glibc 2.42→2.43 + C23-const FTBFS fixes (make stack + gcc) by bryan-minimal · Pull Request #238 · gominimal/pkgs

bryan-minimal · 2026-06-11T01:49:32Z

What

glibc 2.42 → 2.43, plus every FTBFS fix the full-fleet rebuild surfaced. Reshaped to glibc-only (binutils 2.46.1 already landed separately on main, so it's dropped here) and rebased onto current main (picks up the harnesses → stacks/ migration). Builds clean on amd64 + arm64.

Why

glibc 2.43 clears the glibc CVE backlog (CVE-2025-15281 + CVE-2026-5450 CRIT, etc.). Because glibc is libc, the bump rebuilds the entire fleet — which is what surfaced the FTBFS wave below.

The C23-const FTBFS wave

glibc 2.43 implements ISO C23 const-preserving string macros — strchr/strrchr/strstr/memchr/bsearch/etc. now return const char * for a const char * argument. Pre-C23 code assigning the result to a plain char * trips -Werror=discarded-qualifiers and FTBFS. This is an ecosystem-wide wave, and our fleet hits it in two distinct shapes:

1. The make-stack lever — stacks/make/stack.ncl appends -Wno-error=discarded-qualifiers to the shared make-stack CFLAGS; it lands after a package's own -Werror (later flag wins), downgrading that one warning.

⚠️ Honest scope note: this lever only reaches packages that actually use the make stack. Most C packages here have a hand-written build.sh that assigns its own CFLAGS (an assignment, not an append), so they bypass the make-stack lever entirely. There's no in-repo universal interception point (the global -fno-plt etc. come from the build sandbox, not this repo). So the lever is correct + durable for make-stack packages, but the packages below needed individual fixes — surfaced one-at-a-time by the buildbot as it rebuilt the fleet.

2. Per-package build.sh fixes — for the hand-written-build.sh packages that also self-enable -Werror (their -Werror lives in upstream configure, invisible to a grep), the same -Wno-error=discarded-qualifiers goes into each package's own CFLAGS. It lands after their -Werror on the compile line (or after $(AM_CFLAGS) under automake), so the later flag wins.

3. or-tools — a different class entirely (not -Werror, a hard compile error). or-tools FetchContent's SCIP v10.0.0, which bundles tinycthread (a C11 <threads.h> shim doing #define once_flag pthread_once_t). glibc 2.43 surfaces its native once_flag/call_once into SCIP's translation unit, so that macro corrupts glibc's typedef __once_flag once_flag; into a pthread_once_t redefinition → "conflicting types". -Wno-error can't fix a hard error. or-tools 9.15 is already the latest release and hard-FORCEs SCIP TPI=tny in its bundled cmake (so a -DTPI=... is overridden), so we sed that to TPI=omp before cmake — SCIP uses its OpenMP task interface instead of tinycthread, staying parallel + THREADSAFE (libgomp is present from our gcc build). A grep guard fails the build loudly if upstream renames the line.

Changes

File	Change	Kind
`packages/glibc/build.ncl`	2.42 → 2.43 (+ sha256)	the bump
`stacks/make/stack.ncl`	fleet `-Wno-error=discarded-qualifiers`	durable (make-stack pkgs)
`packages/gcc/build.sh`	same downgrade via `CFLAGS_FOR_TARGET` (libgomp)	interim (retired by gcc-16)
`packages/libxcrypt/build.sh`	`-Wno-error=discarded-qualifiers` in CFLAGS (crypt-{gost,sm3}-yescrypt.c)	per-package
`packages/elfutils/build.sh`	`-Wno-error=discarded-qualifiers` in CFLAGS (libcpu/riscv_disasm.c)	per-package
`packages/strace/build.sh`	`-Wno-error=discarded-qualifiers` in CFLAGS (src/ioctl.c)	per-package
`packages/or-tools/build.sh`	SCIP `TPI` tny → omp (drop tinycthread)	tinycthread collision

