wallet: reroll fake outs selection on local tx_sanity_check failure

[xiphon]

No description provided.

[moneromooo-monero]

Does this reuse the fake outs it requested on the first attempt (the wallet asks for more than it needs in the first place in case some are not usable) ?

[xiphon]

Does this reuse the fake outs it requested on the first attempt (the wallet asks for more than it needs in the first place in case some are not usable) ?

Updated. Have to clear ringdb and outs cache on failed tx sanity check to force fetching new fake outs on the next attempt.

That said if generated tx passes sanity check, the following transfer_selected_rct call will reuse the outs.

Does this cover the case you mentioned? If not, could you please provide more details.

[moneromooo-monero]

It should not fetch new outputs, unless it really needs to. My point is that since it already requests more outputs than necessary from the daemon, it should not request other outputs if it can rearrange its pick from the outputs it already has. Failing that, it should ask additional outputs (rather than a full replacement) to avoid the daemon being able to tell which output is the real spend.

[xiphon]

Updated. Attempts to reroll fake outputs from the same outputs set it initially received from the daemon.

[moneromooo-monero]

Are you sure ? The top level loop calls get_outs repeatedly (if the sanity check fails) and the first thing get_outs does is clear the outs, and will call the daemon again to get outputs. What am I missing ?

[xiphon]

get_outs clears previously selected outs that were picked from rct_offsets, rct_offsets is outputs set the daemon returned on the first get_outs call and it doesn't get updated during all the attempts (see line 7809)

[moneromooo-monero]

Great. I thought rct_offets were the per block distribution on reading. All good then, thanks.