Since dd849ffc, _rodata_start label has been moved to a different
linker output section from where the TLS templates (.tdata, .tbss)
are located. Since link-time addresses of thread-local variables are
calculated relative to the section start address, this resulted in
incorrect calculation of THREADPTR/$tp registers.
Fix by introducing new linker label, _flash_rodata_start, which points
to the .flash.rodata output section where TLS variables are located,
and use it when calculating THREADPTR/$tp.
Also remove the hardcoded rodata section alignment for Xtensa targets.
Alignment of rodata can be affected by the user application, which is
the issue dd849ffc was fixing. To accommodate any possible alignment,
save it in a linker label (_flash_rodata_align) and then use when
calculating THREADPTR. Note that this is not required on RISC-V, since
this target doesn't use TPOFF.
It is now possible to have any alignment restriction on rodata in the user
applicaiton. It will not affect the first section which must be aligned
on a 16-byte bound.
Closes https://github.com/espressif/esp-idf/issues/6719
rodata dummy section has now the same alignment as flash text section,
and at least the same size. For these reasons, the cache will map
correctly the following rodata section.
When `DIS_USB_JTAG` eFuse is NOT burned (`False`), it is not possible
to set pins 18 and 19 as GPIOs. This commit solves this by manually
disabling USB JTAG when using pins 18 or 19.
The functions shall use `gpio_hal_iomux_func_sel` instead of
`PIN_FUNC_SELELECT`.
The CPU might prefetch instructions, which means it in some cases
will try to fetch instruction located after the last instruction in
flash.text.
Add dummy bytes to ensure fetching these wont result in an error,
e.g. MMU exceptions
Removed the old dynamically allocated GDMA channel approach.
It proved too unreliable as we couldn't not ensure consumers of the mbedtls
would properly free the channels after use.
Replaced by a single shared GDMA channel for AES and SHA, which won't be
released unless user specifically calls API for releasing it.
On S2 the brown out detector would occasionally trigger erroneously during deep sleep.
Disable it before sleeping to circumvent this issue.
Closes https://github.com/espressif/esp-idf/issues/6179
* Target components pull in xtensa component directly
* Use CPU HAL where applicable
* Remove unnecessary xtensa headers
* Compilation changes necessary to support non-xtensa gcc types (ie int32_t/uint32_t is no
longer signed/unsigned int).
Changes come from internal branch commit a6723fc
During HAL layer refactoring and new chip bringup, we have several
caps.h for each part, to reduce the conflicts to minimum. But this is
The capabilities headers will be relataive stable once completely
written (maybe after the featues are supported by drivers).
Now ESP32 and ESP32-S2 drivers are relative stable, making it a good
time to combine all these caps.h into one soc_caps.h
This cleanup also move HAL config and pin config into separated files,
to make the responsibilities of these headers more clear. This is
helpful for the stabilities of soc_caps.h because we want to make it
public some day.
