]> git.ipfire.org Git - thirdparty/kernel/linux.git/commitdiff
projects / thirdparty / kernel / linux.git / commitdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | patch | inline | side by side (parent: c7c6363)
hfsplus: fix slab-out-of-bounds read in hfsplus_uni2asc()
authorViacheslav Dubeyko <slava@dubeyko.com>
Thu, 10 Jul 2025 23:08:30 +0000 (16:08 -0700)
committerViacheslav Dubeyko <slava@dubeyko.com>
Fri, 25 Jul 2025 22:27:21 +0000 (15:27 -0700)
The hfsplus_readdir() method is capable to crash by calling
hfsplus_uni2asc():

[  667.121659][ T9805] ==================================================================
[  667.122651][ T9805] BUG: KASAN: slab-out-of-bounds in hfsplus_uni2asc+0x902/0xa10
[  667.123627][ T9805] Read of size 2 at addr ffff88802592f40c by task repro/9805
[  667.124578][ T9805]
[  667.124876][ T9805] CPU: 3 UID: 0 PID: 9805 Comm: repro Not tainted 6.16.0-rc3 #1 PREEMPT(full)
[  667.124886][ T9805] Hardware name: QEMU Ubuntu 24.04 PC (i440FX + PIIX, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
[  667.124890][ T9805] Call Trace:
[  667.124893][ T9805]  <TASK>
[  667.124896][ T9805]  dump_stack_lvl+0x10e/0x1f0
[  667.124911][ T9805]  print_report+0xd0/0x660
[  667.124920][ T9805]  ? __virt_addr_valid+0x81/0x610
[  667.124928][ T9805]  ? __phys_addr+0xe8/0x180
[  667.124934][ T9805]  ? hfsplus_uni2asc+0x902/0xa10
[  667.124942][ T9805]  kasan_report+0xc6/0x100
[  667.124950][ T9805]  ? hfsplus_uni2asc+0x902/0xa10
[  667.124959][ T9805]  hfsplus_uni2asc+0x902/0xa10
[  667.124966][ T9805]  ? hfsplus_bnode_read+0x14b/0x360
[  667.124974][ T9805]  hfsplus_readdir+0x845/0xfc0
[  667.124984][ T9805]  ? __pfx_hfsplus_readdir+0x10/0x10
[  667.124994][ T9805]  ? stack_trace_save+0x8e/0xc0
[  667.125008][ T9805]  ? iterate_dir+0x18b/0xb20
[  667.125015][ T9805]  ? trace_lock_acquire+0x85/0xd0
[  667.125022][ T9805]  ? lock_acquire+0x30/0x80
[  667.125029][ T9805]  ? iterate_dir+0x18b/0xb20
[  667.125037][ T9805]  ? down_read_killable+0x1ed/0x4c0
[  667.125044][ T9805]  ? putname+0x154/0x1a0
[  667.125051][ T9805]  ? __pfx_down_read_killable+0x10/0x10
[  667.125058][ T9805]  ? apparmor_file_permission+0x239/0x3e0
[  667.125069][ T9805]  iterate_dir+0x296/0xb20
[  667.125076][ T9805]  __x64_sys_getdents64+0x13c/0x2c0
[  667.125084][ T9805]  ? __pfx___x64_sys_getdents64+0x10/0x10
[  667.125091][ T9805]  ? __x64_sys_openat+0x141/0x200
[  667.125126][ T9805]  ? __pfx_filldir64+0x10/0x10
[  667.125134][ T9805]  ? do_user_addr_fault+0x7fe/0x12f0
[  667.125143][ T9805]  do_syscall_64+0xc9/0x480
[  667.125151][ T9805]  entry_SYSCALL_64_after_hwframe+0x77/0x7f
[  667.125158][ T9805] RIP: 0033:0x7fa8753b2fc9
[  667.125164][ T9805] Code: 00 c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 48
[  667.125172][ T9805] RSP: 002b:00007ffe96f8e0f8 EFLAGS: 00000217 ORIG_RAX: 00000000000000d9
[  667.125181][ T9805] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fa8753b2fc9
[  667.125185][ T9805] RDX: 0000000000000400 RSI: 00002000000063c0 RDI: 0000000000000004
[  667.125190][ T9805] RBP: 00007ffe96f8e110 R08: 00007ffe96f8e110 R09: 00007ffe96f8e110
[  667.125195][ T9805] R10: 0000000000000000 R11: 0000000000000217 R12: 0000556b1e3b4260
[  667.125199][ T9805] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
[  667.125207][ T9805]  </TASK>
[  667.125210][ T9805]
[  667.145632][ T9805] Allocated by task 9805:
[  667.145991][ T9805]  kasan_save_stack+0x20/0x40
[  667.146352][ T9805]  kasan_save_track+0x14/0x30
[  667.146717][ T9805]  __kasan_kmalloc+0xaa/0xb0
[  667.147065][ T9805]  __kmalloc_noprof+0x205/0x550
[  667.147448][ T9805]  hfsplus_find_init+0x95/0x1f0
[  667.147813][ T9805]  hfsplus_readdir+0x220/0xfc0
[  667.148174][ T9805]  iterate_dir+0x296/0xb20
