Message ID | 20240823104310.4076479-1-aha310510@gmail.com (mailing list archive) |
---|---|
State | Changes Requested |
Delegated to: | BPF |
Headers | show |
Series | [bpf] bpf: add check for invalid name in btf_name_valid_section() | expand |
On Fri, Aug 23, 2024 at 3:43 AM Jeongjun Park <aha310510@gmail.com> wrote: > > If the length of the name string is 1 and the value of name[0] is NULL > byte, an OOB vulnerability occurs in btf_name_valid_section() and the > return value is true, so the invalid name passes the check. > > To solve this, you need to check if the first position is NULL byte. > > Fixes: bd70a8fb7ca4 ("bpf: Allow all printable characters in BTF DATASEC names") > Signed-off-by: Jeongjun Park <aha310510@gmail.com> > --- > kernel/bpf/btf.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c > index 520f49f422fe..5c24ea1a65a4 100644 > --- a/kernel/bpf/btf.c > +++ b/kernel/bpf/btf.c > @@ -823,6 +823,9 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) > const char *src = btf_str_by_offset(btf, offset); > const char *src_limit; > > + if (!*src) > + return false; > + We've talked about it. Quote: "Pls add a selftest that demonstrates the issue and produce a patch to fix just that." length == 1 and name[0] = 0 is a hypothesis. Demonstrate that such a scenario is possible then this patch will be worth applying. pw-bot: cr
Alexei Starovoitov wrote: > > On Fri, Aug 23, 2024 at 3:43 AM Jeongjun Park <aha310510@gmail.com> wrote: > > > > If the length of the name string is 1 and the value of name[0] is NULL > > byte, an OOB vulnerability occurs in btf_name_valid_section() and the > > return value is true, so the invalid name passes the check. > > > > To solve this, you need to check if the first position is NULL byte. > > > > Fixes: bd70a8fb7ca4 ("bpf: Allow all printable characters in BTF DATASEC names") > > Signed-off-by: Jeongjun Park <aha310510@gmail.com> > > --- > > kernel/bpf/btf.c | 3 +++ > > 1 file changed, 3 insertions(+) > > > > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c > > index 520f49f422fe..5c24ea1a65a4 100644 > > --- a/kernel/bpf/btf.c > > +++ b/kernel/bpf/btf.c > > @@ -823,6 +823,9 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) > > const char *src = btf_str_by_offset(btf, offset); > > const char *src_limit; > > > > + if (!*src) > > + return false; > > + > > We've talked about it. Quote: > "Pls add a selftest that demonstrates the issue > and produce a patch to fix just that." > > length == 1 and name[0] = 0 is a hypothesis. > Demonstrate that such a scenario is possible then this patch will be > worth applying. > > pw-bot: cr Sorry for the omission, I still don't know how to write selftest. But I can give you the C repro and KASAN log that trigger this vulnerability. I would appreciate it if you could look at it and make a judgment. Regards, Jeongjun Park C repro: // gcc -o repro repro.c #define _GNU_SOURCE #include <endian.h> #include <stdint.