[PATCH AUTOSEL 6.18-6.6] regmap: maple: Workaround for another false-positive compiler warning
From: Sasha Levin
Date: Mon Oct 19 2026 - 10:29:37 EST
From: Uwe Kleine-König (The Capable Hub) <u.kleine-koenig@xxxxxxxxxxxx>
[ Upstream commit dffd32eb2b0e714a8ae23f98518ad7152cb06369 ]
Similar to the issue fixed in commit 542440fd7b30 ("regmap: maple: work
around gcc-14.1 false-positive warning"), I see an uninitialized warning
for upper_index and upper_last.
Happens with arc-linux-gnu-gcc (Debian 15.3.0-1) 15.3.0 on an
allmodconfig for ARCH=arc.
Signed-off-by: Uwe Kleine-König (The Capable Hub) <u.kleine-koenig@xxxxxxxxxxxx>
Link: https://patch.msgid.link/20260626060427.2548578-2-u.kleine-koenig@xxxxxxxxxxxx
Signed-off-by: Mark Brown <broonie@xxxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
LLM Generated explanations, may be completely bogus:
I have the tree version, the pre-image at v6.18.53, the lore thread, and
the call-path context. Compiling the nine-phase analysis next.
---
## Phase 1: Commit message forensics
**Step 1.1: Subject**
Record: `[regmap: maple:] [Workaround] [silence a gcc false-positive
-Wuninitialized on upper_index/upper_last]`
**Step 1.2: Tags**
Record:
- Link: https://patch.msgid.link/20260626060427.2548578-2-u.kleine-
koenig@xxxxxxxxxxxx
- Signed-off-by: Uwe Kleine-König (author)
- Signed-off-by: Mark Brown (regmap maintainer)
- No Fixes:, Reported-by, Tested-by, Reviewed-by, Acked-by, or Cc:
stable (absence expected)
- Mentions prior commit 542440fd7b30
**Step 1.3: Body**
Record: [Bug] gcc 15.3.0 on ARCH=arc allmodconfig reports
-Wuninitialized for `upper_index` and `upper_last` in
`regcache_maple_drop()`. Same class as the gcc-14.1 false-positive
already worked around for `lower_index`/`lower_last`. [Symptom] compiler
warning; with `CONFIG_WERROR` it is a hard build failure
(`-Werror=uninitialized`). [Versions] gcc 15.3.0 (Debian), ARCH=arc
allmodconfig. [Root cause] compiler cannot see that those locals are
only read after they are assigned (guarded by `if (upper)`).
**Step 1.4: Hidden bug fix?**
Record: Not a hidden runtime bug. It is an explicit compiler-warning
workaround. It is a **build fix**, which the stable rules treat as an
allowed exception.
---
## Phase 2: Diff analysis
**Step 2.1: Inventory**
Record: `drivers/base/regmap/regcache-maple.c` — 1 line changed
(`+1/-1`). Function: `regcache_maple_drop()`. Scope: single-file, one-
line surgical change.
**Step 2.2: Flow**
Record: Before: `unsigned long upper_index, upper_last;` uninitialized.
After: `= 0` on both. No control-flow change. `upper_index`/`upper_last`
are assigned at lines 146–147 only when `mas.last > max`, which is also
when `upper` is allocated; they are consumed only under `if (upper)`
(line 171). Initializing to 0 is a dead store on the real path.
**Step 2.3: Bug mechanism**
Record: [Category] Build / compiler false-positive (not a runtime
uninitialized use). [Mechanism] gcc 15 cannot prove initialization
through the maple-tree loop and warns; with Werror that fails the build.
Fix: same dummy init already used for `lower_index`/`lower_last`.
**Step 2.4: Fix quality**
Record: Obviously correct; matches the existing comment and the gcc-14
workaround. Zero runtime behavior change. Regression risk: negligible (0
is never consumed unless `upper` is set, and that path always assigns
first).
---
## Phase 3: Git history
**Step 3.1: Blame**
Record: `regcache_maple_drop()` introduced in f033c26de5a57 (“regmap:
Add maple tree based register cache”, 2023-03-30, ~v6.4). `lower_index`
dummy init: 542440fd7b309 (2024-07-19, in this tree). `upper_index` was
left uninitialized until this commit.
**Step 3.2: Fixes: tag**
Record: None. Related original: 542440fd7b309, already an ancestor of
HEAD and of v6.18.53.
**Step 3.3: File history**
Record: Recent real fixes in this file (`00bb549d7d63a` cache
corruption, `eaa03486d9325` uninitialized `ret`, `f08f2d2907675` leak on
store failure) are independent. This commit is standalone, not patch X/Y
of a functional series.
**Step 3.4: Author**
Record: Uwe Kleine-König is a regular driver-core contributor, not the
regmap maintainer. Mark Brown applied it. No other related maple commits
by this author.
