core/vm: for tracing, do not report post-op memory

[holiman]

closes #24109

Currently, if you try to trace a transaction which MSTOREs something at an offset previously not written to, then the debug_traceTransaction will pad the memory with zeros. In the trace, usually the reported values are what happens before running the opcode
...

On master

$ go run . --code 60406040526001 --json run 
{"pc":0,"op":96,"gas":"0x2540be400","gasCost":"0x3","memSize":0,"stack":[],"depth":1,"refund":0,"opName":"PUSH1"}
{"pc":2,"op":96,"gas":"0x2540be3fd","gasCost":"0x3","memSize":0,"stack":["0x40"],"depth":1,"refund":0,"opName":"PUSH1"}
{"pc":4,"op":82,"gas":"0x2540be3fa","gasCost":"0xc","memSize":96,"stack":["0x40","0x40"],"depth":1,"refund":0,"opName":"MSTORE"}
{"pc":5,"op":96,"gas":"0x2540be3ee","gasCost":"0x3","memSize":96,"stack":[],"depth":1,"refund":0,"opName":"PUSH1"}
{"pc":7,"op":0,"gas":"0x2540be3eb","gasCost":"0x0","memSize":96,"stack":["0x1"],"depth":1,"refund":0,"opName":"STOP"}
{"output":"","gasUsed":"0x15","time":739636}

With this PR:

[user@work evm]$ go run . --code 60406040526001 --json run 
{"pc":0,"op":96,"gas":"0x2540be400","gasCost":"0x3","memSize":0,"stack":[],"depth":1,"refund":0,"opName":"PUSH1"}
{"pc":2,"op":96,"gas":"0x2540be3fd","gasCost":"0x3","memSize":0,"stack":["0x40"],"depth":1,"refund":0,"opName":"PUSH1"}
{"pc":4,"op":82,"gas":"0x2540be3fa","gasCost":"0xc","memSize":0,"stack":["0x40","0x40"],"depth":1,"refund":0,"opName":"MSTORE"}
{"pc":5,"op":96,"gas":"0x2540be3ee","gasCost":"0x3","memSize":96,"stack":[],"depth":1,"refund":0,"opName":"PUSH1"}
{"pc":7,"op":0,"gas":"0x2540be3eb","gasCost":"0x0","memSize":96,"stack":["0x1"],"depth":1,"refund":0,"opName":"STOP"}
{"output":"","gasUsed":"0x15","time":128972}

[s1na]

LGTM

[alcuadrado]

Sorry if these questions are obvious, I want to make sure that our implementation in Hardhat is equivalent to yours.

Does this PR affect the output of debug_traceTransaction? Will this be released with the next version of geth?

[s1na]

Hey @alcuadrado! Yes it'll affect the output of debug_traceTransaction slightly. The memory reported for each step is currently after the expansion necessary to execute that op. This PR changes it so that the step reports memory before expansion.

I'm not sure about the second question. There was a plan to cut the next release early this week. I'll bring up the PR tomorrow morning for discussion, see if we can get it merged.

[alcuadrado]

Hey, @s1na! Thanks for your reply! Knowing that this is planned for an upcoming release, even if it's not the next one, is really helpful for our planning. We've been wanting to implement this same change for some time but didn't want to do it before geth.