Re: linux-next: KUnit failure after merge of the mm-nonmm-unstable tree

[Mark Brown]

On Fri, Jul 03, 2026 at 03:43:35PM +0100, Mark Brown wrote:
> Hi all,
> 
> After merging the mm-nonmm-unstable tree, today's linux-next failed
> KUnit testing like this:
> 
> [14:20:52] Configuring KUnit Kernel ...
> [14:20:52] Building KUnit Kernel ...
> Populating config with:
> $ make ARCH=arm64 O=/tmp/next/arm64_kunit olddefconfig CROSS_COMPILE=aarch64-linux-gnu-
> Building with:
> $ make all compile_commands.json scripts_gdb ARCH=arm64 O=/tmp/next/arm64_kunit --jobs=192 CROSS_COMPILE=aarch64-linux-gnu-
> [14:21:06] Starting KUnit Kernel (1/1)...
> Running tests with:
> $ qemu-system-aarch64 -nodefaults -m 1024 -kernel /tmp/next/arm64_kunit/arch/arm64/boot/Image.gz -append 'kunit.enable=1 console=ttyAMA0 kunit_shutdown=reboot' -no-reboot -nographic -accel kvm -accel hvf -accel tcg -serial stdio -machine virt -cpu max
> [14:22:11] ============================================================
> [14:22:11] =================== cmdline (8 subtests) ===================
> [14:22:11] # cmdline_test_memparse: EXPECTATION FAILED at lib/tests/cmdline_kunit.c:249
> [14:22:11] Expected ret == e->result, but
> [14:22:11]     ret == -16 (0xfffffffffffffff0)
> [14:22:11]     e->result == -1 (0xffffffffffffffff)
> [14:22:11] when parsing '0xffffffffffffffff0'
> [14:22:11] [FAILED] cmdline_test_memparse
> [14:22:11]     # module: cmdline_kunit
> [14:22:11] # cmdline: pass:7 fail:1 skip:0 total:8
> [14:22:11] # Totals: pass:7 fail:1 skip:0 total:8
> [14:22:11] ===================== [FAILED] cmdline =====================
> [14:22:11] ============================================================
> [14:22:11] Testing complete. Ran 9225 tests: passed: 9104, failed: 1, skipped: 120
> [14:22:11] Failures: cmdline.cmdline_test_memparse
> [14:22:11] Test results stored in /tmp/next/arm64_kunit.json
> [14:22:11] Elapsed time: 79.318s total, 0.002s configuring, 14.632s building, 64.682s running
> 
> I honestly have no idea what's triggering this, nothing jumps out from
> the commits introduced in the merge of the branch.  If I have time later
> I'll try to bisect the issue.  I have used the tree from next-20260602
> instead.

bisect tells me it's due to:

  e595545005616 (Revert "lib: fix _parse_integer_limit() to handle overflow")

log:

git bisect start
# status: waiting for both good and bad commits
# bad: [604ad02caddab12905643d5191b4a301d87033ef] ocfs2: fix hung task in orphan recovery
git bisect bad 604ad02caddab12905643d5191b4a301d87033ef
# status: waiting for good commit(s), bad commit known
# good: [f548f2851b9323e7d0834d0c655a46644beeda64] MAINTAINERS: add Usama as a THP reviewer
git bisect good f548f2851b9323e7d0834d0c655a46644beeda64
# good: [1c5b1ee7d53126e8b0f027a7e7277a77900bb29f] signal: change sys_kill() to use SEND_SIG_NOINFO
git bisect good 1c5b1ee7d53126e8b0f027a7e7277a77900bb29f
# good: [a4db696aedb5f7ce197978df4a3c9e60c5c132cc] riscv: mm: fix concurrency in mark_new_valid_map()
git bisect good a4db696aedb5f7ce197978df4a3c9e60c5c132cc
# bad: [e595545005616ea8664184a6c81413fa52d46a49] Revert "lib: fix _parse_integer_limit() to handle overflow"
git bisect bad e595545005616ea8664184a6c81413fa52d46a49
# good: [8264c1b8de74dd1215ac2e35166db29562f16e23] tools/compiler: match glibc 2.42 definition of __attribute_const__
git bisect good 8264c1b8de74dd1215ac2e35166db29562f16e23
# first bad commit: [e595545005616ea8664184a6c81413fa52d46a49] Revert "lib: fix _parse_integer_limit() to handle overflow"

Attachment: signature.asc
Description: PGP signature