Update GHC to 9.6

[arcz]

Description

Around 30 symbolic tests fail due to this update but all concrete tests pass. There is a significant runtime performance improvement (over 30% on concrete tests) probably due to better compiler optimizations. I fixed some cosmetic stuff from hlint on the way.

TODO: debug symbolic tests, update nixpkgs branch once merged into master

Checklist

• tested locally
• added automated tests
• updated the docs
• updated the changelog

[msooseth]

There seems to be two issues:

• Bitwuzla is broken, probably its nix derivation forgot to include some dependency. Nix is far from the magical land it is supposed to be. We are getting: bitwuzla: /nix/store/1x4ijm9r1a88qk7zcmbbfza324gx1aac-glibc-2.37-8/lib/libc.so.6: version GLIBC_2.38' not found (required by /nix/store/wx2ngazi4zyx1gi55kz3p9q8y6497fsk-gmp-with-cxx-6.3.0/lib/libgmp.so.10)
• There seems to be some JSON parsing issue. See my comment below. This is due to forge being updated.

Notes:

$ ldd /nix/store/f1x01z8lk7qb6fm3djbd7h1rbxjg3lxm-bitwuzla-0.3.0/bin/bitwuzla
/nix/store/f1x01z8lk7qb6fm3djbd7h1rbxjg3lxm-bitwuzla-0.3.0/bin/bitwuzla: /nix/store/1x4ijm9r1a88qk7zcmbbfza324gx1aac-glibc-2.37-8/lib/libc.so.6: version `GLIBC_2.38' not found (required by /nix/store/wx2ngazi4zyx1gi55kz3p9q8y6497fsk-gmp-with-cxx-6.3.0/lib/libgmp.so.10)
        linux-vdso.so.1 (0x00007ffc6cf39000)
        libbitwuzla.so.0 => /nix/store/f1x01z8lk7qb6fm3djbd7h1rbxjg3lxm-bitwuzla-0.3.0/lib/libbitwuzla.so.0 (0x00007fd2ffa00000)
        libstdc++.so.6 => /nix/store/2fpmbk0g0ggm9zq89af7phvvvv8dnm7n-gcc-12.3.0-lib/lib/libstdc++.so.6 (0x00007fd2ff600000)
        libgcc_s.so.1 => /nix/store/2fpmbk0g0ggm9zq89af7phvvvv8dnm7n-gcc-12.3.0-lib/lib/libgcc_s.so.1 (0x00007fd2ffdec000)
        libc.so.6 => /nix/store/1x4ijm9r1a88qk7zcmbbfza324gx1aac-glibc-2.37-8/lib/libc.so.6 (0x00007fd2ff41a000)
        libbitwuzlabb.so => /nix/store/f1x01z8lk7qb6fm3djbd7h1rbxjg3lxm-bitwuzla-0.3.0/lib/libbitwuzlabb.so (0x00007fd2ffddd000)
        libbitwuzlabv.so => /nix/store/f1x01z8lk7qb6fm3djbd7h1rbxjg3lxm-bitwuzla-0.3.0/lib/libbitwuzlabv.so (0x00007fd2ffdc7000)
        libbitwuzlals.so => /nix/store/f1x01z8lk7qb6fm3djbd7h1rbxjg3lxm-bitwuzla-0.3.0/lib/libbitwuzlals.so (0x00007fd2ff9a9000)
        libgmp.so.10 => /nix/store/wx2ngazi4zyx1gi55kz3p9q8y6497fsk-gmp-with-cxx-6.3.0/lib/libgmp.so.10 (0x00007fd2ff906000)
        libm.so.6 => /nix/store/1x4ijm9r1a88qk7zcmbbfza324gx1aac-glibc-2.37-8/lib/libm.so.6 (0x00007fd2ff33a000)
        /nix/store/1x4ijm9r1a88qk7zcmbbfza324gx1aac-glibc-2.37-8/lib/ld-linux-x86-64.so.2 => /nix/store/ksk3rnb0ljx8gngzk19jlmbjyvac4hw6-glibc-2.38-44/lib64/ld-linux-x86-64.so.2 (0x00007fd2ffe0f000)

[msooseth]

@arcz can you check what's going on with the build system doing this?

2024-03-11T16:52:22.4496150Z hevm> /nix/store/3frbd0spychry5hg7jfajaiga271nwk1-elfutils-0.190/lib/libdw.a(dwfl_end.o):function dwfl_end:(.text+0x7b): error: undefined reference to 'elf_end'

I'll tackle the bitwuzla thing :)

[msooseth]

@arcz We need this patch to make it work with the new forge that this PR also is introducing. To be more precise, this adds a bit of cleanup AND the --ast flag that is strictly necessary. I think the cleanup is sane, though.

Note that this is due to the updated forge that's part of this nix lock update. Also, this will break ALL previous forge users. Not a big deal, but could be annoying to some users. Old forge compains: --ast is not a flag.

commit c1ce457823f67a7b92139b02b65f494c1e42569f (HEAD -> new-forge, origin/new-forge)
Author: Mate Soos <mate.soos@ethereum.org>
Date:   Wed Mar 13 16:03:32 2024 +0100

    Fixing forge, and some minor code simplification

diff --git a/src/EVM/Solidity.hs b/src/EVM/Solidity.hs
index b1b33da2..45a69a13 100644
--- a/src/EVM/Solidity.hs
+++ b/src/EVM/Solidity.hs
@@ -347,10 +347,10 @@ readSolc pt root fp = do
   -- NOTE: we cannot and must not use Data.Text.IO.readFile because that takes the locale
   --       and may fail with very strange errors when the JSON it's reading
   --       contains any UTF-8 character -- which it will with foundry
-  let fileContents = fmap Data.Text.Encoding.decodeUtf8 (Data.ByteString.readFile fp)
-  (readJSON pt (T.pack $ takeBaseName fp) <$> fileContents) >>=
-    \case
-      Nothing -> pure . Left $ "unable to parse: " <> fp
+  fileContents <- fmap Data.Text.Encoding.decodeUtf8 $ Data.ByteString.readFile fp
+  let contractName = T.pack $ takeBaseName fp
+  case readJSON pt contractName fileContents of
+      Nothing -> pure . Left $ "unable to parse " <> show pt <> " project JSON: " <> fp
       Just (contracts, asts, sources) -> do
         sourceCache <- makeSourceCache root sources asts
         pure (Right (BuildOutput contracts sourceCache))
diff --git a/test/EVM/Test/Utils.hs b/test/EVM/Test/Utils.hs
index 523f67b1..16688ec6 100644
--- a/test/EVM/Test/Utils.hs
+++ b/test/EVM/Test/Utils.hs
@@ -79,7 +79,7 @@ compile foundryType root src = do
   case foundryType of
     FoundryStdLib -> initStdForgeDir (root <> "/lib/forge-std")
     Foundry -> pure ()
-  r@(res,_,_) <- readProcessWithExitCode "forge" ["build", "--root", root] ""
+  r@(res,_,_) <- readProcessWithExitCode "forge" ["build", "--ast", "--root", root] ""
   case res of
     ExitFailure _ -> pure . Left $ "compilation failed: " <> show r
     ExitSuccess -> readBuildOutput root Foundry

[arcz]

@msooseth I pinned bitwuzla and forge and there was just one test failing. I ignored some yul optimization tests which appear to be stuck if you want to take a look.