For the reviewer (parked / notable)

gcc's fix is interim. A gcc-16 bump retires packages/gcc/build.sh (gcc-16 patched its own libgomp upstream) but does not retire the fleet/per-package fixes — every other C program still trips the same break under any gcc against glibc 2.43, because the language semantics are unchanged.
The real toolchain-policy call is -Wno-error=discarded-qualifiers (downgrade the one warning, used here) vs -std=gnu17 (restores C17 prototypes so strchr returns char * — no warning at all, but a broader language-standard change). This PR takes the surgical -Wno-error route; the -std=gnu17 switch can replace all of these levers later if the team prefers it.
or-tools TPI=omp tradeoff: SCIP keeps parallelism via OpenMP instead of tinycthread. If OpenMP were ever unavailable it fails fast at configure (not after the long compile), and the fallback is TPI=none (single-threaded SCIP).
elfutils gotcha (for anyone touching it later): its configure has no --disable-werror in this version — the CFLAGS flag is the reliable knob.

Status

Draft — builds clean on both arches, but it's a toolchain-policy change (touches the C-build warning posture fleet-wide), so leaving it for explicit sign-off on the -Wno-error vs -std=gnu17 direction before merge.

🤖 Generated with Claude Code

coderabbitai · 2026-06-11T01:49:42Z

📝 Walkthrough

Walkthrough

Bumps glibc from version 2.42 to 2.43 with an updated sha256, then applies compatibility fixes across the build system: -Wno-error=discarded-qualifiers is added fleet-wide to the make stack's CFLAGS and individually to gcc, elfutils, libxcrypt, and strace build scripts. The or-tools build script gains a sed-based patch forcing SCIP's TPI setting to "omp".

Changes

glibc 2.43 upgrade and downstream compatibility fixes

Layer / File(s)	Summary
glibc version and checksum bump `packages/glibc/build.ncl`	Updates the `version` field from `2.42` to `2.43` and replaces the `sha256` for the downloaded tarball.
Fleet-wide CFLAGS fix in make stack `stacks/make/stack.ncl`	Appends `-Wno-error=discarded-qualifiers` to `build_env_vars.CFLAGS` with an explanatory comment block, applying the warning downgrade to all autotools builds in the make stack.
Per-package CFLAGS workarounds `packages/gcc/build.sh`, `packages/elfutils/build.sh`, `packages/libxcrypt/build.sh`, `packages/strace/build.sh`	Adds `-Wno-error=discarded-qualifiers` to `CFLAGS` in each package's build script; gcc additionally propagates the flag into `CXXFLAGS`, `CFLAGS_FOR_TARGET`, and `CXXFLAGS_FOR_TARGET`.
or-tools SCIP TPI CMake patch `packages/or-tools/build.sh`	Patches `cmake/dependencies/CMakeLists.txt` via `sed` to force SCIP's `TPI` from `"tny"` to `"omp"`, followed by a `grep` verification that exits with an error if the replacement did not apply.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

🐇 Hop hop, glibc leapt to two-four-three,
Its C23 macros changed qualifiers with glee.
-Wno-error flags were sprinkled all around,
And SCIP's TPI got patched to stay sound.
The rabbit nods — the build stack holds its ground! 🌿

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately and concisely summarizes the main changes: glibc version bump from 2.42 to 2.43 and C23-const related FTBFS (Fails To Build From Source) fixes targeting the make stack and gcc components.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

Create stacked PR
Commit on current branch

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch update-base-soup-2026-06-11

