regs[0 ] = env->isp = env->usp =0 ; env->fpsw =0 ;
Signature · 1 line
2.21.1
Quoted 1 line · Philippe Mathieu-Daudé
From: Philippe Mathieu-Daudé <[email protected]>
Hmm author email should be <[email protected]>...
Quoted 51 lines
The RX code flash is not a Masked ROM but a EEPROM (electrically erasable programmable flash memory).
When implementing the flash hardware, therom_ptr()returns NULL and the reset vector is not set.
Instead, use theaddress_spaceld/st API to fetch the reset vector address from the code flash.Signed-off-by Philippe Mathieu-Daudé <[email protected]>---
Based-on: <[email protected]>Same issue might occurs in Cortex-M
arm_cpu_reset()--- target/rx/cpu.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)target/rx/cpu.c +4
-6@@ -26,6 +26,8 @@#include "hw/loader.h" #include "fpu/softfloat.h" +#define CPU_RESET_VECTOR 0xfffffffc +static void rx_cpu_set_pc (CPUState *cs,vaddr value) {RXCPU *cpu =RXCPU (cs);@@ -51,17 +53,13 @@ static void rx_cpu_reset(CPUState *s)RXCPU *cpu =RXCPU (s);RXCPUClass *rcc =RXCPU_GET_CLASS (cpu);CPURXState *env = &cpu->env; -uint32_t *resetvec; rcc->parent_reset (s);memset (env,0 , offsetof(CPURXState , end_reset_fields)); - resetvec =rom_ptr (0xfffffffc ,4 ); -if (resetvec) { -/* In the case of kernel, it is ignored because it is not set. */ - env->pc =ldl_p (resetvec); - } + env->pc =address_space_ldl (cpu_get_address_space (s,0 ), +CPU_RESET_VECTOR ,MEMTXATTRS_UNSPECIFIED ,NULL );rx_cpu_unpack_psw (env,0 ,1 ); env->regs[0 ] = env->isp = env->usp =0 ; env->fpsw =0 ;
Quoted 9 lines · Philippe Mathieu-Daudé
From: Philippe Mathieu-Daudé <[email protected]>
The RX code flash is not a Masked ROM but a EEPROM (electrically erasable programmable flash memory).
When implementing the flash hardware, therom_ptr()returns NULL and the reset vector is not set.
Instead, use theaddress_spaceld/st API to fetch the reset vector address from the code flash.
Quoted 8 lines
Signed-off-by Philippe Mathieu-Daudé <[email protected]>---
Based-on: <[email protected]>Same issue might occurs in Cortex-M
arm_cpu_reset()--- target/rx/cpu.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)
r~
Quoted 10 lines · Philippe Mathieu-Daudé
From: Philippe Mathieu-Daudé <[email protected]>
The RX code flash is not a Masked ROM but a EEPROM (electrically erasable programmable flash memory).
When implementing the flash hardware, therom_ptr()returns NULL and the reset vector is not set.
Instead, use theaddress_spaceld/st API to fetch the reset vector address from the code flash.
Quoted 5 lines
Signed-off-by Philippe Mathieu-Daudé <[email protected]>---
Based-on: <[email protected]>Same issue might occurs in Cortex-M
arm_cpu_reset()
rom_ptr() does not mean "I'm trying to get this from ROM", it means "I'm trying to get this from a user-supplied ELF file or similar which hasn't been loaded into guest memory yet". (This is a workaround for a reset ordering issue where CPU reset happens before rom_reset() runs.)
Removing the usage of rom_ptr() altogether here doesn't look right -- have you tested the case where the initial reset vector contents are provided via -kernel or -device loader,... ?
thanks
-- PMM