[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <aSBm8+ZXPfHHvozl@xsang-OptiPlex-9020>
Date: Fri, 21 Nov 2025 21:19:47 +0800
From: Oliver Sang <oliver.sang@...el.com>
To: Mikulas Patocka <mpatocka@...hat.com>
CC: kernel test robot <lkp@...el.com>, Dongsheng Yang
<dongsheng.yang@...ux.dev>, <oe-kbuild-all@...ts.linux.dev>,
<linux-kernel@...r.kernel.org>, <oliver.sang@...el.com>
Subject: Re: include/linux/fortify-string.h:571:4: warning: call to
'__write_overflow_field' declared with attribute warning: detected write
beyond size of field (1st parameter); maybe use struct_group()?
hi, Mikulas,
On Tue, Nov 18, 2025 at 07:24:49PM +0100, Mikulas Patocka wrote:
> Hi
>
> Would you explain what's wrong there?
>
> at drivers/md/dm-pcache/cache_key.c:816 there's
> cache_pos_copy(&cache->key_head, pos);
>
> cache_pos_copy is defined as:
> static inline void cache_pos_copy(struct pcache_cache_pos *dst, struct pcache_cache_pos *src)
> {
> memcpy(dst, src, sizeof(struct pcache_cache_pos));
> }
>
> cache->key_head has the type struct pcache_cache_pos
> pos has the type struct pcache_cache_pos *
> pos points to the local variable pos_tail, that has the type
> pcache_cache_pos.
>
> So, I don't see any access beyond end of structure there. Am I missing
> something?
by more tests, this could be compiler issue. if we change to use gcc-15, the
issue gone.
we will configure the bot to avoid sending report for this warning.
thanks
>
> Mikulas
>
>
> On Mon, 17 Nov 2025, kernel test robot wrote:
>
> > tree: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> > head: 6a23ae0a96a600d1d12557add110e0bb6e32730c
> > commit: 1d57628ff95b32d5cfa8d8f50e07690c161e9cf0 dm-pcache: add persistent cache target in device-mapper
> > date: 3 months ago
> > config: riscv-randconfig-r113-20251117 (https://download.01.org/0day-ci/archive/20251117/202511171758.8thF02dJ-lkp@intel.com/config)
> > compiler: riscv64-linux-gcc (GCC) 9.5.0
> > reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251117/202511171758.8thF02dJ-lkp@intel.com/reproduce)
> >
> > If you fix the issue in a separate patch/commit (i.e. not just a new version of
> > the same patch/commit), kindly add following tags
> > | Reported-by: kernel test robot <lkp@...el.com>
> > | Closes: https://lore.kernel.org/oe-kbuild-all/202511171758.8thF02dJ-lkp@intel.com/
> >
> > All warnings (new ones prefixed by >>):
> >
> > In file included from include/linux/string.h:382,
> > from include/linux/bitmap.h:13,
> > from include/linux/cpumask.h:12,
> > from include/linux/smp.h:13,
> > from include/linux/lockdep.h:14,
> > from include/linux/spinlock.h:63,
> > from include/linux/sched.h:37,
> > from include/linux/mempool.h:8,
> > from include/linux/bio.h:8,
> > from drivers/md/dm-pcache/segment.h:5,
> > from drivers/md/dm-pcache/cache.h:5,
> > from drivers/md/dm-pcache/cache_key.c:2:
> > In function 'fortify_memcpy_chk',
> > inlined from 'cache_pos_copy' at drivers/md/dm-pcache/cache.h:403:2,
> > inlined from 'cache_replay' at drivers/md/dm-pcache/cache_key.c:816:2:
> > >> include/linux/fortify-string.h:571:4: warning: call to '__write_overflow_field' declared with attribute warning: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Wattribute-warning]
> > 571 | __write_overflow_field(p_size_field, size);
> > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> >
> >
> > vim +/__write_overflow_field +571 include/linux/fortify-string.h
> >
> > a28a6e860c6cf2 Francis Laniel 2021-02-25 515
> > f68f2ff91512c1 Kees Cook 2021-04-20 516 /*
> > f68f2ff91512c1 Kees Cook 2021-04-20 517 * To make sure the compiler can enforce protection against buffer overflows,
> > f68f2ff91512c1 Kees Cook 2021-04-20 518 * memcpy(), memmove(), and memset() must not be used beyond individual
> > f68f2ff91512c1 Kees Cook 2021-04-20 519 * struct members. If you need to copy across multiple members, please use
> > f68f2ff91512c1 Kees Cook 2021-04-20 520 * struct_group() to create a named mirror of an anonymous struct union.