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <sys/syscall.h> #include <sys/types.h> #include <unistd.h> #ifndef __NR_bpf #define __NR_bpf 321 #endif int main(void) { syscall(__NR_mmap, /*addr=*/0x1ffff000ul, /*len=*/0x1000ul, /*prot=*/0ul, /*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/0x32ul, /*fd=*/-1, /*offset=*/0ul); syscall(__NR_mmap, /*addr=*/0x20000000ul, /*len=*/0x1000000ul, /*prot=PROT_WRITE|PROT_READ|PROT_EXEC*/7ul, /*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/0x32ul, /*fd=*/-1, /*offset=*/0ul); syscall(__NR_mmap, /*addr=*/0x21000000ul, /*len=*/0x1000ul, /*prot=*/0ul, /*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/0x32ul, /*fd=*/-1, /*offset=*/0ul); const char* reason; (void)reason; if (write(1, "executing program\n", sizeof("executing program\n") - 1)) {} *(uint64_t*)0x20000340 = 0x20000140; *(uint16_t*)0x20000140 = 0xeb9f; *(uint8_t*)0x20000142 = 1; *(uint8_t*)0x20000143 = 0; *(uint32_t*)0x20000144 = 0x18; *(uint32_t*)0x20000148 = 0; *(uint32_t*)0x2000014c = 0xc4; *(uint32_t*)0x20000150 = 0xc4; *(uint32_t*)0x20000154 = 5; *(uint32_t*)0x20000158 = 4; *(uint16_t*)0x2000015c = 0xa; *(uint8_t*)0x2000015e = 0; *(uint8_t*)0x2000015f = 0xf; *(uint32_t*)0x20000160 = 1; *(uint32_t*)0x20000164 = 3; *(uint32_t*)0x20000168 = 1; *(uint32_t*)0x2000016c = 0x9247; *(uint32_t*)0x20000170 = 3; *(uint32_t*)0x20000174 = 0xfffffff9; *(uint32_t*)0x20000178 = 9; *(uint32_t*)0x2000017c = 5; *(uint32_t*)0x20000180 = 0x7d; *(uint32_t*)0x20000184 = 0x48; *(uint32_t*)0x20000188 = 3; *(uint32_t*)0x2000018c = 9; *(uint32_t*)0x20000190 = 2; *(uint32_t*)0x20000194 = 1; *(uint32_t*)0x20000198 = 0xe; *(uint32_t*)0x2000019c = 5; *(uint32_t*)0x200001a0 = 3; *(uint32_t*)0x200001a4 = 0xfffffa64; *(uint32_t*)0x200001a8 = 0xffff; *(uint32_t*)0x200001ac = 3; *(uint32_t*)0x200001b0 = 3; *(uint32_t*)0x200001b4 = 0xfff; *(uint32_t*)0x200001b8 = 5; *(uint32_t*)0x200001bc = 0x8fd; *(uint32_t*)0x200001c0 = 0xa16; *(uint32_t*)0x200001c4 = 2; *(uint32_t*)0x200001c8 = 7; *(uint32_t*)0x200001cc = 6; *(uint32_t*)0x200001d0 = 2; *(uint32_t*)0x200001d4 = 0xe821; *(uint32_t*)0x200001d8 = 5; memset((void*)0x200001dc, 104, 1); *(uint32_t*)0x200001dd = 0xf; *(uint16_t*)0x200001e1 = 0; *(uint8_t*)0x200001e3 = 0; *(uint8_t*)0x200001e4 = 8; *(uint32_t*)0x200001e5 = 4; *(uint32_t*)0x200001e9 = 0; *(uint16_t*)0x200001ed = 0; *(uint8_t*)0x200001ef = 0; *(uint8_t*)0x200001f0 = 3; *(uint32_t*)0x200001f1 = 0; *(uint32_t*)0x200001f5 = 4; *(uint32_t*)0x200001f9 = 3; *(uint32_t*)0x200001fd = 2; *(uint32_t*)0x20000201 = 7; *(uint16_t*)0x20000205 = 0; *(uint8_t*)0x20000207 = 0; *(uint8_t*)0x20000208 = 0xa; *(uint32_t*)0x20000209 = 5; *(uint32_t*)0x2000020d = 3; *(uint16_t*)0x20000211 = 0; *(uint8_t*)0x20000213 = 0; *(uint8_t*)0x20000214 = 0xf; *(uint32_t*)0x20000215 = 3; memcpy((void*)0x20000219, "\x65\xd8\x38", 3); *(uint8_t*)0x2000021c = 0; *(uint8_t*)0x2000021d = 0; *(uint8_t*)0x2000021e = 0; *(uint8_t*)0x2000021f = 0; *(uint8_t*)0x20000220 = 0; *(uint64_t*)0x20000348 = 0; *(uint32_t*)0x20000350 = 0xe1; *(uint32_t*)0x20000354 = 0; *(uint32_t*)0x20000358 = 0; *(uint32_t*)0x2000035c = 0; syscall(__NR_bpf, /*cmd=*/0x12ul, /*arg=*/0x20000340ul, /*size=*/0x20ul); return 0; } KASAN log: ================================================================== BUG: KASAN: slab-out-of-bounds in btf_name_valid_section kernel/bpf/btf.c:829 [inline] BUG: KASAN: slab-out-of-bounds in btf_datasec_check_meta+0x7f8/0x880 kernel/bpf/btf.c:4698 Read of size 1 at addr ffff888012faaeb8 by task syz.3.2008/25618 CPU: 0 UID: 0 PID: 25618 Comm: syz.3.2008 Not tainted 6.11.0-rc5 #12 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <TASK> __dump_stack lib/dump_stack.c:93 [inline] dump_stack_lvl+0x116/0x1b0 