XOR existing mempools #32372

1. 
   Sjors commented at 9:20 am on April 29, 2025: member

   #32359 allows for larger contiguous payloads to end up in the mempool than what can be achieved with inscriptions. This might trip a naive virus scanner into false positives.

   It seems like a good precaution to XOR existing mempools, rather than only new mempools. Assuming this feature has been tested well enough by now, a migration seems relatively safe.

2. laanwj added the label Mempool on Apr 29, 2025
3. 
   laanwj commented at 10:53 am on April 29, 2025: member

   Why don’t we? Isn’t the mempool.dat always written from scratch and not changed in place? i’m probably missing something but it seems candidate for “use a new random XOR key on every rewrite”.
4. 
   Sjors commented at 12:35 pm on April 29, 2025: member

   Isn’t the mempool.dat always written from scratch and not changed in place?

   I just realized I should check that…

5. 
   Sjors commented at 12:54 pm on April 29, 2025: member

   Ah, so we have an optional -persistmempoolv1 that’s used to maintain the old version.

   Without that FastRandomContext{}.fillrand(xor_key); is used every time. So indeed it does:

   it seems candidate for “use a new random XOR key on every rewrite”.

6. Sjors closed this on Apr 29, 2025

   ---

Contributors
Sjors laanwj

Labels
Mempool