> > f68f2ff91512c1 Kees Cook 2021-04-20 521 * (e.g. see struct sk_buff.) Read overflow checking is currently only
> > f68f2ff91512c1 Kees Cook 2021-04-20 522 * done when a write overflow is also present, or when building with W=1.
> > f68f2ff91512c1 Kees Cook 2021-04-20 523 *
> > f68f2ff91512c1 Kees Cook 2021-04-20 524 * Mitigation coverage matrix
> > f68f2ff91512c1 Kees Cook 2021-04-20 525 * Bounds checking at:
> > f68f2ff91512c1 Kees Cook 2021-04-20 526 * +-------+-------+-------+-------+
> > f68f2ff91512c1 Kees Cook 2021-04-20 527 * | Compile time | Run time |
> > f68f2ff91512c1 Kees Cook 2021-04-20 528 * memcpy() argument sizes: | write | read | write | read |
> > f68f2ff91512c1 Kees Cook 2021-04-20 529 * dest source length +-------+-------+-------+-------+
> > f68f2ff91512c1 Kees Cook 2021-04-20 530 * memcpy(known, known, constant) | y | y | n/a | n/a |
> > f68f2ff91512c1 Kees Cook 2021-04-20 531 * memcpy(known, unknown, constant) | y | n | n/a | V |
> > f68f2ff91512c1 Kees Cook 2021-04-20 532 * memcpy(known, known, dynamic) | n | n | B | B |
> > f68f2ff91512c1 Kees Cook 2021-04-20 533 * memcpy(known, unknown, dynamic) | n | n | B | V |
> > f68f2ff91512c1 Kees Cook 2021-04-20 534 * memcpy(unknown, known, constant) | n | y | V | n/a |
> > f68f2ff91512c1 Kees Cook 2021-04-20 535 * memcpy(unknown, unknown, constant) | n | n | V | V |
> > f68f2ff91512c1 Kees Cook 2021-04-20 536 * memcpy(unknown, known, dynamic) | n | n | V | B |
> > f68f2ff91512c1 Kees Cook 2021-04-20 537 * memcpy(unknown, unknown, dynamic) | n | n | V | V |
> > f68f2ff91512c1 Kees Cook 2021-04-20 538 * +-------+-------+-------+-------+
> > f68f2ff91512c1 Kees Cook 2021-04-20 539 *
> > f68f2ff91512c1 Kees Cook 2021-04-20 540 * y = perform deterministic compile-time bounds checking
> > f68f2ff91512c1 Kees Cook 2021-04-20 541 * n = cannot perform deterministic compile-time bounds checking
> > f68f2ff91512c1 Kees Cook 2021-04-20 542 * n/a = no run-time bounds checking needed since compile-time deterministic
> > f68f2ff91512c1 Kees Cook 2021-04-20 543 * B = can perform run-time bounds checking (currently unimplemented)
> > f68f2ff91512c1 Kees Cook 2021-04-20 544 * V = vulnerable to run-time overflow (will need refactoring to solve)
> > f68f2ff91512c1 Kees Cook 2021-04-20 545 *
> > f68f2ff91512c1 Kees Cook 2021-04-20 546 */
> > 54d9469bc515dc Kees Cook 2021-06-24 547 __FORTIFY_INLINE bool fortify_memcpy_chk(__kernel_size_t size,
> > f68f2ff91512c1 Kees Cook 2021-04-20 548 const size_t p_size,
> > f68f2ff91512c1 Kees Cook 2021-04-20 549 const size_t q_size,
> > f68f2ff91512c1 Kees Cook 2021-04-20 550 const size_t p_size_field,
> > f68f2ff91512c1 Kees Cook 2021-04-20 551 const size_t q_size_field,
> > 475ddf1fce1ec4 Kees Cook 2023-04-07 552 const u8 func)
> > a28a6e860c6cf2 Francis Laniel 2021-02-25 553 {
> > a28a6e860c6cf2 Francis Laniel 2021-02-25 554 if (__builtin_constant_p(size)) {
> > f68f2ff91512c1 Kees Cook 2021-04-20 555 /*
> > f68f2ff91512c1 Kees Cook 2021-04-20 556 * Length argument is a constant expression, so we
> > f68f2ff91512c1 Kees Cook 2021-04-20 557 * can perform compile-time bounds checking where
> > fa35198f39571b Kees Cook 2022-09-19 558 * buffer sizes are also known at compile time.