[  667.148549][ T9805]  __x64_sys_getdents64+0x13c/0x2c0
[  667.148937][ T9805]  do_syscall_64+0xc9/0x480
[  667.149291][ T9805]  entry_SYSCALL_64_after_hwframe+0x77/0x7f
[  667.149809][ T9805]
[  667.150030][ T9805] The buggy address belongs to the object at ffff88802592f000
[  667.150030][ T9805]  which belongs to the cache kmalloc-2k of size 2048
[  667.151282][ T9805] The buggy address is located 0 bytes to the right of
[  667.151282][ T9805]  allocated 1036-byte region [ffff88802592f000, ffff88802592f40c)
[  667.152580][ T9805]
[  667.152798][ T9805] The buggy address belongs to the physical page:
[  667.153373][ T9805] page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x25928
[  667.154157][ T9805] head: order:3 mapcount:0 entire_mapcount:0 nr_pages_mapped:0 pincount:0
[  667.154916][ T9805] anon flags: 0xfff00000000040(head|node=0|zone=1|lastcpupid=0x7ff)
[  667.155631][ T9805] page_type: f5(slab)
[  667.155997][ T9805] raw: 00fff00000000040 ffff88801b442f00 0000000000000000 dead000000000001
[  667.156770][ T9805] raw: 0000000000000000 0000000080080008 00000000f5000000 0000000000000000
[  667.157536][ T9805] head: 00fff00000000040 ffff88801b442f00 0000000000000000 dead000000000001
[  667.158317][ T9805] head: 0000000000000000 0000000080080008 00000000f5000000 0000000000000000
[  667.159088][ T9805] head: 00fff00000000003 ffffea0000964a01 00000000ffffffff 00000000ffffffff
[  667.159865][ T9805] head: ffffffffffffffff 0000000000000000 00000000ffffffff 0000000000000008
[  667.160643][ T9805] page dumped because: kasan: bad access detected
[  667.161216][ T9805] page_owner tracks the page as allocated
[  667.161732][ T9805] page last allocated via order 3, migratetype Unmovable, gfp_mask 0xd20c0(__GFP_IO|__GFP_FS|__GFP_NOWARN9
[  667.163566][ T9805]  post_alloc_hook+0x1c0/0x230
[  667.164003][ T9805]  get_page_from_freelist+0xdeb/0x3b30
[  667.164503][ T9805]  __alloc_frozen_pages_noprof+0x25c/0x2460
[  667.165040][ T9805]  alloc_pages_mpol+0x1fb/0x550
[  667.165489][ T9805]  new_slab+0x23b/0x340
[  667.165872][ T9805]  ___slab_alloc+0xd81/0x1960
[  667.166313][ T9805]  __slab_alloc.isra.0+0x56/0xb0
[  667.166767][ T9805]  __kmalloc_cache_noprof+0x255/0x3e0
[  667.167255][ T9805]  psi_cgroup_alloc+0x52/0x2d0
[  667.167693][ T9805]  cgroup_mkdir+0x694/0x1210
[  667.168118][ T9805]  kernfs_iop_mkdir+0x111/0x190
[  667.168568][ T9805]  vfs_mkdir+0x59b/0x8d0
[  667.168956][ T9805]  do_mkdirat+0x2ed/0x3d0
[  667.169353][ T9805]  __x64_sys_mkdir+0xef/0x140
[  667.169784][ T9805]  do_syscall_64+0xc9/0x480
[  667.170195][ T9805]  entry_SYSCALL_64_after_hwframe+0x77/0x7f
[  667.170730][ T9805] page last free pid 1257 tgid 1257 stack trace:
[  667.171304][ T9805]  __free_frozen_pages+0x80c/0x1250
[  667.171770][ T9805]  vfree.part.0+0x12b/0xab0
[  667.172182][ T9805]  delayed_vfree_work+0x93/0xd0
[  667.172612][ T9805]  process_one_work+0x9b5/0x1b80
[  667.173067][ T9805]  worker_thread+0x630/0xe60
[  667.173486][ T9805]  kthread+0x3a8/0x770
[  667.173857][ T9805]  ret_from_fork+0x517/0x6e0
[  667.174278][ T9805]  ret_from_fork_asm+0x1a/0x30
[  667.174703][ T9805]
[  667.174917][ T9805] Memory state around the buggy address:
[  667.175411][ T9805]  ffff88802592f300: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[  667.176114][ T9805]  ffff88802592f380: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[  667.176830][ T9805] >ffff88802592f400: 00 04 fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[  667.177547][ T9805]                       ^
[  667.177933][ T9805]  ffff88802592f480: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[  667.178640][ T9805]  ffff88802592f500: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[  667.179350][ T9805] ==================================================================