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

The base-soup rebuild recompiles gcc-15.2.0 against the new glibc 2.43 headers, which implement ISO C23 const-preserving string macros (strchr et al. now return `const char *` for a `const char *` input). gcc-15.2.0 predates this, so libgomp/affinity-fmt.c (`char *q = strchr(p+1,'}')`) discards const and fails -Werror=discarded-qualifiers — the FTBFS the buildbot hit. The pointer is only used for `q - p`, never written, so it's a source-pedantry mismatch, not a runtime bug. Downgrade only that single warning (everything else stays -Werror) via CFLAGS_FOR_TARGET (the documented knob for target libs like libgomp) + CFLAGS, for gcc's own build only — the shipped compiler is byte-identical. Interim escape hatch; gcc-16.1.0 fixes this upstream and builds cleanly against glibc 2.43 (LFS pairs them with no workaround), so the clean follow-up is bumping gcc to 16. This is the same C23 wave hitting the whole ecosystem (openssl, grub, libevent, ...). Refs in build.sh. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

bryan-minimal · 2026-06-11T02:38:36Z

Added: gcc glibc-2.43 build fix (commit `4cf505f`)

The first buildbot run failed building gcc-15.2.0 against the new glibc 2.43 headers — not glibc/binutils themselves. glibc 2.43 implements ISO C23 const-preserving string macros: strchr/strrchr/strstr/memchr/… now return const char * for a const char * input. gcc-15.2.0 predates this, so libgomp/affinity-fmt.c (char *q = strchr(p+1,'}')) discards const → -Werror=discarded-qualifiers → FTBFS.

This is the same C23 wave hitting the whole ecosystem right now (openssl, grub, libevent, criu, …) — a known, documented event, not anything specific to us.

Fix here (interim, surgical): -Wno-error=discarded-qualifiers via CFLAGS_FOR_TARGET (the documented knob for target libs like libgomp) for gcc's own build only. It downgrades exactly that one warning — everything else stays -Werror — and the shipped compiler is byte-identical. The pointer is only used for q - p, never written, so it's source-pedantry, not a runtime bug.

Clean follow-up: bump gcc → 16.1.0, which fixes this upstream and builds against glibc 2.43 with no workaround (LFS pairs them directly). That retires the band-aid; tracked separately as it's a major-version migration.

Refs are also inline in packages/gcc/build.sh.

bryan-minimal · 2026-06-11T03:03:19Z

🚧 Moved to draft — needs a team decision on the glibc-2.43 C23 wave

Second buildbot run got past gcc but then libxcrypt failed with the identical pattern:

lib/crypt-gost-yescrypt.c:134: error: initialization discards 'const' qualifier ... [-Werror=discarded-qualifiers]
  char *hptr = strchr ((const char *) intbuf->retval + 3, '$');

So this is confirmed fleet-wide, not a gcc one-off (see the previous comment for root cause: glibc 2.43's ISO C23 const-preserving string macros). Patching package-by-package is whack-a-mole, and gcc 16.1.0 would not help — it only fixes gcc's own source, not libxcrypt/openssl/grub/etc.

The systematic fix: one line in the shared make harness

The flags that land after each package's -Werror come from harnesses/make/harness.ncl:

# harnesses/make/harness.ncl:9
CFLAGS = "-O2 -pipe -gno-record-gcc-switches",

Because these are appended last on every compile (confirmed in libxcrypt's command line: … -Werror -fno-plt -march=… -O2 -pipe -gno-record-gcc-switches …), adding the override here downgrades the one warning fleet-wide for every make/autotools package, in a single line:

-    CFLAGS = "-O2 -pipe -gno-record-gcc-switches",
+    CFLAGS = "-O2 -pipe -gno-record-gcc-switches -Wno-error=discarded-qualifiers",

gcc has a custom build.sh, so it's handled separately (the commit already in this PR). A handful of CMake/meson packages have their own harnesses — if any of those break, the same one-line override applies to harnesses/cmake / harnesses/meson.

Why this is safe (but is a band-aid)

Downgrades exactly one warning (discarded-qualifiers) error→warning; every other -Werror stays.
The flagged code is runtime-safe — it's the old C17 char *q = strchr(const…) pattern where the pointer isn't written; byte-identical output.
It does not change the language standard or any ABI.
Tech debt: it masks the not-yet-const-correct code rather than fixing it. We unwind it when we adopt C23 properly.

