AI review of 2026-11-16_11-38-45-65abef8608a2-results.txt PR: #9035 https://github.com/cppcheck-opensource/cppcheck/pull/9035 Tested: 65abef8608a2829217b03a85b1ca5afc5fe0921a Merge base: e6188446626b9e3c0cd16ffe0e49c73b18e52793 Reviewed: 2026-11-16 16:04:47 UTC Model: claude-opus-5-5 (effort high) Results: 50 reviewed of 58 in the report (random sample) Verdicts: 50 improvement Tokens: 162763 input, 122304 cache read, 2496 cache write, 30702 output The verdicts are written by AI and can be wrong. ---- 1 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/c/cctbx/cctbx_2026.9+ds2+~3.30.1+ds1.orig.tar.xz Result: main cctbx-cctbx_project-4b6acc4/simtbx/nanoBragg/nanoBragg_standalone.c:4110:62: error: Uninitialized struct member: frame.header_size [uninitStructMember] Explanation: At line 4110 `frame` is passed by value to `ValueOf()` while several members (header_size, width, height, swap_bytes, mmapdata) are not yet set. The PR only suppresses uninitStructMember when an uninitvar diagnostic was already recorded for the same token, so a warning still remains at this spot. The five per-member messages are redundant duplicates. They are also of doubtful value: `ValueOf` almost certainly only reads `frame.header`, and copying a partly initialized struct is not itself a bug. Removing the noise is an improvement. ---- 2 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/c/cctbx/cctbx_2026.9+ds2+~3.30.1+ds1.orig.tar.xz Result: main cctbx-cctbx_project-4b6acc4/simtbx/nanoBragg/nanoBragg_standalone.c:4110:62: error: Uninitialized struct member: frame.height [uninitStructMember] Explanation: At line 4110 the struct `frame` is passed by value to `ValueOf` while some members (height, width, etc.) are still unset. The PR suppresses uninitStructMember when the same token has already been diagnosed by uninitvar. The lines are removed only for this location, so an uninitvar diagnostic for `frame` here most likely still exists. The five per-member uninitStructMember messages were therefore duplicate noise. They are also questionable on their own: copying a C struct with indeterminate members is harmless, and `ValueOf` likely reads only `frame.header`. Removing them reduces redundant or dubious warnings. ---- 3 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/c/cctbx/cctbx_2026.9+ds2+~3.30.1+ds1.orig.tar.xz Result: main cctbx-cctbx_project-4b6acc4/simtbx/nanoBragg/nanoBragg_standalone.c:4110:62: error: Uninitialized struct member: frame.mmapdata [uninitStructMember] Explanation: At line 4110, `frame` (col 62) is passed by value to `ValueOf` before `header_size`, `width`, `height`, `swap_bytes` and `mmapdata` are set. The PR only drops uninitStructMember when the same token has already been diagnosed through `diag()`, which normally means an uninitvar warning was emitted there. That warning is not in the diff, so it is still reported in both runs. The five per-member messages were therefore redundant noise on the same expression. They are also doubtful as real bugs: copying a partially initialized struct is harmless if `ValueOf` only reads `frame.header`, which is likely given how it is called here. Removing the redundant `frame.mmapdata` message is an improvement. ---- 4 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/c/cctbx/cctbx_2026.9+ds2+~3.30.1+ds1.orig.tar.xz Result: main cctbx-cctbx_project-4b6acc4/simtbx/nanoBragg/nanoBragg_standalone.c:4110:62: error: Uninitialized struct member: frame.swap_bytes [uninitStructMember] Explanation: At line 4110, `frame` is passed by value to `ValueOf` while several members (header_size, width, height, swap_bytes, mmapdata) are still unset, so the issue itself is real. The PR only suppresses uninitStructMember when an uninit diagnostic has already been recorded for the same token (`diag(tok, false)`). So this token is still reported, almost certainly by an uninitvar warning that appears unchanged and is therefore not in the diff. The per-member uninitStructMember messages, including this frame.swap_bytes one, were redundant duplicates. Removing them cuts noise without losing the finding. ---- 5 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/c/cctbx/cctbx_2026.9+ds2+~3.30.1+ds1.orig.tar.xz Result: main cctbx-cctbx_project-4b6acc4/simtbx/nanoBragg/nanoBragg_standalone.c:4110:62: error: Uninitialized struct member: frame.width [uninitStructMember] Explanation: The struct `frame` is passed by value to `ValueOf()` while several members (width, height, swap_bytes, mmapdata, header_size) are still unset. The PR only skips an uninitStructMember warning when an uninitvar warning has