| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
When comparing current and cached states verifier should consider
bpf_func_state->callback_depth. Current state cannot be pruned against
cached state, when current states has more iterations left compared to
cached state. Current state has more iterations left when it's
callback_depth is smaller.
Below is an example illustrating this bug, minimized from mailing list
discussion [0] (assume that BPF_F_TEST_STATE_FREQ is set).
The example is not a safe program: if loop_cb point (1) is followed by
loop_cb point (2), then division by zero is possible at point (4).
struct ctx {
__u64 a;
__u64 b;
__u64 c;
};
static void loop_cb(int i, struct ctx *ctx)
{
/* assume that generated code is "fallthrough-first":
* if ... == 1 goto
* if ... == 2 goto
* <default>
*/
switch (bpf_get_prandom_u32()) {
case 1: /* 1 */ ctx->a = 42; return 0; break;
case 2: /* 2 */ ctx->b = 42; return 0; break;
default: /* 3 */ ctx->c = 42; return 0; break;
}
}
SEC("tc")
__failure
__flag(BPF_F_TEST_STATE_FREQ)
int test(struct __sk_buff *skb)
{
struct ctx ctx = { 7, 7, 7 };
bpf_loop(2, loop_cb, &ctx, 0); /* 0 */
/* assume generated checks are in-order: .a first */
if (ctx.a == 42 && ctx.b == 42 && ctx.c == 7)
asm volatile("r0 /= 0;":::"r0"); /* 4 */
return 0;
}
Prior to this commit verifier built the following checkpoint tree for
this example:
.------------------------------------- Checkpoint / State name
| .-------------------------------- Code point number
| | .---------------------------- Stack state {ctx.a,ctx.b,ctx.c}
| | | .------------------- Callback depth in frame #0
v v v v
- (0) {7P,7P,7},depth=0
- (3) {7P,7P,7},depth=1
- (0) {7P,7P,42},depth=1
- (3) {7P,7,42},depth=2
- (0) {7P,7,42},depth=2 loop terminates because of depth limit
- (4) {7P,7,42},depth=0 predicted false, ctx.a marked precise
- (6) exit
(a) - (2) {7P,7,42},depth=2
- (0) {7P,42,42},depth=2 loop terminates because of depth limit
- (4) {7P,42,42},depth=0 predicted false, ctx.a marked precise
- (6) exit
(b) - (1) {7P,7P,42},depth=2
- (0) {42P,7P,42},depth=2 loop terminates because of depth limit
- (4) {42P,7P,42},depth=0 predicted false, ctx.{a,b} marked precise
- (6) exit
- (2) {7P,7,7},depth=1 considered safe, pruned using checkpoint (a)
(c) - (1) {7P,7P,7},depth=1 considered safe, pruned using checkpoint (b)
Here checkpoint (b) has callback_depth of 2, meaning that it would
never reach state {42,42,7}.
While checkpoint (c) has callback_depth of 1, and thus
could yet explore the state {42,42,7} if not pruned prematurely.
This commit makes forbids such premature pruning,
allowing verifier to explore states sub-tree starting at (c):
(c) - (1) {7,7,7P},depth=1
- (0) {42P,7,7P},depth=1
...
- (2) {42,7,7},depth=2
- (0) {42,42,7},depth=2 loop terminates because of depth limit
- (4) {42,42,7},depth=0 predicted true, ctx.{a,b,c} marked precise
- (5) division by zero
[0] https://lore.kernel.org/bpf/9b251840-7cb8-4d17-bd23-1fc8071d8eef@linux.dev/
Fixes: bb124da69c47 ("bpf: keep track of max number of bpf_loop callback iterations")
Suggested-by: Yonghong Song <yonghong.song@linux.dev>
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Acked-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20240222154121.6991-2-eddyz87@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
The BPF struct_ops previously only allowed one page of trampolines.
Each function pointer of a struct_ops is implemented by a struct_ops
bpf program. Each struct_ops bpf program requires a trampoline.
The following selftest patch shows each page can hold a little more
than 20 trampolines.
