nuttx/esp32s3: apply ibus/dbus adjustment to internal ram 1 as well #3421
Conversation
| @@ -19,6 +19,11 @@ | |||
| #define IRAM0_CACHE_ADDRESS_HIGH 0x44000000 | |||
| #define IRAM_ATTR locate_data(".iram1") | |||
|
|
|||
| #define INTERNAL_SRAM_1_DBUS_ADDRESS_LOW 0x3fc88000 | |||
| #define INTERNAL_SRAM_1_DBUS_ADDRESS_HIGH 0x3fcf0000 | |||
There was a problem hiding this comment.
INTERNAL_SRAM_1_DBUS_ADDRESS_HIGH is not used?
There was a problem hiding this comment.
right. it's just for completeness.
| @@ -182,6 +187,11 @@ os_get_dbus_mirror(void *ibus) | |||
| if (in_ibus_ext(ibus)) { | |||
| return (void *)((uint8 *)ibus - MEM_DUAL_BUS_OFFSET); | |||
| } | |||
| else if (INTERNAL_SRAM_1_IBUS_ADDRESS_LOW <= (uintptr_t)ibus | |||
| && (uintptr_t)ibus < INTERNAL_SRAM_1_IBUS_ADDRESS_HIGH) { | |||
| return (void *)((uintptr_t)ibus - INTERNAL_SRAM_1_IBUS_ADDRESS_LOW | |||
There was a problem hiding this comment.
do you need special handling like this for the mirrored memmory allocated by os_mmap and os_mumap?
There was a problem hiding this comment.
yes.
for some configurations, os_mmap can return memory from the internal ram 1.
There was a problem hiding this comment.
Will the special handling be implemented in this PR?
There was a problem hiding this comment.
sorry, which special handling are you talking about?
There was a problem hiding this comment.
Can we move out these arch specific impls from common place?
There was a problem hiding this comment.
sorry, which special handling are you talking about?
I mean if the os_mmap returns memory from the internal ram1, does that also need convert it into its mirrored addr once the feature enabled?
There was a problem hiding this comment.
up_textheap_memalign already returns i-bus address.
i have no idea about the WASM_MEM_DUAL_BUS_MIRROR block in os_mmap.
to me, it seems like a dead code because CONFIG_ARCH_USE_TEXT_HEAP is always set for
esp32s3.
i guess it's different for the private fork of nuttx?
There was a problem hiding this comment.
Can we move out these arch specific impls from common place?
i guess it can be eventually moved to nuttx.
There was a problem hiding this comment.
to me, it seems like a dead code because CONFIG_ARCH_USE_TEXT_HEAP is always set for
OK, agree to refactor this into nuttx.
|
LGTM |
No description provided.