already been issued for the same token. So `frame` at 4110:62 is still flagged by the uninitvar warning, which appears in both runs and therefore not in the diff. The removed per-member warnings were redundant duplicates on the same expression. Removing them cuts noise without losing the diagnostic. ---- 6 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/k/kjs/kjs_5.103.0.orig.tar.xz Result: main kjs-5.103.0/src/kjs/bytecode/generator/types.cpp:274:12: error: Uninitialized struct member: t.flags [uninitStructMember] Explanation: `resolveType` returns a default-constructed `Type t` whose `flags` member is never initialized, so a warning here is valid. The PR only suppresses `uninitStructMember` when an uninitvar diagnostic was already recorded for the same token. An uninitvar warning about the same uninitialized read is therefore presumably still reported, unchanged, and so does not appear in the diff. Removing the duplicate `uninitStructMember` report reduces noise without losing the finding. The remaining uninitvar warning is not shown, so this is not fully certain. ---- 7 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:1056:64: error: Uninitialized struct member: h.bitrate [uninitStructMember] Explanation: Line 1056 passes the uninitialized local `struct splt_header h` by value to `splt_mp3_makehead`. The PR skips uninitStructMember when a diagnostic (uninitvar) has already been reported on the same token. The uninitvar warning for `h` does not appear in the diff, so it is still reported. Only the redundant per-member messages (h.bitrate and the others) go away. The underlying issue is still reported once, so this is a reduction of duplicate noise. ---- 8 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:1056:64: error: Uninitialized struct member: h.frame_data_space [uninitStructMember] Explanation: On line 1056, `h` is declared without initialization and passed by value to `splt_mp3_makehead` on the first loop iteration. The pull request suppresses uninitStructMember when the same token already has an uninitvar diagnostic. Only the per-member warnings disappear here; an uninitvar for `h` on this line is presumably still reported, because it does not appear in the diff. So the eight duplicate member messages (h.bitrate, h.frame_data_space, ...) are just noise for one issue, and removing them is a deduplication improvement. ---- 9 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:1056:64: error: Uninitialized struct member: h.framesize [uninitStructMember] Explanation: In `splt_mp3_findvalidhead`, `h` is declared without initialization and passed by value to `splt_mp3_makehead` at line 1056. The PR only suppresses `uninitStructMember` when an `uninitvar` diagnostic was already recorded for the same token (`mUninitDiags`). So an `uninitvar` warning for `h` at this location is still reported by both versions and does not appear in the diff. The removed per-member message (`h.framesize`) is a redundant duplicate of that warning, part of a burst of eight member messages for one problem. Less noise with no lost information is an improvement. ---- 10 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:1056:64: error: Uninitialized struct member: h.has_crc [uninitStructMember] Explanation: At line 1056 the uninitialized struct `h` is passed by value to `splt_mp3_makehead` on the first loop iteration. The PR only suppresses `uninitStructMember` when an `uninitvar` diagnostic has already been recorded for the same token. So this problem is still reported once as `uninitvar`. The removed lines were eight redundant per-member duplicates (`h.has_crc` among them). Removing that noise without losing the finding is an improvement. ---- 11 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:1056:64: error: Uninitialized struct member: h.padding [uninitStructMember] Explanation: On line 1056 the uninitialized local `struct splt_header h` is passed by value to `splt_mp3_makehead` on the first loop iteration. The PR only suppresses `uninitStructMember` when an `uninitvar` diagnostic was already recorded for the same token. So the `uninitvar` warning for `h` at this spot is presumably still reported; it is not in the diff, so it is unchanged. Only the eight redundant per-member messages (`h.padding`, `h.bitrate`, …) are dropped. This removes duplicate noise without losing information, so it is an improvement. ---- 12 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:1056:64: error: Uninitialized struct member: h.ptr [uninitStructMember] Explanation: At line 1056 the uninitialized local `struct splt_header h` is