While one page is more than enough for the tcp-cc usecase,
the sched_ext use case shows that one page is not always enough and hits
the one page limit. This patch overcomes the one page limit by allocating
another page when needed and it is limited to a total of
MAX_IMAGE_PAGES (8) pages which is more than enough for
reasonable usages.
The variable st_map->image has been changed to st_map->image_pages, and
its type has been changed to an array of pointers to pages.
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240224223418.526631-3-thinker.li@gmail.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
|
|
Perform all validations when updating values of struct_ops maps. Doing
validation in st_ops->reg() and st_ops->update() is not necessary anymore.
However, tcp_register_congestion_control() has been called in various
places. It still needs to do validations.
Cc: netdev@vger.kernel.org
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240224223418.526631-2-thinker.li@gmail.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
|
|
This is a cleanup patch, making code a bit more concise.
1) Use skb_network_offset(skb) in place of
(skb_network_header(skb) - skb->data)
2) Use -skb_network_offset(skb) in place of
(skb->data - skb_network_header(skb))
3) Use skb_transport_offset(skb) in place of
(skb_transport_header(skb) - skb->data)
4) Use skb_inner_transport_offset(skb) in place of
(skb_inner_transport_header(skb) - skb->data)
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Edward Cree <ecree.xilinx@gmail.com> # for sfc
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
Daniel Borkmann says:
====================
pull-request: bpf-next 2024-02-29
We've added 119 non-merge commits during the last 32 day(s) which contain
a total of 150 files changed, 3589 insertions(+), 995 deletions(-).
The main changes are:
1) Extend the BPF verifier to enable static subprog calls in spin lock
critical sections, from Kumar Kartikeya Dwivedi.
2) Fix confusing and incorrect inference of PTR_TO_CTX argument type
in BPF global subprogs, from Andrii Nakryiko.
3) Larger batch of riscv BPF JIT improvements and enabling inlining
of the bpf_kptr_xchg() for RV64, from Pu Lehui.
4) Allow skeleton users to change the values of the fields in struct_ops
maps at runtime, from Kui-Feng Lee.
5) Extend the verifier's capabilities of tracking scalars when they
are spilled to stack, especially when the spill or fill is narrowing,
from Maxim Mikityanskiy & Eduard Zingerman.
6) Various BPF selftest improvements to fix errors under gcc BPF backend,
from Jose E. Marchesi.
7) Avoid module loading failure when the module trying to register
a struct_ops has its BTF section stripped, from Geliang Tang.
8) Annotate all kfuncs in .BTF_ids section which eventually allows
for automatic kfunc prototype generation from bpftool, from Daniel Xu.
9) Several updates to the instruction-set.rst IETF standardization
document, from Dave Thaler.
10) Shrink the size of struct bpf_map resp. bpf_array,
from Alexei Starovoitov.
11) Initial small subset of BPF verifier prepwork for sleepable bpf_timer,
from Benjamin Tissoires.
12) Fix bpftool to be more portable to musl libc by using POSIX's
basename(), from Arnaldo Carvalho de Melo.
13) Add libbpf support to gcc in CORE macro definitions,
from Cupertino Miranda.
14) Remove a duplicate type check in perf_event_bpf_event,
from Florian Lehner.
15) Fix bpf_spin_{un,}lock BPF helpers to actually annotate them
with notrace correctly, from Yonghong Song.
16) Replace the deprecated bpf_lpm_trie_key 0-length array with flexible
array to fix build warnings, from Kees Cook.
17) Fix resolve_btfids cross-compilation to non host-native endianness,
from Viktor Malik.
* tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: (119 commits)
selftests/bpf: Test if shadow types work correctly.
bpftool: Add an example for struct_ops map and shadow type.
bpftool: Generated shadow variables for struct_ops maps.
libbpf: Convert st_ops->data to shadow type.
libbpf: Set btf_value_type_id of struct bpf_map for struct_ops.
bpf: Replace bpf_lpm_trie_key 0-length array with flexible array
bpf, arm64: use bpf_prog_pack for memory management
arm64: patching: implement text_poke API
bpf, arm64: support exceptions
arm64: stacktrace: Implement arch_bpf_stack_walk() for the BPF JIT
bpf: add is_async_callback_calling_insn() helper
bpf: introduce in_sleepable() helper
bpf: allow more maps in sleepable bpf programs
selftests/bpf: Test case for lacking CFI stub functions.
bpf: Check cfi_stubs before registering a struct_ops type.
bpf: Clarify batch lookup/lookup_and_delete semantics
bpf, docs: specify which BPF_ABS and BPF_IND fields were zero
bpf, docs: Fix typos in instruction-set.rst
selftests/bpf: update tcp_custom_syncookie to use scalar packet offset
bpf: Shrink size of struct bpf_map/bpf_array.
...
====================
Link: https://lore.kernel.org/r/20240301001625.8800-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Replace deprecated 0-length array in struct bpf_lpm_trie_key with
flexible array. Found with GCC 13:
../kernel/bpf/lpm_trie.c:207:51: warning: array subscript i is outside array bounds of 'const __u8[0]' {aka 'const unsigned char[]'} [-Warray-bounds=]
207 | *(__be16 *)&key->data[i]);
| ^~~~~~~~~~~~~
../include/uapi/linux/swab.h:102:54: note: in definition of macro '__swab16'
102 | #define __swab16(x) (__u16)__builtin_bswap16((__u16)(x))
| ^
../include/linux/byteorder/generic.h:97:21: note: in expansion of macro '__be16_to_cpu'
97 | #define be16_to_cpu __be16_to_cpu
| ^~~~~~~~~~~~~
../kernel/bpf/lpm_trie.c:206:28: note: in expansion of macro 'be16_to_cpu'
206 | u16 diff = be16_to_cpu(*(__be16 *)&node->data[i]
^
| ^~~~~~~~~~~
In file included from ../include/linux/bpf.h:7:
../include/uapi/linux/bpf.h:82:17: note: while referencing 'data'
82 | __u8 data[0]; /* Arbitrary size */
| ^~~~
And found at run-time under CONFIG_FORTIFY_SOURCE:
UBSAN: array-index-out-of-bounds in kernel/bpf/lpm_trie.c:218:49
index 0 is out of range for type '__u8 [*]'
Changing struct bpf_lpm_trie_key is difficult since has been used by
userspace. For example, in Cilium:
struct egress_gw_policy_key {
struct bpf_lpm_trie_key lpm_key;
__u32 saddr;
__u32 daddr;
};
While direct references to the "data" member haven't been found, there
are static initializers what include the final member. For example,
the "{}" here:
struct egress_gw_policy_key in_key = {
.lpm_key = { 32 + 24, {} },
.saddr = CLIENT_IP,
.daddr = EXTERNAL_SVC_IP & 0Xffffff,
};
To avoid the build time and run time warnings seen with a 0-sized
trailing array for struct bpf_lpm_trie_key, introduce a new struct
that correctly uses a flexible array for the trailing bytes,
struct bpf_lpm_trie_key_u8. As part of this, include the "header"
portion (which is just the "prefixlen" member), so it can be used
by anything building a bpf_lpr_trie_key that has trailing members that
aren't a u8 flexible array (like the self-test[1]), which is named
struct bpf_lpm_trie_key_hdr.
Unfortunately, C++ refuses to parse the __struct_group() helper, so
it is not possible to define struct bpf_lpm_trie_key_hdr directly in
struct bpf_lpm_trie_key_u8, so we must open-code the union directly.
Adjust the kernel code to use struct bpf_lpm_trie_key_u8 through-out,
and for the selftest to use struct bpf_lpm_trie_key_hdr. Add a comment
to the UAPI header directing folks to the two new options.
