core/txpool/blobpool: avoid possible zero index panic

[mask-pp]

This situation(len(txs) == 0) rarely occurs, but if it does, it will panic.

[holiman]

One minor nit, otherwise this looks like a correct fix to me!

[holiman]

Thanks!

[minh-bq]

In my opinion, it's impossible to cause the panic. The panic only happens when all transactions in txs has nonce lower than next. But if that is the case, it will fall into filled case above this logic and the function returns. Do I miss anything?

[holiman]

In my opinion, it's impossible to cause the panic.

I would suspect you are correct, but I didn't spend a lot of time investigating. The thing is, with the change in this PR, it is trivial to ascertain that code is correct and will not panic. Without it, one has to analyze the preconditions that lead up to a potential crash.

So, code is clearer and more robust with this PR (more robust since any changes in events leading up to the vulnerable line can no longer cause the bug to trigger).