The hfsplus_uni2asc() method operates by struct hfsplus_unistr:

struct hfsplus_unistr {
__be16 length;
hfsplus_unichr unicode[HFSPLUS_MAX_STRLEN];
} __packed;

where HFSPLUS_MAX_STRLEN is 255 bytes. The issue happens if length
of the structure instance has value bigger than 255 (for example,
65283). In such case, pointer on unicode buffer is going beyond of
the allocated memory.

The patch fixes the issue by checking the length value of
hfsplus_unistr instance and using 255 value in the case if length
value is bigger than HFSPLUS_MAX_STRLEN. Potential reason of such
situation could be a corruption of Catalog File b-tree's node.

Reported-by: Wenzhi Wang <wenzhi.wang@uwaterloo.ca>
Signed-off-by: Liu Shixin <liushixin2@huawei.com>
Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
cc: Yangtao Li <frank.li@vivo.com>
cc: linux-fsdevel@vger.kernel.org
Reviewed-by: Yangtao Li <frank.li@vivo.com>
Link: https://lore.kernel.org/r/20250710230830.110500-1-slava@dubeyko.com
Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
fs/hfsplus/unicode.c patch | blob | blame | history

diff --git a/fs/hfsplus/unicode.c b/fs/hfsplus/unicode.c
index 73342c925a4b6efb900885e849fa797fdc63782e..36b6cf2a3abba4aa39fdea1890ad42df982f8845 100644 (file)
--- a/fs/hfsplus/unicode.c
+++ b/fs/hfsplus/unicode.c
@@ -132,7 +132,14 @@ int hfsplus_uni2asc(struct super_block *sb,
 
        op = astr;
        ip = ustr->unicode;
+
        ustrlen = be16_to_cpu(ustr->length);
+       if (ustrlen > HFSPLUS_MAX_STRLEN) {
+               ustrlen = HFSPLUS_MAX_STRLEN;
+               pr_err("invalid length %u has been corrected to %d\n",
+                       be16_to_cpu(ustr->length), ustrlen);
+       }
+
        len = *len_p;
        ce1 = NULL;
        compose = !test_bit(HFSPLUS_SB_NODECOMPOSE, &HFSPLUS_SB(sb)->flags);
A mirror of Linus' kernel repository