Reported-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Closes: https://paste.debian.net/hidden/ca500597/
Link: https://lore.kernel.org/all/202206281009.4332AA33@keescook/ [1]
Link: https://lore.kernel.org/bpf/20240222155612.it.533-kees@kernel.org
|
|
Currently we have a special case for BPF_FUNC_timer_set_callback,
let's introduce a helper we can extend for the kfunc that will come in
a later patch
Signed-off-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240221-hid-bpf-sleepable-v3-3-1fb378ca6301@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
No code change, but it'll allow to have only one place to change
everything when we add in_sleepable in cur_state.
Signed-off-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240221-hid-bpf-sleepable-v3-2-1fb378ca6301@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
These 2 maps types are required for HID-BPF when a user wants to do
IO with a device from a sleepable tracing point.
Allowing BPF_MAP_TYPE_QUEUE (and therefore BPF_MAP_TYPE_STACK) allows
for a BPF program to prepare from an IRQ the list of HID commands to send
back to the device and then these commands can be retrieved from the
sleepable trace point.
Signed-off-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240221-hid-bpf-sleepable-v3-1-1fb378ca6301@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
Cross-merge networking fixes after downstream PR.
Conflicts:
net/ipv4/udp.c
f796feabb9f5 ("udp: add local "peek offset enabled" flag")
56667da7399e ("net: implement lockless setsockopt(SO_PEEK_OFF)")
Adjacent changes:
net/unix/garbage.c
aa82ac51d633 ("af_unix: Drop oob_skb ref before purging queue in GC.")
11498715f266 ("af_unix: Remove io_uring code for GC.")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Recently, st_ops->cfi_stubs was introduced. However, the upcoming new
struct_ops support (e.g. sched_ext) is not aware of this and does not
provide its own cfi_stubs. The kernel ends up NULL dereferencing the
st_ops->cfi_stubs.
Considering struct_ops supports kernel module now, this NULL check
is necessary. This patch is to reject struct_ops registration
that does not provide a cfi_stubs.
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240222021105.1180475-2-thinker.li@gmail.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
|
|
Failure to initialize it->pos, coupled with the presence of an invalid
value in the flags variable, can lead to it->pos referencing an invalid
task, potentially resulting in a kernel panic. To mitigate this risk, it's
crucial to ensure proper initialization of it->pos to NULL.
Fixes: ac8148d957f5 ("bpf: bpf_iter_task_next: use next_task(kit->task) rather than next_task(kit->pos)")
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yonghong.song@linux.dev>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Link: https://lore.kernel.org/bpf/20240217114152.1623-2-laoar.shao@gmail.com
|
|
The following race is possible between bpf_timer_cancel_and_free
and bpf_timer_cancel. It will lead a UAF on the timer->timer.
bpf_timer_cancel();
spin_lock();
t = timer->time;
spin_unlock();
bpf_timer_cancel_and_free();
spin_lock();
t = timer->timer;
timer->timer = NULL;
spin_unlock();
hrtimer_cancel(&t->timer);
kfree(t);
/* UAF on t */
hrtimer_cancel(&t->timer);
In bpf_timer_cancel_and_free, this patch frees the timer->timer
after a rcu grace period. This requires a rcu_head addition
to the "struct bpf_hrtimer". Another kfree(t) happens in bpf_timer_init,
this does not need a kfree_rcu because it is still under the
spin_lock and timer->timer has not been visible by others yet.
In bpf_timer_cancel, rcu_read_lock() is added because this helper
can be used in a non rcu critical section context (e.g. from
a sleepable bpf prog). Other timer->timer usages in helpers.c
have been audited, bpf_timer_cancel() is the only place where
timer->timer is used outside of the spin_lock.
Another solution considered is to mark a t->flag in bpf_timer_cancel
and clear it after hrtimer_cancel() is done. In bpf_timer_cancel_and_free,
it busy waits for the flag to be cleared before kfree(t). This patch
goes with a straight forward solution and frees timer->timer after
a rcu grace period.
Fixes: b00628b1c7d5 ("bpf: Introduce bpf timers.")
