14:00:29 <achow101> #startmeeting 14:00:31 <Sjors[m]11> Hi 14:00:33 <achow101> #bitcoin -core-dev Meeting: achow101 _aj_ amiti ariard aureleoules b10c BlueMatt brunoerg cfields darosior dergoegge dongcarl fanquake fjahr furszy gleb glozow hebasto instagibbs jamesob jarolrod jonatack josibake kallewoof kanzure kouloumos kvaciral laanwj LarryRuane lightlike luke-jr MacroFake Murch phantomcircuit pinheadmz promag provoostenator ryanofsky sdaftuar S3RK stickies-v sipa sr_gi theStack TheCharlatan vasild 14:00:38 <Murch[m]> hi 14:00:40 <brunoerg> hi 14:00:41 <vasild> hi 14:00:41 <gleb> hi 14:00:46 <dzxzg> hi 14:00:49 <josie> hi 14:00:49 <achow101> There are no pre-proposed meeting topics this week. Any last minute ones to add? 14:00:51 <TheCharlatan> hi 14:00:54 <hebasto> hi 14:01:21 <laanwj> hi 14:01:23 <pinheadmz> hi 14:01:25 <lightlike> Hi 14:01:59 <achow101> #topic Ad-hoc high priority for review 14:02:00 <achow101> Anything to add or remove from https://github.com/orgs/bitcoin/projects/1/views/4 14:02:11 <glozow> hi 14:02:35 <sipa> hi 14:02:40 <Sjors[m]11> #29868 is draft now 14:02:42 <gribble> https://github.com/bitcoin/bitcoin/issues/29868 | Reintroduce external signer support for Windows by hebasto ÷ Pull Request #29868 ÷ bitcoin/bitcoin ÷ GitHub 14:02:50 <kanzure> hi 14:02:57 <bitcoin-git> [bitcoin] willcl-ark opened pull request #30879: test: re-bucket p2p_node_network_limited (master...ci-p2p_node_network_limited) https://github.com/bitcoin/bitcoin/pull/30879 14:02:58 <achow101> a bunch of those need rebasing too 14:03:19 <Sjors[m]11> And as muchas I like #30043 it needs laanwj 14:03:21 <gribble> https://github.com/bitcoin/bitcoin/issues/30043 | net: Replace libnatpmp with built-in PCP+NATPMP implementation by laanwj ÷ Pull Request #30043 ÷ bitcoin/bitcoin ÷ GitHub 14:03:30 <laanwj> i'm here 14:03:43 <Sjors[m]11> ð 14:03:57 <achow101> removing #30043 #29868 and #30116 for now 14:03:59 <gribble> https://github.com/bitcoin/bitcoin/issues/30043 | net: Replace libnatpmp with built-in PCP+NATPMP implementation by laanwj ÷ Pull Request #30043 ÷ bitcoin/bitcoin ÷ GitHub 14:03:59 <hebasto> yes, 29868 can be removed 14:04:00 <gribble> https://github.com/bitcoin/bitcoin/issues/29868 | Reintroduce external signer support for Windows by hebasto ÷ Pull Request #29868 ÷ bitcoin/bitcoin ÷ GitHub 14:04:02 <gribble> https://github.com/bitcoin/bitcoin/issues/30116 | p2p: Fill reconciliation sets (Erlay) attempt 2 by sr-gi ÷ Pull Request #30116 ÷ bitcoin/bitcoin ÷ GitHub 14:04:03 <laanwj> but yes it needs rebase, will get to it one of these dayd 14:04:15 <laanwj> sorry for the delay 14:04:51 <Sjors[m]11> Maybe add #30409 14:04:53 <gribble> https://github.com/bitcoin/bitcoin/issues/30409 | Introduce waitTipChanged() mining interface, replace RPCNotifyBlockChange, drop CRPCSignals & g_best_block by Sjors ÷ Pull Request #30409 ÷ bitcoin/bitcoin ÷ GitHub 14:05:14 <Sjors[m]11> Already has two ACKs but could use more eyes. 14:05:14 <achow101> Sjors[m]11: done 14:06:21 <achow101> #topic 28.0 release priorities 14:06:33 <achow101> Milestone is https://github.com/bitcoin/bitcoin/milestone/66 14:06:44 <achow101> Current backports pr #30827 14:06:46 <gribble> https://github.com/bitcoin/bitcoin/issues/30827 | [28.x] Further backports and rc2 by achow101 ÷ Pull Request #30827 ÷ bitcoin/bitcoin ÷ GitHub 14:07:08 <achow101> The remaining active issue is this ibd regression on Windows 14:07:26 <achow101> any thoughts on fixing that? 14:08:00 <jonatack> hi 14:08:46 <Sjors[m]11> xor causing a slowdown? 