lib/dump_stack.c:119 print_address_description mm/kasan/report.c:377 [inline] print_report+0xc0/0x5e0 mm/kasan/report.c:488 kasan_report+0xbd/0xf0 mm/kasan/report.c:601 btf_name_valid_section kernel/bpf/btf.c:829 [inline] btf_datasec_check_meta+0x7f8/0x880 kernel/bpf/btf.c:4698 btf_check_meta kernel/bpf/btf.c:5180 [inline] btf_check_all_metas kernel/bpf/btf.c:5204 [inline] btf_parse_type_sec kernel/bpf/btf.c:5340 [inline] btf_parse kernel/bpf/btf.c:5732 [inline] btf_new_fd+0x1764/0x4c30 kernel/bpf/btf.c:7650 bpf_btf_load kernel/bpf/syscall.c:5035 [inline] __sys_bpf+0x1dc4/0x5040 kernel/bpf/syscall.c:5755 __do_sys_bpf kernel/bpf/syscall.c:5817 [inline] __se_sys_bpf kernel/bpf/syscall.c:5815 [inline] __x64_sys_bpf+0x78/0xc0 kernel/bpf/syscall.c:5815 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xcb/0x250 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7f6dee39712d Code: 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 a8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007f6def181fa8 EFLAGS: 00000246 ORIG_RAX: 0000000000000141 RAX: ffffffffffffffda RBX: 00007f6dee535f80 RCX: 00007f6dee39712d RDX: 0000000000000020 RSI: 0000000020000600 RDI: 0000000000000012 RBP: 00007f6dee41bd8a R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 000000000000000b R14: 00007f6dee535f80 R15: 00007f6def162000 </TASK> Allocated by task 25618: kasan_save_stack+0x24/0x50 mm/kasan/common.c:47 kasan_save_track+0x14/0x30 mm/kasan/common.c:68 poison_kmalloc_redzone mm/kasan/common.c:370 [inline] __kasan_kmalloc+0xaa/0xb0 mm/kasan/common.c:387 kasan_kmalloc include/linux/kasan.h:211 [inline] __do_kmalloc_node mm/slub.c:4158 [inline] __kmalloc_node_noprof+0x20a/0x450 mm/slub.c:4164 __kvmalloc_node_noprof+0x148/0x1b0 mm/util.c:650 btf_parse kernel/bpf/btf.c:5708 [inline] btf_new_fd+0x5f9/0x4c30 kernel/bpf/btf.c:7650 bpf_btf_load kernel/bpf/syscall.c:5035 [inline] __sys_bpf+0x1dc4/0x5040 kernel/bpf/syscall.c:5755 __do_sys_bpf kernel/bpf/syscall.c:5817 [inline] __se_sys_bpf kernel/bpf/syscall.c:5815 [inline] __x64_sys_bpf+0x78/0xc0 kernel/bpf/syscall.c:5815 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xcb/0x250 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f The buggy address belongs to the object at ffff888012faae00 which belongs to the cache kmalloc-192 of size 192 The buggy address is located 0 bytes to the right of allocated 184-byte region [ffff888012faae00, ffff888012faaeb8) The buggy address belongs to the physical page: page: refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x12faa flags: 0xfff00000000000(node=0|zone=1|lastcpupid=0x7ff) page_type: 0xfdffffff(slab) raw: 00fff00000000000 ffff8880154413c0 ffffea00004b78c0 dead000000000002 raw: 0000000000000000 0000000080100010 00000001fdffffff 0000000000000000 page dumped because: kasan: bad access detected page_owner tracks the page as allocated page last allocated via order 0, migratetype Unmovable, gfp_mask 0x52cc0(GFP_KERNEL|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP), pid 8387, tgid 8387 (syz-executor), ts 57078807121, free_ts 54812229064 set_page_owner include/linux/page_owner.h:32 [inline] post_alloc_hook+0x2e7/0x350 mm/page_alloc.c:1493 prep_new_page mm/page_alloc.c:1501 [inline] get_page_from_freelist+0xbf3/0x2850 mm/page_alloc.c:3439 __alloc_pages_noprof+0x214/0x21e0 mm/page_alloc.c:4695 __alloc_pages_node_noprof include/linux/gfp.h:269 [inline] alloc_pages_node_noprof include/linux/gfp.h:296 [inline] alloc_slab_page+0x55/0x100 mm/slub.c:2321 allocate_slab mm/slub.c:2484 [inline] new_slab+0x83/0x260 mm/slub.c:2537 ___slab_alloc+0xbb7/0x1850 mm/slub.c:3723 __slab_alloc.constprop.0+0x56/0xb0 mm/slub.c:3813 __slab_alloc_node mm/slub.c:3866 [inline] slab_alloc_node mm/slub.c:4025 [inline] __kmalloc_cache_noprof+0x2be/0x310 mm/slub.c:4184 kmalloc_noprof include/linux/slab.h:681 [inline] kzalloc_noprof include/linux/slab.h:807 [inline] kset_create lib/kobject.c:965 [inline] kset_create_and_add+0x4f/0x1a0 lib/kobject.c:1008 register_queue_kobjects net/core/net-sysfs.c:1887 [inline] netdev_register_kobject+0x1d2/0x400 net/core/net-sysfs.c:2140 register_netdevice+0x1329/0x1bf0 net/core/dev.c:10444 veth_newlink+0x4bd/0x960 drivers/net/veth.c:1860 rtnl_newlink_create net/core/rtnetlink.c:3510 [inline] __rtnl_newlink+0x1138/0x18e0 net/core/rtnetlink.c:3730 rtnl_newlink+0x68/0xa0 net/core/rtnetlink.c:3743 rtnetlink_rcv_msg+0x3c9/0xfb0 net/core/rtnetlink.c:6647 netlink_rcv_skb+0x165/0x440 net/netlink/af_netlink.c:2550 page last free pid 96 tgid 96 stack trace: reset_page_owner include/linux/page_owner.h:25 [inline] free_pages_prepare mm/page_alloc.c:1094 [inline] free_unref_folios+0x9c6/0x1320 mm/page_alloc.c:2660 shrink_folio_list+0x29f0/0x4260 mm/vmscan.c:1530 evict_folios+0x71e/0x1b50 mm/vmscan.c:4580 try_to_shrink_lruvec+0x62b/0x9a0 mm/vmscan.c:4775 shrink_one+0x417/0x7c0 mm/vmscan.c:4813 shrink_many mm/vmscan.c:4876 [inline] lru_gen_shrink_node mm/vmscan.c:4954 [inline] shrink_node+0x244d/0x3830 mm/vmscan.c:5934 kswapd_shrink_node mm/vmscan.c:6762 [inline] balance_pgdat+0xba2/0x1880 mm/vmscan.c:6954 kswapd+0x702/0xd50 mm/vmscan.c:7223 kthread+0x2c7/0x3b0 kernel/kthread.c:389 ret_from_fork+0x45/0x80 arch/x86/kernel/process.c:147 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:244 Memory state around the buggy address: ffff888012faad80: 00 fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc ffff888012faae00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 >ffff888012faae80: 00 00 00 00 00 00 00 fc fc fc fc fc fc fc fc fc ^ ffff888012faaf00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ffff888012faaf80: 00 00 00 00 00 fc fc fc fc fc fc fc fc fc fc fc ==================================================================
On Thu, 2024-08-29 at 12:45 +0900, Jeongjun Park wrote: > Alexei Starovoitov wrote: > > > > On Fri, Aug 23, 2024 at 3:43 AM Jeongjun Park <aha310510@gmail.com> wrote: > > > > > > If the length of the name string is 1 and the value of name[0] is NULL > > > byte, an OOB vulnerability occurs in btf_name_valid_section() and the > > > return value is true, so the invalid name passes the check. > > > > > > To solve this, you need to check if the first position is NULL byte. > > > > > > Fixes: bd70a8fb7ca4 ("bpf: Allow all printable characters in BTF DATASEC names") > > > Signed-off-by: Jeongjun Park <aha310510@gmail.com> > > > --- > > > kernel/bpf/btf.c | 3 +++ > > > 1 file changed, 3 insertions(+) > > > > > > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c > > > index 520f49f422fe..5c24ea1a65a4 100644 > > > --- a/kernel/bpf/btf.c > > > +++ b/kernel/bpf/btf.c > > > @@ -823,6 +823,9 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) > > > const char *src = btf_str_by_offset(btf, offset); > > > const char *src_limit; > > > > > > + if (!