Suggested-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/bpf/20240215211218.990808-1-martin.lau@linux.dev
|
|
With latest llvm19, I hit the following selftest failures with
$ ./test_progs -j
libbpf: prog 'on_event': BPF program load failed: Permission denied
libbpf: prog 'on_event': -- BEGIN PROG LOAD LOG --
combined stack size of 4 calls is 544. Too large
verification time 1344153 usec
stack depth 24+440+0+32
processed 51008 insns (limit 1000000) max_states_per_insn 19 total_states 1467 peak_states 303 mark_read 146
-- END PROG LOAD LOG --
libbpf: prog 'on_event': failed to load: -13
libbpf: failed to load object 'strobemeta_subprogs.bpf.o'
scale_test:FAIL:expect_success unexpected error: -13 (errno 13)
#498 verif_scale_strobemeta_subprogs:FAIL
The verifier complains too big of the combined stack size (544 bytes) which
exceeds the maximum stack limit 512. This is a regression from llvm19 ([1]).
In the above error log, the original stack depth is 24+440+0+32.
To satisfy interpreter's need, in verifier the stack depth is adjusted to
32+448+32+32=544 which exceeds 512, hence the error. The same adjusted
stack size is also used for jit case.
But the jitted codes could use smaller stack size.
$ egrep -r stack_depth | grep round_up
arm64/net/bpf_jit_comp.c: ctx->stack_size = round_up(prog->aux->stack_depth, 16);
loongarch/net/bpf_jit.c: bpf_stack_adjust = round_up(ctx->prog->aux->stack_depth, 16);
powerpc/net/bpf_jit_comp.c: cgctx.stack_size = round_up(fp->aux->stack_depth, 16);
riscv/net/bpf_jit_comp32.c: round_up(ctx->prog->aux->stack_depth, STACK_ALIGN);
riscv/net/bpf_jit_comp64.c: bpf_stack_adjust = round_up(ctx->prog->aux->stack_depth, 16);
s390/net/bpf_jit_comp.c: u32 stack_depth = round_up(fp->aux->stack_depth, 8);
sparc/net/bpf_jit_comp_64.c: stack_needed += round_up(stack_depth, 16);
x86/net/bpf_jit_comp.c: EMIT3_off32(0x48, 0x81, 0xEC, round_up(stack_depth, 8));
x86/net/bpf_jit_comp.c: int tcc_off = -4 - round_up(stack_depth, 8);
x86/net/bpf_jit_comp.c: round_up(stack_depth, 8));
x86/net/bpf_jit_comp.c: int tcc_off = -4 - round_up(stack_depth, 8);
x86/net/bpf_jit_comp.c: EMIT3_off32(0x48, 0x81, 0xC4, round_up(stack_depth, 8));
In the above, STACK_ALIGN in riscv/net/bpf_jit_comp32.c is defined as 16.
So stack is aligned in either 8 or 16, x86/s390 having 8-byte stack alignment and
the rest having 16-byte alignment.
This patch calculates total stack depth based on 16-byte alignment if jit is requested.
For the above failing case, the new stack size will be 32+448+0+32=512 and no verification
failure. llvm19 regression will be discussed separately in llvm upstream.
The verifier change caused three test failures as these tests compared messages
with stack size. More specifically,
- test_global_funcs/global_func1: fail with interpreter mode and success with jit mode.
Adjusted stack sizes so both jit and interpreter modes will fail.
- async_stack_depth/{pseudo_call_check, async_call_root_check}: since jit and interpreter
will calculate different stack sizes, the failure msg is adjusted to omit those
specific stack size numbers.
[1] https://lore.kernel.org/bpf/32bde0f0-1881-46c9-931a-673be566c61d@linux.dev/
Suggested-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20240214232951.4113094-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
Verifier log avoids printing the same source code line multiple times
when a consecutive block of BPF assembly instructions are covered by the
same original (C) source code line. This greatly improves verifier log
legibility.