14:08:53 <dzxzg> I thought maybe it could be the size of the array that is being allocated, have not tested yet though: https://github.com/davidgumberg/bitcoin/commit/64b7313437b0cadce69678efdaacb5c03c912786 14:08:54 <achow101> it looks like the ibd regression is probably caused by blocksdir xor, so perhaps we should revert that? 14:08:57 <sipa> i don't think we have a good theory what could even cause it 14:09:03 <abubakarsadiq> hi 14:09:13 <laanwj> looking at the issue, https://github.com/bitcoin/bitcoin/issues/30833 , some people report not seeing any difference 14:09:14 <lightlike> maybe revert it just for windows? 14:09:21 <lightlike> (or disable) 14:09:49 <achow101> i didn't observe a difference, so it might be hardware dependent too 14:10:15 <laanwj> it's extremely unlikely to be an xor operation causing such a slowdown 14:10:15 <sipa> unfortunately, windows is probably the only platform where people might want the xoring... (due to the prevalance of virus scanners there) 14:10:22 <Sjors[m]11> Turning it off for Windows until it's clear what's happening seems like a safe option. 14:10:37 <Sjors[m]11> But also it indeed tends to flag stuff 14:10:44 <pablomartin> sure is it only Win? https://github.com/bitcoin/bitcoin/issues/30878#issuecomment-2346309699 14:10:57 <achow101> pablomartin: that's a different problem 14:10:57 <sipa> davidgumberg has a patch that fixes the slowdown if xoring is disabled at runtime 14:11:02 <pablomartin> ok 14:11:05 <sipa> we could combine that with defaulting it to off on windows 14:11:16 <laanwj> might add a flag to turn it off so that the person reporting the issue can benchmark with it on and off? 14:12:04 <achow101> sipa: that seems reasonable 14:12:11 <laanwj> until it's confirmed that it's really that which is the problem, turning it permanently off for any platform makes no sense 14:13:00 <achow101> laanwj: it looks like they bisected to the commit adding it 14:13:09 <laanwj> oh, didn't see that 14:13:26 <achow101> https://github.com/bitcoin/bitcoin/issues/30833#issuecomment-2335184072 14:13:27 <dzxzg> There is a cli flag `-blocksxor=0` for disabling but it doesn't currently stop bitcoin core from avoiding the xor path, I've a commit here: https://github.com/davidgumberg/bitcoin/commit/aa0be4031cbe5bed407bc7a1753875cc1c4ddccb 14:13:28 <laanwj> curious, why would xoring (which id assume, is one assembler instruction) cause such a slowdown 14:13:28 <dzxzg> And I've reproduced that on 28.x with xoring disabled the regression is gone 14:13:49 <achow101> dzxzg: can you open a pr? 14:13:52 <sipa> laanwj: there are different code paths in AutoFile that get used when xoring is used vs not used 14:13:54 <dzxzg> I suspect it's probably not the x'oring, it's something to do with the array and copying 14:13:54 <laanwj> it should be neglible compared to any i/o 14:14:10 <dzxzg> achow101: sure 14:14:31 <laanwj> sipa: okay, thanks 14:14:35 <sipa> given the situation, i think that dzxzg commit doesn't hurt 14:14:50 <sipa> then we can disable on enable on windows easily, and revert if fixed 14:15:33 <achow101> Has anyone noticed any other issues with 28.0rc1? 14:15:52 <achow101> There's a testing guide now: https://github.com/bitcoin-core/bitcoin-devwiki/wiki/28.0-Release-Candidate-Testing-Guide 14:16:58 <achow101> or any other topics to discuss today? 14:18:20 <achow101> #endmeeting