pkg/ifaceprobe: optimize cache

Instead of storing real PC values store indexes into the PCs table.
This significantly reduces size of the cache (in my case from
1823 MB to 473 MB) and actually makes use of the cache simpler
(don't need separate map).

3 files changed, 30 insertions, 35 deletions

diff --git a/pkg/declextract/fileops.go b/pkg/declextract/fileops.go
index 78016d57b..d28c48337 100644
--- a/pkg/declextract/fileops.go
+++ b/pkg/declextract/fileops.go
@@ -89,10 +89,6 @@ func (ctx *context) mapFopsToFiles() map[*FileOps][]string {
 			uniqueFuncs[fn]++
 		}
 	}
-	pcToFunc := make(map[uint64]string)
-	for _, pc := range ctx.probe.PCs {
-		pcToFunc[pc.PC] = pc.Func
-	}
 	// matchedFuncs holds functions are present in any file_operations callbacks
 	// (lots of coverage is not related to any file_operations at all).
 	matchedFuncs := make(map[string]bool)
@@ -102,7 +98,7 @@ func (ctx *context) mapFopsToFiles() map[*FileOps][]string {
 		funcs := make(map[string]bool)
 		fileToFuncs[file.Name] = funcs
 		for _, pc := range file.Cover {
-			fn := pcToFunc[pc]
+			fn := ctx.probe.PCs[pc].Func
 			if len(funcToFops[fn]) != 0 {
 				funcs[fn] = true
 				matchedFuncs[fn] = true
diff --git a/pkg/ifaceprobe/ifaceprobe.go b/pkg/ifaceprobe/ifaceprobe.go
index b2b3569df..f16be42a4 100644
--- a/pkg/ifaceprobe/ifaceprobe.go
+++ b/pkg/ifaceprobe/ifaceprobe.go
@@ -30,12 +30,11 @@ type Info struct {
 }
 
 type FileInfo struct {
-	Name  string   // Full file name, e.g. /dev/null.
-	Cover []uint64 // Combined coverage for operations on the file.
+	Name  string // Full file name, e.g. /dev/null.
+	Cover []int  // Combined coverage for operations on the file.
 }
 
 type PCInfo struct {
-	PC   uint64
 	Func string
 	File string
 }
@@ -82,7 +81,7 @@ func (pr *prober) run() (*Info, error) {
 	}()
 
 	info := &Info{}
-	dedup := make(map[uint64]bool)
+	pcIndexes := make(map[uint64]int)
 	kernelObj := filepath.Join(pr.cfg.KernelObj, pr.cfg.SysTarget.KernelObject)
 	sourceBase := filepath.Clean(pr.cfg.KernelSrc) + string(filepath.Separator)
 	i := 0
@@ -102,25 +101,28 @@ func (pr *prober) run() (*Info, error) {
 					continue
 				}
 				fileDedup[pc] = true
-				fi.Cover = append(fi.Cover, pc)
-				if dedup[pc] {
-					continue
-				}
-				dedup[pc] = true
-				frames, err := symb.Symbolize(kernelObj, pc)
-				if err != nil {
-					return nil, err
+				pcIndex, ok := pcIndexes[pc]
+				if !ok {
+					pcIndex = -1
+					frames, err := symb.Symbolize(kernelObj, pc)
+					if err != nil {
+						return nil, err
+					}
+					if len(frames) != 0 {
+						// Look only at the non-inline frame,
+						// callbacks we are looking for can't be inlined.
+						frame := frames[len(frames)-1]
+						pcIndex = len(info.PCs)
+						info.PCs = append(info.PCs, PCInfo{
+							Func: frame.Func,
+							File: strings.TrimPrefix(filepath.Clean(frame.File), sourceBase),
+						})
+					}
+					pcIndexes[pc] = pcIndex
 				}
-				if len(frames) == 0 {
-					continue
+				if pcIndex >= 0 {
+					fi.Cover = append(fi.Cover, pcIndex)
 				}
-				// Look only at the non-inline frame, callbacks we are looking for can't be inlined.
-				frame := frames[len(frames)-1]
-				info.PCs = append(info.PCs, PCInfo{
-					PC:   pc,
-					Func: frame.Func,
-					File: strings.TrimPrefix(filepath.Clean(frame.File), sourceBase),
-				})
 			}
 		}
 		slices.Sort(fi.Cover)
@@ -129,9 +131,6 @@ func (pr *prober) run() (*Info, error) {
 	slices.SortFunc(info.Files, func(a, b FileInfo) int {
 		return strings.Compare(a.Name, b.Name)
 	})
-	slices.SortFunc(info.PCs, func(a, b PCInfo) int {
-		return int(a.PC - b.PC)
-	})
 	select {
 	case err := <-pr.errc:
 		return nil, err
diff --git a/tools/syz-declextract/testdata/file_operations.c.probe b/tools/syz-declextract/testdata/file_operations.c.probe
index 6ff4a213a..dbc15f826 100644
--- a/tools/syz-declextract/testdata/file_operations.c.probe
+++ b/tools/syz-declextract/testdata/file_operations.c.probe
@@ -2,14 +2,14 @@
 	"Files": [
 		{
 			"Name": "/dev/foo",
-			"Cover": [1, 2, 3, 4, 5]
+			"Cover": [0, 1, 2, 3, 4]
 		}
 	],
 	"PCs": [
-		{"PC": 1, "Func": "foo_open"},
-		{"PC": 2, "Func": "foo_read"},
-		{"PC": 3, "Func": "foo_write"},
-		{"PC": 4, "Func": "foo_mmap"},
-		{"PC": 5, "Func": "foo_ioctl"}
+		{"Func": "foo_open"},
+		{"Func": "foo_read"},
+		{"Func": "foo_write"},
+		{"Func": "foo_mmap"},
+		{"Func": "foo_ioctl"}
 	]
 }