*src) > > > + return false; > > > + > > > > We've talked about it. Quote: > > "Pls add a selftest that demonstrates the issue > > and produce a patch to fix just that." > > > > length == 1 and name[0] = 0 is a hypothesis. > > Demonstrate that such a scenario is possible then this patch will be > > worth applying. > > > > pw-bot: cr > > Sorry for the omission, I still don't know how to write selftest. > > But I can give you the C repro and KASAN log that trigger this vulnerability. > I would appreciate it if you could look at it and make a judgment. I will prepare a test case. Probably tomorrow.
On Wed, 2024-08-28 at 22:45 -0700, Eduard Zingerman wrote: [...] > I will prepare a test case. > Probably tomorrow. Please find test in the attachment. This test triggers KASAN error report as in another attachment. (I enabled CONFIG_KASAN using menuconfig on top of regular selftest config). On Fri, Aug 23, 2024 at 3:43 AM Jeongjun Park <aha310510@gmail.com> wrote: [...] > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c > index 520f49f422fe..5c24ea1a65a4 100644 > --- a/kernel/bpf/btf.c > +++ b/kernel/bpf/btf.c > @@ -823,6 +823,9 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) > const char *src = btf_str_by_offset(btf, offset); > const char *src_limit; > > + if (!*src) > + return false; > + I think that correct fix would be as follows: --- diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c index edad152cee8e..d583d76fcace 100644 --- a/kernel/bpf/btf.c +++ b/kernel/bpf/btf.c @@ -820,7 +820,6 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) /* set a limit on identifier length */ src_limit = src + KSYM_NAME_LEN; - src++; while (*src && src < src_limit) { if (!isprint(*src)) return false; [ 7.025728] ================================================================== [ 7.025977] BUG: KASAN: slab-out-of-bounds in btf_datasec_check_meta+0x73c/0x7a0 [ 7.026185] Read of size 1 at addr ffff888108eaa5d4 by task test_progs/172 [ 7.026358] [ 7.026430] CPU: 2 UID: 0 PID: 172 Comm: test_progs Tainted: G OE 6.11.0-rc4-00225-g0673888b8469-dirty #23 [ 7.026673] Tainted: [O]=OOT_MODULE, [E]=UNSIGNED_MODULE [ 7.026673] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-2.fc40 04/01/2014 [ 7.026673] Call Trace: [ 7.026673] <TASK> [ 7.026673] dump_stack_lvl+0x5d/0x80 [ 7.026673] ? btf_datasec_check_meta+0x73c/0x7a0 [ 7.026673] print_report+0x171/0x502 [ 7.026673] ? btf_datasec_check_meta+0x73c/0x7a0 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] ? __virt_addr_valid+0x20c/0x410 [ 7.026673] ? btf_datasec_check_meta+0x73c/0x7a0 [ 7.026673] kasan_report+0xda/0x1b0 [ 7.026673] ? btf_datasec_check_meta+0x73c/0x7a0 [ 7.026673] btf_datasec_check_meta+0x73c/0x7a0 [ 7.026673] btf_check_all_metas+0x32b/0xac0 [ 7.026673] btf_new_fd+0x6fb/0x2dc0 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] ? cred_has_capability.isra.0+0x139/0x230 [ 7.026673] ? __pfx_cred_has_capability.isra.0+0x10/0x10 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] ? __pfx_btf_new_fd+0x10/0x10 [ 7.026673] ? __pfx_lock_release+0x10/0x10 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] ? security_capable+0x6f/0xb0 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] __sys_bpf+0x1251/0x4650 