Improvements to the DTO_WAIT_METHODS

[byrnedj]

1. use busypoll and autotune as defaults
2. add tpause instruction
3. change to use compiler intrinsics for UMWAIT and TPAUSE
4. update the documentation for using UMWAIT
5. clean up wait_method paths
6. update min size to 32K

[Copilot]

Pull Request Overview

This PR improves the DTO wait methods by updating default wait behaviors, adding support for the TPAUSE instruction, and updating intrinsics and documentation accordingly.

• Change default wait method from yield to busypoll
• Introduce TPAUSE wait method and update UMWAIT intrinsics usage
• Update documentation in README and add build flag changes in the Makefile

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
dto.c	Updates wait methods, uses compiler intrinsics, and adds TPAUSE support.
README.md	Updates documentation for new wait methods including TPAUSE.
Makefile	Adds the -mwaitpkg flag needed for the new intrinsics.

Comments suppressed due to low confidence (2)

dto.c:394

• There's an inconsistency in the TPAUSE wake-up state between 'dsa_wait_tpause' (using C02_STATE) and the TPAUSE branch in '__dsa_wait' (using C01_STATE). Consider aligning these values or documenting the rationale for the difference.

_tpause( C01_STATE, _rdtsc() + TPAUSE_C01_DELAY_NS);

dto.c:1148

• The condition for enabling the TPAUSE wait method relies on the 'umwait_support' flag. If TPAUSE has separate hardware or library support requirements, consider verifying TPAUSE support explicitly.

if (!strncmp(env_str, wait_names[WAIT_TPAUSE], strlen(wait_names[WAIT_TPAUSE]))) {

[jsydir]

Is tpause supported if umwait is supported? Should there be a separate check for tpause support?

[jsydir]

This is in reference to line 1149