bpf: expose is_mptcp flag to bpf_tcp_sock #49
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Master branch: 2bab48c patch https://patchwork.ozlabs.org/project/netdev/patch/[email protected]/ applied successfully |
|
Master branch: 2bab48c patch https://patchwork.ozlabs.org/project/netdev/patch/[email protected]/ applied successfully |
kernel-patches-bot
force-pushed
the
series/201180
branch
from
b964f69 to
76a2ea3
Compare
|
Master branch: 2bab48c patch https://patchwork.ozlabs.org/project/netdev/patch/[email protected]/ applied successfully |
kernel-patches-bot
force-pushed
the
series/201180
branch
from
76a2ea3 to
09f66c8
Compare
|
Master branch: 2bab48c patch https://patchwork.ozlabs.org/project/netdev/patch/[email protected]/ applied successfully |
kernel-patches-bot
force-pushed
the
series/201180
branch
from
09f66c8 to
5b9e08b
Compare
|
Master branch: d72714c patch https://patchwork.ozlabs.org/project/netdev/patch/[email protected]/ applied successfully |
kernel-patches-bot
force-pushed
the
series/201180
branch
from
5b9e08b to
5a49940
Compare
|
Master branch: bf74a37 patch https://patchwork.ozlabs.org/project/netdev/patch/[email protected]/ applied successfully |
kernel-patches-bot
force-pushed
the
series/201180
branch
from
5a49940 to
33f70e4
Compare
current tcp_sock is part of the MPTCP protocol. In this protocol, a first socket (mptcp_sock) is created with sk_protocol set to IPPROTO_MPTCP (=262) for control purpose but it isn't directly on the wire. This is the role of the subflow (kernel) sockets which are classical tcp_sock with sk_protocol set to IPPROTO_TCP. The only way to differentiate such sockets from plain TCP sockets is the is_mptcp field from tcp_sock. Such an exposure in BPF is thus required to be able to differentiate plain TCP sockets from MPTCP subflow sockets in BPF_PROG_TYPE_SOCK_OPS programs. The choice has been made to silently pass the case when CONFIG_MPTCP is unset by defaulting is_mptcp to 0 in order to make BPF independent of the MPTCP configuration. Another solution is to make the verifier fail in 'bpf_tcp_sock_is_valid_ctx_access' but this will add an additional '#ifdef CONFIG_MPTCP' in the BPF code and a same injected BPF program will not run if MPTCP is not set. An example use-case is provided in https://github.com/multipath-tcp/mptcp_net-next/tree/scripts/bpf/examples Suggested-by: Matthieu Baerts <[email protected]> Acked-by: Matthieu Baerts <[email protected]> Acked-by: Mat Martineau <[email protected]> Signed-off-by: Nicolas Rybowski <[email protected]> --- include/uapi/linux/bpf.h | 1 + net/core/filter.c | 9 ++++++++- tools/include/uapi/linux/bpf.h | 1 + 3 files changed, 10 insertions(+), 1 deletion(-)
(except the first one) are not in the same cgroup as their parents. That's because the additional subflows are created by kernel workers. This is a problem with eBPF programs attached to the parent's cgroup won't be executed for the children. But also with any other features of CGroup linked to a sk. This patch fixes this behaviour. As the subflow sockets are created by the kernel, we can't use 'mem_cgroup_sk_alloc' because of the current context being the one of the kworker. This is why we have to do low level memcg manipulation, if required. Suggested-by: Matthieu Baerts <[email protected]> Suggested-by: Paolo Abeni <[email protected]> Acked-by: Matthieu Baerts <[email protected]> Reviewed-by: Mat Martineau <[email protected]> Signed-off-by: Nicolas Rybowski <[email protected]> --- net/mptcp/subflow.c | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+)
it is required to access the mptcp_sock's token which uniquely identify a MPTCP connection. This patch adds a new structure 'bpf_mptcp_sock' exposing the 'token' field of the 'mptcp_sock' extracted from a subflow's 'tcp_sock'. It also adds the declaration of a new BPF helper of the same name to expose the newly defined structure in the userspace BPF API. This is the foundation to expose more MPTCP-specific fields through BPF. Currently, it is limited to the field 'token' of the msk but it is easily extensible. Acked-by: Matthieu Baerts <[email protected]> Acked-by: Mat Martineau <[email protected]> Signed-off-by: Nicolas Rybowski <[email protected]> --- include/linux/bpf.h | 33 ++++++++++++++++ include/uapi/linux/bpf.h | 14 +++++++ kernel/bpf/verifier.c | 30 ++++++++++++++ net/core/filter.c | 4 ++ net/mptcp/Makefile | 2 + net/mptcp/bpf.c | 72 ++++++++++++++++++++++++++++++++++ scripts/bpf_helpers_doc.py | 2 + tools/include/uapi/linux/bpf.h | 14 +++++++ 8 files changed, 171 insertions(+) create mode 100644 net/mptcp/bpf.c
It is currently limited to the is_mptcp field in case of plain TCP connection because for the moment there is no easy way to get the subflow sk from a msk in userspace. This implies that we cannot lookup the sk_storage attached to the subflow sk in the sockops program. Acked-by: Matthieu Baerts <[email protected]> Signed-off-by: Nicolas Rybowski <[email protected]> --- Notes: v1 -> v2: - new patch: mandatory selftests (Alexei) tools/testing/selftests/bpf/config | 1 + tools/testing/selftests/bpf/network_helpers.c | 37 +++++- tools/testing/selftests/bpf/network_helpers.h | 3 + .../testing/selftests/bpf/prog_tests/mptcp.c | 119 ++++++++++++++++++ tools/testing/selftests/bpf/progs/mptcp.c | 48 +++++++ 5 files changed, 203 insertions(+), 5 deletions(-) create mode 100644 tools/testing/selftests/bpf/prog_tests/mptcp.c create mode 100644 tools/testing/selftests/bpf/progs/mptcp.c
Here are the new tests : - NULL bpf_sock is correctly handled - We cannot access a field from bpf_mptcp_sock if the latter is NULL - We can access a field from bpf_mptcp_sock if the latter is not NULL - We cannot modify a field from bpf_mptcp_sock. Note that "token" is currently the only field in bpf_mptcp_sock. Currently, there is no easy way to test the token field since we cannot get back the mptcp_sock in userspace, this could be a future amelioration. Acked-by: Matthieu Baerts <[email protected]> Signed-off-by: Nicolas Rybowski <[email protected]> --- Notes: v1 -> v2: - new patch: mandatory selftests (Alexei) tools/testing/selftests/bpf/verifier/sock.c | 63 +++++++++++++++++++++ 1 file changed, 63 insertions(+)
|
Master branch: d317b0a patch https://patchwork.ozlabs.org/project/netdev/patch/[email protected]/ applied successfully |
kernel-patches-bot
force-pushed
the
series/201180
branch
from
33f70e4 to
8e33d5c
Compare
|
At least one diff in series https://patchwork.ozlabs.org/project/netdev/list/?series=201180 expired. Closing PR. |
kernel-patches-bot
pushed a commit
that referenced
this pull request
Apr 23, 2021
Fix BPF_CORE_READ_BITFIELD() macro used for reading CO-RE-relocatable bitfields. Missing breaks in a switch caused 8-byte reads always. This can confuse libbpf because it does strict checks that memory load size corresponds to the original size of the field, which in this case quite often would be wrong. After fixing that, we run into another problem, which quite subtle, so worth documenting here. The issue is in Clang optimization and CO-RE relocation interactions. Without that asm volatile construct (also known as barrier_var()), Clang will re-order BYTE_OFFSET and BYTE_SIZE relocations and will apply BYTE_OFFSET 4 times for each switch case arm. This will result in the same error from libbpf about mismatch of memory load size and original field size. I.e., if we were reading u32, we'd still have *(u8 *), *(u16 *), *(u32 *), and *(u64 *) memory loads, three of which will fail. Using barrier_var() forces Clang to apply BYTE_OFFSET relocation first (and once) to calculate p, after which value of p is used without relocation in each of switch case arms, doing appropiately-sized memory load. Here's the list of relevant relocations and pieces of generated BPF code before and after this patch for test_core_reloc_bitfields_direct selftests. BEFORE ===== #45: core_reloc: insn #160 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 #46: core_reloc: insn #167 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #47: core_reloc: insn #174 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #48: core_reloc: insn #178 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #49: core_reloc: insn #182 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 157: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 159: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 160: b7 02 00 00 04 00 00 00 r2 = 4 ; BYTE_SIZE relocation here ^^^ 161: 66 02 07 00 03 00 00 00 if w2 s> 3 goto +7 <LBB0_63> 162: 16 02 0d 00 01 00 00 00 if w2 == 1 goto +13 <LBB0_65> 163: 16 02 01 00 02 00 00 00 if w2 == 2 goto +1 <LBB0_66> 164: 05 00 12 00 00 00 00 00 goto +18 <LBB0_69> 0000000000000528 <LBB0_66>: 165: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 167: 69 11 08 00 00 00 00 00 r1 = *(u16 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 168: 05 00 0e 00 00 00 00 00 goto +14 <LBB0_69> 0000000000000548 <LBB0_63>: 169: 16 02 0a 00 04 00 00 00 if w2 == 4 goto +10 <LBB0_67> 170: 16 02 01 00 08 00 00 00 if w2 == 8 goto +1 <LBB0_68> 171: 05 00 0b 00 00 00 00 00 goto +11 <LBB0_69> 0000000000000560 <LBB0_68>: 172: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 174: 79 11 08 00 00 00 00 00 r1 = *(u64 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 175: 05 00 07 00 00 00 00 00 goto +7 <LBB0_69> 0000000000000580 <LBB0_65>: 176: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 178: 71 11 08 00 00 00 00 00 r1 = *(u8 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 179: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 00000000000005a0 <LBB0_67>: 180: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 182: 61 11 08 00 00 00 00 00 r1 = *(u32 *)(r1 + 8) ; BYTE_OFFSET relo here w/ RIGHT size ^^^^^^^^^^^^^^^^ 00000000000005b8 <LBB0_69>: 183: 67 01 00 00 20 00 00 00 r1 <<= 32 184: b7 02 00 00 00 00 00 00 r2 = 0 185: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 186: c7 01 00 00 20 00 00 00 r1 s>>= 32 187: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000005e0 <LBB0_71>: 188: 77 01 00 00 20 00 00 00 r1 >>= 32 AFTER ===== #30: core_reloc: insn #132 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #31: core_reloc: insn #134 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 129: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 131: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 132: b7 01 00 00 08 00 00 00 r1 = 8 ; BYTE_OFFSET relo here ^^^ ; no size check for non-memory dereferencing instructions 133: 0f 12 00 00 00 00 00 00 r2 += r1 134: b7 03 00 00 04 00 00 00 r3 = 4 ; BYTE_SIZE relocation here ^^^ 135: 66 03 05 00 03 00 00 00 if w3 s> 3 goto +5 <LBB0_63> 136: 16 03 09 00 01 00 00 00 if w3 == 1 goto +9 <LBB0_65> 137: 16 03 01 00 02 00 00 00 if w3 == 2 goto +1 <LBB0_66> 138: 05 00 0a 00 00 00 00 00 goto +10 <LBB0_69> 0000000000000458 <LBB0_66>: 139: 69 21 00 00 00 00 00 00 r1 = *(u16 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 140: 05 00 08 00 00 00 00 00 goto +8 <LBB0_69> 0000000000000468 <LBB0_63>: 141: 16 03 06 00 04 00 00 00 if w3 == 4 goto +6 <LBB0_67> 142: 16 03 01 00 08 00 00 00 if w3 == 8 goto +1 <LBB0_68> 143: 05 00 05 00 00 00 00 00 goto +5 <LBB0_69> 0000000000000480 <LBB0_68>: 144: 79 21 00 00 00 00 00 00 r1 = *(u64 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 145: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 0000000000000490 <LBB0_65>: 146: 71 21 00 00 00 00 00 00 r1 = *(u8 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 147: 05 00 01 00 00 00 00 00 goto +1 <LBB0_69> 00000000000004a0 <LBB0_67>: 148: 61 21 00 00 00 00 00 00 r1 = *(u32 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 00000000000004a8 <LBB0_69>: 149: 67 01 00 00 20 00 00 00 r1 <<= 32 150: b7 02 00 00 00 00 00 00 r2 = 0 151: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 152: c7 01 00 00 20 00 00 00 r1 s>>= 32 153: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000004d0 <LBB0_71>: 154: 77 01 00 00 20 00 00 00 r1 >>= 323 Fixes: ee26dad ("libbpf: Add support for relocatable bitfields") Signed-off-by: Andrii Nakryiko <[email protected]>
kernel-patches-bot
pushed a commit
that referenced
this pull request
Apr 24, 2021