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] ? __pfx___sys_bpf+0x10/0x10 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] ? find_held_lock+0x2d/0x110 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] ? lock_release+0x45a/0x7a0 [ 7.026673] ? __pfx_lock_release+0x10/0x10 [ 7.026673] ? __pfx___rseq_handle_notify_resume+0x10/0x10 [ 7.026673] __x64_sys_bpf+0x7b/0xc0 [ 7.026673] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7.026673] ? lockdep_hardirqs_on+0x7b/0x100 [ 7.026673] do_syscall_64+0x68/0x140 [ 7.026673] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 7.026673] RIP: 0033:0x7f9398d281dd [ 7.026673] Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 0b dc 0c 00 f7 d8 64 89 01 48 [ 7.026673] RSP: 002b:00007ffddc11b488 EFLAGS: 00000206 ORIG_RAX: 0000000000000141 [ 7.026673] RAX: ffffffffffffffda RBX: 00007ffddc11b9f8 RCX: 00007f9398d281dd [ 7.026673] RDX: 0000000000000028 RSI: 00007ffddc11b500 RDI: 0000000000000012 [ 7.026673] RBP: 00007ffddc11b4a0 R08: 00007ffddc11b320 R09: 00007ffddc11b500 [ 7.026673] R10: 0000000000000007 R11: 0000000000000206 R12: 0000000000000003 [ 7.026673] R13: 0000000000000000 R14: 00007f93a0114000 R15: 0000000001090d90 [ 7.026673] </TASK> [ 7.026673] [ 7.026673] Allocated by task 172: [ 7.026673] kasan_save_stack+0x30/0x50 [ 7.026673] kasan_save_track+0x14/0x30 [ 7.026673] __kasan_kmalloc+0x8f/0xa0 [ 7.026673] __kmalloc_node_noprof+0x1ac/0x490 [ 7.026673] btf_new_fd+0x3e4/0x2dc0 [ 7.026673] __sys_bpf+0x1251/0x4650 [ 7.026673] __x64_sys_bpf+0x7b/0xc0 [ 7.026673] do_syscall_64+0x68/0x140 [ 7.026673] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 7.026673] [ 7.026673] The buggy address belongs to the object at ffff888108eaa580 [ 7.026673] which belongs to the cache kmalloc-96 of size 96 [ 7.026673] The buggy address is located 0 bytes to the right of [ 7.026673] allocated 84-byte region [ffff888108eaa580, ffff888108eaa5d4) [ 7.026673] [ 7.026673] The buggy address belongs to the physical page: [ 7.026673] page: refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x108eaa [ 7.026673] flags: 0x2fffe0000000000(node=0|zone=2|lastcpupid=0x7fff) [ 7.026673] page_type: 0xfdffffff(slab) [ 7.026673] raw: 02fffe0000000000 ffff888100042280 dead000000000122 0000000000000000 [ 7.026673] raw: 0000000000000000 0000000080200020 00000001fdffffff 0000000000000000 [ 7.026673] page dumped because: kasan: bad access detected [ 7.026673] [ 7.026673] Memory state around the buggy address: [ 7.026673] ffff888108eaa480: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc [ 7.026673] ffff888108eaa500: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc [ 7.026673] >ffff888108eaa580: 00 00 00 00 00 00 00 00 00 00 04 fc fc fc fc fc [ 7.026673] ^ [ 7.026673] ffff888108eaa600: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc [ 7.026673] ffff888108eaa680: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc [ 7.026673] ==================================================================
