libbpf: bound /proc/<pid>/maps scansets in parse_vma_segs()
PATCH 1/1 libbpf: bound /proc/<pid>/maps scansets in parse_vma_segs()
parse_vma_segs() in tools/lib/bpf/usdt.c parses /proc/<pid>/maps with two widthless scansets, "%s" into mode[16] and "%[^\n]" into line[PATH_MAX]. Both assume the kernel caps maps records to PATH_MAX, but it does not.
show_map_vma() emits the path via seq_path() against the seq buffer, which doubles on overflow (m->size <<= 1 in fs/seq_file.c), so a VMA whose backing path is a deeply nested directory tree produces a single maps record longer than PATH_MAX. scanf "%s" / "%[^\n]" without a width writes until the field terminator regardless of destination size, so a bpf_program__attach_usdt() consumer attaching against an attacker-controlled PID overflows its own stack inside parse_vma_segs().
Bound both scansets to the declared buffer sizes ("%15s" for mode[16] and "%4095[^\n]" for line[PATH_MAX]) and drain any residue past line[4094] with "%*[^\n]" before the trailing "\n", matching the libbpf-local fscanf style. Without the drain the residue of an over-long record would stay in the stream and break the next "%zx-%zx" parse, so the loop would exit early and any maps records after the over-long entry would be silently skipped.
---
Reproduced with Debian 12 on rootless podman: an unprivileged container process mkdirs 50 nested 200-char directories and mmaps a file at the bottom, producing a 10403-byte /proc/<host-pid>/maps line. A harness on the host then calls the real parse_vma_segs() against the container's PID; libbpf is built with -fsanitize=address and the only local source change is dropping the "static" keyword on parse_vma_segs so the symbol is linkable from the harness.
Stock libbpf reports:
==ERROR: AddressSanitizer: stack-buffer-overflow WRITE of size 10349 at <line> thread T0
#0 scanf_common -> #1 __isoc99_fscanf
#3 parse_vma_segs tools/lib/bpf/usdt.c:509 Address ... in frame parse_vma_segs at offset 8512, just past line[PATH_MAX].
Patched libbpf parses the same maps cleanly. Follow-up calls return 0 with seg_cnt > 0 for libc.so.6 and for ld-linux-x86-64.so.2 (format drain), which appears in maps after the over-long entry.
On normal hardened builds the stack canary aborts the consumer; on builds without stack protector the bytes past line[] are attacker-influenced path bytes.
Selftest gate
=============
tools/testing/selftests/bpf/test_progs -t usdt under QEMU x86_64 (KVM) on the patched kernel: all 6 subtests pass (usdt/basic, basic_optimized, optimized_attach, multispec, urand_auto_attach, urand_pid_attach) on both stock and patched libbpf, diff-clean. The in-tree selftest does not itself exercise long maps records.
tools/lib/bpf/usdt.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
tools/lib/bpf/usdt.c +4 -1
@@ -504,8 +504,11 @@ static int parse_vma_segs(int pid, const char *lib_path, struct elf_seg **segs,* 7f5c6f5d1000-7f5c6f5d3000 rw-p 001c7000 08:04 21238613 /usr/lib64/libc-2.17.so * 7f5c6f5d3000-7f5c6f5d8000 rw-p 00000000 00:00 0 * 7f5c6f5d8000-7f5c6f5d9000 r-xp 00000000 103:01 362990598 /data/users/andriin/linux/tools/bpf/usdt/libhello_usdt.so +* +* Bound the writes and drain residue: maps lines can exceed +* PATH_MAX when d_path() expands beyond it. */ -while (fscanf (f,"%zx-%zx %s %zx %*s %*d%[^\n]\n" , +while (fscanf (f,"%zx-%zx %15s %zx %*s %*d%4095[^\n]%*[^\n]\n" , &seg_start, &seg_end, mode, &seg_off, line) ==5 ) {void *tmp;
base-commit: 5200f5f493f79f14bbdc349e402a40dfb32f23c8
Signature · 1 line
2.53.0
[email protected]
AI reviewThank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [Critical] Unbounded %s format specifier in sscanf allows stack buffer overflow and truncates paths with spaces. --
Note: The format of this report is altered due to recitation restrictions. I am providing a free-form summary of the findings without quoting the original patch code.
commit 4f7490fd6defafa70e5f169165e4d30d7e6cabcd
Author: Michael Bommarito <[email protected]>
Subject: libbpf: bound /proc/<pid>/maps scansets in parse_vma_segs()
This commit addresses a stack buffer overflow in parse_vma_segs() by adding bounds to the fscanf calls used to parse /proc/<pid>/maps. It ensures that deeply nested directory trees do not overflow the path and mode buffers.
[Severity: Critical]
This is a pre-existing issue, but while this patch successfully adds bounds to the fscanf call reading maps entries, is it possible that the earlier sscanf call in parse_vma_segs() still has an unbounded width issue?
The function uses sscanf on lib_path with the format string "/proc/%d/root%s" to extract the remainder into a local stack buffer sized to PATH_MAX. If a tracing tool accepts an untrusted lib_path longer than PATH_MAX that starts with the root prefix, won't this sscanf write out of bounds and overflow the stack buffer?
Additionally, since the %s format specifier stops at the first whitespace, wouldn't any valid paths containing spaces be truncated here? If so, this might cause the later strcmp against the maps file to fail silently, potentially breaking USDT attachment for libraries residing in directories with spaces.
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1