Fix BPF_CORE_READ_BITFIELD() macro used for reading CO-RE-relocatable bitfields. Missing breaks in a switch caused 8-byte reads always. This can confuse libbpf because it does strict checks that memory load size corresponds to the original size of the field, which in this case quite often would be wrong. After fixing that, we run into another problem, which quite subtle, so worth documenting here. The issue is in Clang optimization and CO-RE relocation interactions. Without that asm volatile construct (also known as barrier_var()), Clang will re-order BYTE_OFFSET and BYTE_SIZE relocations and will apply BYTE_OFFSET 4 times for each switch case arm. This will result in the same error from libbpf about mismatch of memory load size and original field size. I.e., if we were reading u32, we'd still have *(u8 *), *(u16 *), *(u32 *), and *(u64 *) memory loads, three of which will fail. Using barrier_var() forces Clang to apply BYTE_OFFSET relocation first (and once) to calculate p, after which value of p is used without relocation in each of switch case arms, doing appropiately-sized memory load. Here's the list of relevant relocations and pieces of generated BPF code before and after this patch for test_core_reloc_bitfields_direct selftests. BEFORE ===== #45: core_reloc: insn #160 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 #46: core_reloc: insn #167 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #47: core_reloc: insn #174 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #48: core_reloc: insn #178 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #49: core_reloc: insn #182 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 157: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 159: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 160: b7 02 00 00 04 00 00 00 r2 = 4 ; BYTE_SIZE relocation here ^^^ 161: 66 02 07 00 03 00 00 00 if w2 s> 3 goto +7 <LBB0_63> 162: 16 02 0d 00 01 00 00 00 if w2 == 1 goto +13 <LBB0_65> 163: 16 02 01 00 02 00 00 00 if w2 == 2 goto +1 <LBB0_66> 164: 05 00 12 00 00 00 00 00 goto +18 <LBB0_69> 0000000000000528 <LBB0_66>: 165: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 167: 69 11 08 00 00 00 00 00 r1 = *(u16 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 168: 05 00 0e 00 00 00 00 00 goto +14 <LBB0_69> 0000000000000548 <LBB0_63>: 169: 16 02 0a 00 04 00 00 00 if w2 == 4 goto +10 <LBB0_67> 170: 16 02 01 00 08 00 00 00 if w2 == 8 goto +1 <LBB0_68> 171: 05 00 0b 00 00 00 00 00 goto +11 <LBB0_69> 0000000000000560 <LBB0_68>: 172: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 174: 79 11 08 00 00 00 00 00 r1 = *(u64 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 175: 05 00 07 00 00 00 00 00 goto +7 <LBB0_69> 0000000000000580 <LBB0_65>: 176: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 178: 71 11 08 00 00 00 00 00 r1 = *(u8 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 179: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 00000000000005a0 <LBB0_67>: 180: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 182: 61 11 08 00 00 00 00 00 r1 = *(u32 *)(r1 + 8) ; BYTE_OFFSET relo here w/ RIGHT size ^^^^^^^^^^^^^^^^ 00000000000005b8 <LBB0_69>: 183: 67 01 00 00 20 00 00 00 r1 <<= 32 184: b7 02 00 00 00 00 00 00 r2 = 0 185: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 186: c7 01 00 00 20 00 00 00 r1 s>>= 32 187: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000005e0 <LBB0_71>: 188: 77 01 00 00 20 00 00 00 r1 >>= 32 AFTER ===== #30: core_reloc: insn #132 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #31: core_reloc: insn #134 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 129: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 131: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 132: b7 01 00 00 08 00 00 00 r1 = 8 ; BYTE_OFFSET relo here ^^^ ; no size check for non-memory dereferencing instructions 133: 0f 12 00 00 00 00 00 00 r2 += r1 134: b7 03 00 00 04 00 00 00 r3 = 4 ; BYTE_SIZE relocation here ^^^ 135: 66 03 05 00 03 00 00 00 if w3 s> 3 goto +5 <LBB0_63> 136: 16 03 09 00 01 00 00 00 if w3 == 1 goto +9 <LBB0_65> 137: 16 03 01 00 02 00 00 00 if w3 == 2 goto +1 <LBB0_66> 138: 05 00 0a 00 00 00 00 00 goto +10 <LBB0_69> 0000000000000458 <LBB0_66>: 139: 69 21 00 00 00 00 00 00 r1 = *(u16 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 140: 05 00 08 00 00 00 00 00 goto +8 <LBB0_69> 0000000000000468 <LBB0_63>: 141: 16 03 06 00 04 00 00 00 if w3 == 4 goto +6 <LBB0_67> 142: 16 03 01 00 08 00 00 00 if w3 == 8 goto +1 <LBB0_68> 143: 05 00 05 00 00 00 00 00 goto +5 <LBB0_69> 0000000000000480 <LBB0_68>: 144: 79 21 00 00 00 00 00 00 r1 = *(u64 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 145: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 0000000000000490 <LBB0_65>: 146: 71 21 00 00 00 00 00 00 r1 = *(u8 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 147: 05 00 01 00 00 00 00 00 goto +1 <LBB0_69> 00000000000004a0 <LBB0_67>: 148: 61 21 00 00 00 00 00 00 r1 = *(u32 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 00000000000004a8 <LBB0_69>: 149: 67 01 00 00 20 00 00 00 r1 <<= 32 150: b7 02 00 00 00 00 00 00 r2 = 0 151: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 152: c7 01 00 00 20 00 00 00 r1 s>>= 32 153: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000004d0 <LBB0_71>: 154: 77 01 00 00 20 00 00 00 r1 >>= 323 Fixes: ee26dad ("libbpf: Add support for relocatable bitfields") Signed-off-by: Andrii Nakryiko <[email protected]>
kernel-patches-bot
pushed a commit
that referenced
this pull request
Apr 26, 2021
Fix BPF_CORE_READ_BITFIELD() macro used for reading CO-RE-relocatable bitfields. Missing breaks in a switch caused 8-byte reads always. This can confuse libbpf because it does strict checks that memory load size corresponds to the original size of the field, which in this case quite often would be wrong. After fixing that, we run into another problem, which quite subtle, so worth documenting here. The issue is in Clang optimization and CO-RE relocation interactions. Without that asm volatile construct (also known as barrier_var()), Clang will re-order BYTE_OFFSET and BYTE_SIZE relocations and will apply BYTE_OFFSET 4 times for each switch case arm. This will result in the same error from libbpf about mismatch of memory load size and original field size. I.e., if we were reading u32, we'd still have *(u8 *), *(u16 *), *(u32 *), and *(u64 *) memory loads, three of which will fail. Using barrier_var() forces Clang to apply BYTE_OFFSET relocation first (and once) to calculate p, after which value of p is used without relocation in each of switch case arms, doing appropiately-sized memory load. Here's the list of relevant relocations and pieces of generated BPF code before and after this patch for test_core_reloc_bitfields_direct selftests. BEFORE ===== #45: core_reloc: insn #160 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 #46: core_reloc: insn #167 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #47: core_reloc: insn #174 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #48: core_reloc: insn #178 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #49: core_reloc: insn #182 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 157: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 159: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 160: b7 02 00 00 04 00 00 00 r2 = 4 ; BYTE_SIZE relocation here ^^^ 161: 66 02 07 00 03 00 00 00 if w2 s> 3 goto +7 <LBB0_63> 162: 16 02 0d 00 01 00 00 00 if w2 == 1 goto +13 <LBB0_65> 163: 16 02 01 00 02 00 00 00 if w2 == 2 goto +1 <LBB0_66> 164: 05 00 12 00 00 00 00 00 goto +18 <LBB0_69> 0000000000000528 <LBB0_66>: 165: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 167: 69 11 08 00 00 00 00 00 r1 = *(u16 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 168: 05 00 0e 00 00 00 00 00 goto +14 <LBB0_69> 0000000000000548 <LBB0_63>: 169: 16 02 0a 00 04 00 00 00 if w2 == 4 goto +10 <LBB0_67> 170: 16 02 01 00 08 00 00 00 if w2 == 8 goto +1 <LBB0_68> 171: 05 00 0b 00 00 00 00 00 goto +11 <LBB0_69> 0000000000000560 <LBB0_68>: 172: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 174: 79 11 08 00 00 00 00 00 r1 = *(u64 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 175: 05 00 07 00 00 00 00 00 goto +7 <LBB0_69> 0000000000000580 <LBB0_65>: 176: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 178: 71 11 08 00 00 00 00 00 r1 = *(u8 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 179: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 00000000000005a0 <LBB0_67>: 180: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 182: 61 11 08 00 00 00 00 00 r1 = *(u32 *)(r1 + 8) ; BYTE_OFFSET relo here w/ RIGHT size ^^^^^^^^^^^^^^^^ 00000000000005b8 <LBB0_69>: 183: 67 01 00 00 20 00 00 00 r1 <<= 32 184: b7 02 00 00 00 00 00 00 r2 = 0 185: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 186: c7 01 00 00 20 00 00 00 r1 s>>= 32 187: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000005e0 <LBB0_71>: 188: 77 01 00 00 20 00 00 00 r1 >>= 32 AFTER ===== #30: core_reloc: insn #132 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #31: core_reloc: insn #134 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 129: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 131: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 132: b7 01 00 00 08 00 00 00 r1 = 8 ; BYTE_OFFSET relo here ^^^ ; no size check for non-memory dereferencing instructions 133: 0f 12 00 00 00 00 00 00 r2 += r1 134: b7 03 00 00 04 00 00 00 r3 = 4 ; BYTE_SIZE relocation here ^^^ 135: 66 03 05 00 03 00 00 00 if w3 s> 3 goto +5 <LBB0_63> 136: 16 03 09 00 01 00 00 00 if w3 == 1 goto +9 <LBB0_65> 137: 16 03 01 00 02 00 00 00 if w3 == 2 goto +1 <LBB0_66> 138: 05 00 0a 00 00 00 00 00 goto +10 <LBB0_69> 0000000000000458 <LBB0_66>: 139: 69 21 00 00 00 00 00 00 r1 = *(u16 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 140: 05 00 08 00 00 00 00 00 goto +8 <LBB0_69> 0000000000000468 <LBB0_63>: 141: 16 03 06 00 04 00 00 00 if w3 == 4 goto +6 <LBB0_67> 142: 16 03 01 00 08 00 00 00 if w3 == 8 goto +1 <LBB0_68> 143: 05 00 05 00 00 00 00 00 goto +5 <LBB0_69> 0000000000000480 <LBB0_68>: 144: 79 21 00 00 00 00 00 00 r1 = *(u64 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 145: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 0000000000000490 <LBB0_65>: 146: 71 21 00 00 00 00 00 00 r1 = *(u8 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 147: 05 00 01 00 00 00 00 00 goto +1 <LBB0_69> 00000000000004a0 <LBB0_67>: 148: 61 21 00 00 00 00 00 00 r1 = *(u32 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 00000000000004a8 <LBB0_69>: 149: 67 01 00 00 20 00 00 00 r1 <<= 32 150: b7 02 00 00 00 00 00 00 r2 = 0 151: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 152: c7 01 00 00 20 00 00 00 r1 s>>= 32 153: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000004d0 <LBB0_71>: 154: 77 01 00 00 20 00 00 00 r1 >>= 323 Fixes: ee26dad ("libbpf: Add support for relocatable bitfields") Signed-off-by: Andrii Nakryiko <[email protected]>
kernel-patches-bot
pushed a commit
that referenced
this pull request
Apr 26, 2021