passed by value to `splt_mp3_makehead` before any member has been set. The PR only skips `uninitStructMember` when an `uninitvar` diagnostic was already recorded for the same token. That means `h` is still reported there as an uninitialized variable. The per-member messages (h.ptr, h.bitrate, etc.) were redundant duplicates of that one warning. Removing them cuts noise without losing information. ---- 13 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:392:56: error: Uninitialized struct member: first_frame_header.bitrate [uninitStructMember] Explanation: At line 392 the uninitialized struct `first_frame_header` is passed by value to `splt_mp3_makehead`. The PR suppresses `uninitStructMember` when an `uninitvar` diagnostic was already reported for the same token. The `uninitvar` warning for this expression does not appear in the diff, so it is still reported by both versions. Removing the eight redundant per-member messages for the same expression is deduplication of noise, not a lost true positive. ---- 14 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:392:56: error: Uninitialized struct member: first_frame_header.frame_data_space [uninitStructMember] Explanation: At line 392 the uninitialized struct `first_frame_header` is passed by value to `splt_mp3_makehead`. The PR suppresses uninitStructMember only when a diagnostic (normally uninitvar) was already recorded for the same token. That uninitvar warning is not in the diff, so it is still reported unchanged. The removed per-member uninitStructMember message for `frame_data_space` is therefore a redundant duplicate of an existing warning. Removing it cuts noise without losing the finding. ---- 15 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:392:56: error: Uninitialized struct member: first_frame_header.has_crc [uninitStructMember] Explanation: At line 392, `first_frame_header` is still uninitialized when it is passed by value to `splt_mp3_makehead`. The PR suppresses `uninitStructMember` only on tokens that already received a `uninitvar` diagnostic. The removed per-member warning (`has_crc`, along with the seven other members) is therefore redundant. The underlying issue is still reported once by `uninitvar` at the same location. Removing this duplicate noise is an improvement and loses no real detection. ---- 16 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:392:56: error: Uninitialized struct member: first_frame_header.main_data_begin [uninitStructMember] Explanation: Line 392 passes the uninitialized struct `first_frame_header` by value to `splt_mp3_makehead`. The PR only suppresses uninitStructMember when the same token is already in mUninitDiags, which means an uninitvar warning was already reported there. That uninitvar warning is not in the diff, so it is still emitted. The per-member uninitStructMember lines (main_data_begin and the others) were redundant duplicates of the same problem. Removing them cuts noise without losing the diagnostic. ---- 17 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:392:56: error: Uninitialized struct member: first_frame_header.ptr [uninitStructMember] Explanation: At line 392 the uninitialized struct `first_frame_header` is passed by value to `splt_mp3_makehead()`, so a problem does exist there. The PR only suppresses `uninitStructMember` when a diagnostic has already been recorded for the same token. The diff shows no new `uninitvar` line for this location, which suggests an `uninitvar` warning for `first_frame_header` at 392:56 is reported by both versions. If so, the issue is still reported once, and the eight per-member warnings removed here were redundant duplicates. Less noise with the same coverage is an improvement. ---- 18 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:400:64: error: Uninitialized struct member: h.bitrate [uninitStructMember] Explanation: Line 400 passes the uninitialized struct `h` by value to `splt_mp3_makehead` on the first loop iteration. The PR only suppresses `uninitStructMember` when an `uninitvar` diagnostic was already recorded for the same token. So an `uninitvar` warning for `h` is still reported at this location; it does not appear in the diff because both versions report it. The removed per-member message (one of eight for the same expression) was redundant noise. Deduplicating it makes the output cleaner without losing the finding. ---- 19 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:400:64: error: Uninitialized struct member: h.frame_data_space [uninitStructMember] Explanation: At line 400, `h` is declared without initialization and then