Config type fixes

[nmorey]

While debugging mpich 4.3.2 with ch4:ucx on s390x, several issues popped up.
Found, and tested on 1.19.0

First:

sles@ucx-debug:~/mpich/mpich-4.3.2/src/mpi/romio/test> mpirun -np 2 ./file_info -fname test
[1762529195.399132] [ucx-debug:464075:0]     ucp_context.c:2055 UCX  ERROR UCX_DYNAMIC_TL_PROGRESS_FACTOR must be > 0
[1762529195.399238] [ucx-debug:464074:0]     ucp_context.c:2055 UCX  ERROR UCX_DYNAMIC_TL_PROGRESS_FACTOR must be > 0

This fixed by the first patch. Config parser was wrongly looking for a time instead of a regular int. Not sure why I only see this on s390 though.

Second:

sles@ucx-debug:~/mpich/mpich-4.3.2/src/mpi/romio/test> ./file_info -fname test
[1762533043.448147] [ucx-debug:488518:0]            self.c:276  UCX  ERROR failed to allocate device resource

This is caused by UCT/SELF scaning the option to a uint while it is a size_t in memory.

Summary by CodeRabbit

• Chores
  • Updated configuration parameter type handling to improve internal parsing and validation consistency.

[coderabbitai]

Walkthrough

Configuration parameter types were modified in two transport modules. The UCP core's DYNAMIC_TL_PROGRESS_FACTOR changed from time units to unsigned integer type. The UCT self transport's NUM_DEVICES changed from signed integer to unsigned long type. These alterations affect how configuration values are parsed and validated during initialization.

Changes

Cohort / File(s)	Summary
UCP Core Configuration src/ucp/core/ucp_context.c	Changed DYNAMIC_TL_PROGRESS_FACTOR config entry type from UCS_CONFIG_TYPE_TIME_UNITS to UCS_CONFIG_TYPE_UINT
UCT Self Transport Configuration src/uct/sm/self/self.c	Changed NUM_DEVICES config entry type from UCS_CONFIG_TYPE_INT to UCS_CONFIG_TYPE_ULONG

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10–15 minutes

• Verify that config type changes align with parameter semantics and intended value ranges
• Confirm downstream code consuming these configuration values remains compatible with the new types
• Check that default values and validation constraints are appropriate for the new types

Poem

🐰 From time units to integers plain,
Config types dance through the lane,
Unsigned long and uint so keen,
Values parse in a different scene,
Transport flows now smooth and light!

Pre-merge checks and finishing touches

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'Config type fixes' directly summarizes the main changes: correcting configuration type declarations in two files (ucp_context.c and self.c).
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

📜 Recent review details

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between e29e5f5 and da4da6d.

📒 Files selected for processing (2)

• src/ucp/core/ucp_context.c (1 hunks)
• src/uct/sm/self/self.c (1 hunks)

🔇 Additional comments (2)

src/uct/sm/self/self.c (1)

59-60: LGTM! Type correction fixes allocation error.

The change from UCS_CONFIG_TYPE_INT to UCS_CONFIG_TYPE_ULONG correctly aligns the configuration parser with the in-memory size_t representation, resolving the "failed to allocate device resource" error described in the PR.

src/ucp/core/ucp_context.c (1)

444-448: LGTM! Correct type fixes parser error.

The change from UCS_CONFIG_TYPE_TIME_UNITS to UCS_CONFIG_TYPE_UINT is correct. The configuration entry represents a numeric factor/count ("Number of usage tracker rounds"), not a time duration. This fixes the runtime parser error reported in the PR.

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}