Fix BPF_CORE_READ_BITFIELD() macro used for reading CO-RE-relocatable bitfields. Missing breaks in a switch caused 8-byte reads always. This can confuse libbpf because it does strict checks that memory load size corresponds to the original size of the field, which in this case quite often would be wrong. After fixing that, we run into another problem, which quite subtle, so worth documenting here. The issue is in Clang optimization and CO-RE relocation interactions. Without that asm volatile construct (also known as barrier_var()), Clang will re-order BYTE_OFFSET and BYTE_SIZE relocations and will apply BYTE_OFFSET 4 times for each switch case arm. This will result in the same error from libbpf about mismatch of memory load size and original field size. I.e., if we were reading u32, we'd still have *(u8 *), *(u16 *), *(u32 *), and *(u64 *) memory loads, three of which will fail. Using barrier_var() forces Clang to apply BYTE_OFFSET relocation first (and once) to calculate p, after which value of p is used without relocation in each of switch case arms, doing appropiately-sized memory load. Here's the list of relevant relocations and pieces of generated BPF code before and after this patch for test_core_reloc_bitfields_direct selftests. BEFORE ===== #45: core_reloc: insn #160 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 #46: core_reloc: insn #167 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #47: core_reloc: insn #174 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #48: core_reloc: insn #178 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #49: core_reloc: insn #182 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 157: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 159: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 160: b7 02 00 00 04 00 00 00 r2 = 4 ; BYTE_SIZE relocation here ^^^ 161: 66 02 07 00 03 00 00 00 if w2 s> 3 goto +7 <LBB0_63> 162: 16 02 0d 00 01 00 00 00 if w2 == 1 goto +13 <LBB0_65> 163: 16 02 01 00 02 00 00 00 if w2 == 2 goto +1 <LBB0_66> 164: 05 00 12 00 00 00 00 00 goto +18 <LBB0_69> 0000000000000528 <LBB0_66>: 165: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 167: 69 11 08 00 00 00 00 00 r1 = *(u16 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 168: 05 00 0e 00 00 00 00 00 goto +14 <LBB0_69> 0000000000000548 <LBB0_63>: 169: 16 02 0a 00 04 00 00 00 if w2 == 4 goto +10 <LBB0_67> 170: 16 02 01 00 08 00 00 00 if w2 == 8 goto +1 <LBB0_68> 171: 05 00 0b 00 00 00 00 00 goto +11 <LBB0_69> 0000000000000560 <LBB0_68>: 172: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 174: 79 11 08 00 00 00 00 00 r1 = *(u64 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 175: 05 00 07 00 00 00 00 00 goto +7 <LBB0_69> 0000000000000580 <LBB0_65>: 176: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 178: 71 11 08 00 00 00 00 00 r1 = *(u8 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 179: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 00000000000005a0 <LBB0_67>: 180: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 182: 61 11 08 00 00 00 00 00 r1 = *(u32 *)(r1 + 8) ; BYTE_OFFSET relo here w/ RIGHT size ^^^^^^^^^^^^^^^^ 00000000000005b8 <LBB0_69>: 183: 67 01 00 00 20 00 00 00 r1 <<= 32 184: b7 02 00 00 00 00 00 00 r2 = 0 185: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 186: c7 01 00 00 20 00 00 00 r1 s>>= 32 187: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000005e0 <LBB0_71>: 188: 77 01 00 00 20 00 00 00 r1 >>= 32 AFTER ===== #30: core_reloc: insn #132 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #31: core_reloc: insn #134 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 129: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 131: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 132: b7 01 00 00 08 00 00 00 r1 = 8 ; BYTE_OFFSET relo here ^^^ ; no size check for non-memory dereferencing instructions 133: 0f 12 00 00 00 00 00 00 r2 += r1 134: b7 03 00 00 04 00 00 00 r3 = 4 ; BYTE_SIZE relocation here ^^^ 135: 66 03 05 00 03 00 00 00 if w3 s> 3 goto +5 <LBB0_63> 136: 16 03 09 00 01 00 00 00 if w3 == 1 goto +9 <LBB0_65> 137: 16 03 01 00 02 00 00 00 if w3 == 2 goto +1 <LBB0_66> 138: 05 00 0a 00 00 00 00 00 goto +10 <LBB0_69> 0000000000000458 <LBB0_66>: 139: 69 21 00 00 00 00 00 00 r1 = *(u16 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 140: 05 00 08 00 00 00 00 00 goto +8 <LBB0_69> 0000000000000468 <LBB0_63>: 141: 16 03 06 00 04 00 00 00 if w3 == 4 goto +6 <LBB0_67> 142: 16 03 01 00 08 00 00 00 if w3 == 8 goto +1 <LBB0_68> 143: 05 00 05 00 00 00 00 00 goto +5 <LBB0_69> 0000000000000480 <LBB0_68>: 144: 79 21 00 00 00 00 00 00 r1 = *(u64 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 145: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 0000000000000490 <LBB0_65>: 146: 71 21 00 00 00 00 00 00 r1 = *(u8 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 147: 05 00 01 00 00 00 00 00 goto +1 <LBB0_69> 00000000000004a0 <LBB0_67>: 148: 61 21 00 00 00 00 00 00 r1 = *(u32 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 00000000000004a8 <LBB0_69>: 149: 67 01 00 00 20 00 00 00 r1 <<= 32 150: b7 02 00 00 00 00 00 00 r2 = 0 151: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 152: c7 01 00 00 20 00 00 00 r1 s>>= 32 153: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000004d0 <LBB0_71>: 154: 77 01 00 00 20 00 00 00 r1 >>= 323 Fixes: ee26dad ("libbpf: Add support for relocatable bitfields") Signed-off-by: Andrii Nakryiko <[email protected]>
kernel-patches-bot
pushed a commit
that referenced
this pull request
Apr 26, 2021
Fix BPF_CORE_READ_BITFIELD() macro used for reading CO-RE-relocatable bitfields. Missing breaks in a switch caused 8-byte reads always. This can confuse libbpf because it does strict checks that memory load size corresponds to the original size of the field, which in this case quite often would be wrong. After fixing that, we run into another problem, which quite subtle, so worth documenting here. The issue is in Clang optimization and CO-RE relocation interactions. Without that asm volatile construct (also known as barrier_var()), Clang will re-order BYTE_OFFSET and BYTE_SIZE relocations and will apply BYTE_OFFSET 4 times for each switch case arm. This will result in the same error from libbpf about mismatch of memory load size and original field size. I.e., if we were reading u32, we'd still have *(u8 *), *(u16 *), *(u32 *), and *(u64 *) memory loads, three of which will fail. Using barrier_var() forces Clang to apply BYTE_OFFSET relocation first (and once) to calculate p, after which value of p is used without relocation in each of switch case arms, doing appropiately-sized memory load. Here's the list of relevant relocations and pieces of generated BPF code before and after this patch for test_core_reloc_bitfields_direct selftests. BEFORE ===== #45: core_reloc: insn #160 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 #46: core_reloc: insn #167 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #47: core_reloc: insn #174 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #48: core_reloc: insn #178 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #49: core_reloc: insn #182 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 157: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 159: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 160: b7 02 00 00 04 00 00 00 r2 = 4 ; BYTE_SIZE relocation here ^^^ 161: 66 02 07 00 03 00 00 00 if w2 s> 3 goto +7 <LBB0_63> 162: 16 02 0d 00 01 00 00 00 if w2 == 1 goto +13 <LBB0_65> 163: 16 02 01 00 02 00 00 00 if w2 == 2 goto +1 <LBB0_66> 164: 05 00 12 00 00 00 00 00 goto +18 <LBB0_69> 0000000000000528 <LBB0_66>: 165: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 167: 69 11 08 00 00 00 00 00 r1 = *(u16 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 168: 05 00 0e 00 00 00 00 00 goto +14 <LBB0_69> 0000000000000548 <LBB0_63>: 169: 16 02 0a 00 04 00 00 00 if w2 == 4 goto +10 <LBB0_67> 170: 16 02 01 00 08 00 00 00 if w2 == 8 goto +1 <LBB0_68> 171: 05 00 0b 00 00 00 00 00 goto +11 <LBB0_69> 0000000000000560 <LBB0_68>: 172: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 174: 79 11 08 00 00 00 00 00 r1 = *(u64 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 175: 05 00 07 00 00 00 00 00 goto +7 <LBB0_69> 0000000000000580 <LBB0_65>: 176: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 178: 71 11 08 00 00 00 00 00 r1 = *(u8 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 179: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 00000000000005a0 <LBB0_67>: 180: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 182: 61 11 08 00 00 00 00 00 r1 = *(u32 *)(r1 + 8) ; BYTE_OFFSET relo here w/ RIGHT size ^^^^^^^^^^^^^^^^ 00000000000005b8 <LBB0_69>: 183: 67 01 00 00 20 00 00 00 r1 <<= 32 184: b7 02 00 00 00 00 00 00 r2 = 0 185: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 186: c7 01 00 00 20 00 00 00 r1 s>>= 32 187: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000005e0 <LBB0_71>: 188: 77 01 00 00 20 00 00 00 r1 >>= 32 AFTER ===== #30: core_reloc: insn #132 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #31: core_reloc: insn #134 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 129: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 131: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 132: b7 01 00 00 08 00 00 00 r1 = 8 ; BYTE_OFFSET relo here ^^^ ; no size check for non-memory dereferencing instructions 133: 0f 12 00 00 00 00 00 00 r2 += r1 134: b7 03 00 00 04 00 00 00 r3 = 4 ; BYTE_SIZE relocation here ^^^ 135: 66 03 05 00 03 00 00 00 if w3 s> 3 goto +5 <LBB0_63> 136: 16 03 09 00 01 00 00 00 if w3 == 1 goto +9 <LBB0_65> 137: 16 03 01 00 02 00 00 00 if w3 == 2 goto +1 <LBB0_66> 138: 05 00 0a 00 00 00 00 00 goto +10 <LBB0_69> 0000000000000458 <LBB0_66>: 139: 69 21 00 00 00 00 00 00 r1 = *(u16 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 140: 05 00 08 00 00 00 00 00 goto +8 <LBB0_69> 0000000000000468 <LBB0_63>: 141: 16 03 06 00 04 00 00 00 if w3 == 4 goto +6 <LBB0_67> 142: 16 03 01 00 08 00 00 00 if w3 == 8 goto +1 <LBB0_68> 143: 05 00 05 00 00 00 00 00 goto +5 <LBB0_69> 0000000000000480 <LBB0_68>: 144: 79 21 00 00 00 00 00 00 r1 = *(u64 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 145: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 0000000000000490 <LBB0_65>: 146: 71 21 00 00 00 00 00 00 r1 = *(u8 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 147: 05 00 01 00 00 00 00 00 goto +1 <LBB0_69> 00000000000004a0 <LBB0_67>: 148: 61 21 00 00 00 00 00 00 r1 = *(u32 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 00000000000004a8 <LBB0_69>: 149: 67 01 00 00 20 00 00 00 r1 <<= 32 150: b7 02 00 00 00 00 00 00 r2 = 0 151: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 152: c7 01 00 00 20 00 00 00 r1 s>>= 32 153: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000004d0 <LBB0_71>: 154: 77 01 00 00 20 00 00 00 r1 >>= 323 Acked-by: Lorenz Bauer <[email protected]> Fixes: ee26dad ("libbpf: Add support for relocatable bitfields") Signed-off-by: Andrii Nakryiko <[email protected]>
kernel-patches-bot
pushed a commit
that referenced
this pull request
Apr 27, 2021
Fix BPF_CORE_READ_BITFIELD() macro used for reading CO-RE-relocatable bitfields. Missing breaks in a switch caused 8-byte reads always. This can confuse libbpf because it does strict checks that memory load size corresponds to the original size of the field, which in this case quite often would be wrong. After fixing that, we run into another problem, which quite subtle, so worth documenting here. The issue is in Clang optimization and CO-RE relocation interactions. Without that asm volatile construct (also known as barrier_var()), Clang will re-order BYTE_OFFSET and BYTE_SIZE relocations and will apply BYTE_OFFSET 4 times for each switch case arm. This will result in the same error from libbpf about mismatch of memory load size and original field size. I.e., if we were reading u32, we'd still have *(u8 *), *(u16 *), *(u32 *), and *(u64 *) memory loads, three of which will fail. Using barrier_var() forces Clang to apply BYTE_OFFSET relocation first (and once) to calculate p, after which value of p is used without relocation in each of switch case arms, doing appropiately-sized memory load. Here's the list of relevant relocations and pieces of generated BPF code before and after this patch for test_core_reloc_bitfields_direct selftests. BEFORE ===== #45: core_reloc: insn #160 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 #46: core_reloc: insn #167 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #47: core_reloc: insn #174 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #48: core_reloc: insn #178 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #49: core_reloc: insn #182 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 157: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 159: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 160: b7 02 00 00 04 00 00 00 r2 = 4 ; BYTE_SIZE relocation here ^^^ 161: 66 02 07 00 03 00 00 00 if w2 s> 3 goto +7 <LBB0_63> 162: 16 02 0d 00 01 00 00 00 if w2 == 1 goto +13 <LBB0_65> 163: 16 02 01 00 02 00 00 00 if w2 == 2 goto +1 <LBB0_66> 164: 05 00 12 00 00 00 00 00 goto +18 <LBB0_69> 0000000000000528 <LBB0_66>: 165: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 167: 69 11 08 00 00 00 00 00 r1 = *(u16 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 168: 05 00 0e 00 00 00 00 00 goto +14 <LBB0_69> 0000000000000548 <LBB0_63>: 169: 16 02 0a 00 04 00 00 00 if w2 == 4 goto +10 <LBB0_67> 170: 16 02 01 00 08 00 00 00 if w2 == 8 goto +1 <LBB0_68> 171: 05 00 0b 00 00 00 00 00 goto +11 <LBB0_69> 0000000000000560 <LBB0_68>: 172: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 174: 79 11 08 00 00 00 00 00 r1 = *(u64 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 175: 05 00 07 00 00 00 00 00 goto +7 <LBB0_69> 0000000000000580 <LBB0_65>: 176: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 178: 71 11 08 00 00 00 00 00 r1 = *(u8 *)(r1 + 8) ; BYTE_OFFSET relo here w/ WRONG size ^^^^^^^^^^^^^^^^ 179: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 00000000000005a0 <LBB0_67>: 180: 18 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r1 = 0 ll 182: 61 11 08 00 00 00 00 00 r1 = *(u32 *)(r1 + 8) ; BYTE_OFFSET relo here w/ RIGHT size ^^^^^^^^^^^^^^^^ 00000000000005b8 <LBB0_69>: 183: 67 01 00 00 20 00 00 00 r1 <<= 32 184: b7 02 00 00 00 00 00 00 r2 = 0 185: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 186: c7 01 00 00 20 00 00 00 r1 s>>= 32 187: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000005e0 <LBB0_71>: 188: 77 01 00 00 20 00 00 00 r1 >>= 32 AFTER ===== #30: core_reloc: insn #132 --> [5] + 0:5: byte_off --> struct core_reloc_bitfields.u32 #31: core_reloc: insn #134 --> [5] + 0:5: byte_sz --> struct core_reloc_bitfields.u32 129: 18 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 r2 = 0 ll 131: 7b 12 20 01 00 00 00 00 *(u64 *)(r2 + 288) = r1 132: b7 01 00 00 08 00 00 00 r1 = 8 ; BYTE_OFFSET relo here ^^^ ; no size check for non-memory dereferencing instructions 133: 0f 12 00 00 00 00 00 00 r2 += r1 134: b7 03 00 00 04 00 00 00 r3 = 4 ; BYTE_SIZE relocation here ^^^ 135: 66 03 05 00 03 00 00 00 if w3 s> 3 goto +5 <LBB0_63> 136: 16 03 09 00 01 00 00 00 if w3 == 1 goto +9 <LBB0_65> 137: 16 03 01 00 02 00 00 00 if w3 == 2 goto +1 <LBB0_66> 138: 05 00 0a 00 00 00 00 00 goto +10 <LBB0_69> 0000000000000458 <LBB0_66>: 139: 69 21 00 00 00 00 00 00 r1 = *(u16 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 140: 05 00 08 00 00 00 00 00 goto +8 <LBB0_69> 0000000000000468 <LBB0_63>: 141: 16 03 06 00 04 00 00 00 if w3 == 4 goto +6 <LBB0_67> 142: 16 03 01 00 08 00 00 00 if w3 == 8 goto +1 <LBB0_68> 143: 05 00 05 00 00 00 00 00 goto +5 <LBB0_69> 0000000000000480 <LBB0_68>: 144: 79 21 00 00 00 00 00 00 r1 = *(u64 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 145: 05 00 03 00 00 00 00 00 goto +3 <LBB0_69> 0000000000000490 <LBB0_65>: 146: 71 21 00 00 00 00 00 00 r1 = *(u8 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 147: 05 00 01 00 00 00 00 00 goto +1 <LBB0_69> 00000000000004a0 <LBB0_67>: 148: 61 21 00 00 00 00 00 00 r1 = *(u32 *)(r2 + 0) ; NO CO-RE relocation here ^^^^^^^^^^^^^^^^ 00000000000004a8 <LBB0_69>: 149: 67 01 00 00 20 00 00 00 r1 <<= 32 150: b7 02 00 00 00 00 00 00 r2 = 0 151: 16 02 02 00 00 00 00 00 if w2 == 0 goto +2 <LBB0_71> 152: c7 01 00 00 20 00 00 00 r1 s>>= 32 153: 05 00 01 00 00 00 00 00 goto +1 <LBB0_72> 00000000000004d0 <LBB0_71>: 154: 77 01 00 00 20 00 00 00 r1 >>= 323 Fixes: ee26dad ("libbpf: Add support for relocatable bitfields") Signed-off-by: Andrii Nakryiko <[email protected]> Signed-off-by: Alexei Starovoitov <[email protected]> Acked-by: Lorenz Bauer <[email protected]> Link: https://lore.kernel.org/bpf/[email protected]
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Jun 8, 2023
SMCRv1 has a similar issue to SMCRv2 (see link below) that may access invalid MRs of RMBs when construct LLC ADD LINK CONT messages. BUG: kernel NULL pointer dereference, address: 0000000000000014 #PF: supervisor read access in kernel mode #PF: error_code(0x0000) - not-present page PGD 0 P4D 0 Oops: 0000 [#1] PREEMPT SMP PTI CPU: 5 PID: 48 Comm: kworker/5:0 Kdump: loaded Tainted: G W E 6.4.0-rc3+ #49 Workqueue: events smc_llc_add_link_work [smc] RIP: 0010:smc_llc_add_link_cont+0x160/0x270 [smc] RSP: 0018:ffffa737801d3d50 EFLAGS: 00010286 RAX: ffff964f82144000 RBX: ffffa737801d3dd8 RCX: 0000000000000000 RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff964f81370c30 RBP: ffffa737801d3dd4 R08: ffff964f81370000 R09: ffffa737801d3db0 R10: 0000000000000001 R11: 0000000000000060 R12: ffff964f82e70000 R13: ffff964f81370c38 R14: ffffa737801d3dd3 R15: 0000000000000001 FS: 0000000000000000(0000) GS:ffff9652bfd40000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000014 CR3: 000000008fa20004 CR4: 00000000003706e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: <TASK> smc_llc_srv_rkey_exchange+0xa7/0x190 [smc] smc_llc_srv_add_link+0x3ae/0x5a0 [smc] smc_llc_add_link_work+0xb8/0x140 [smc] process_one_work+0x1e5/0x3f0 worker_thread+0x4d/0x2f0 ? __pfx_worker_thread+0x10/0x10 kthread+0xe5/0x120 ? __pfx_kthread+0x10/0x10 ret_from_fork+0x2c/0x50 </TASK> When an alernate RNIC is available in system, SMC will try to add a new link based on the RNIC for resilience. All the RMBs in use will be mapped to the new link. Then the RMBs' MRs corresponding to the new link will be filled into LLC messages. For SMCRv1, they are ADD LINK CONT messages. However smc_llc_add_link_cont() may mistakenly access to unused RMBs which haven't been mapped to the new link and have no valid MRs, thus causing a crash. So this patch fixes it. Fixes: 87f88cd ("net/smc: rkey processing for a new link as SMC client") Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Wen Gu <[email protected]> Reviewed-by: Wenjia Zhang <[email protected]> Reviewed-by: Tony Lu <[email protected]> Signed-off-by: David S. Miller <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Oct 27, 2023
Christoph Paasch reported a panic in TCP stack [1] Indeed, we should not call sk_dst_reset() without holding the socket lock, as __sk_dst_get() callers do not all rely on bare RCU. [1] BUG: kernel NULL pointer dereference, address: 0000000000000000 PGD 12bad6067 P4D 12bad6067 PUD 12bad5067 PMD 0 Oops: 0000 [#1] PREEMPT SMP CPU: 1 PID: 2750 Comm: syz-executor.5 Not tainted 6.6.0-rc4-g7a5720a344e7 #49 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7 04/01/2014 RIP: 0010:tcp_get_metrics+0x118/0x8f0 net/ipv4/tcp_metrics.c:321 Code: c7 44 24 70 02 00 8b 03 89 44 24 48 c7 44 24 4c 00 00 00 00 66 c7 44 24 58 02 00 66 ba 02 00 b1 01 89 4c 24 04 4c 89 7c 24 10 <49> 8b 0f 48 8b 89 50 05 00 00 48 89 4c 24 30 33 81 00 02 00 00 69 RSP: 0018:ffffc90000af79b8 EFLAGS: 00010293 RAX: 000000000100007f RBX: ffff88812ae8f500 RCX: ffff88812b5f8f01 RDX: 0000000000000002 RSI: ffffffff8300f080 RDI: 0000000000000002 RBP: 0000000000000002 R08: 0000000000000003 R09: ffffffff8205eca0 R10: 0000000000000002 R11: ffff88812b5f8f00 R12: ffff88812a9e0580 R13: 0000000000000000 R14: ffff88812ae8fbd2 R15: 0000000000000000 FS: 00007f70a006b640(0000) GS:ffff88813bd00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000000 CR3: 000000012bad7003 CR4: 0000000000170ee0 Call Trace: <TASK> tcp_fastopen_cache_get+0x32/0x140 net/ipv4/tcp_metrics.c:567 tcp_fastopen_cookie_check+0x28/0x180 net/ipv4/tcp_fastopen.c:419 tcp_connect+0x9c8/0x12a0 net/ipv4/tcp_output.c:3839 tcp_v4_connect+0x645/0x6e0 net/ipv4/tcp_ipv4.c:323 __inet_stream_connect+0x120/0x590 net/ipv4/af_inet.c:676 tcp_sendmsg_fastopen+0x2d6/0x3a0 net/ipv4/tcp.c:1021 tcp_sendmsg_locked+0x1957/0x1b00 net/ipv4/tcp.c:1073 tcp_sendmsg+0x30/0x50 net/ipv4/tcp.c:1336 __sock_sendmsg+0x83/0xd0 net/socket.c:730 __sys_sendto+0x20a/0x2a0 net/socket.c:2194 __do_sys_sendto net/socket.c:2206 [inline] Fixes: e08d0b3 ("inet: implement lockless IP_TOS") Reported-by: Christoph Paasch <[email protected]> Signed-off-by: Eric Dumazet <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Paolo Abeni <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Mar 3, 2024
bnxt_alloc_mem() dereferences ::vnic_info in the variable declaration block, but allocates it much later. As a result, the following crash happens on my setup: BUG: kernel NULL pointer dereference, address: 0000000000000090 fbcon: Taking over console #PF: supervisor write access in kernel mode #PF: error_code (0x0002) - not-present page PGD 12f382067 P4D 0 Oops: 8002 [#1] PREEMPT SMP NOPTI CPU: 47 PID: 2516 Comm: NetworkManager Not tainted 6.8.0-rc5-libeth+ #49 Hardware name: Intel Corporation M50CYP2SBSTD/M58CYP2SBSTD, BIOS SE5C620.86B.01.01.0088.2305172341 05/17/2023 RIP: 0010:bnxt_alloc_mem+0x1609/0x1910 [bnxt_en] Code: 81 c8 48 83 c8 08 31 c9 e9 d7 fe ff ff c7 44 24 Oc 00 00 00 00 49 89 d5 e9 2d fe ff ff 41 89 c6 e9 88 00 00 00 48 8b 44 24 50 <80> 88 90 00 00 00 Od 8b 43 74 a8 02 75 1e f6 83 14 02 00 00 80 74 RSP: 0018:ff3f25580f3432c8 EFLAGS: 00010246 RAX: 0000000000000000 RBX: ff15a5cfc45249e0 RCX: 0000002079777000 RDX: ff15a5dfb9767000 RSI: 0000000000000000 RDI: 0000000000000000 RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000 R10: ff15a5dfb9777000 R11: ffffff8000000000 R12: 0000000000000000 R13: 0000000000000000 R14: 0000000000000020 R15: ff15a5cfce34f540 FS: 000007fb9a160500(0000) GS:ff15a5dfbefc0000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CRO: 0000000080050033 CR2: 0000000000000090 CR3: 0000000109efc00Z CR4: 0000000000771ef0 DR0: 0000000000000000 DR1: 0000000000000000 DRZ: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 PKRU: 55555554 Call Trace: <TASK> ? __die_body+0x68/0xb0 ? page_fault_oops+0x3a6/0x400 ? exc_page_fault+0x7a/0x1b0 ? asm_exc_page_fault+0x26/8x30 ? bnxt_alloc_mem+0x1609/0x1910 [bnxt_en] ? bnxt_alloc_mem+0x1389/8x1918 [bnxt_en] _bnxt_open_nic+0x198/0xa50 [bnxt_en] ? bnxt_hurm_if_change+0x287/0x3d0 [bnxt_en] bnxt_open+0xeb/0x1b0 [bnxt_en] _dev_open+0x12e/0x1f0 _dev_change_flags+0xb0/0x200 dev_change_flags+0x25/0x60 do_setlink+0x463/0x1260 ? sock_def_readable+0x14/0xc0 ? rtnl_getlink+0x4b9/0x590 ? _nla_validate_parse+0x91/0xfa0 rtnl_newlink+0xbac/0xe40 <...> Don't create a variable and dereference the first array member directly since it's used only once in the code. Fixes: ef4ee64 ("bnxt_en: Define BNXT_VNIC_DEFAULT for the default vnic index") Signed-off-by: Alexander Lobakin <[email protected]> Reviewed-by: Jiri Pirko <[email protected]> Reviewed-by: Michael Chan <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
kuba-moo
pushed a commit
to linux-netdev/testing-bpf-ci
that referenced
this pull request
May 2, 2024
If stack_depot_save_flags() allocates memory it always drops __GFP_NOLOCKDEP flag. So when KASAN tries to track __GFP_NOLOCKDEP allocation we may end up with lockdep splat like bellow: ====================================================== WARNING: possible circular locking dependency detected 6.9.0-rc3+ kernel-patches#49 Not tainted ------------------------------------------------------ kswapd0/149 is trying to acquire lock: ffff88811346a920 (&xfs_nondir_ilock_class){++++}-{4:4}, at: xfs_reclaim_inode+0x3ac/0x590 [xfs] but task is already holding lock: ffffffff8bb33100 (fs_reclaim){+.+.}-{0:0}, at: balance_pgdat+0x5d9/0xad0 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> kernel-patches#1 (fs_reclaim){+.+.}-{0:0}: __lock_acquire+0x7da/0x1030 lock_acquire+0x15d/0x400 fs_reclaim_acquire+0xb5/0x100 prepare_alloc_pages.constprop.0+0xc5/0x230 __alloc_pages+0x12a/0x3f0 alloc_pages_mpol+0x175/0x340 stack_depot_save_flags+0x4c5/0x510 kasan_save_stack+0x30/0x40 kasan_save_track+0x10/0x30 __kasan_slab_alloc+0x83/0x90 kmem_cache_alloc+0x15e/0x4a0 __alloc_object+0x35/0x370 __create_object+0x22/0x90 __kmalloc_node_track_caller+0x477/0x5b0 krealloc+0x5f/0x110 xfs_iext_insert_raw+0x4b2/0x6e0 [xfs] xfs_iext_insert+0x2e/0x130 [xfs] xfs_iread_bmbt_block+0x1a9/0x4d0 [xfs] xfs_btree_visit_block+0xfb/0x290 [xfs] xfs_btree_visit_blocks+0x215/0x2c0 [xfs] xfs_iread_extents+0x1a2/0x2e0 [xfs] xfs_buffered_write_iomap_begin+0x376/0x10a0 [xfs] iomap_iter+0x1d1/0x2d0 iomap_file_buffered_write+0x120/0x1a0 xfs_file_buffered_write+0x128/0x4b0 [xfs] vfs_write+0x675/0x890 ksys_write+0xc3/0x160 do_syscall_64+0x94/0x170 entry_SYSCALL_64_after_hwframe+0x71/0x79 Always preserve __GFP_NOLOCKDEP to fix this. Link: https://lkml.kernel.org/r/[email protected] Fixes: cd11016 ("mm, kasan: stackdepot implementation. Enable stackdepot for SLAB") Signed-off-by: Andrey Ryabinin <[email protected]> Reported-by: Xiubo Li <[email protected]> Closes: https://lore.kernel.org/all/[email protected]/ Reported-by: Damien Le Moal <[email protected]> Closes: https://lore.kernel.org/all/[email protected]/ Suggested-by: Dave Chinner <[email protected]> Tested-by: Xiubo Li <[email protected]> Cc: Christoph Hellwig <[email protected]> Cc: Alexander Potapenko <[email protected]> Cc: <[email protected]> Signed-off-by: Andrew Morton <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Nov 6, 2024