**Step 3.5: Dependencies**
Record: Needs the maple cache (present since v6.4) and the existing
`lower_*` init comment (present). Applies standalone. No new helpers or
struct fields.
---
## Phase 4: Mailing list
**Step 4.1: Original discussion**
Record: [Lore]
https://patch.msgid.link/20260626060427.2548578-2-u.kleine-
koenig@xxxxxxxxxxxx
[Revisions] v1 only (`b4 dig -a`). Thread: submission + Mark Brown
“Applied to broonie/regmap.git for-7.3”. No NAKs, no stable nomination,
no reviewer risk flags.
**Step 4.2: Recipients (`b4 dig -w`)**
Record: To: Mark Brown. Cc: Greg Kroah-Hartman, Rafael Wysocki, Danilo
Krummrich, linux-kernel, driver-core. Maintainer received and applied
it.
**Step 4.3: Bug report**
Record: Author-reported compiler warning on arc-linux-gnu-gcc 15.3.0
allmodconfig. No syzbot/bugzilla. Severity: build warning/error, not a
runtime crash.
**Step 4.4: Series**
Record: msgid `…2548578-2` but subject is `[PATCH v1]` and Mark applied
it as `[1/1]`. Standalone.
**Step 4.5: Stable list**
Record: No lore.kernel.org/stable discussion of this specific fix found.
---
## Phase 5: Semantic analysis
**Step 5.1: Functions**
Record: `regcache_maple_drop()` only.
**Step 5.2: Callers**
Record: Wired as `regcache_ops.drop`. Reached via
`regcache_drop_region()` (exported). Callers: regmap core, kunit tests,
various drivers. The *warning* triggers at compile time whenever this
file is built, independent of runtime callers.
**Step 5.3: Callees**
Record: maple-tree `mas_*`, `kmemdup_array`, `kfree`. Unchanged.
**Step 5.4: Reachability**
Record: Compile-time: `drivers/base/regmap/regcache-maple.c` is built
with `CONFIG_REGMAP` (common). Runtime drop path is not relevant to the
failure mode.
**Step 5.5: Similar patterns**
Record: Same function already initializes `lower_index`/`lower_last` for
the same reason (542440fd7b309) and `ret` for smatch (eaa03486d9325).
This completes the remaining pair.
---
## Phase 6: Local tree (this checkout)
**Step 6.1: Does the buggy code exist?**
Record: Local tree is **6.18.53** (`git describe`:
v6.18.53-5049-g0f0bdc2b0dffb; Makefile VERSION=6 PATCHLEVEL=18
SUBLEVEL=53). At **v6.18.53**:
```
unsigned long upper_index, upper_last;
```
exists; `lower_index = 0` is already present. Upstream dffd32eb2b0e is
**not** an ancestor of v6.18.53. Maple cache has been here since ~v6.4.
**Step 6.2: Backport difficulty**
Record: Clean apply on v6.18.53. Surrounding function matches the hunk.
No refactor in the way.
**Step 6.3: Related fixes already here?**
Record: gcc-14 `lower_*` workaround is in v6.18.53. Nothing already
initializes `upper_*`.
---
## Phase 7: Subsystem context
**Step 7.1**
Record: [Subsystem] drivers/base/regmap (maple register cache).
[Criticality] IMPORTANT — used by many hardware drivers; this change is
compile-time only.
**Step 7.2**
Record: Mature cache implementation (since v6.4), modest ongoing churn.
Not staging.
---
## Phase 8: Impact and risk
**Step 8.1: Who is affected**
Record: Anyone compiling this file with gcc that emits the warning.
Documented trigger: ARCH=arc, gcc 15.3.0, allmodconfig. `CONFIG_WERROR`
defaults to `COMPILE_TEST` (`init/Kconfig`); allmodconfig enables that,
so the warning becomes `-Werror=uninitialized` (same diagnostic the
gcc-14 commit showed as a hard error). Distro production configs usually
have WERROR=n, so they see a warning rather than a failed build.
**Step 8.2: Trigger**
Record: Compile of `regcache-maple.c`. Not a runtime race. Not
userspace-triggerable.
**Step 8.3: Failure mode**
Record: Build warning, or build *error* when WERROR is on. Severity:
MEDIUM for CI/allmodconfig (broken compile), LOW at runtime (no
functional bug).
**Step 8.4: Risk-benefit**
Record: Benefit: restores allmodconfig/WERROR builds on gcc 15 (and
likely similar compilers). Risk: very low — one-line initializer, proven
pattern in the same function. Ratio: favorable.
---
## Phase 9: Synthesis
**Step 9.1: Evidence**
FOR:
- Explicit **build-fix** exception (compilation warnings/errors).
- Reproduced: gcc 15.3.0, ARCH=arc allmodconfig.
- With `CONFIG_WERROR` this is a hard build failure
(`-Werror=uninitialized`), same class as 542440fd7b309, which is
already in this tree.
- One line, obviously correct, no runtime change.