passed by value to `splt_mp3_makehead(..., h, 0)`. Reading the whole uninitialized struct is a real issue. The PR only suppresses `uninitStructMember` when an uninit diagnostic was already recorded for the same token, so a `uninitvar` warning for `h` should still be reported here (no 'your' line removes it). The removed per-member messages (h.frame_data_space and seven others at the same spot) are redundant duplicates of that warning. Dropping them reduces noise without losing the finding. ---- 20 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:400:64: error: Uninitialized struct member: h.framesize [uninitStructMember] Explanation: At line 400 the uninitialized struct `h` is passed by value to `splt_mp3_makehead`. The PR suppresses an uninitStructMember warning only when an uninitvar diagnostic was already recorded for the same token, so Cppcheck still reports the uninitialized use of `h` there. The removed per-member warnings (h.framesize and the other members) only duplicated that report. Removing this duplicate noise improves the output without losing information. ---- 21 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:400:64: error: Uninitialized struct member: h.main_data_begin [uninitStructMember] Explanation: At line 400, `h` is declared without initialization and then passed by value to `splt_mp3_makehead(..., h, 0)` on the first loop iteration. Main reported one uninitStructMember warning per member of `h` at this token. The PR suppresses uninitStructMember when a diagnostic has already been recorded for the same token (normally the uninitvar warning for `h`). That uninitvar warning is not in the diff, so it is presumably still reported. Removing this h.main_data_begin message therefore only drops a redundant duplicate and reduces noise without losing information. ---- 22 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:400:64: error: Uninitialized struct member: h.padding [uninitStructMember] Explanation: On the first loop iteration, `h` is passed by value to `splt_mp3_makehead` before it has been initialized. The PR only suppresses `uninitStructMember` when an uninitialized-variable diagnostic has already been recorded for the same token. So the defect at line 400 is still reported, presumably by the `uninitvar` warning, which does not appear in the diff because it is unchanged. What disappears is a set of redundant per-member warnings (`h.padding` and the others) for the same issue. This reduces noise without losing the finding. ---- 23 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:400:64: error: Uninitialized struct member: h.ptr [uninitStructMember] Explanation: On the first loop iteration, `h` is passed by value to `splt_mp3_makehead()` while still uninitialized. The PR only suppresses `uninitStructMember` when an `uninitvar` diagnostic was already recorded for the same token. So the issue is still reported once as `uninitvar` for `h`. The eight redundant per-member messages (`h.ptr`, `h.bitrate`, …) at this location are noise. Removing this duplicate makes the output cleaner without losing the finding. ---- 24 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:400:64: error: Uninitialized struct member: h.sideinfo_size [uninitStructMember] Explanation: On the first loop iteration, `h` is uninitialized when it is passed by value to `splt_mp3_makehead` at line 400. Cppcheck flags this via the `uninitvar` diagnostic. That warning is not in the diff, so it is unchanged and presumably still reported at the same token. Main also emitted one `uninitStructMember` warning per member (eight lines here), which is redundant. The PR suppresses `uninitStructMember` when `uninitvar` was already reported for the same token. Removing `h.sideinfo_size` therefore deduplicates noise without losing the finding. ---- 25 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:952:54: error: Uninitialized struct member: h.bitrate [uninitStructMember] Explanation: Line 952 passes the uninitialized `struct splt_header h` by value to `splt_mp3_makehead()`, which fills in the header and returns it. The PR only suppresses `uninitStructMember` when a diagnostic (e.g. `uninitvar`) was already reported for the same token. So this removal is deduplication: eight per-member warnings collapse into the existing single report. The pattern itself is benign, since the callee overwrites `h`. Either way, removing these redundant member-by-member duplicates reduces noise without losing information. ---- 26 / 50 ---- Verdict: IMPROVEMENT (high confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:952:54: error: Uninitialized struct member: h.framesize [uninitStructMember] Explanation: At line 952 the uninitialized local `struct splt_header h` is passed by value to `splt_mp3_makehead`. The PR only suppresses `uninitStructMember` when `diag()` shows an uninitvar diagnostic was already recorded for the same token, so an `uninitvar` warning for `h` is still reported there. The removed per-member message (`h.framesize`, and the seven others) was a duplicate of that warning. Fewer redundant messages with no lost information is an improvement. ---- 27 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:952:54: error: Uninitialized struct member: h.has_crc [uninitStructMember] Explanation: Line 952 passes the uninitialized local `struct splt_header h` by value to `splt_mp3_makehead`. The PR only suppresses `uninitStructMember` when an `uninitvar` diagnostic has already been recorded for the same token. That `uninitvar` warning for `h` is not in the diff, so it is still reported in both versions. Dropping the per-member duplicates such as `h.has_crc` reduces noise without losing the finding. ---- 28 / 50 ---- Verdict: IMPROVEMENT (high confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:952:54: error: Uninitialized struct member: h.padding [uninitStructMember] Explanation: At line 952 the uninitialized struct `h` is passed by value to `splt_mp3_makehead`. The PR only suppresses `uninitStructMember` when an `uninitvar` diagnostic was already reported for the same token. So the issue is still reported once as uninitvar at this location. The per-member messages (h.padding and seven others) were redundant duplicates, and removing them reduces noise without losing the finding. ---- 29 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:952:54: error: Uninitialized struct member: h.ptr [uninitStructMember] Explanation: At line 952 the uninitialized local `struct splt_header h` is passed by value to splt_mp3_makehead, so reading it is a real problem. The PR suppresses uninitStructMember only when an uninitvar diagnostic was already recorded for the same token (diag(tok, false)). So an uninitvar warning for `h` is still reported at this location, and since it does not appear in the diff it is unchanged. The removed `h.ptr` line is one of eight redundant per-member duplicates of that warning. Removing it cuts noise without losing the finding. ---- 30 / 50 ---- Verdict: IMPROVEMENT (high confidence) Package: https://ftp.debian.org/debian/pool/main/m/mp3splt/mp3splt_2.6.2+20170630.orig.tar.gz Result: main mp3splt-2.6.2+20170630/libmp3splt/plugins/mp3_utils.c:952:54: error: Uninitialized struct member: h.sideinfo_size [uninitStructMember] Explanation: At line 952 the uninitialized local `struct splt_header h` is passed by value to `splt_mp3_makehead(header, ..., h, 0)`. The PR only skips `uninitStructMember` when an uninitialized-variable diagnostic was already recorded for the same token (`diag(tok, false)`). So an `uninitvar` warning for `h` at this spot is presumably still reported by both versions, which is why it does not appear in the diff. The removed line is one of eight per-member duplicates (`h.bitrate`, `h.framesize`, …, `h.sideinfo_size`) of that one issue. Dropping these redundant messages cuts noise without losing the finding. ---- 31 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/epc-gtpc-header.cc:733:43: error: Uninitialized struct member: bearerContext.cause [uninitStructMember] Explanation: `bearerContext` is a local `BearerContextCreated` whose `cause` member is never set before it is copied by `push_back`. The PR only suppresses `uninitStructMember` when `diag()` has already recorded the same token. That only happens when an uninitvar error was reported at that token, and no new or removed uninitvar line shows up for this location. So an uninitvar warning for the same use still exists, and the removed `uninitStructMember` message was a duplicate of it. Less redundant output is an improvement, though the remaining message may name the member less precisely. ---- 32 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/epc-pgw-application.cc:330:41: error: Uninitialized struct member: bearerContextOut.cause [uninitStructMember] Explanation: `bearerContextOut` is declared without an initializer. Its `cause` member is never set before the struct is copied into the list by `push_back`. The PR only skips `uninitStructMember` when the same token is already in `mUninitDiags`. Tokens are added there when a `uninitvar` diagnostic is reported at that location. No new `your` lines appear, so that `uninitvar` warning is unchanged