For htab of maps, when the map is removed from the htab, it may hold the last reference of the map. bpf_map_fd_put_ptr() will invoke bpf_map_free_id() to free the id of the removed map element. However, bpf_map_fd_put_ptr() is invoked while holding a bucket lock (raw_spin_lock_t), and bpf_map_free_id() attempts to acquire map_idr_lock (spinlock_t), triggering the following lockdep warning: ============================= [ BUG: Invalid wait context ] 6.11.0-rc4+ #49 Not tainted ----------------------------- test_maps/4881 is trying to lock: ffffffff84884578 (map_idr_lock){+...}-{3:3}, at: bpf_map_free_id.part.0+0x21/0x70 other info that might help us debug this: context-{5:5} 2 locks held by test_maps/4881: #0: ffffffff846caf60 (rcu_read_lock){....}-{1:3}, at: bpf_fd_htab_map_update_elem+0xf9/0x270 #1: ffff888149ced148 (&htab->lockdep_key#2){....}-{2:2}, at: htab_map_update_elem+0x178/0xa80 stack backtrace: CPU: 0 UID: 0 PID: 4881 Comm: test_maps Not tainted 6.11.0-rc4+ #49 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), ... Call Trace: <TASK> dump_stack_lvl+0x6e/0xb0 dump_stack+0x10/0x20 __lock_acquire+0x73e/0x36c0 lock_acquire+0x182/0x450 _raw_spin_lock_irqsave+0x43/0x70 bpf_map_free_id.part.0+0x21/0x70 bpf_map_put+0xcf/0x110 bpf_map_fd_put_ptr+0x9a/0xb0 free_htab_elem+0x69/0xe0 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 bpf_map_update_value+0x266/0x380 __sys_bpf+0x21bb/0x36b0 __x64_sys_bpf+0x45/0x60 x64_sys_call+0x1b2a/0x20d0 do_syscall_64+0x5d/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e One way to fix the lockdep warning is using raw_spinlock_t for map_idr_lock as well. However, bpf_map_alloc_id() invokes idr_alloc_cyclic() after acquiring map_idr_lock, it will trigger a similar lockdep warning because the slab's lock (s->cpu_slab->lock) is still a spinlock. Instead of changing map_idr_lock's type, fix the issue by invoking htab_put_fd_value() after htab_unlock_bucket(). However, only deferring the invocation of htab_put_fd_value() is not enough, because the old map pointers in htab of maps can not be saved during batched deletion. Therefore, also defer the invocation of free_htab_elem(), so these to-be-freed elements could be linked together similar to lru map. There are four callers for ->map_fd_put_ptr: (1) alloc_htab_elem() (through htab_put_fd_value()) It invokes ->map_fd_put_ptr() under a raw_spinlock_t. The invocation of htab_put_fd_value() can not simply move after htab_unlock_bucket(), because the old element has already been stashed in htab->extra_elems. It may be reused immediately after htab_unlock_bucket() and the invocation of htab_put_fd_value() after htab_unlock_bucket() may release the newly-added element incorrectly. Therefore, saving the map pointer of the old element for htab of maps before unlocking the bucket and releasing the map_ptr after unlock. Beside the map pointer in the old element, should do the same thing for the special fields in the old element as well. (2) free_htab_elem() (through htab_put_fd_value()) Its caller includes __htab_map_lookup_and_delete_elem(), htab_map_delete_elem() and __htab_map_lookup_and_delete_batch(). For htab_map_delete_elem(), simply invoke free_htab_elem() after htab_unlock_bucket(). For __htab_map_lookup_and_delete_batch(), just like lru map, linking the to-be-freed element into node_to_free list and invoking free_htab_elem() for these element after unlock. It is safe to reuse batch_flink as the link for node_to_free, because these elements have been removed from the hash llist. Because htab of maps doesn't support lookup_and_delete operation, __htab_map_lookup_and_delete_elem() doesn't have the problem, so kept it as is. (3) fd_htab_map_free() It invokes ->map_fd_put_ptr without raw_spinlock_t. (4) bpf_fd_htab_map_update_elem() It invokes ->map_fd_put_ptr without raw_spinlock_t. After moving free_htab_elem() outside htab bucket lock scope, using pcpu_freelist_push() instead of __pcpu_freelist_push() to disable the irq before freeing elements, and protecting the invocations of bpf_mem_cache_free() with migrate_{disable|enable} pair. Signed-off-by: Hou Tao <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Nov 6, 2024
For htab of maps, when the map is removed from the htab, it may hold the last reference of the map. bpf_map_fd_put_ptr() will invoke bpf_map_free_id() to free the id of the removed map element. However, bpf_map_fd_put_ptr() is invoked while holding a bucket lock (raw_spin_lock_t), and bpf_map_free_id() attempts to acquire map_idr_lock (spinlock_t), triggering the following lockdep warning: ============================= [ BUG: Invalid wait context ] 6.11.0-rc4+ #49 Not tainted ----------------------------- test_maps/4881 is trying to lock: ffffffff84884578 (map_idr_lock){+...}-{3:3}, at: bpf_map_free_id.part.0+0x21/0x70 other info that might help us debug this: context-{5:5} 2 locks held by test_maps/4881: #0: ffffffff846caf60 (rcu_read_lock){....}-{1:3}, at: bpf_fd_htab_map_update_elem+0xf9/0x270 #1: ffff888149ced148 (&htab->lockdep_key#2){....}-{2:2}, at: htab_map_update_elem+0x178/0xa80 stack backtrace: CPU: 0 UID: 0 PID: 4881 Comm: test_maps Not tainted 6.11.0-rc4+ #49 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), ... Call Trace: <TASK> dump_stack_lvl+0x6e/0xb0 dump_stack+0x10/0x20 __lock_acquire+0x73e/0x36c0 lock_acquire+0x182/0x450 _raw_spin_lock_irqsave+0x43/0x70 bpf_map_free_id.part.0+0x21/0x70 bpf_map_put+0xcf/0x110 bpf_map_fd_put_ptr+0x9a/0xb0 free_htab_elem+0x69/0xe0 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 bpf_map_update_value+0x266/0x380 __sys_bpf+0x21bb/0x36b0 __x64_sys_bpf+0x45/0x60 x64_sys_call+0x1b2a/0x20d0 do_syscall_64+0x5d/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e One way to fix the lockdep warning is using raw_spinlock_t for map_idr_lock as well. However, bpf_map_alloc_id() invokes idr_alloc_cyclic() after acquiring map_idr_lock, it will trigger a similar lockdep warning because the slab's lock (s->cpu_slab->lock) is still a spinlock. Instead of changing map_idr_lock's type, fix the issue by invoking htab_put_fd_value() after htab_unlock_bucket(). However, only deferring the invocation of htab_put_fd_value() is not enough, because the old map pointers in htab of maps can not be saved during batched deletion. Therefore, also defer the invocation of free_htab_elem(), so these to-be-freed elements could be linked together similar to lru map. There are four callers for ->map_fd_put_ptr: (1) alloc_htab_elem() (through htab_put_fd_value()) It invokes ->map_fd_put_ptr() under a raw_spinlock_t. The invocation of htab_put_fd_value() can not simply move after htab_unlock_bucket(), because the old element has already been stashed in htab->extra_elems. It may be reused immediately after htab_unlock_bucket() and the invocation of htab_put_fd_value() after htab_unlock_bucket() may release the newly-added element incorrectly. Therefore, saving the map pointer of the old element for htab of maps before unlocking the bucket and releasing the map_ptr after unlock. Beside the map pointer in the old element, should do the same thing for the special fields in the old element as well. (2) free_htab_elem() (through htab_put_fd_value()) Its caller includes __htab_map_lookup_and_delete_elem(), htab_map_delete_elem() and __htab_map_lookup_and_delete_batch(). For htab_map_delete_elem(), simply invoke free_htab_elem() after htab_unlock_bucket(). For __htab_map_lookup_and_delete_batch(), just like lru map, linking the to-be-freed element into node_to_free list and invoking free_htab_elem() for these element after unlock. It is safe to reuse batch_flink as the link for node_to_free, because these elements have been removed from the hash llist. Because htab of maps doesn't support lookup_and_delete operation, __htab_map_lookup_and_delete_elem() doesn't have the problem, so kept it as is. (3) fd_htab_map_free() It invokes ->map_fd_put_ptr without raw_spinlock_t. (4) bpf_fd_htab_map_update_elem() It invokes ->map_fd_put_ptr without raw_spinlock_t. After moving free_htab_elem() outside htab bucket lock scope, using pcpu_freelist_push() instead of __pcpu_freelist_push() to disable the irq before freeing elements, and protecting the invocations of bpf_mem_cache_free() with migrate_{disable|enable} pair. Signed-off-by: Hou Tao <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Nov 7, 2024
For htab of maps, when the map is removed from the htab, it may hold the last reference of the map. bpf_map_fd_put_ptr() will invoke bpf_map_free_id() to free the id of the removed map element. However, bpf_map_fd_put_ptr() is invoked while holding a bucket lock (raw_spin_lock_t), and bpf_map_free_id() attempts to acquire map_idr_lock (spinlock_t), triggering the following lockdep warning: ============================= [ BUG: Invalid wait context ] 6.11.0-rc4+ #49 Not tainted ----------------------------- test_maps/4881 is trying to lock: ffffffff84884578 (map_idr_lock){+...}-{3:3}, at: bpf_map_free_id.part.0+0x21/0x70 other info that might help us debug this: context-{5:5} 2 locks held by test_maps/4881: #0: ffffffff846caf60 (rcu_read_lock){....}-{1:3}, at: bpf_fd_htab_map_update_elem+0xf9/0x270 #1: ffff888149ced148 (&htab->lockdep_key#2){....}-{2:2}, at: htab_map_update_elem+0x178/0xa80 stack backtrace: CPU: 0 UID: 0 PID: 4881 Comm: test_maps Not tainted 6.11.0-rc4+ #49 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), ... Call Trace: <TASK> dump_stack_lvl+0x6e/0xb0 dump_stack+0x10/0x20 __lock_acquire+0x73e/0x36c0 lock_acquire+0x182/0x450 _raw_spin_lock_irqsave+0x43/0x70 bpf_map_free_id.part.0+0x21/0x70 bpf_map_put+0xcf/0x110 bpf_map_fd_put_ptr+0x9a/0xb0 free_htab_elem+0x69/0xe0 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 bpf_map_update_value+0x266/0x380 __sys_bpf+0x21bb/0x36b0 __x64_sys_bpf+0x45/0x60 x64_sys_call+0x1b2a/0x20d0 do_syscall_64+0x5d/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e One way to fix the lockdep warning is using raw_spinlock_t for map_idr_lock as well. However, bpf_map_alloc_id() invokes idr_alloc_cyclic() after acquiring map_idr_lock, it will trigger a similar lockdep warning because the slab's lock (s->cpu_slab->lock) is still a spinlock. Instead of changing map_idr_lock's type, fix the issue by invoking htab_put_fd_value() after htab_unlock_bucket(). However, only deferring the invocation of htab_put_fd_value() is not enough, because the old map pointers in htab of maps can not be saved during batched deletion. Therefore, also defer the invocation of free_htab_elem(), so these to-be-freed elements could be linked together similar to lru map. There are four callers for ->map_fd_put_ptr: (1) alloc_htab_elem() (through htab_put_fd_value()) It invokes ->map_fd_put_ptr() under a raw_spinlock_t. The invocation of htab_put_fd_value() can not simply move after htab_unlock_bucket(), because the old element has already been stashed in htab->extra_elems. It may be reused immediately after htab_unlock_bucket() and the invocation of htab_put_fd_value() after htab_unlock_bucket() may release the newly-added element incorrectly. Therefore, saving the map pointer of the old element for htab of maps before unlocking the bucket and releasing the map_ptr after unlock. Beside the map pointer in the old element, should do the same thing for the special fields in the old element as well. (2) free_htab_elem() (through htab_put_fd_value()) Its caller includes __htab_map_lookup_and_delete_elem(), htab_map_delete_elem() and __htab_map_lookup_and_delete_batch(). For htab_map_delete_elem(), simply invoke free_htab_elem() after htab_unlock_bucket(). For __htab_map_lookup_and_delete_batch(), just like lru map, linking the to-be-freed element into node_to_free list and invoking free_htab_elem() for these element after unlock. It is safe to reuse batch_flink as the link for node_to_free, because these elements have been removed from the hash llist. Because htab of maps doesn't support lookup_and_delete operation, __htab_map_lookup_and_delete_elem() doesn't have the problem, so kept it as is. (3) fd_htab_map_free() It invokes ->map_fd_put_ptr without raw_spinlock_t. (4) bpf_fd_htab_map_update_elem() It invokes ->map_fd_put_ptr without raw_spinlock_t. After moving free_htab_elem() outside htab bucket lock scope, using pcpu_freelist_push() instead of __pcpu_freelist_push() to disable the irq before freeing elements, and protecting the invocations of bpf_mem_cache_free() with migrate_{disable|enable} pair. Signed-off-by: Hou Tao <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Nov 7, 2024