Decision for the team

Option	Effect	Cost
Global `-Wno-error=discarded-qualifiers` in harness (proposed)	Whole wave gone, glibc 2.43 + CVEs land now	Band-aid / tech debt
`-std=gnu17` in harness	Broader — turns off all C23, not just this warning	Bigger hammer; also masks future C23 issues
Per-package `const` patches (what distros/upstreams do)	"Correct" C23-forward	Whack-a-mole across N packages
Hold glibc 2.43	Safe, no churn	glibc CVEs stay unpatched
Bump gcc → 16.1.0	Fixes gcc only — does NOT address libxcrypt/others	Major-version migration, doesn't solve the wave

My recommendation: the global -Wno-error=discarded-qualifiers harness override is the pragmatic way to land glibc 2.43 (and clear the CVEs) now without whack-a-mole, then do the const-correct per-package fixes (or a gcc-16 bump) as deliberate follow-up. But this touches every C build, so it's a deliberate toolchain-policy call — hence draft + this writeup for discussion.

Refs: glibc 2.43 C23 const macros · GCC CFLAGS_FOR_TARGET docs · LFS GCC-16.1.0

…-06-11

…2.43 C23 rebuild glibc 2.43's ISO C23 const-preserving string macros (strchr/strstr/memchr/... return `const char *` for const input) cause an ecosystem-wide -Werror=discarded-qualifiers FTBFS wave. The make stack's shared CFLAGS land after each package's own -Werror on the compile line, so appending the downgrade here fixes the whole make/autotools fleet in one place. Interim until a gcc-16 bump (which fixes it upstream). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

…ires gcc/build.sh Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

…TBFS libxcrypt has a hand-written build.sh that *assigns* CFLAGS (not the make stack), so the stacks/make/stack.ncl fleet lever from this PR can't reach it. Its crypt-{gost,sm3}-yescrypt.c assign const-qualified strchr() results to a plain char*, tripping libxcrypt's own configure-enabled -Werror under glibc 2.43's ISO C23 const-preserving string macros. Downgrade only that one warning, appended after -Werror so the later flag wins. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

elfutils hardcodes CFLAGS in its build.sh (bypasses the make stack) and defaults -Werror on. glibc 2.43's ISO C23 const-preserving bsearch/strchr-family lookups discard const into plain pointers (libcpu/riscv_disasm.c known_csrs bsearch), tripping -Werror=discarded-qualifiers. elfutils is notorious for multiple distinct -Werror trips across toolchain bumps, so use its own --disable-werror off-switch rather than chase warnings one flag at a time. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

…disable-werror unrecognized) elfutils' configure has no --disable-werror in this version (it warned 'unrecognized options: --disable-werror' and kept -Werror on, so the C23 discarded-qualifiers trip in libcpu/riscv_disasm.c persisted). Use the same surgical CFLAGS flag as libxcrypt instead: automake emits $(AM_CFLAGS) $(CFLAGS), and elfutils' -Werror lives in AM_CFLAGS, so a -Wno-error in CFLAGS lands after it on the compile line and wins. The whole libelf tree compiled clean before the single trip, so the one warning is the only break. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

strace hardcodes CFLAGS in build.sh (bypasses make stack) and self-enables -Werror. glibc 2.43's ISO C23 const-preserving bsearch returns const for a const arg; src/ioctl.c assigns it to a plain pointer (iop = bsearch(..., ioctlent, ...)). Downgrade only that warning, appended after strace's -Werror on the compile line so the later flag wins. Same glibc-2.43 C23 class as #238. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