and still emitted. The removed line is therefore a duplicate report of the same issue. Dropping it reduces noise without losing information. ---- 33 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/epc-x2.cc:334:47: error: Uninitialized struct member: params.sourceCellId [uninitStructMember] Explanation: `params` is a UeContextReleaseParams with only oldEnbUeX2apId and newEnbUeX2apId assigned. sourceCellId and targetCellId are never set before `params` is passed at line 334, so the underlying problem is real. The PR only suppresses uninitStructMember when another uninit diagnostic (normally uninitvar) has already been recorded for the same token. That warning is in neither diff line, so it was presumably reported by both main and the PR. This removal therefore drops a duplicate report at the same spot rather than hiding the bug, which matches the PR's intent. ---- 34 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/epc-x2.cc:334:47: error: Uninitialized struct member: params.targetCellId [uninitStructMember] Explanation: `params` (`UeContextReleaseParams`) has only `oldEnbUeX2apId` and `newEnbUeX2apId` set before it is passed to `RecvUeContextRelease`. `sourceCellId` and `targetCellId` stay uninitialized, so the underlying problem is real. The PR only suppresses `uninitStructMember` when the same token was already recorded in `mUninitDiags`. That set is filled when an `uninitvar` error is reported for that token. So the defect is still reported once as `uninitvar` at line 334, and this change just removes a duplicate warning for the same use. ---- 35 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-enb-phy.cc:864:12: error: Uninitialized struct member: ulcqi.m_sfnSf [uninitStructMember] Explanation: In `CreatePuschCqiReport`, `ulcqi` is a struct whose member `m_sfnSf` is never set before `return ulcqi;`. So the underlying finding is real. The PR only suppresses `uninitStructMember` when the same token is already in `mUninitDiags`. That set is filled by the uninitvar/valueFlowUninit path just before it reports an error. So line 864 should still carry an uninitvar error in both runs (unchanged, hence not in the diff). The removed `uninitStructMember` line is a duplicate report of the same problem. Dropping it reduces noise without losing the diagnosis, which is the intent of the PR (see the gtk.c test change). ---- 36 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-enb-rrc.cc:2863:45: error: Uninitialized struct member: i.dlGtpTeid [uninitStructMember] Explanation: `EpcX2Sap::ErabAdmittedItem i` only has `erabId` set before it is copied into `push_back(i)`, so `ulGtpTeid` and `dlGtpTeid` really are uninitialized. The PR only suppresses `uninitStructMember` when an uninit diagnostic was already recorded for the same token, i.e. when `uninitvar` already fired at this location. No new line appears in the diff, which fits that: the `uninitvar` report at 2863 is presumably still there. So this is a duplicate report being removed, not a lost true positive. The issue is still reported once, which is the intended noise reduction. ---- 37 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-enb-rrc.cc:2863:45: error: Uninitialized struct member: i.ulGtpTeid [uninitStructMember] Explanation: `EpcX2Sap::ErabAdmittedItem i` only has `erabId` set before it is copied by `push_back(i)`, so `ulGtpTeid` is uninitialized at that point. The PR suppresses uninitStructMember only when an uninit diagnostic has already been recorded for the same token (via `mUninitDiags`). So this issue is still reported by the uninitvar check at line 2863, and the removed message was a duplicate. Less duplicate noise with no lost information makes this an improvement. ---- 38 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-enb-rrc.cc:827:60: error: Uninitialized struct member: i.dlHfn [uninitStructMember] Explanation: Only `dlPdcpSn` and `ulPdcpSn` of the local `ErabsSubjectToStatusTransferItem i` are set before it is copied into the list by `push_back(i)`, so `dlHfn` really is uninitialized. The PR only skips uninitStructMember when the same token is already in `mUninitDiags`. That set is filled when an uninit diagnostic (uninitvar) has already been reported for that token. The uninitvar warning at 827:60 stays the same in both runs, so it does not appear in the diff. The removed line is therefore a duplicate report of the same defect at the same spot. The user still gets one warning for this location, so removing the redundant one reduces noise. ---- 39 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-enb-rrc.cc:827:60: