diff options
Diffstat (limited to 'pkg/subsystem/extractor.go')
| -rw-r--r-- | pkg/subsystem/extractor.go | 50 |
1 files changed, 35 insertions, 15 deletions
diff --git a/pkg/subsystem/extractor.go b/pkg/subsystem/extractor.go index 9794642d3..fe1c720aa 100644 --- a/pkg/subsystem/extractor.go +++ b/pkg/subsystem/extractor.go @@ -3,6 +3,8 @@ package subsystem +import "math" + // Extractor deduces the subsystems from the list of crashes. type Extractor struct { raw rawExtractorInterface @@ -37,6 +39,11 @@ func (e *Extractor) Extract(crashes []*Crash) []*Subsystem { reproCount++ } } + subsystems = removeParents(subsystems) + counts := make(map[*Subsystem]int) + for _, entry := range subsystems { + counts[entry]++ + } // If all reproducers hint at the same subsystem, take it as well. reproCounts := map[*Subsystem]int{} @@ -54,15 +61,14 @@ func (e *Extractor) Extract(crashes []*Crash) []*Subsystem { } // It can be the case that guilty paths point to several subsystems, but the reproducer // can clearly point to one of them. + // Let's consider it to be the strongest singal. if len(fromRepro) > 0 { - // If we do not exclude parents, there'll always be a root subsystem in there - // and, as a result, subsystems reproducers will always replace all other ones. - withoutParents := removeParents(subsystems) + fromRepro = removeParents(fromRepro) newSubsystems := []*Subsystem{} for _, reproSubsystem := range fromRepro { parents := reproSubsystem.ReachableParents() parents[reproSubsystem] = struct{}{} // also include the subsystem itself - for _, subsystem := range withoutParents { + for _, subsystem := range subsystems { if _, ok := parents[subsystem]; ok { newSubsystems = append(newSubsystems, reproSubsystem) break @@ -70,24 +76,38 @@ func (e *Extractor) Extract(crashes []*Crash) []*Subsystem { } } if len(newSubsystems) > 0 { - subsystems = newSubsystems + // Just pick those subsystems. + return newSubsystems } - } - // And calculate counts. - counts := make(map[*Subsystem]int) - maxCount := 0 - for _, entry := range removeParents(append(subsystems, fromRepro...)) { - counts[entry]++ - if counts[entry] > maxCount { - maxCount = counts[entry] + // If there are sufficiently many reproducers that point to subsystems other than + // those from guilty paths, there's a chance we just didn't parse report correctly. + const cutOff = 3 + if reproCount >= cutOff { + // But if the guilty paths are non-controversial, also take the leading candidate. + return append(fromRepro, mostVoted(counts, 0.66)...) } } - // Pick the most prevalent ones. + // Take subsystems from reproducers into account. + for _, entry := range fromRepro { + counts[entry] += reproCount + } + + // Let's pick all subsystems that received >= 33% of votes (thus no more than 3). + return removeParents(mostVoted(counts, 0.33)) +} + +// mostVoted picks subsystems that have received >= share votes. +func mostVoted(counts map[*Subsystem]int, share float64) []*Subsystem { + total := 0 + for _, count := range counts { + total += count + } + cutOff := int(math.Ceil(share * float64(total))) ret := []*Subsystem{} for entry, count := range counts { - if count < maxCount { + if count < cutOff { continue } ret = append(ret, entry) |