So, apparently i missed some action in these couple of days. 2 of my nodes (including a seednode) were either ddosed or had the +2 problem. Using the latest banlist seem to have solved for now.

i saw selsta's post. Any PR i should be running?

maybe try 7189 and 7191

Ok. They both have failing checks. I guess it's a problem of the buildbot

.merge+ 7188

Added

.merge+ 7189

Added

ping luigi1111w Snipa we would need merges today

k

branch with 7189 and 7191 seems to be working fine without blocklist.

moneromooo: we need both 7154 and 7188 right?

Yes.

.merge+ 7190 7191

Added

we need one review for 7192 afterwards we have everything for the next point release

I pushed a bit more to 717[56].

Got bored waiting to push after merges and it was conflicting so...

ok 7175 and 7192 need a review

moneromooo: error: ‘num_txes’ was not declared in this scope

meh

7175

Guess I should have held out more for merges.

reverted, I'll PR separately later

this was with everything merged

soo something is missing

moneromooo in 7192 you changed it to "buffer_ssl_init_fill = bytes_transferred;" because you use message_peek, so data is not removed from the buffer?

Yes.

in 7175, where did m_requested_fluffy_block_txes.insert() go, lol

it doesn't insert stuff into this set anymore

I removed it ? AFAICT it was never here...

I saw some commit where you added m_requested_fluffy_block_txes.insert() somewhere, but it's not there now

The extra patch above fixed that fwiw. But it seems to rely on another patch.

Where do you add elements to this set?

Yes.

AFAIK it was always broken.

so how does 7175 work? That set is always empty

It uses a separate set from the one it was hijacking.

It might be it was fine in practice, but why chance it.

so before if(!context.m_requested_objects.empty()) could trigger, now it can't trigger because new set is always empty

something like for future use?

Maybe.

I've not tried to see if it could end up in crosstalk.

I just thought that (1) it's possible it could, seeing the code and (2) using a separate set is basically free.

Let's rephrase 1:

.merge+ 7192 7193

Added

Not having tried to make it do so, I estimate that having crosstalk has a non zero probability.

only if block hash = tx hash for some block and tx

Maybe someone can prove it can't happen, in which case I can drop the patch.

probability is so small it will never happen

OK, at this point I'll just drop it and avoid all the extra talk.

having separate set is still better

because it's different hash we're looking for there

ideally we would use different C++ types for these hashes across the codebase

Well, yes. But several people have asked about the patch, despite it being trivially simple from my pov so..

that would catch all errors where we pass tx hash to function working with block hashes, for example

Note that it's still broken, even reusing the same var.

I noticed, yes

at least it will not waste CPU cycles running this code now

.merge+ 6931

Added

.merges

rapid blocks lately too... really notice if you have --block-notify on to a sound like an alarm bell...

Kronovestan: community.xmr.to/blocks/frequencies

ugh. for static, how do i fix these. (.text+0x2c): undefined reference to `udev_device_get_action'

I only see a shared lib in libudev-dev on Ubuntu 18.04.

yeah it was libudev. didn't have it installed.

@selsta did Luigi get to em?

.merges

not yet

Doesn't look like it. Lemme wake up and I'll get merging once my brain's kicked in.

thanks

.merge+ 7196 7197

Added

.merge- 7196 7197

Removed

.merge- 7175 7176

Removed

Both were closed by Moo 3 hours ago

K, first pass on master is done. Waiting for the merge list to update and I'll do release-17

.merges

Ah, delightful.

Apologies I didn't get to it yesterday, was w/ the fam for most of yesterday, and ran way later than expected.

np we were busy with current attack anyway

moneromooo: 7196 7197 need rebase

7193 is pending a second confirm, it's technically the same as 7192, but want to verify that vtnerd's issues are satisifified.

selsta, can add 159.89.114.14 to the list it's syncing+2 for me

Kronovestan: thanks, done

OK

done

.merge+ 7196 7197

Added

.merges