- Buggy (uninitialized) locals exist in v6.18.53; patch applies cleanly.
- Maintainer (Mark Brown) applied it; no NAKs.
AGAINST:
- Not a runtime crash/corruption/security bug.
- Applied to for-7.3 (merge window), not a -fixes pull, so not treated
as urgent by the maintainer.
- Narrowest documented trigger is ARC + gcc 15 + allmodconfig.
- No Cc: stable (expected; not a negative).
UNRESOLVED:
- Whether gcc 15 also warns on x86/arm allmodconfig for 6.18 was not
reproduced here (author only cited ARC).
- lore.kernel.org HTTP fetch returned 403; discussion recovered via `b4
dig -m` and openwall/patchew mirrors.
**Step 9.2: Stable rules**
1. Obviously correct and tested? **PASS** — dummy init; author built ARC
allmodconfig; maintainer applied.
2. Real bug affecting users? **PASS** as a build break under
WERROR/allmodconfig; not a runtime bug.
3. Important (crash/security/corruption/deadlock)? **FAIL** as a runtime
issue; **PASS** under the build-fix exception.
4. Small and contained? **PASS** — 1 line, 1 file.
5. No new features/APIs? **PASS**.
6. Can apply to this tree? **PASS** — v6.18.53 has the pre-image.
**Step 9.3: Exception**
Record: **Build fix** (compiler warning that becomes a compile error
with WERROR).
**Step 9.4: Decision**
This is a one-line, zero-risk workaround for a real gcc 15 false-
positive that breaks `CONFIG_WERROR` builds of a file this 6.18.y tree
already ships. It matches the gcc-14 workaround already present here. It
meets the build-fix exception and applies cleanly.
---
### What the commit solves for this tree
v6.18.53 still declares `upper_index`/`upper_last` uninitialized in
`regcache_maple_drop()`. gcc 15.3.0 (ARC allmodconfig) warns; with
WERROR that fails the kernel build. Initializing them to 0, as already
done for `lower_index`/`lower_last`, silences the diagnostic without
changing runtime behavior.
### Risk vs benefit
Benefit is restoring compiles for WERROR/allmodconfig (and silencing the
warning otherwise). Risk is a dead store of 0. No prerequisites beyond
code already in 6.18.y.
### Verification
- [Phase 1] Parsed tags: Link to msgid, SOBs from Uwe Kleine-König and
Mark Brown; no Fixes/Reported-by/Cc: stable
- [Phase 2] Diff: 1 line in `regcache_maple_drop()`; `upper_index,
upper_last` → `= 0`
- [Phase 2] Read function: `upper_*` assigned only when `mas.last >
max`, used only under `if (upper)`
- [Phase 3] blame: maple cache f033c26de5a57 (v6.4); lower_* init
542440fd7b309 (in this tree)
- [Phase 3] `git merge-base --is-ancestor 542440fd7b309 HEAD` → yes;
`dffd32eb2b0e` vs v6.18.53 → not ancestor
- [Phase 3] `git show 542440fd7b309`: same -Werror=uninitialized class
for `lower_index`
- [Phase 4] `b4 dig -c dffd32eb2b0e`: lore msgid
20260626060427.2548578-2; v1 only; applied for-7.3
- [Phase 4] `b4 dig -w`: Mark Brown, Greg KH, driver-core CC’d
- [Phase 4] mbox: Mark Brown applied [1/1]; no NAK, no stable Cc
- [Phase 4] WebFetch lore/msgid.link: 403; recovered via openwall +
patchew + mbox
- [Phase 5] callers: `regcache_ops.drop` → `regcache_drop_region()`;
warning is compile-time
- [Phase 6] `git describe HEAD` / `make kernelversion`: 6.18.53
- [Phase 6] `v6.18.53:drivers/base/regmap/regcache-maple.c` still has
`unsigned long upper_index, upper_last;`
- [Phase 7] init/Kconfig: `config WERROR` default `COMPILE_TEST`;
Makefile.extrawarn adds `-Werror`
- [Phase 8] Failure mode: compile warning/error, not runtime oops
- UNVERIFIED: whether gcc 15 warns on non-ARC 6.18 allmodconfig (author
only reported ARC)
**YES**
drivers/base/regmap/regcache-maple.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/base/regmap/regcache-maple.c b/drivers/base/regmap/regcache-maple.c
index 9cf0384ce7b95..615ecb89d5377 100644
--- a/drivers/base/regmap/regcache-maple.c
+++ b/drivers/base/regmap/regcache-maple.c
@@ -112,7 +112,7 @@ static int regcache_maple_drop(struct regmap *map, unsigned int min,
unsigned long *entry, *lower, *upper;
/* initialized to work around false-positive -Wuninitialized warning */
unsigned long lower_index = 0, lower_last = 0;
- unsigned long upper_index, upper_last;
+ unsigned long upper_index = 0, upper_last = 0;
int ret = 0;
lower = NULL;
--
2.53.0