Unfortunately, this check is imperfect and in production applications it
quite often happens that verifier log will have multiple duplicated
source lines emitted, for no apparently good reason. E.g., this is
excerpt from a real-world BPF application (with register states omitted
for clarity):
BEFORE
======
; for (int i = 0; i < STROBE_MAX_MAP_ENTRIES; ++i) { @ strobemeta_probe.bpf.c:394
5369: (07) r8 += 2 ;
5370: (07) r7 += 16 ;
; for (int i = 0; i < STROBE_MAX_MAP_ENTRIES; ++i) { @ strobemeta_probe.bpf.c:394
5371: (07) r9 += 1 ;
5372: (79) r4 = *(u64 *)(r10 -32) ;
; for (int i = 0; i < STROBE_MAX_MAP_ENTRIES; ++i) { @ strobemeta_probe.bpf.c:394
5373: (55) if r9 != 0xf goto pc+2
; if (i >= map->cnt) @ strobemeta_probe.bpf.c:396
5376: (79) r1 = *(u64 *)(r10 -40) ;
5377: (79) r1 = *(u64 *)(r1 +8) ;
; if (i >= map->cnt) @ strobemeta_probe.bpf.c:396
5378: (dd) if r1 s<= r9 goto pc-5 ;
; descr->key_lens[i] = 0; @ strobemeta_probe.bpf.c:398
5379: (b4) w1 = 0 ;
5380: (6b) *(u16 *)(r8 -30) = r1 ;
; task, data, off, STROBE_MAX_STR_LEN, map->entries[i].key); @ strobemeta_probe.bpf.c:400
5381: (79) r3 = *(u64 *)(r7 -8) ;
5382: (7b) *(u64 *)(r10 -24) = r6 ;
; task, data, off, STROBE_MAX_STR_LEN, map->entries[i].key); @ strobemeta_probe.bpf.c:400
5383: (bc) w6 = w6 ;
; barrier_var(payload_off); @ strobemeta_probe.bpf.c:280
5384: (bf) r2 = r6 ;
5385: (bf) r1 = r4 ;
As can be seen, line 394 is emitted thrice, 396 is emitted twice, and
line 400 is duplicated as well. Note that there are no intermingling
other lines of source code in between these duplicates, so the issue is
not compiler reordering assembly instruction such that multiple original
source code lines are in effect.
It becomes more obvious what's going on if we look at *full* original line info
information (using btfdump for this, [0]):
#2764: line: insn #5363 --> 394:3 @ ./././strobemeta_probe.bpf.c
for (int i = 0; i < STROBE_MAX_MAP_ENTRIES; ++i) {
#2765: line: insn #5373 --> 394:21 @ ./././strobemeta_probe.bpf.c
for (int i = 0; i < STROBE_MAX_MAP_ENTRIES; ++i) {
#2766: line: insn #5375 --> 394:47 @ ./././strobemeta_probe.bpf.c
for (int i = 0; i < STROBE_MAX_MAP_ENTRIES; ++i) {
#2767: line: insn #5377 --> 394:3 @ ./././strobemeta_probe.bpf.c
for (int i = 0; i < STROBE_MAX_MAP_ENTRIES; ++i) {
#2768: line: insn #5378 --> 414:10 @ ./././strobemeta_probe.bpf.c
return off;
We can see that there are four line info records covering
instructions #5363 through #5377 (instruction indices are shifted due to
subprog instruction being appended to main program), all of them are
pointing to the same C source code line #394. But each of them points to
a different part of that line, which is denoted by differing column
numbers (3, 21, 47, 3).
But verifier log doesn't distinguish between parts of the same source code line
and doesn't emit this column number information, so for end user it's just a
repetitive visual noise. So let's improve the detection of repeated source code
line and avoid this.