> > f68f2ff91512c1 Kees Cook 2021-04-20 559 */
> > f68f2ff91512c1 Kees Cook 2021-04-20 560
> > f68f2ff91512c1 Kees Cook 2021-04-20 561 /* Error when size is larger than enclosing struct. */
> > fa35198f39571b Kees Cook 2022-09-19 562 if (__compiletime_lessthan(p_size_field, p_size) &&
> > fa35198f39571b Kees Cook 2022-09-19 563 __compiletime_lessthan(p_size, size))
> > a28a6e860c6cf2 Francis Laniel 2021-02-25 564 __write_overflow();
> > fa35198f39571b Kees Cook 2022-09-19 565 if (__compiletime_lessthan(q_size_field, q_size) &&
> > fa35198f39571b Kees Cook 2022-09-19 566 __compiletime_lessthan(q_size, size))
> > a28a6e860c6cf2 Francis Laniel 2021-02-25 567 __read_overflow2();
> > f68f2ff91512c1 Kees Cook 2021-04-20 568
> > f68f2ff91512c1 Kees Cook 2021-04-20 569 /* Warn when write size argument larger than dest field. */
> > fa35198f39571b Kees Cook 2022-09-19 570 if (__compiletime_lessthan(p_size_field, size))
> > f68f2ff91512c1 Kees Cook 2021-04-20 @571 __write_overflow_field(p_size_field, size);
> > f68f2ff91512c1 Kees Cook 2021-04-20 572 /*
> > f68f2ff91512c1 Kees Cook 2021-04-20 573 * Warn for source field over-read when building with W=1
> > f68f2ff91512c1 Kees Cook 2021-04-20 574 * or when an over-write happened, so both can be fixed at
> > f68f2ff91512c1 Kees Cook 2021-04-20 575 * the same time.
> > f68f2ff91512c1 Kees Cook 2021-04-20 576 */
> > fa35198f39571b Kees Cook 2022-09-19 577 if ((IS_ENABLED(KBUILD_EXTRA_WARN1) ||
> > fa35198f39571b Kees Cook 2022-09-19 578 __compiletime_lessthan(p_size_field, size)) &&
> > fa35198f39571b Kees Cook 2022-09-19 579 __compiletime_lessthan(q_size_field, size))
> > f68f2ff91512c1 Kees Cook 2021-04-20 580 __read_overflow2_field(q_size_field, size);
> > a28a6e860c6cf2 Francis Laniel 2021-02-25 581 }
> > f68f2ff91512c1 Kees Cook 2021-04-20 582 /*
> > f68f2ff91512c1 Kees Cook 2021-04-20 583 * At this point, length argument may not be a constant expression,
> > f68f2ff91512c1 Kees Cook 2021-04-20 584 * so run-time bounds checking can be done where buffer sizes are
> > f68f2ff91512c1 Kees Cook 2021-04-20 585 * known. (This is not an "else" because the above checks may only
> > f68f2ff91512c1 Kees Cook 2021-04-20 586 * be compile-time warnings, and we want to still warn for run-time
> > f68f2ff91512c1 Kees Cook 2021-04-20 587 * overflows.)
