ci: Switch to ubuntu:26.04 images#5175
Open
tgross35 wants to merge 8 commits into
Open
Conversation
Collaborator
|
Some changes occurred in an Android module cc @maurer |
tgross35
force-pushed
the
docker-image-updates
branch
from
0691db8 to
294ac73
Compare
This comment has been minimized.
This comment has been minimized.
tgross35
force-pushed
the
docker-image-updates
branch
4 times, most recently
from
58e90cc to
2df722a
Compare
This comment has been minimized.
This comment has been minimized.
tgross35
force-pushed
the
docker-image-updates
branch
10 times, most recently
from
48101b5 to
da91ee5
Compare
Contributor
Author
|
🎉 that took some effort |
tgross35
changed the title
ubuntu:26.04 images
This was referenced Jun 24, 2026
This comment has been minimized.
This comment has been minimized.
tgross35
force-pushed
the
docker-image-updates
branch
from
da91ee5 to
1d9ef30
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
tgross35
force-pushed
the
docker-image-updates
branch
from
1d9ef30 to
e922061
Compare
This comment has been minimized.
This comment has been minimized.
When there are additions or updates that are only available with newer platform versions, we don't have a great way to handle testing. The only thing we can really do is unconditionally skip anything that doesn't pass on the oldest versions we have on CI. Improve this by adding a preprocess-only compiler invocation that collects versions from the headers we are building against. These are then used to skip This initial patch only uses the versions for skips on Linux and replacing the NetBSD `uname` invocation.
PowerPC, Sparc, and MIPS (32- and 64-bit for all) have different `_IOC`
numbers, meaning the constants were inaccurate. Example error on
PowerPC64:
bad `EPIOCSPARAMS` value at byte 4: rust: 64 (0x40) != c 128 (0x80)
rust bytes: 00 00 00 00 40 08 8a 01
c bytes: 00 00 00 00 80 08 8a 01
bad `EPIOCGPARAMS` value at byte 4: rust: 128 (0x80) != c 64 (0x40)
rust bytes: 00 00 00 00 80 08 8a 02
c bytes: 00 00 00 00 40 08 8a 02
Resolve this by using `_IOW` and `_IOR`.
Source: https://github.com/torvalds/linux/blob/502d801f0ab03e4f32f9a33d203154ce84887921/include/uapi/linux/eventpoll.h#L97-L99
Fixes: fb58c01 ("epoll: add busy polling parameters")
Where possible, change to the latest Ubuntu release. The 32-bit images still aren't bumped because of the time64 mismatch.
Glibc 2.42 updated this definition to an unsigned integer so it no longer overflows. Source: https://sourceware.org/git/?p=glibc.git;a=commit;h=3263675250cbcbbcc76ede4f7c660418bd345a11
Glibc 2.43 adds the additional fields from Linux 6.17 to `struct tcp_info`. Source: https://sourceware.org/git/?p=glibc.git;a=commit;h=7e46c2aae47d3284d4eb0845ddcc3951e987d681
In Ubuntu 26.04, the musl build hits:
In file included from /checkout/target/aarch64-unknown-linux-musl/debug/build/libc-test-1eed7f417ef9e9ac/out/ctest_output.c:51:
/musl-aarch64/include/string.h:97:7: error: conflicting types for 'basename'; have 'char *(void)'
97 | char *basename();
| ^~~~~~~~
In file included from /checkout/target/aarch64-unknown-linux-musl/debug/build/libc-test-1eed7f417ef9e9ac/out/ctest_output.c:19:
/musl-aarch64/include/libgen.h:9:7: note: previous declaration of 'basename' with type 'char *(char *)'
9 | char *basename(char *);
| ^~~~~~~~
cc1: note: unrecognized command-line option '-Wno-unknown-warning-option' may have been intended to silence earlier diagnostics
I believe this may be due to a new error in recent versions of GCC,
though I have not actually gone through the effort to verify this. In
any case, musl received a patch in 1.2.5 that resolves the issue; apply
it to our 1.1.x build to keep it working.
Link: kraj/musl@725e17e
Ubuntu 26.04 transitioned PowerPC64LE from the IBM long double format to
IEEE binary128 [1], which mean a handful of variadic functions needed to
switch to a binary128-compatible version. These were available starting
with glibc 2.32 (2020-08) per the abilist [2], which is quite a bit
newer than what we say we support at [3] (2.17, 2012-12).
This explains test failures, so skip checking the functions per now. We
should migrate to the `__<name>ieee128` functions at some point, which
need not be related to our documented minimum, but there is no reason to
do it now since Rust doesn't even stably support either IBM `long
double` or binary128.
Demo on Ubuntu 24.04:
$ cat demo.c
#include <stdio.h>
#include <syslog.h>
void demo() {
char buf1[20] = { 0 };
char buf2[20] = { 0 };
printf("Hello, world! %d\n", 1);
fprintf(stderr, "Hello, world! %d\n", 2);
sprintf(buf1, "Hello, world! %d\n", 3);
snprintf(buf2, 20, "Hello, world! %d\n", 4);
syslog(0, "Hello, world! %d\n", 5);
}
$ powerpc64le-linux-gnu-gcc demo.c -c
$ powerpc64le-linux-gnu-nm demo.o
U .TOC.
U __stack_chk_fail
0000000000000000 T demo
U fprintf
U printf
U snprintf
U sprintf
U stderr
U syslog
And the same demo on Ubuntu 26.04:
$ cat demo.c
#include <stdio.h>
#include <syslog.h>
void demo() {
char buf1[20] = { 0 };
char buf2[20] = { 0 };
printf("Hello, world! %d\n", 1);
fprintf(stderr, "Hello, world! %d\n", 2);
sprintf(buf1, "Hello, world! %d\n", 3);
snprintf(buf2, 20, "Hello, world! %d\n", 4);
syslog(0, "Hello, world! %d\n", 5);
}
$ powerpc64le-linux-gnu-gcc demo.c -c
$ powerpc64le-linux-gnu-nm demo.o
U .TOC.
U __fprintfieee128
U __printfieee128
U __snprintfieee128
U __sprintfieee128
U __stack_chk_fail
U __syslogieee128
0000000000000000 T demo
U stderr
[1]: https://bugs.launchpad.net/ubuntu/+bug/2132257
[2]: https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/powerpc/powerpc64/le/libc.abilist;h=65d78e50760b5d6b1a9cb39016a7bf8064224794;hb=HEAD#l2519
[3]: https://doc.rust-lang.org/1.96.0/rustc/platform-support.html
tgross35
force-pushed
the
docker-image-updates
branch
from
e922061 to
1ec51ca
Compare
Collaborator
|
This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed. Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Where possible, change to the latest Ubuntu release. The 32-bit images still aren't bumped because of the time64 mismatch.
There are a handful of supporting changes needed, see individual commit messages for more details. The first two commits are #5188.
Closes: #5126