For htab of maps, when the map is removed from the htab, it may hold the last reference of the map. bpf_map_fd_put_ptr() will invoke bpf_map_free_id() to free the id of the removed map element. However, bpf_map_fd_put_ptr() is invoked while holding a bucket lock (raw_spin_lock_t), and bpf_map_free_id() attempts to acquire map_idr_lock (spinlock_t), triggering the following lockdep warning: ============================= [ BUG: Invalid wait context ] 6.11.0-rc4+ #49 Not tainted ----------------------------- test_maps/4881 is trying to lock: ffffffff84884578 (map_idr_lock){+...}-{3:3}, at: bpf_map_free_id.part.0+0x21/0x70 other info that might help us debug this: context-{5:5} 2 locks held by test_maps/4881: #0: ffffffff846caf60 (rcu_read_lock){....}-{1:3}, at: bpf_fd_htab_map_update_elem+0xf9/0x270 #1: ffff888149ced148 (&htab->lockdep_key#2){....}-{2:2}, at: htab_map_update_elem+0x178/0xa80 stack backtrace: CPU: 0 UID: 0 PID: 4881 Comm: test_maps Not tainted 6.11.0-rc4+ #49 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), ... Call Trace: <TASK> dump_stack_lvl+0x6e/0xb0 dump_stack+0x10/0x20 __lock_acquire+0x73e/0x36c0 lock_acquire+0x182/0x450 _raw_spin_lock_irqsave+0x43/0x70 bpf_map_free_id.part.0+0x21/0x70 bpf_map_put+0xcf/0x110 bpf_map_fd_put_ptr+0x9a/0xb0 free_htab_elem+0x69/0xe0 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 bpf_map_update_value+0x266/0x380 __sys_bpf+0x21bb/0x36b0 __x64_sys_bpf+0x45/0x60 x64_sys_call+0x1b2a/0x20d0 do_syscall_64+0x5d/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e One way to fix the lockdep warning is using raw_spinlock_t for map_idr_lock as well. However, bpf_map_alloc_id() invokes idr_alloc_cyclic() after acquiring map_idr_lock, it will trigger a similar lockdep warning because the slab's lock (s->cpu_slab->lock) is still a spinlock. Instead of changing map_idr_lock's type, fix the issue by invoking htab_put_fd_value() after htab_unlock_bucket(). However, only deferring the invocation of htab_put_fd_value() is not enough, because the old map pointers in htab of maps can not be saved during batched deletion. Therefore, also defer the invocation of free_htab_elem(), so these to-be-freed elements could be linked together similar to lru map. There are four callers for ->map_fd_put_ptr: (1) alloc_htab_elem() (through htab_put_fd_value()) It invokes ->map_fd_put_ptr() under a raw_spinlock_t. The invocation of htab_put_fd_value() can not simply move after htab_unlock_bucket(), because the old element has already been stashed in htab->extra_elems. It may be reused immediately after htab_unlock_bucket() and the invocation of htab_put_fd_value() after htab_unlock_bucket() may release the newly-added element incorrectly. Therefore, saving the map pointer of the old element for htab of maps before unlocking the bucket and releasing the map_ptr after unlock. Beside the map pointer in the old element, should do the same thing for the special fields in the old element as well. (2) free_htab_elem() (through htab_put_fd_value()) Its caller includes __htab_map_lookup_and_delete_elem(), htab_map_delete_elem() and __htab_map_lookup_and_delete_batch(). For htab_map_delete_elem(), simply invoke free_htab_elem() after htab_unlock_bucket(). For __htab_map_lookup_and_delete_batch(), just like lru map, linking the to-be-freed element into node_to_free list and invoking free_htab_elem() for these element after unlock. It is safe to reuse batch_flink as the link for node_to_free, because these elements have been removed from the hash llist. Because htab of maps doesn't support lookup_and_delete operation, __htab_map_lookup_and_delete_elem() doesn't have the problem, so kept it as is. (3) fd_htab_map_free() It invokes ->map_fd_put_ptr without raw_spinlock_t. (4) bpf_fd_htab_map_update_elem() It invokes ->map_fd_put_ptr without raw_spinlock_t. After moving free_htab_elem() outside htab bucket lock scope, using pcpu_freelist_push() instead of __pcpu_freelist_push() to disable the irq before freeing elements, and protecting the invocations of bpf_mem_cache_free() with migrate_{disable|enable} pair. Signed-off-by: Hou Tao <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Nov 8, 2024
For htab of maps, when the map is removed from the htab, it may hold the last reference of the map. bpf_map_fd_put_ptr() will invoke bpf_map_free_id() to free the id of the removed map element. However, bpf_map_fd_put_ptr() is invoked while holding a bucket lock (raw_spin_lock_t), and bpf_map_free_id() attempts to acquire map_idr_lock (spinlock_t), triggering the following lockdep warning: ============================= [ BUG: Invalid wait context ] 6.11.0-rc4+ #49 Not tainted ----------------------------- test_maps/4881 is trying to lock: ffffffff84884578 (map_idr_lock){+...}-{3:3}, at: bpf_map_free_id.part.0+0x21/0x70 other info that might help us debug this: context-{5:5} 2 locks held by test_maps/4881: #0: ffffffff846caf60 (rcu_read_lock){....}-{1:3}, at: bpf_fd_htab_map_update_elem+0xf9/0x270 #1: ffff888149ced148 (&htab->lockdep_key#2){....}-{2:2}, at: htab_map_update_elem+0x178/0xa80 stack backtrace: CPU: 0 UID: 0 PID: 4881 Comm: test_maps Not tainted 6.11.0-rc4+ #49 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), ... Call Trace: <TASK> dump_stack_lvl+0x6e/0xb0 dump_stack+0x10/0x20 __lock_acquire+0x73e/0x36c0 lock_acquire+0x182/0x450 _raw_spin_lock_irqsave+0x43/0x70 bpf_map_free_id.part.0+0x21/0x70 bpf_map_put+0xcf/0x110 bpf_map_fd_put_ptr+0x9a/0xb0 free_htab_elem+0x69/0xe0 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 bpf_map_update_value+0x266/0x380 __sys_bpf+0x21bb/0x36b0 __x64_sys_bpf+0x45/0x60 x64_sys_call+0x1b2a/0x20d0 do_syscall_64+0x5d/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e One way to fix the lockdep warning is using raw_spinlock_t for map_idr_lock as well. However, bpf_map_alloc_id() invokes idr_alloc_cyclic() after acquiring map_idr_lock, it will trigger a similar lockdep warning because the slab's lock (s->cpu_slab->lock) is still a spinlock. Instead of changing map_idr_lock's type, fix the issue by invoking htab_put_fd_value() after htab_unlock_bucket(). However, only deferring the invocation of htab_put_fd_value() is not enough, because the old map pointers in htab of maps can not be saved during batched deletion. Therefore, also defer the invocation of free_htab_elem(), so these to-be-freed elements could be linked together similar to lru map. There are four callers for ->map_fd_put_ptr: (1) alloc_htab_elem() (through htab_put_fd_value()) It invokes ->map_fd_put_ptr() under a raw_spinlock_t. The invocation of htab_put_fd_value() can not simply move after htab_unlock_bucket(), because the old element has already been stashed in htab->extra_elems. It may be reused immediately after htab_unlock_bucket() and the invocation of htab_put_fd_value() after htab_unlock_bucket() may release the newly-added element incorrectly. Therefore, saving the map pointer of the old element for htab of maps before unlocking the bucket and releasing the map_ptr after unlock. Beside the map pointer in the old element, should do the same thing for the special fields in the old element as well. (2) free_htab_elem() (through htab_put_fd_value()) Its caller includes __htab_map_lookup_and_delete_elem(), htab_map_delete_elem() and __htab_map_lookup_and_delete_batch(). For htab_map_delete_elem(), simply invoke free_htab_elem() after htab_unlock_bucket(). For __htab_map_lookup_and_delete_batch(), just like lru map, linking the to-be-freed element into node_to_free list and invoking free_htab_elem() for these element after unlock. It is safe to reuse batch_flink as the link for node_to_free, because these elements have been removed from the hash llist. Because htab of maps doesn't support lookup_and_delete operation, __htab_map_lookup_and_delete_elem() doesn't have the problem, so kept it as is. (3) fd_htab_map_free() It invokes ->map_fd_put_ptr without raw_spinlock_t. (4) bpf_fd_htab_map_update_elem() It invokes ->map_fd_put_ptr without raw_spinlock_t. After moving free_htab_elem() outside htab bucket lock scope, using pcpu_freelist_push() instead of __pcpu_freelist_push() to disable the irq before freeing elements, and protecting the invocations of bpf_mem_cache_free() with migrate_{disable|enable} pair. Signed-off-by: Hou Tao <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Alexei Starovoitov <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Nov 11, 2024
For htab of maps, when the map is removed from the htab, it may hold the last reference of the map. bpf_map_fd_put_ptr() will invoke bpf_map_free_id() to free the id of the removed map element. However, bpf_map_fd_put_ptr() is invoked while holding a bucket lock (raw_spin_lock_t), and bpf_map_free_id() attempts to acquire map_idr_lock (spinlock_t), triggering the following lockdep warning: ============================= [ BUG: Invalid wait context ] 6.11.0-rc4+ #49 Not tainted ----------------------------- test_maps/4881 is trying to lock: ffffffff84884578 (map_idr_lock){+...}-{3:3}, at: bpf_map_free_id.part.0+0x21/0x70 other info that might help us debug this: context-{5:5} 2 locks held by test_maps/4881: #0: ffffffff846caf60 (rcu_read_lock){....}-{1:3}, at: bpf_fd_htab_map_update_elem+0xf9/0x270 #1: ffff888149ced148 (&htab->lockdep_key#2){....}-{2:2}, at: htab_map_update_elem+0x178/0xa80 stack backtrace: CPU: 0 UID: 0 PID: 4881 Comm: test_maps Not tainted 6.11.0-rc4+ #49 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), ... Call Trace: <TASK> dump_stack_lvl+0x6e/0xb0 dump_stack+0x10/0x20 __lock_acquire+0x73e/0x36c0 lock_acquire+0x182/0x450 _raw_spin_lock_irqsave+0x43/0x70 bpf_map_free_id.part.0+0x21/0x70 bpf_map_put+0xcf/0x110 bpf_map_fd_put_ptr+0x9a/0xb0 free_htab_elem+0x69/0xe0 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 bpf_map_update_value+0x266/0x380 __sys_bpf+0x21bb/0x36b0 __x64_sys_bpf+0x45/0x60 x64_sys_call+0x1b2a/0x20d0 do_syscall_64+0x5d/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e One way to fix the lockdep warning is using raw_spinlock_t for map_idr_lock as well. However, bpf_map_alloc_id() invokes idr_alloc_cyclic() after acquiring map_idr_lock, it will trigger a similar lockdep warning because the slab's lock (s->cpu_slab->lock) is still a spinlock. Instead of changing map_idr_lock's type, fix the issue by invoking htab_put_fd_value() after htab_unlock_bucket(). However, only deferring the invocation of htab_put_fd_value() is not enough, because the old map pointers in htab of maps can not be saved during batched deletion. Therefore, also defer the invocation of free_htab_elem(), so these to-be-freed elements could be linked together similar to lru map. There are four callers for ->map_fd_put_ptr: (1) alloc_htab_elem() (through htab_put_fd_value()) It invokes ->map_fd_put_ptr() under a raw_spinlock_t. The invocation of htab_put_fd_value() can not simply move after htab_unlock_bucket(), because the old element has already been stashed in htab->extra_elems. It may be reused immediately after htab_unlock_bucket() and the invocation of htab_put_fd_value() after htab_unlock_bucket() may release the newly-added element incorrectly. Therefore, saving the map pointer of the old element for htab of maps before unlocking the bucket and releasing the map_ptr after unlock. Beside the map pointer in the old element, should do the same thing for the special fields in the old element as well. (2) free_htab_elem() (through htab_put_fd_value()) Its caller includes __htab_map_lookup_and_delete_elem(), htab_map_delete_elem() and __htab_map_lookup_and_delete_batch(). For htab_map_delete_elem(), simply invoke free_htab_elem() after htab_unlock_bucket(). For __htab_map_lookup_and_delete_batch(), just like lru map, linking the to-be-freed element into node_to_free list and invoking free_htab_elem() for these element after unlock. It is safe to reuse batch_flink as the link for node_to_free, because these elements have been removed from the hash llist. Because htab of maps doesn't support lookup_and_delete operation, __htab_map_lookup_and_delete_elem() doesn't have the problem, so kept it as is. (3) fd_htab_map_free() It invokes ->map_fd_put_ptr without raw_spinlock_t. (4) bpf_fd_htab_map_update_elem() It invokes ->map_fd_put_ptr without raw_spinlock_t. After moving free_htab_elem() outside htab bucket lock scope, using pcpu_freelist_push() instead of __pcpu_freelist_push() to disable the irq before freeing elements, and protecting the invocations of bpf_mem_cache_free() with migrate_{disable|enable} pair. Signed-off-by: Hou Tao <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Alexei Starovoitov <[email protected]> Signed-off-by: Andrii Nakryiko <[email protected]>
kernel-patches-daemon-bpf bot
pushed a commit
that referenced
this pull request
Nov 25, 2024
open_cached_dir() may either race with the tcon reconnection even before
compound_send_recv() or directly trigger a reconnection via
SMB2_open_init() or SMB_query_info_init().