> > f68f2ff91512c1 Kees Cook 2021-04-20 588 */
> > f68f2ff91512c1 Kees Cook 2021-04-20 589
> > f68f2ff91512c1 Kees Cook 2021-04-20 590 /*
> > f68f2ff91512c1 Kees Cook 2021-04-20 591 * Always stop accesses beyond the struct that contains the
> > f68f2ff91512c1 Kees Cook 2021-04-20 592 * field, when the buffer's remaining size is known.
> > 311fb40aa0569a Kees Cook 2022-09-02 593 * (The SIZE_MAX test is to optimize away checks where the buffer
> > f68f2ff91512c1 Kees Cook 2021-04-20 594 * lengths are unknown.)
> > f68f2ff91512c1 Kees Cook 2021-04-20 595 */
> > 475ddf1fce1ec4 Kees Cook 2023-04-07 596 if (p_size != SIZE_MAX && p_size < size)
> > 3d965b33e40d97 Kees Cook 2023-04-07 597 fortify_panic(func, FORTIFY_WRITE, p_size, size, true);
> > 475ddf1fce1ec4 Kees Cook 2023-04-07 598 else if (q_size != SIZE_MAX && q_size < size)
> > 94fd44648dae2a Kees Cook 2025-07-29 599 fortify_panic(func, FORTIFY_READ, q_size, size, true);
> > 54d9469bc515dc Kees Cook 2021-06-24 600
> > 54d9469bc515dc Kees Cook 2021-06-24 601 /*
> > 54d9469bc515dc Kees Cook 2021-06-24 602 * Warn when writing beyond destination field size.
> > 54d9469bc515dc Kees Cook 2021-06-24 603 *
> > 2003e483a81cc2 Kees Cook 2024-06-19 604 * Note the implementation of __builtin_*object_size() behaves
> > 54d9469bc515dc Kees Cook 2021-06-24 605 * like sizeof() when not directly referencing a flexible
> > 54d9469bc515dc Kees Cook 2021-06-24 606 * array member, which means there will be many bounds checks
> > 54d9469bc515dc Kees Cook 2021-06-24 607 * that will appear at run-time, without a way for them to be
> > 54d9469bc515dc Kees Cook 2021-06-24 608 * detected at compile-time (as can be done when the destination
> > 54d9469bc515dc Kees Cook 2021-06-24 609 * is specifically the flexible array member).
> > 54d9469bc515dc Kees Cook 2021-06-24 610 * https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101832
> > 54d9469bc515dc Kees Cook 2021-06-24 611 */
> > 2003e483a81cc2 Kees Cook 2024-06-19 612 if (p_size_field != SIZE_MAX &&
> > 54d9469bc515dc Kees Cook 2021-06-24 613 p_size != p_size_field && p_size_field < size)
> > 54d9469bc515dc Kees Cook 2021-06-24 614 return true;
> > 54d9469bc515dc Kees Cook 2021-06-24 615
> > 54d9469bc515dc Kees Cook 2021-06-24 616 return false;
> > a28a6e860c6cf2 Francis Laniel 2021-02-25 617 }
> > a28a6e860c6cf2 Francis Laniel 2021-02-25 618
> >
> > :::::: The code at line 571 was first introduced by commit
> > :::::: f68f2ff91512c199ec24883001245912afc17873 fortify: Detect struct member overflows in memcpy() at compile-time
> >
> > :::::: TO: Kees Cook <keescook@...omium.org>
> > :::::: CC: Kees Cook <keescook@...omium.org>
> >
> > --
> > 0-DAY CI Kernel Test Service
> > https://github.com/intel/lkp-tests/wiki
> >
>
Powered by blists - more mailing lists