< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #12408: wallet: Change output type globals to members (master...Mf1802-walletChangeTypeMember) https://github.com/bitcoin/bitcoin/pull/12408
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #12409: rpc: Reject deprecated reserveChangeKey in fundrawtransaction (master...Mf1802-rpcFundRawDeprecatedOption) https://github.com/bitcoin/bitcoin/pull/12409
< wumpus>
usually that's some kind of i/o corruption, detected by leveldb as it CRCs everything
< wumpus>
you could ask the user to send the ldb to see if something can be learnt from the kind of corruption (e.g. was a block overwritten, or a bit toggled, etc)
< provoostenator>
you mean chainstate/*.ldb? Where should they upload that?
< provoostenator>
Ah, ok, I'll ask. Github shold be able to handle 2MB
< provoostenator>
(or less if it's the last one)
< provoostenator>
Shouldn't Windows create a wallets directory if no wallet.dat is found?
< wumpus>
no, it creates a wallets directory for a new datadir
< provoostenator>
Ok, I'll try that.
< wumpus>
creating it when wallet.dat is not found is problematic because the user might have a wallets with other names around
< wumpus>
the only time we know it's safe to use wallets/ is for a new data dir
< provoostenator>
I'll do a couple of IBD's to see if more high energy particles from the universe hit that VM
< provoostenator>
That makes sense
< provoostenator>
Would be nice though if it worked on an _empty_ data dir as well as a new one, so the user can create bitcoin.conf before the first launch.
< wumpus>
they can make wallets directory themselves if they want
< wumpus>
and move the wallets, if any, there
< wumpus>
the data directory is not empty if a bitcoin.conf is there, it'd have to have some specialized, twisted definition of 'empty'
< provoostenator>
Yeah, I just realized that I was trying to turn the idea in a todo. Probably too convoluted. And there's the walletdir option now too.
< provoostenator>
*that as I was
< sipa>
murrayn: it's verifying all transactions in history
< bitcoin-git>
[bitcoin] jameshilliard opened pull request #12412: Enable spectre mitigations when supported by compiler. (master...spectre-mitigation) https://github.com/bitcoin/bitcoin/pull/12412
< Lightsword>
hmm, so apparently retpolineplt linker is causes breakage test_bitcoin: key.cpp:324: void ECC_Start(): Assertion `secp256k1_context_sign == nullptr' failed.
< BlueMatt>
fun!
< luke-jr>
Lightsword: no problems with GCC's..
< * luke-jr>
wonders how to deal with Bech32 for libblkmaker; add it to libbase58? rename libbase58?
< luke-jr>
Why does the linker need to do anything?
< luke-jr>
I'm talking about GCC
< luke-jr>
retpolineplt seems to be some LLVM thing
< Lightsword>
hmm, not sure, I kinda just copied the autoconf stuff from openssh
< * luke-jr>
ponders if a linker-side implementation could turn everything into retpolines without compiler involvement, and also doing it for assembly at the same time
< sipa>
it looks like it can be done by the linker
< sipa>
but it basically introduces a stub function between every caller and callee
< Lightsword>
“For dynamic executables, you must compile *all* libraries with -mretpoline and additionally link the dynamic executable and all shared libraries with LLD and pass -z retpolineplt (or use similar functionality from some other linker).”
< Lightsword>
“For kernels or other freestanding or fully static executables, the compiler switch -mretpoline is sufficient to fully mitigate this particular attack.“
< Lightsword>
do we build fully statically or still do some dynamic linking that would need -retpolineplt?
< sipa>
ping cfields
< sasha_>
hello, i am having an issue with bitcoin core
< sasha_>
i started synchronizing the block but sent bitcoin to the wallet before it was done
< sasha_>
since it has to synchronize before i can use my bitcoin, i want to move it to omniwallet
< sasha_>
however, it wont let me dump my private key
< luke-jr>
Lightsword: that's LLVM; GCC doesn't use the same options at all
< Lightsword>
so travis is using LLVM’s linker?
< Lightsword>
but GCC compiler?
< Lightsword>
“checking if the linker (/usr/bin/ld) is GNU ld... yes”
< ProfMac>
Is there an archive of the discussion when the version 2 block was defined, and the blockheight was included?
< sipa>
bip34
< ProfMac>
thanks.
< luke-jr>
Lightsword: no idea what Travis is doing.. GNU ld doesn't support retpolineplt though..
< luke-jr>
I'm reading that without -O2, the compiler options don't do the job, so maybe my test is inconclusive
< ProfMac>
Do I understand correctly that the ExtraNonce in a block is located in the coinbase vin?
< sipa>
yes
< ProfMac>
thank you.
< luke-jr>
well, it's not a special field… it *can* be anywhere
< sipa>
right
< luke-jr>
some pools use the generation transaction's lock time for example
< sipa>
it's just conventionally what is used to add variance to a block
< ProfMac>
I just had the idea that it was an extra field in the block header, but then I was looking at this code snippet.