The reconnection process invokes invalidate_all_cached_dirs() via
cifs_mark_open_files_invalid(), which removes all cfids from the
cfids->entries list but doesn't drop a ref if has_lease isn't true. This
results in the currently-being-constructed cfid not being on the list,
but still having a refcount of 2. It leaks if returned from
open_cached_dir().
Fix this by setting cfid->has_lease when the ref is actually taken; the
cfid will not be used by other threads until it has a valid time.
Addresses these kmemleaks:
unreferenced object 0xffff8881090c4000 (size 1024):
comm "bash", pid 1860, jiffies 4295126592
hex dump (first 32 bytes):
00 01 00 00 00 00 ad de 22 01 00 00 00 00 ad de ........".......
00 ca 45 22 81 88 ff ff f8 dc 4f 04 81 88 ff ff ..E"......O.....
backtrace (crc 6f58c20f):
[<ffffffff8b895a1e>] __kmalloc_cache_noprof+0x2be/0x350
[<ffffffff8bda06e3>] open_cached_dir+0x993/0x1fb0
[<ffffffff8bdaa750>] cifs_readdir+0x15a0/0x1d50
[<ffffffff8b9a853f>] iterate_dir+0x28f/0x4b0
[<ffffffff8b9a9aed>] __x64_sys_getdents64+0xfd/0x200
[<ffffffff8cf6da05>] do_syscall_64+0x95/0x1a0
[<ffffffff8d00012f>] entry_SYSCALL_64_after_hwframe+0x76/0x7e
unreferenced object 0xffff8881044fdcf8 (size 8):
comm "bash", pid 1860, jiffies 4295126592
hex dump (first 8 bytes):
00 cc cc cc cc cc cc cc ........
backtrace (crc 10c106a9):
[<ffffffff8b89a3d3>] __kmalloc_node_track_caller_noprof+0x363/0x480
[<ffffffff8b7d7256>] kstrdup+0x36/0x60
[<ffffffff8bda0700>] open_cached_dir+0x9b0/0x1fb0
[<ffffffff8bdaa750>] cifs_readdir+0x15a0/0x1d50
[<ffffffff8b9a853f>] iterate_dir+0x28f/0x4b0
[<ffffffff8b9a9aed>] __x64_sys_getdents64+0xfd/0x200
[<ffffffff8cf6da05>] do_syscall_64+0x95/0x1a0
[<ffffffff8d00012f>] entry_SYSCALL_64_after_hwframe+0x76/0x7e
And addresses these BUG splats when unmounting the SMB filesystem:
BUG: Dentry ffff888140590ba0{i=1000000000080,n=/} still in use (2) [unmount of cifs cifs]
WARNING: CPU: 3 PID: 3433 at fs/dcache.c:1536 umount_check+0xd0/0x100
Modules linked in:
CPU: 3 UID: 0 PID: 3433 Comm: bash Not tainted 6.12.0-rc4-g850925a8133c-dirty #49
Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 11/12/2020
RIP: 0010:umount_check+0xd0/0x100
Code: 8d 7c 24 40 e8 31 5a f4 ff 49 8b 54 24 40 41 56 49 89 e9 45 89 e8 48 89 d9 41 57 48 89 de 48 c7 c7 80 e7 db ac e8 f0 72 9a ff <0f> 0b 58 31 c0 5a 5b 5d 41 5c 41 5d 41 5e 41 5f e9 2b e5 5d 01 41
RSP: 0018:ffff88811cc27978 EFLAGS: 00010286
RAX: 0000000000000000 RBX: ffff888140590ba0 RCX: ffffffffaaf20bae
RDX: dffffc0000000000 RSI: 0000000000000008 RDI: ffff8881f6fb6f40
RBP: ffff8881462ec000 R08: 0000000000000001 R09: ffffed1023984ee3
R10: ffff88811cc2771f R11: 00000000016cfcc0 R12: ffff888134383e08
R13: 0000000000000002 R14: ffff8881462ec668 R15: ffffffffaceab4c0
FS: 00007f23bfa98740(0000) GS:ffff8881f6f80000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000556de4a6f808 CR3: 0000000123c80000 CR4: 0000000000350ef0
Call Trace:
<TASK>
d_walk+0x6a/0x530
shrink_dcache_for_umount+0x6a/0x200
generic_shutdown_super+0x52/0x2a0
kill_anon_super+0x22/0x40
cifs_kill_sb+0x159/0x1e0
deactivate_locked_super+0x66/0xe0
cleanup_mnt+0x140/0x210
task_work_run+0xfb/0x170
syscall_exit_to_user_mode+0x29f/0x2b0
do_syscall_64+0xa1/0x1a0
entry_SYSCALL_64_after_hwframe+0x76/0x7e
RIP: 0033:0x7f23bfb93ae7
Code: ff ff ff ff c3 66 0f 1f 44 00 00 48 8b 0d 11 93 0d 00 f7 d8 64 89 01 b8 ff ff ff ff eb bf 0f 1f 44 00 00 b8 50 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d e9 92 0d 00 f7 d8 64 89 01 48
RSP: 002b:00007ffee9138598 EFLAGS: 00000246 ORIG_RAX: 0000000000000050
RAX: 0000000000000000 RBX: 0000558f1803e9a0 RCX: 00007f23bfb93ae7
RDX: 0000000000000000 RSI: 0000000000000004 RDI: 0000558f1803e9a0
RBP: 0000558f1803e600 R08: 0000000000000007 R09: 0000558f17fab610
R10: d91d5ec34ab757b0 R11: 0000000000000246 R12: 0000000000000001
R13: 0000000000000000 R14: 0000000000000015 R15: 0000000000000000
</TASK>
irq event stamp: 1163486
hardirqs last enabled at (1163485): [<ffffffffac98d344>] _raw_spin_unlock_irqrestore+0x34/0x60
hardirqs last disabled at (1163486): [<ffffffffac97dcfc>] __schedule+0xc7c/0x19a0
softirqs last enabled at (1163482): [<ffffffffab79a3ee>] __smb_send_rqst+0x3de/0x990
softirqs last disabled at (1163480): [<ffffffffac2314f1>] release_sock+0x21/0xf0
---[ end trace 0000000000000000 ]---
VFS: Busy inodes after unmount of cifs (cifs)
------------[ cut here ]------------
kernel BUG at fs/super.c:661!
Oops: invalid opcode: 0000 [#1] PREEMPT SMP KASAN NOPTI
CPU: 1 UID: 0 PID: 3433 Comm: bash Tainted: G W 6.12.0-rc4-g850925a8133c-dirty #49
Tainted: [W]=WARN
Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 11/12/2020
RIP: 0010:generic_shutdown_super+0x290/0x2a0
Code: e8 15 7c f7 ff 48 8b 5d 28 48 89 df e8 09 7c f7 ff 48 8b 0b 48 89 ee 48 8d 95 68 06 00 00 48 c7 c7 80 7f db ac e8 00 69 af ff <0f> 0b 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 90 90 90 90 90 90
RSP: 0018:ffff88811cc27a50 EFLAGS: 00010246
RAX: 000000000000003e RBX: ffffffffae994420 RCX: 0000000000000027
RDX: 0000000000000000 RSI: ffffffffab06180e RDI: ffff8881f6eb18c8
RBP: ffff8881462ec000 R08: 0000000000000001 R09: ffffed103edd6319
R10: ffff8881f6eb18cb R11: 00000000016d3158 R12: ffff8881462ec9c0
R13: ffff8881462ec050 R14: 0000000000000001 R15: 0000000000000000
FS: 00007f23bfa98740(0000) GS:ffff8881f6e80000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f8364005d68 CR3: 0000000123c80000 CR4: 0000000000350ef0
Call Trace:
<TASK>
kill_anon_super+0x22/0x40
cifs_kill_sb+0x159/0x1e0
deactivate_locked_super+0x66/0xe0
cleanup_mnt+0x140/0x210
task_work_run+0xfb/0x170
syscall_exit_to_user_mode+0x29f/0x2b0
do_syscall_64+0xa1/0x1a0
entry_SYSCALL_64_after_hwframe+0x76/0x7e
RIP: 0033:0x7f23bfb93ae7
</TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---
RIP: 0010:generic_shutdown_super+0x290/0x2a0
Code: e8 15 7c f7 ff 48 8b 5d 28 48 89 df e8 09 7c f7 ff 48 8b 0b 48 89 ee 48 8d 95 68 06 00 00 48 c7 c7 80 7f db ac e8 00 69 af ff <0f> 0b 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 90 90 90 90 90 90
RSP: 0018:ffff88811cc27a50 EFLAGS: 00010246
RAX: 000000000000003e RBX: ffffffffae994420 RCX: 0000000000000027
RDX: 0000000000000000 RSI: ffffffffab06180e RDI: ffff8881f6eb18c8
RBP: ffff8881462ec000 R08: 0000000000000001 R09: ffffed103edd6319
R10: ffff8881f6eb18cb R11: 00000000016d3158 R12: ffff8881462ec9c0
R13: ffff8881462ec050 R14: 0000000000000001 R15: 0000000000000000
FS: 00007f23bfa98740(0000) GS:ffff8881f6e80000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f8364005d68 CR3: 0000000123c80000 CR4: 0000000000350ef0
This reproduces eventually with an SMB mount and two shells running
these loops concurrently
- while true; do
cd ~; sleep 1;
for i in {1..3}; do cd /mnt/test/subdir;
echo $PWD; sleep 1; cd ..; echo $PWD; sleep 1;
done;
echo ...;
done
- while true; do
iptables -F OUTPUT; mount -t cifs -a;
for _ in {0..2}; do ls /mnt/test/subdir/ | wc -l; done;
iptables -I OUTPUT -p tcp --dport 445 -j DROP;
sleep 10
echo "unmounting"; umount -l -t cifs -a; echo "done unmounting";
sleep 20
echo "recovering"; iptables -F OUTPUT;
sleep 10;
done
Fixes: ebe98f1 ("cifs: enable caching of directories for which a lease is held")
Fixes: 5c86919 ("smb: client: fix use-after-free in smb2_query_info_compound()")
Cc: [email protected]
Signed-off-by: Paul Aurich <[email protected]>
Signed-off-by: Steve French <[email protected]>
shunghsiyu
pushed a commit
to shunghsiyu/bpf
that referenced
this pull request
Dec 18, 2024
commit 7afb867 upstream. open_cached_dir() may either race with the tcon reconnection even before compound_send_recv() or directly trigger a reconnection via SMB2_open_init() or SMB_query_info_init(). The reconnection process invokes invalidate_all_cached_dirs() via cifs_mark_open_files_invalid(), which removes all cfids from the cfids->entries list but doesn't drop a ref if has_lease isn't true. This results in the currently-being-constructed cfid not being on the list, but still having a refcount of 2. It leaks if returned from open_cached_dir(). Fix this by setting cfid->has_lease when the ref is actually taken; the cfid will not be used by other threads until it has a valid time. Addresses these kmemleaks: unreferenced object 0xffff8881090c4000 (size 1024): comm "bash", pid 1860, jiffies 4295126592 hex dump (first 32 bytes): 00 01 00 00 00 00 ad de 22 01 00 00 00 00 ad de ........"....... 00 ca 45 22 81 88 ff ff f8 dc 4f 04 81 88 ff ff ..E"......O..... backtrace (crc 6f58c20f): [<ffffffff8b895a1e>] __kmalloc_cache_noprof+0x2be/0x350 [<ffffffff8bda06e3>] open_cached_dir+0x993/0x1fb0 [<ffffffff8bdaa750>] cifs_readdir+0x15a0/0x1d50 [<ffffffff8b9a853f>] iterate_dir+0x28f/0x4b0 [<ffffffff8b9a9aed>] __x64_sys_getdents64+0xfd/0x200 [<ffffffff8cf6da05>] do_syscall_64+0x95/0x1a0 [<ffffffff8d00012f>] entry_SYSCALL_64_after_hwframe+0x76/0x7e unreferenced object 0xffff8881044fdcf8 (size 8): comm "bash", pid 1860, jiffies 4295126592 hex dump (first 8 bytes): 00 cc cc cc cc cc cc cc ........ backtrace (crc 10c106a9): [<ffffffff8b89a3d3>] __kmalloc_node_track_caller_noprof+0x363/0x480 [<ffffffff8b7d7256>] kstrdup+0x36/0x60 [<ffffffff8bda0700>] open_cached_dir+0x9b0/0x1fb0 [<ffffffff8bdaa750>] cifs_readdir+0x15a0/0x1d50 [<ffffffff8b9a853f>] iterate_dir+0x28f/0x4b0 [<ffffffff8b9a9aed>] __x64_sys_getdents64+0xfd/0x200 [<ffffffff8cf6da05>] do_syscall_64+0x95/0x1a0 [<ffffffff8d00012f>] entry_SYSCALL_64_after_hwframe+0x76/0x7e And addresses these BUG splats when unmounting the SMB filesystem: BUG: Dentry ffff888140590ba0{i=1000000000080,n=/} still in use (2) [unmount of cifs cifs] WARNING: CPU: 3 PID: 3433 at fs/dcache.c:1536 umount_check+0xd0/0x100 Modules linked in: CPU: 3 UID: 0 PID: 3433 Comm: bash Not tainted 6.12.0-rc4-g850925a8133c-dirty kernel-patches#49 Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 11/12/2020 RIP: 0010:umount_check+0xd0/0x100 Code: 8d 7c 24 40 e8 31 5a f4 ff 49 8b 54 24 40 41 56 49 89 e9 45 89 e8 48 89 d9 41 57 48 89 de 48 c7 c7 80 e7 db ac e8 f0 72 9a ff <0f> 0b 58 31 c0 5a 5b 5d 41 5c 41 5d 41 5e 41 5f e9 2b e5 5d 01 41 RSP: 0018:ffff88811cc27978 EFLAGS: 00010286 RAX: 0000000000000000 RBX: ffff888140590ba0 RCX: ffffffffaaf20bae RDX: dffffc0000000000 RSI: 0000000000000008 RDI: ffff8881f6fb6f40 RBP: ffff8881462ec000 R08: 0000000000000001 R09: ffffed1023984ee3 R10: ffff88811cc2771f R11: 00000000016cfcc0 R12: ffff888134383e08 R13: 0000000000000002 R14: ffff8881462ec668 R15: ffffffffaceab4c0 FS: 00007f23bfa98740(0000) GS:ffff8881f6f80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000556de4a6f808 CR3: 0000000123c80000 CR4: 0000000000350ef0 Call Trace: <TASK> d_walk+0x6a/0x530 shrink_dcache_for_umount+0x6a/0x200 generic_shutdown_super+0x52/0x2a0 kill_anon_super+0x22/0x40 cifs_kill_sb+0x159/0x1e0 deactivate_locked_super+0x66/0xe0 cleanup_mnt+0x140/0x210 task_work_run+0xfb/0x170 syscall_exit_to_user_mode+0x29f/0x2b0 do_syscall_64+0xa1/0x1a0 entry_SYSCALL_64_after_hwframe+0x76/0x7e RIP: 0033:0x7f23bfb93ae7 Code: ff ff ff ff c3 66 0f 1f 44 00 00 48 8b 0d 11 93 0d 00 f7 d8 64 89 01 b8 ff ff ff ff eb bf 0f 1f 44 00 00 b8 50 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d e9 92 0d 00 f7 d8 64 89 01 48 RSP: 002b:00007ffee9138598 EFLAGS: 00000246 ORIG_RAX: 0000000000000050 RAX: 0000000000000000 RBX: 0000558f1803e9a0 RCX: 00007f23bfb93ae7 RDX: 0000000000000000 RSI: 0000000000000004 RDI: 0000558f1803e9a0 RBP: 0000558f1803e600 R08: 0000000000000007 R09: 0000558f17fab610 R10: d91d5ec34ab757b0 R11: 0000000000000246 R12: 0000000000000001 R13: 0000000000000000 R14: 0000000000000015 R15: 0000000000000000 </TASK> irq event stamp: 1163486 hardirqs last enabled at (1163485): [<ffffffffac98d344>] _raw_spin_unlock_irqrestore+0x34/0x60 hardirqs last disabled at (1163486): [<ffffffffac97dcfc>] __schedule+0xc7c/0x19a0 softirqs last enabled at (1163482): [<ffffffffab79a3ee>] __smb_send_rqst+0x3de/0x990 softirqs last disabled at (1163480): [<ffffffffac2314f1>] release_sock+0x21/0xf0 ---[ end trace 0000000000000000 ]--- VFS: Busy inodes after unmount of cifs (cifs) ------------[ cut here ]------------ kernel BUG at fs/super.c:661! Oops: invalid opcode: 0000 [kernel-patches#1] PREEMPT SMP KASAN NOPTI CPU: 1 UID: 0 PID: 3433 Comm: bash Tainted: G W 6.12.0-rc4-g850925a8133c-dirty kernel-patches#49 Tainted: [W]=WARN Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 11/12/2020 RIP: 0010:generic_shutdown_super+0x290/0x2a0 Code: e8 15 7c f7 ff 48 8b 5d 28 48 89 df e8 09 7c f7 ff 48 8b 0b 48 89 ee 48 8d 95 68 06 00 00 48 c7 c7 80 7f db ac e8 00 69 af ff <0f> 0b 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 90 90 90 90 90 90 RSP: 0018:ffff88811cc27a50 EFLAGS: 00010246 RAX: 000000000000003e RBX: ffffffffae994420 RCX: 0000000000000027 RDX: 0000000000000000 RSI: ffffffffab06180e RDI: ffff8881f6eb18c8 RBP: ffff8881462ec000 R08: 0000000000000001 R09: ffffed103edd6319 R10: ffff8881f6eb18cb R11: 00000000016d3158 R12: ffff8881462ec9c0 R13: ffff8881462ec050 R14: 0000000000000001 R15: 0000000000000000 FS: 00007f23bfa98740(0000) GS:ffff8881f6e80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f8364005d68 CR3: 0000000123c80000 CR4: 0000000000350ef0 Call Trace: <TASK> kill_anon_super+0x22/0x40 cifs_kill_sb+0x159/0x1e0 deactivate_locked_super+0x66/0xe0 cleanup_mnt+0x140/0x210 task_work_run+0xfb/0x170 syscall_exit_to_user_mode+0x29f/0x2b0 do_syscall_64+0xa1/0x1a0 entry_SYSCALL_64_after_hwframe+0x76/0x7e RIP: 0033:0x7f23bfb93ae7 </TASK> Modules linked in: ---[ end trace 0000000000000000 ]--- RIP: 0010:generic_shutdown_super+0x290/0x2a0 Code: e8 15 7c f7 ff 48 8b 5d 28 48 89 df e8 09 7c f7 ff 48 8b 0b 48 89 ee 48 8d 95 68 06 00 00 48 c7 c7 80 7f db ac e8 00 69 af ff <0f> 0b 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 90 90 90 90 90 90 RSP: 0018:ffff88811cc27a50 EFLAGS: 00010246 RAX: 000000000000003e RBX: ffffffffae994420 RCX: 0000000000000027 RDX: 0000000000000000 RSI: ffffffffab06180e RDI: ffff8881f6eb18c8 RBP: ffff8881462ec000 R08: 0000000000000001 R09: ffffed103edd6319 R10: ffff8881f6eb18cb R11: 00000000016d3158 R12: ffff8881462ec9c0 R13: ffff8881462ec050 R14: 0000000000000001 R15: 0000000000000000 FS: 00007f23bfa98740(0000) GS:ffff8881f6e80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f8364005d68 CR3: 0000000123c80000 CR4: 0000000000350ef0 This reproduces eventually with an SMB mount and two shells running these loops concurrently - while true; do cd ~; sleep 1; for i in {1..3}; do cd /mnt/test/subdir; echo $PWD; sleep 1; cd ..; echo $PWD; sleep 1; done; echo ...; done - while true; do iptables -F OUTPUT; mount -t cifs -a; for _ in {0..2}; do ls /mnt/test/subdir/ | wc -l; done; iptables -I OUTPUT -p tcp --dport 445 -j DROP; sleep 10 echo "unmounting"; umount -l -t cifs -a; echo "done unmounting"; sleep 20 echo "recovering"; iptables -F OUTPUT; sleep 10; done Fixes: ebe98f1 ("cifs: enable caching of directories for which a lease is held") Fixes: 5c86919 ("smb: client: fix use-after-free in smb2_query_info_compound()") Cc: [email protected] Signed-off-by: Paul Aurich <[email protected]> Signed-off-by: Steve French <[email protected]> Signed-off-by: Greg Kroah-Hartman <[email protected]>
shunghsiyu
pushed a commit
to shunghsiyu/bpf
that referenced
this pull request
Dec 18, 2024
[ Upstream commit b9e9ed9 ] For htab of maps, when the map is removed from the htab, it may hold the last reference of the map. bpf_map_fd_put_ptr() will invoke bpf_map_free_id() to free the id of the removed map element. However, bpf_map_fd_put_ptr() is invoked while holding a bucket lock (raw_spin_lock_t), and bpf_map_free_id() attempts to acquire map_idr_lock (spinlock_t), triggering the following lockdep warning: ============================= [ BUG: Invalid wait context ] 6.11.0-rc4+ kernel-patches#49 Not tainted ----------------------------- test_maps/4881 is trying to lock: ffffffff84884578 (map_idr_lock){+...}-{3:3}, at: bpf_map_free_id.part.0+0x21/0x70 other info that might help us debug this: context-{5:5} 2 locks held by test_maps/4881: #0: ffffffff846caf60 (rcu_read_lock){....}-{1:3}, at: bpf_fd_htab_map_update_elem+0xf9/0x270 kernel-patches#1: ffff888149ced148 (&htab->lockdep_key#2){....}-{2:2}, at: htab_map_update_elem+0x178/0xa80 stack backtrace: CPU: 0 UID: 0 PID: 4881 Comm: test_maps Not tainted 6.11.0-rc4+ kernel-patches#49 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), ... Call Trace: <TASK> dump_stack_lvl+0x6e/0xb0 dump_stack+0x10/0x20 __lock_acquire+0x73e/0x36c0 lock_acquire+0x182/0x450 _raw_spin_lock_irqsave+0x43/0x70 bpf_map_free_id.part.0+0x21/0x70 bpf_map_put+0xcf/0x110 bpf_map_fd_put_ptr+0x9a/0xb0 free_htab_elem+0x69/0xe0 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 htab_map_update_elem+0x50f/0xa80 bpf_fd_htab_map_update_elem+0x131/0x270 bpf_map_update_value+0x266/0x380 __sys_bpf+0x21bb/0x36b0 __x64_sys_bpf+0x45/0x60 x64_sys_call+0x1b2a/0x20d0 do_syscall_64+0x5d/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e One way to fix the lockdep warning is using raw_spinlock_t for map_idr_lock as well. However, bpf_map_alloc_id() invokes idr_alloc_cyclic() after acquiring map_idr_lock, it will trigger a similar lockdep warning because the slab's lock (s->cpu_slab->lock) is still a spinlock. Instead of changing map_idr_lock's type, fix the issue by invoking htab_put_fd_value() after htab_unlock_bucket(). However, only deferring the invocation of htab_put_fd_value() is not enough, because the old map pointers in htab of maps can not be saved during batched deletion. Therefore, also defer the invocation of free_htab_elem(), so these to-be-freed elements could be linked together similar to lru map. There are four callers for ->map_fd_put_ptr: (1) alloc_htab_elem() (through htab_put_fd_value()) It invokes ->map_fd_put_ptr() under a raw_spinlock_t. The invocation of htab_put_fd_value() can not simply move after htab_unlock_bucket(), because the old element has already been stashed in htab->extra_elems. It may be reused immediately after htab_unlock_bucket() and the invocation of htab_put_fd_value() after htab_unlock_bucket() may release the newly-added element incorrectly. Therefore, saving the map pointer of the old element for htab of maps before unlocking the bucket and releasing the map_ptr after unlock. Beside the map pointer in the old element, should do the same thing for the special fields in the old element as well. (2) free_htab_elem() (through htab_put_fd_value()) Its caller includes __htab_map_lookup_and_delete_elem(), htab_map_delete_elem() and __htab_map_lookup_and_delete_batch(). For htab_map_delete_elem(), simply invoke free_htab_elem() after htab_unlock_bucket(). For __htab_map_lookup_and_delete_batch(), just like lru map, linking the to-be-freed element into node_to_free list and invoking free_htab_elem() for these element after unlock. It is safe to reuse batch_flink as the link for node_to_free, because these elements have been removed from the hash llist. Because htab of maps doesn't support lookup_and_delete operation, __htab_map_lookup_and_delete_elem() doesn't have the problem, so kept it as is. (3) fd_htab_map_free() It invokes ->map_fd_put_ptr without raw_spinlock_t. (4) bpf_fd_htab_map_update_elem() It invokes ->map_fd_put_ptr without raw_spinlock_t. After moving free_htab_elem() outside htab bucket lock scope, using pcpu_freelist_push() instead of __pcpu_freelist_push() to disable the irq before freeing elements, and protecting the invocations of bpf_mem_cache_free() with migrate_{disable|enable} pair. Signed-off-by: Hou Tao <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Alexei Starovoitov <[email protected]> Signed-off-by: Andrii Nakryiko <[email protected]> Signed-off-by: Sasha Levin <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Pull request for series with
subject: bpf: expose is_mptcp flag to bpf_tcp_sock
version: 2
url: https://patchwork.ozlabs.org/project/netdev/list/?series=201180