…reads collision or-tools FetchContent's SCIP v10.0.0, which bundles tinycthread (a C11 <threads.h> shim doing `#define once_flag pthread_once_t`). glibc 2.43 surfaces its native once_flag/call_once into SCIP's TU, so that macro corrupts glibc's `typedef __once_flag once_flag;` into a pthread_once_t redefinition -> hard "conflicting types" FTBFS (not the -Werror class; -Wno-error can't help). or-tools 9.15 is already the latest release and pins current SCIP, so no version bump escapes it. or-tools hard-FORCEs SCIP TPI=tny in its bundled cmake, so a -DTPI=... is overridden. sed the forced TPI to omp before cmake: SCIP uses its OpenMP task interface (no tinycthread), staying parallel + THREADSAFE (SCIP auto-forces THREADSAFE on for non-none TPI). libgomp is present from gcc. A grep guard fails the build loudly if upstream renames the line. Leaf package (nothing depends on it) but in the glibc closure, so it must build. glibc-2.43 class, #238. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

msample

LGTM

bryan-minimal added 2 commits June 10, 2026 18:49

Update glibc to 2.43

5d656de

Update binutils to 2.46.1

30cef78

bryan-minimal marked this pull request as draft June 11, 2026 03:03

bryan-minimal and others added 2 commits June 16, 2026 12:57

Merge remote-tracking branch 'origin/main' into update-base-soup-2026…

cdafc7b

…-06-11

bryan-minimal changed the title ~~Update base-soup: 2 packages~~ glibc 2.42→2.43 + C23-const FTBFS fixes (make stack + gcc) Jun 16, 2026

bryan-minimal and others added 6 commits June 16, 2026 13:00

make stack: correct comment — fleet lever is durable, gcc-16 only ret…

40e3d4a

…ires gcc/build.sh Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

bryan-minimal marked this pull request as ready for review June 17, 2026 15:04

bryan-minimal requested a review from twitchyliquid64 June 17, 2026 15:04

msample approved these changes Jun 17, 2026

View reviewed changes

twitchyliquid64 approved these changes Jun 17, 2026

View reviewed changes

bryan-minimal added this pull request to the merge queue Jun 17, 2026

github-merge-queue Bot removed this pull request from the merge queue due to no response for status checks Jun 17, 2026

bryan-minimal added this pull request to the merge queue Jun 18, 2026

Merged via the queue into main with commit 724c7ad Jun 18, 2026
4 checks passed

bryan-minimal deleted the update-base-soup-2026-06-11 branch June 18, 2026 01:38

Navigation Menu

Uh oh!

glibc 2.42→2.43 + C23-const FTBFS fixes (make stack + gcc) - #238

glibc 2.42→2.43 + C23-const FTBFS fixes (make stack + gcc)#238
bryan-minimal merged 11 commits into
mainfrom
update-base-soup-2026-06-11

bryan-minimal commented Jun 11, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Jun 11, 2026 •

edited

Loading

Walkthrough

Changes

Estimated code review effort

Poem

Uh oh!

bryan-minimal commented Jun 11, 2026

Uh oh!

bryan-minimal commented Jun 11, 2026

Uh oh!

msample left a comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

bryan-minimal commented Jun 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What

Why

The C23-const FTBFS wave

Changes

For the reviewer (parked / notable)

Status

Uh oh!

coderabbitai Bot commented Jun 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Estimated code review effort

Poem

Uh oh!

bryan-minimal commented Jun 11, 2026

Added: gcc glibc-2.43 build fix (commit 4cf505f)

Uh oh!

bryan-minimal commented Jun 11, 2026

🚧 Moved to draft — needs a team decision on the glibc-2.43 C23 wave

The systematic fix: one line in the shared make harness

Why this is safe (but is a band-aid)

Decision for the team

Uh oh!

msample left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

bryan-minimal commented Jun 11, 2026 •

edited

Loading

coderabbitai Bot commented Jun 11, 2026 •

edited

Loading

Added: gcc glibc-2.43 build fix (commit `4cf505f`)