error: Uninitialized struct member: i.erabId [uninitStructMember] Explanation: `i` is an `ErabsSubjectToStatusTransferItem` with only `dlPdcpSn` and `ulPdcpSn` set. It is then copied into the list by `push_back(i)`, so `erabId`, `dlHfn` and `ulHfn` really are uninitialized when copied. The PR does not decide that this is fine. It only skips `uninitStructMember` when an uninitialized-variable diagnostic was already issued for the same token. So the problem at line 827 should still be reported once through that earlier diagnostic, and the extra per-member messages go away. Removing these duplicates means less noise without losing the finding. Confidence is medium because the surviving warning is not shown in the diff. ---- 40 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-enb-rrc.cc:827:60: error: Uninitialized struct member: i.ulHfn [uninitStructMember] Explanation: At line 827, `i` (an `ErabsSubjectToStatusTransferItem`) is pushed into the list with only `dlPdcpSn` and `ulPdcpSn` set, so members like `ulHfn` really are uninitialized. The PR drops uninitStructMember only when `diag()` already has an uninitvar diagnostic recorded for the same token. So an uninitvar warning is still reported at this spot. Removing this per-member line just removes a duplicate report of the same defect, which reduces noise without losing the finding. ---- 41 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-rrc-header.cc:2316:57: error: Uninitialized struct member: srbToAddMod.logicalChannelConfig [uninitStructMember] Explanation: The PR now skips uninitStructMember when another uninit diagnostic (uninitvar) was already reported at the same token. Here `srbToAddMod.logicalChannelConfig` stays uninitialized when `optionalFields[0]` is false or `sel == 1`, and the struct is then copied at line 2316. No warning appears only in 'your', so the overlapping uninitvar report at this token is presumably still present. The issue is still reported; only the duplicate uninitStructMember message is gone, which reduces noise. ---- 42 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-rrc-header.cc:3888:59: error: Uninitialized struct member: measResultServFreq.measResultBestNeighCell [uninitStructMember] Explanation: The PR suppresses uninitStructMember only when another uninit diagnostic (normally uninitvar) has already been reported for the same token. At line 3888, `measResultServFreq` is pushed into the vector while `measResultBestNeighCell` may be uninitialized if `haveMeasResultBestNeighCell` is false. That case is still covered by the remaining diagnostic on the same token, so the removed message is a duplicate. Copying a partially-initialized aggregate into a vector is also questionable as a real bug, because the member is only read when its have-flag is set. Removing the redundant noise is an improvement. ---- 43 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/model/lte-rrc-header.cc:3888:59: error: Uninitialized struct member: measResultServFreq.measResultSCell [uninitStructMember] Explanation: At line 3888, `measResultServFreq` is copied by `push_back`. Its `measResultSCell` member is only filled in when `haveMeasResultSCell` is true. The PR only skips `uninitStructMember` when `diag()` has already recorded that token, which in practice means an `uninitvar` was already reported there. That `uninitvar` would appear in both runs, so it does not show up in the diff. The removed line is therefore a duplicate report on the same expression, and dropping it cuts noise. The warning itself is also weak: copying a struct whose optional sub-struct is guarded by a `have*` flag is a common, harmless pattern. Users lose nothing meaningful. ---- 44 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/lte/test/test-lte-x2-handover.cc:416:38: error: Uninitialized struct member: ueData.id [uninitStructMember] Explanation: The PR only suppresses uninitStructMember when an uninitvar diagnostic has already been recorded for the same token. There is no matching 'your' line, so the uninitvar finding for `ueData` at 416:38 is still reported and unchanged. Removing this line therefore drops a duplicate report for the same expression rather than losing a finding. That is less noise for users. If the original warning was itself a false positive (the declaration of `ueData` and any assignment of `id` are not shown), its removal is still not a loss. ---- 45 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/wifi/model/rate-control/rraa-wifi-manager.cc:190:12: error: Uninitialized struct member: threshold.m_ewnd [uninitStructMember] Explanation: `threshold` is a local `WifiRraaThresholds` that is never initialized and is returned at line 190. That return is only reached after `NS_ABORT_MSG`, which aborts, so the code is effectively unreachable. The PR only suppresses `uninitStructMember` when an `uninitvar` diagnostic was already recorded for the same token. So an `uninitvar` warning for `threshold` is still reported at this location. The removed lines were redundant per-member duplicates (`m_ewnd`, `m_mtl`, `m_ori`) of that warning. Dropping them reduces noise without losing information, and the reported issue is dubious anyway because the line cannot be reached. ---- 46 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/wifi/model/rate-control/rraa-wifi-manager.cc:190:12: error: Uninitialized struct member: threshold.m_mtl [uninitStructMember] Explanation: `threshold` is a local `WifiRraaThresholds` that is never initialized, and `return threshold;` sits after `NS_ABORT_MSG`, only to silence a compiler warning. The PR skips uninitStructMember when the same token has already been diagnosed through `diag()`, which normally means an uninitvar warning for `threshold` was issued there. That uninitvar line is not in the diff, so it is still reported. The removed per-member warning (`m_mtl`, and likewise `m_ewnd`/`m_ori`) is therefore a duplicate of an existing diagnostic. Also, because NS_ABORT_MSG terminates the program, the return is effectively unreachable, so these per-member reports were noise anyway. Removing them reduces clutter without losing information. ---- 47 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/wifi/model/rate-control/rraa-wifi-manager.cc:190:12: error: Uninitialized struct member: threshold.m_ori [uninitStructMember] Explanation: `threshold` is a default-constructed `WifiRraaThresholds` returned only after `NS_ABORT_MSG`, so the line is effectively unreachable. The PR only drops `uninitStructMember` when the same token is already in `mUninitDiags`, meaning another uninit diagnostic was already issued at 190:12. No uninitvar line for 190:12 appears in the diff, so that report is unchanged and still present in both runs. The removed message is therefore a duplicate, or a false positive if the abort is noreturn. Removing it reduces noise without losing information. ---- 48 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/wifi/model/rate-control/rrpaa-wifi-manager.cc:221:12: error: Uninitialized struct member: threshold.m_ewnd [uninitStructMember] Explanation: The PR suppresses uninitStructMember on a token where uninitvar has already been reported. The removed line is one of three per-member messages for `return threshold;`. No uninitvar line appears in the diff, so the uninitvar report on this token is still emitted by both versions, and this message was a duplicate. In practice, NS_ABORT_MSG aborts, so the return is unreachable and only silences a compiler warning. Either way, dropping the redundant struct-member message reduces noise without losing a distinct finding. ---- 49 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/wifi/model/rate-control/rrpaa-wifi-manager.cc:221:12: error: Uninitialized struct member: threshold.m_mtl [uninitStructMember] Explanation: The PR now skips uninitStructMember when another uninit diagnostic has already been reported for the same token. Here `return threshold;` returns a whole struct that is never initialized. An uninitvar warning on `threshold` at this location remains, which is why this member-level message was dropped. That warning is itself questionable, because `NS_ABORT_MSG` aborts and the return only exists to silence the compiler. Removing a redundant (and likely false-positive) per-member message is an improvement. ---- 50 / 50 ---- Verdict: IMPROVEMENT (medium confidence) Package: https://ftp.debian.org/debian/pool/main/n/ns3/ns3_3.48.orig.tar.bz2 Result: main ns-3.48/src/wifi/model/rate-control/rrpaa-wifi-manager.cc:221:12: error: Uninitialized struct member: threshold.m_ori [uninitStructMember] Explanation: `threshold` is declared without initialization and returned at line 221. That return is only reached after `NS_ABORT_MSG`, so in practice it is unreachable. The PR only suppresses uninitStructMember when an uninitvar diagnostic was already recorded for the same token. So this removal means Cppcheck still reports uninitvar on this `return threshold;`, and the per-member uninitStructMember warnings (m_ori, m_ewnd, m_mtl) were duplicates of it. Whether the remaining warning is a true or false positive, dropping the redundant duplicates reduces noise without losing information.