[skip uplift] Add missing uninits for blackhole

[vvukomanovicTT]

Ticket

Github issue

Problem description

Majority of initis in the LLK are missing unints and leaving HW in random state for the next operation.

What's changed

Second PR which implements missing BH unints. This is the second out of 3 PRs (third is for Quasar).

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update

Checklist

• All post commit CI passes
• Blackhole Post commit CI passes (if applicable)

[github-actions]

Thank you for your contribution! 🚀
If you want to run metal post-commit tests, you can add the metal-post-commit-tests label to this pull request.
📖 For more information, please refer to our CONTRIBUTING guide.

[Copilot]

Pull request overview

This PR adds missing uninitialization functions for Blackhole architecture's Low-Level Kernel (LLK) library. The PR is part of a series (second of three) addressing issue #33825, which identifies that most init functions in the LLK leave hardware in an undefined state without corresponding cleanup. These uninit functions restore hardware registers and counters to their default states after operations complete.

Key changes:

• Added uninit functions to unpack operations (untilize, matmul, reduce, and generic A/AB unpacking) that restore address counters and configuration registers
• Added empty uninit stubs for math operations that use only transient state
• Added empty uninit stub for pack operations where default state matches initialized state

Reviewed changes

Copilot reviewed 16 out of 16 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
tt_llk_blackhole/llk_lib/llk_unpack_untilize.h	Adds uninit function with conditional logic to either restore from saved GPRs or reset to default values
tt_llk_blackhole/llk_lib/llk_unpack_AB_reduce_custom.h	Updates uninit function to accept dimension parameters and restore untilizer state; includes commented code referencing fp32 accumulation requirements
tt_llk_blackhole/llk_lib/llk_unpack_AB_matmul.h	Adds uninit function to reset unpacker A and B address counters to defaults
tt_llk_blackhole/llk_lib/llk_unpack_AB.h	Adds uninit function to reset unpacker A and B address counters to defaults
tt_llk_blackhole/llk_lib/llk_unpack_A.h	Adds templated uninit function to reset unpacker counter and debug feature register
tt_llk_blackhole/llk_lib/llk_pack.h	Adds empty uninit stub noting pack_init already sets default state
tt_llk_blackhole/llk_lib/llk_math_welfords_sfpu.h	Adds empty uninit stub for Welford's algorithm SFPU operations
tt_llk_blackhole/llk_lib/llk_math_transpose_dest.h	Adds empty uninit stub for destination transpose operations
tt_llk_blackhole/llk_lib/llk_math_reduce_custom.h	Adds empty uninit stub for custom block reduce operations
tt_llk_blackhole/llk_lib/llk_math_reduce.h	Adds empty uninit stub for reduce operations
tt_llk_blackhole/llk_lib/llk_math_matmul.h	Adds empty uninit stub for matrix multiplication operations
tt_llk_blackhole/llk_lib/llk_math_eltwise_unary_sfpu.h	Adds empty uninit stub for unary SFPU element-wise operations
tt_llk_blackhole/llk_lib/llk_math_eltwise_unary_datacopy.h	Adds empty uninit stub for unary datacopy operations
tt_llk_blackhole/llk_lib/llk_math_eltwise_ternary_sfpu.h	Adds empty uninit stub for ternary SFPU element-wise operations
tt_llk_blackhole/llk_lib/llk_math_eltwise_binary_sfpu.h	Adds empty uninit stub for binary SFPU element-wise operations
tt_llk_blackhole/llk_lib/llk_math_eltwise_binary.h	Adds empty uninit stub for binary element-wise operations

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

🚀 tt-metal post-commit tests

Branch: vvukomanovic/1225-bh-uninits
Test Configuration:

• Wormhole tests: ⚪ Disabled
• Blackhole tests: 🟢 Enabled

Test Results:

• Wormhole tests: ⚪ Not run
• C++ post-commit tests: 🎯 Passed - View run
• Blackhole tests: 🎯 Passed - View run
• Device perf regressions tests: 🎯 Passed - View run

🔗 Links

📊 Post-commit workflow: #20597970904

[github-actions]

🚀 tt-metal post-commit tests

Branch: vvukomanovic/1225-bh-uninits
Test Configuration:

• Wormhole tests: ⚪ Disabled
• Blackhole tests: 🟢 Enabled

Test Results:

🔗 Links

📊 Post-commit workflow: #20617860802

[github-actions]

🚀 tt-metal post-commit tests

Branch: vvukomanovic/1225-bh-uninits
Test Configuration:

• Wormhole tests: 🟢 Enabled
• Blackhole tests: 🟢 Enabled

Test Results:

🔗 Links

📊 Post-commit workflow: #20619527396

[github-actions]

🚀 tt-metal post-commit tests

Branch: vvukomanovic/1225-bh-uninits
Test Configuration:

• Wormhole tests: 🟢 Enabled
• Blackhole tests: 🟢 Enabled

Test Results:

• Wormhole tests: 🎯 Passed - View run
• C++ post-commit tests: 🛑 Failed - View run
• Blackhole tests: 🛑 Failed - View run
• Device perf regressions tests: 🛑 Failed - View run

🔗 Links

📊 Post-commit workflow: #20624129974

[github-actions]

🚀 tt-metal post-commit tests

Branch: vvukomanovic/1225-bh-uninits
Test Configuration:

• Wormhole tests: 🟢 Enabled
• Blackhole tests: 🟢 Enabled

Test Results:

• Wormhole tests: 🚫 Workflow Cancelled - Check logs for details
• C++ post-commit tests: 🚫 Workflow Cancelled - Check logs for details
• Blackhole tests: 🚫 Workflow Cancelled - Check logs for details
• Device perf regressions tests: 🚫 Workflow Cancelled - Check logs for details

🔗 Links

📊 Post-commit workflow: #20919444135