fix: sealing: Avoid nil dereference in debug log

[magik6k]

Related Issues

Fixes #9769

Proposed Changes

Don't panic, also improve assigner debug logs

Additional Info

Before the Shark upgrade checkDealAssignable would only ever return false if the sector was an upgrade sector.

When that happened we'd try to get sector expiration for the CC update sector %d cannot fit deal... debug log, which would panic when we try that on non-upgrade sectors - which can now happen due to the new TermMax handling logic in checkDealAssignable when the sector has some deals assigned, and we have an incoming deal which can't fit into the sector.

Checklist

Before you mark the PR ready for review, please make sure that:

• Commits have a clear commit message.
• PR title is in the form of of <PR type>: <area>: <change being made>
  • example: fix: mempool: Introduce a cache for valid signatures
  • PR type: fix, feat, build, chore, ci, docs, perf, refactor, revert, style, test
  • area, e.g. api, chain, state, market, mempool, multisig, networking, paych, proving, sealing, wallet, deps
• New features have usage guidelines and / or documentation updates in
  • Lotus Documentation
  • Discussion Tutorials
• Tests exist for new functionality or change in behavior
• CI is green

[RobQuistNL]

I'm trying to manually apply this patch on our production setups as we speak

[RobQuistNL]

Tested on prod machines & fixes issue when applied to 1.18.0

[jennijuju]

thanks @RobQuistNL ! Magik has done. the same and we will try to get this reviewed and do a patch asap!

[geoff-vball]

Instead of fixing this here, I think I would rather fix GetSector() in chain/actors/builtin/miner/vX.go (these are generated files). I think we should return an error from GetSector() if we fail to find it, instead of the nil, nil that we're currently returning. It's not idiomatic and is what caused this issue in the first place.

[magik6k]

This is documented behaviour, we want some way there to indicate “sector not found” that works over jsonrpc - Im pręty sure other parts of the codebase depend on this behaviour

[geoff-vball]

Did Aayush get the typed json-rpc errors PR done? There are other parts of our code that aren't checking for nil, nil, such as warmpup.go that we should also fix if we're going to keep the behaviour the same. We should also throw comments in a couple more places just to make sure as well.

[arajasek]

This looks good to me (i would've preferred to only include the panic fix for the patch release, but I think this is correct).