> Eduard Zingerman wrote: > > On Wed, 2024-08-28 at 22:45 -0700, Eduard Zingerman wrote: > > [...] > >> I will prepare a test case. >> Probably tomorrow. > > Please find test in the attachment. This test triggers KASAN error > report as in another attachment. (I enabled CONFIG_KASAN using > menuconfig on top of regular selftest config). > Thank you for writing the selftest for me. > On Fri, Aug 23, 2024 at 3:43 AM Jeongjun Park <aha310510@gmail.com> wrote: > > [...] > >> diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c >> index 520f49f422fe..5c24ea1a65a4 100644 >> --- a/kernel/bpf/btf.c >> +++ b/kernel/bpf/btf.c >> @@ -823,6 +823,9 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) >> const char *src = btf_str_by_offset(btf, offset); >> const char *src_limit; >> >> + if (!*src) >> + return false; >> + > > I think that correct fix would be as follows: > > --- > > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c > index edad152cee8e..d583d76fcace 100644 > --- a/kernel/bpf/btf.c > +++ b/kernel/bpf/btf.c > @@ -820,7 +820,6 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) > > /* set a limit on identifier length */ > src_limit = src + KSYM_NAME_LEN; > - src++; > while (*src && src < src_limit) { > if (!isprint(*src)) > return false; However, this patch is logically flawed. It will return true for invalid names with length 1 and src[0] being NULL. So I think it's better to stick with the original patch. > > <bad-name-test.patch> > <bad-name-kasan-report.txt>
On Fri, 2024-08-30 at 11:03 +0900, Jeongjun Park wrote: [...] > > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c > > index edad152cee8e..d583d76fcace 100644 > > --- a/kernel/bpf/btf.c > > +++ b/kernel/bpf/btf.c > > @@ -820,7 +820,6 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) > > > > /* set a limit on identifier length */ > > src_limit = src + KSYM_NAME_LEN; > > - src++; > > while (*src && src < src_limit) { > > if (!isprint(*src)) > > return false; > > However, this patch is logically flawed. > It will return true for invalid names with > length 1 and src[0] being NULL. So I think > it's better to stick with the original patch. Fair enough, however the isprint check should be done for the first character. So the full fix is a combination :) --- a/kernel/bpf/btf.c +++ b/kernel/bpf/btf.c @@ -818,9 +818,11 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) const char *src = btf_str_by_offset(btf, offset); const char *src_limit; + if (!*src) + return false; + /* set a limit on identifier length */ src_limit = src + KSYM_NAME_LEN; - src++; while (*src && src < src_limit) { if (!isprint(*src)) return false; And corresponding test cases (tools/testing/selftests/bpf/prog_tests/btf.c): { .descr = "datasec: name with non-printable first char not is ok", .raw_types = { /* int */ BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4), /* [1] */ /* VAR x */ /* [2] */ BTF_TYPE_ENC(1, BTF_INFO_ENC(BTF_KIND_VAR, 0, 0), 1), BTF_VAR_STATIC, /* DATASEC ?.data */ /* [3] */ BTF_TYPE_ENC(3, BTF_INFO_ENC(BTF_KIND_DATASEC, 0, 1), 4), BTF_VAR_SECINFO_ENC(2, 0, 4), BTF_END_RAW, }, BTF_STR_SEC("\0x\0\7foo"), .err_str = "Invalid name", .btf_load_err = true, },{ .descr = "datasec: name '\\0' is not ok", .raw_types = { /* int */ BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4), /* [1] */ /* VAR x */ /* [2] */ BTF_TYPE_ENC(1, BTF_INFO_ENC(BTF_KIND_VAR, 0, 0), 1), BTF_VAR_STATIC, /* DATASEC \0 */ /* [3] */ BTF_TYPE_ENC(3, BTF_INFO_ENC(BTF_KIND_DATASEC, 0, 1), 4), BTF_VAR_SECINFO_ENC(2, 0, 4), BTF_END_RAW, }, BTF_STR_SEC("\0x\0"), .err_str = "Invalid name", .btf_load_err = true, }, Could you please resend your patch as a patch-set fix + selftests update?
