Your message dated Mon, 27 Nov 2023 10:19:47 +0100
with message-id <[email protected]>
and subject line Re: rustc: version in buster fails to build Rust code,
aborting with "undefined symbol: llvm.x86.subborrow.64"
has caused the Debian Bug report #976601,
regarding rustc: version in buster fails to build Rust code, aborting with
"undefined symbol: llvm.x86.subborrow.64"
to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [email protected]
immediately.)
--
976601: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=976601
Debian Bug Tracking System
Contact [email protected] with problems
--- Begin Message ---
Package: rustc
Version: 1.41.1+dfsg1-1~deb10u1
Severity: normal
Dear Maintainer,
while trying to build the new 6.0.1 version of Suricata [1] with the
current Rust toolchain in buster, I noticed that one of the Rust
components in this project [2] fails to build. This can be reproduced on
buster simply by:
$ git clone https://github.com/rusticata/der-parser -b der-parser-4.x
Cloning into 'der-parser'...
remote: Enumerating objects: 61, done.
remote: Counting objects: 100% (61/61), done.
remote: Compressing objects: 100% (38/38), done.
remote: Total 2154 (delta 31), reused 48 (delta 23), pack-reused 2093
Receiving objects: 100% (2154/2154), 516.92 KiB | 1.29 MiB/s, done.
Resolving deltas: 100% (1396/1396), done.
$ cd der-parser
[satta@BLN04NB0421:~/der-parser] $ [satta@BLN04NB0421:~/der-parser] $ cargo
build
[der-parser-4.x] [0]
Updating crates.io index
Downloaded proc-macro-hack v0.5.19
Downloaded num-bigint v0.3.1
Downloaded nom v5.1.2
Downloaded num-integer v0.1.44
Downloaded lexical-core v0.7.4
Compiling autocfg v1.0.1
Compiling ryu v1.0.5
Compiling bitflags v1.2.1
Compiling memchr v2.3.4
Compiling lexical-core v0.7.4
Compiling version_check v0.9.2
Compiling cfg-if v0.1.10
Compiling arrayvec v0.5.2
Compiling static_assertions v1.1.0
Compiling proc-macro-hack v0.5.19
Compiling num-traits v0.2.14
Compiling num-integer v0.1.44
Compiling num-bigint v0.3.1
Compiling nom v5.1.2
Compiling rusticata-macros v2.1.0
Compiling der-oid-macro v0.2.0 (/home/satta/der-parser/der-oid-macro)
Compiling der-parser v4.1.0 (/home/satta/der-parser)
warning: unknown lint: `broken_intra_doc_links`
--> src/lib.rs:142:9
|
142 | #![deny(broken_intra_doc_links)]
| ^^^^^^^^^^^^^^^^^^^^^^
|
= note: `#[warn(unknown_lints)]` on by default
error:
/home/satta/der-parser/target/debug/deps/libder_oid_macro-b8b90f0c61ae3277.so:
undefined symbol: llvm.x86.subborrow.64
--> src/lib.rs:173:9
|
173 | pub use der_oid_macro::oid;
| ^^^^^^^^^^^^^
error: aborting due to previous error
error: could not compile `der-parser`.
To learn more, run the command again with --verbose.
Here's an upstream bug ticket regarding this issue:
https://github.com/rusticata/der-parser/issues/36
Interestingly, the der-parser code mentioned above builds with the Debian
Rust compiler package when built as part of the Suricata 6.0.0 tarball [3],
in which it is vendored in the rust/ subdirectory, but fails with the above
mentioned error when built as part of the vendored code in the Suricata
6.0.1 tarball [4]. No changes have been made to the vendored der-parser
code between those two versions, so it might also be dependent on the mix
of vendored crates in this context and how they might interact.
Also, ARM builds do not seem to be affected, according to the upstream
ticket.
Could the recent move to 1.41.1+dfsg1-1~deb10u1 and the switch to LLVM 7
(see changelog) [5] be related?
We also notice a mismatch between the cargo and Rust versions:
$ which rustc
/usr/bin/rustc
$ rustc -V
rustc 1.41.1
$ which cargo
/usr/bin/cargo
$ cargo -V
cargo 1.42.1
If you have any more questions and I can help to clear them up, feel
free to let me know! This currently blocks backporting Suricata 6.0.1
to buster with the current Rust version, and I start to suspect that
some quirk in the current buster Rust package might be the problem.
(Yes , there are some other things to sort out to get 6.0.x into
unstable/testing first before this becomes an issue, but I just wanted
to test the waters regarding the Rust ecosystem that is usually a bit
sensitive when it comes to versions.)
Thanks
Sascha
[1] https://suricata-ids.org
[2] https://github.com/rusticata/der-parser
[3] https://www.openinfosecfoundation.org/download/suricata-6.0.0.tar.gz
[4] https://www.openinfosecfoundation.org/download/suricata-6.0.1.tar.gz
[5]
https://tracker.debian.org/news/1175959/accepted-rustc-1411dfsg1-1deb10u1-source-amd64-all-into-proposed-updates-stable-new-proposed-updates/
Versions of packages rustc depends on:
ii binutils 2.34-5
ii gcc 4:8.3.0-1
ii libc6 2.30-4
ii libc6-dev [libc-dev] 2.30-4
ii libgcc-s1 [libgcc1] 10-20200411-1
ii libgcc1 1:8.3.0-6
ii libstd-rust-dev 1.41.1+dfsg1-1~deb10u1
Versions of packages rustc recommends:
ii cargo 0.43.1-3~deb10u1
ii rust-gdb 1.41.1+dfsg1-1~deb10u1
Versions of packages rustc suggests:
pn lld-7 <none>
pn rust-doc <none>
pn rust-src <none>
-- no debconf information
--- End Message ---
--- Begin Message ---
Hi Blair,
Since we now have bookworm as stable and rustc 1.63.0 in it, is this bug still
relevant?
I don't think so. I introduced a patch to address this issue back then,
and I assume this problem is also gone now. Thanks for the reminder.
Best regards
Sascha
OpenPGP_signature
Description: OpenPGP digital signature
--- End Message ---
