code and regression

[openroad-ci]

Summary

dbSdcNetwork::findInstancesMatching1 traverses the entire design
hierarchy for every wildcard pattern, matching each instance's full
path against the pattern. On hierarchical loads with millions of
instances, a single read_sdc containing ~100k constraints like
set_false_path -from [get_cells cpu/core0/icache/*_reg] performs
O(patterns x insts) work -- tens of minutes to hours to read an SDC
that other tools load in seconds.

Anchor the DFS at the deepest literal path prefix before any
wildcard. For cpu/core0/icache/*_reg we resolve cpu/core0/icache
via the existing findInstance() hash (which already covers divider
and bracket escape variants) and only DFS under that subtree.
Patterns with a leading wildcard (*/clk) are unchanged.

Correctness

• Match semantics are preserved: path_buffer is seeded with the
  literal prefix, so pattern->match() still sees the full
  hierarchical name.
• Regex patterns skip the optimization entirely (cannot assume /
  is a literal segment boundary under regex).
• The scan honors pathEscape() so escaped dividers in a prefix
  don't misanchor.
• Flat-mode designs (read_def) can have leaf names like b1/r1
  with embedded dividers where no instance b1 exists -- the
  optimization only anchors when the resolved prefix is non-leaf,
  otherwise falls through to the legacy unanchored DFS. Caught by
  the existing sdc_names1 test during development.

Type of Change

• Bug fix
• Performance

Impact

• Hierarchical read_sdc on customer design with ~6M instances and
  ~100k constraints: hours -> seconds.
• All 62 existing dbSta regressions continue to pass.

Verification

• I have verified that the local build succeeds (./etc/Build.sh).
• I have run the relevant tests and they pass.
• My code follows the repository's formatting guidelines.
• I have signed my commits (DCO).

Related Issues

Reported slowdown on hierarchical SDC load for 6M-instance design.

[gemini-code-assist]

Code Review

This pull request optimizes hierarchical instance matching in dbSdcNetwork::findInstancesMatching1 by implementing prefix pruning. The search now identifies the deepest literal prefix in a pattern to anchor the depth-first search at a specific sub-instance rather than the design root, which improves performance for non-regex patterns. A new test case, sdc_hier_wildcard, has been added to verify this logic against various wildcard scenarios. I have no feedback to provide as there were no review comments.

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[dsengupta0628]

/gemini review

[gemini-code-assist]

Code Review

This pull request optimizes hierarchical instance matching in dbSdcNetwork::findInstancesMatching1 by anchoring the depth-first search at the deepest literal prefix of the pattern. This optimization avoids traversing the entire design hierarchy when a specific prefix is provided. Additionally, a new test case sdc_hier_wildcard was added to verify the logic. I have no feedback to provide.

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"