Eduard Zingerman wrote: > > On Fri, 2024-08-30 at 11:03 +0900, Jeongjun Park wrote: > > [...] > > > > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c > > > index edad152cee8e..d583d76fcace 100644 > > > --- a/kernel/bpf/btf.c > > > +++ b/kernel/bpf/btf.c > > > @@ -820,7 +820,6 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) > > > > > > /* set a limit on identifier length */ > > > src_limit = src + KSYM_NAME_LEN; > > > - src++; > > > while (*src && src < src_limit) { > > > if (!isprint(*src)) > > > return false; > > > > However, this patch is logically flawed. > > It will return true for invalid names with > > length 1 and src[0] being NULL. So I think > > it's better to stick with the original patch. > > Fair enough, however the isprint check should be done for the first character. > So the full fix is a combination :) So does that mean it's appropriate to add if(!isprint(*src)) instead of if(!*src)? As far as I know, the first character of name doesn't need isprint() check, so if that's true, it would be appropriate to use isprint. Once this is confirmed, I'll send you a v2 patch that added selftest. Regards, Jeongjun Park > > --- a/kernel/bpf/btf.c > +++ b/kernel/bpf/btf.c > @@ -818,9 +818,11 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) > const char *src = btf_str_by_offset(btf, offset); > const char *src_limit; > > + if (!*src) > + return false; > + > /* set a limit on identifier length */ > src_limit = src + KSYM_NAME_LEN; > - src++; > while (*src && src < src_limit) { > if (!isprint(*src)) > return false; > > > And corresponding test cases (tools/testing/selftests/bpf/prog_tests/btf.c): > > { > .descr = "datasec: name with non-printable first char not is ok", > .raw_types = { > /* int */ > BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4), /* [1] */ > /* VAR x */ /* [2] */ > BTF_TYPE_ENC(1, BTF_INFO_ENC(BTF_KIND_VAR, 0, 0), 1), > BTF_VAR_STATIC, > /* DATASEC ?.data */ /* [3] */ > BTF_TYPE_ENC(3, BTF_INFO_ENC(BTF_KIND_DATASEC, 0, 1), 4), > BTF_VAR_SECINFO_ENC(2, 0, 4), > BTF_END_RAW, > }, > BTF_STR_SEC("\0x\0\7foo"), > .err_str = "Invalid name", > .btf_load_err = true, > },{ > .descr = "datasec: name '\\0' is not ok", > .raw_types = { > /* int */ > BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4), /* [1] */ > /* VAR x */ /* [2] */ > BTF_TYPE_ENC(1, BTF_INFO_ENC(BTF_KIND_VAR, 0, 0), 1), > BTF_VAR_STATIC, > /* DATASEC \0 */ /* [3] */ > BTF_TYPE_ENC(3, BTF_INFO_ENC(BTF_KIND_DATASEC, 0, 1), 4), > BTF_VAR_SECINFO_ENC(2, 0, 4), > BTF_END_RAW, > }, > BTF_STR_SEC("\0x\0"), > .err_str = "Invalid name", > .btf_load_err = true, > }, > > Could you please resend your patch as a patch-set fix + selftests update? >
On Fri, 2024-08-30 at 20:41 +0900, Jeongjun Park wrote: [...] > So does that mean it's appropriate to add if(!isprint(*src)) instead > of if(!*src)? I'd prefer to add "if (!*src) return false" and remove "src++" in order to not repeat isprint(). > As far as I know, the first character of name doesn't need isprint() check, > so if that's true, it would be appropriate to use isprint. Once this > is confirmed, The intent of the buggy commit [1] was to check that all characters in a section name are printable. But I can't even check name for printable characters these days :( [1] bd70a8fb7ca4 ("bpf: Allow all printable characters in BTF DATASEC names") [...]
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c index 520f49f422fe..5c24ea1a65a4 100644 --- a/kernel/bpf/btf.c +++ b/kernel/bpf/btf.c @@ -823,6 +823,9 @@ static bool btf_name_valid_section(const struct btf *btf, u32 offset) const char *src = btf_str_by_offset(btf, offset); const char *src_limit; + if (!*src) + return false; + /* set a limit on identifier length */ src_limit = src + KSYM_NAME_LEN; src++;
If the length of the name string is 1 and the value of name[0] is NULL byte, an OOB vulnerability occurs in btf_name_valid_section() and the return value is true, so the invalid name passes the check. To solve this, you need to check if the first position is NULL byte. Fixes: bd70a8fb7ca4 ("bpf: Allow all printable characters in BTF DATASEC names") Signed-off-by: Jeongjun Park <aha310510@gmail.com> --- kernel/bpf/btf.c | 3 +++ 1 file changed, 3 insertions(+) --