Fix another segfault in the partition deletion logic

[tobim]

📔 Description

This time the index forgot to check the validity of a requested partition chunk. The relevant error handling below was also slightly modified to log the cause of failed mmap(2)s and prevent similar problems in the future.

📝 Checklist

• All user-facing changes have changelog entries.
• The PR description contains instructions for the reviewer, if necessary.

🎯 Review Instructions

By commit.

[lava]

The fix looks correct; a unit test might have been nice but also annoying to write this since ultimately requires a system error to trigger and results in a segfault which isn't really testable so our testbed testing should suffice.

[JoeLoser]

The fix looks correct; a unit test might have been nice but also annoying to write this since ultimately requires a system error to trigger and results in a segfault which isn't really testable so our testbed testing should suffice.

I'd like to see a unit test too. Perhaps we need to build out some testing infrastructure to be able to fake out the mmap call and have it return an error so we can test this code. I'm OK with that coming as a follow-up after this release though given the timing of things.

[tobim]

The fix looks correct; a unit test might have been nice but also annoying to write this since ultimately requires a system error to trigger and results in a segfault which isn't really testable so our testbed testing should suffice.

I'd like to see a unit test too. Perhaps we need to build out some testing infrastructure to be able to fake out the mmap call and have it return an error so we can test this code. I'm OK with that coming as a follow-up after this release though given the timing of things.

I think it shouldn't actually be that hard, we can just mock the filesystem, but we'll have to delay that until after the release.