fix(deps): pin supraseal-c2/sppark to pre-regression versions

[rvagg]

sppark 0.1.12+ leaks GPU memory under FFI_USE_CUDA_SUPRASEAL=1, OOMing long-lived lotus-miner/lotus-worker processes. Hold supraseal-c2 at 0.1.1 and sppark at 0.1.11 until upstream ships a fix.

Ref: filecoin-project/lotus#13634, supranational/sppark#81

---

We don't necessarily have to pull the trigger on this, but we can if we don't get movement upstream and/or this causes problems for SPs and they really need it fixed ASAP.

[Copilot]

Pull request overview

Pins the CUDA SupraSeal dependency chain to known-good versions to avoid a reported GPU memory leak in sppark when building/running with FFI_USE_CUDA_SUPRASEAL=1, protecting long-lived Lotus miner/worker processes from OOM.

Changes:

• Add explicit version pins for supraseal-c2 (=0.1.1) and sppark (=0.1.11) in rust/Cargo.toml.
• Update rust/Cargo.lock to resolve to the pinned supraseal-c2/sppark versions.

Reviewed changes

Copilot reviewed 1 out of 2 changed files in this pull request and generated 1 comment.

File	Description
rust/Cargo.toml	Introduces manifest-level pins intended to hold the SupraSeal GPU stack at pre-regression versions.
rust/Cargo.lock	Updates the resolved dependency graph to use the pinned supraseal-c2/sppark versions (and includes an additional unrelated resolution change).

---

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

[rjan90]

Pinged in stack-leads channel asking if anyone has contacts over at supranational. That said, we can also consider shipping a release with the nv29 in the short-term: #580 and bump version again once fixed upstream

[rvagg]

Got some engagement @ supranational/sppark#81 (comment), will wait and see, interesting point made there but the net result might be that we essentially go down the revert path for that change anyway and end up somewhere close to what this PR is doing. We can wait though unless SPs start to express urgency about this.

[rvagg]

supranational/sppark@01e723f original change got reverted, so we're in a similar state upstream as we would be here with a revert, will wait for a bit to see if we can get a release for it

[rjan90]

Closing this PR since we got a release for it, which was updated in: #582