With the changes in this patch, we get this output for the same piece of BPF
program log:
AFTER
=====
; for (int i = 0; i < STROBE_MAX_MAP_ENTRIES; ++i) { @ strobemeta_probe.bpf.c:394
5369: (07) r8 += 2 ;
5370: (07) r7 += 16 ;
5371: (07) r9 += 1 ;
5372: (79) r4 = *(u64 *)(r10 -32) ;
5373: (55) if r9 != 0xf goto pc+2
; if (i >= map->cnt) @ strobemeta_probe.bpf.c:396
5376: (79) r1 = *(u64 *)(r10 -40) ;
5377: (79) r1 = *(u64 *)(r1 +8) ;
5378: (dd) if r1 s<= r9 goto pc-5 ;
; descr->key_lens[i] = 0; @ strobemeta_probe.bpf.c:398
5379: (b4) w1 = 0 ;
5380: (6b) *(u16 *)(r8 -30) = r1 ;
; task, data, off, STROBE_MAX_STR_LEN, map->entries[i].key); @ strobemeta_probe.bpf.c:400
5381: (79) r3 = *(u64 *)(r7 -8) ;
5382: (7b) *(u64 *)(r10 -24) = r6 ;
5383: (bc) w6 = w6 ;
; barrier_var(payload_off); @ strobemeta_probe.bpf.c:280
5384: (bf) r2 = r6 ;
5385: (bf) r1 = r4 ;
All the duplication is gone and the log is cleaner and less distracting.
[0] https://github.com/anakryiko/btfdump
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240214174100.2847419-1-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
Real-world BPF applications keep growing in size. Medium-sized production
application can easily have 50K+ verified instructions, and its line
info section in .BTF.ext has more than 3K entries.
When verifier emits log with log_level>=1, it annotates assembly code
with matched original C source code. Currently it uses linear search
over line info records to find a match. As complexity of BPF
applications grows, this O(K * N) approach scales poorly.
So, let's instead of linear O(N) search for line info record use faster
equivalent O(log(N)) binary search algorithm. It's not a plain binary
search, as we don't look for exact match. It's an upper bound search
variant, looking for rightmost line info record that starts at or before
given insn_off.
Some unscientific measurements were done before and after this change.
They were done in VM and fluctuate a bit, but overall the speed up is
undeniable.
BASELINE
========
File Program Duration (us) Insns
-------------------------------- ---------------- ------------- ------
katran.bpf.o balancer_ingress 2497130 343552
pyperf600.bpf.linked3.o on_event 12389611 627288
strobelight_pyperf_libbpf.o on_py_event 387399 52445
-------------------------------- ---------------- ------------- ------
BINARY SEARCH
=============
File Program Duration (us) Insns
-------------------------------- ---------------- ------------- ------
katran.bpf.o balancer_ingress 2339312 343552
pyperf600.bpf.linked3.o on_event 5602203 627288
strobelight_pyperf_libbpf.o on_py_event 294761 52445
-------------------------------- ---------------- ------------- ------
While Katran's speed up is pretty modest (about 105ms, or 6%), for
production pyperf BPF program (on_py_event) it's much greater already,
going from 387ms down to 295ms (23% improvement).
Looking at BPF selftests's biggest pyperf example, we can see even more
dramatic improvement, shaving more than 50% of time, going from 12.3s
down to 5.6s.
Different amount of improvement is the function of overall amount of BPF
assembly instructions in .bpf.o files (which contributes to how much
line info records there will be and thus, on average, how much time linear
search will take), among other things:
$ llvm-objdump -d katran.bpf.o | wc -l
3863
$ llvm-objdump -d strobelight_pyperf_libbpf.o | wc -l
6997
$ llvm-objdump -d pyperf600.bpf.linked3.o | wc -l
87854
Granted, this only applies to debugging cases (e.g., using veristat, or
failing verification in production), but seems worth doing to improve
overall developer experience anyways.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/bpf/20240214002311.2197116-1-andrii@kernel.org
|
|
As BPF applications grow in size and complexity and are separated into
multiple .bpf.c files that are statically linked together, it becomes
harder and harder to match verifier's BPF assembly level output to
original C code. While often annotated C source code is unique enough to
be able to identify the file it belongs to, quite often this is actually
problematic as parts of source code can be quite generic.
Long story short, it is very useful to see source code file name and
line number information along with the original C code. Verifier already
knows this information, we just need to output it.
This patch extends verifier log with file name and line number
information, emitted next to original (presumably C) source code,
annotating BPF assembly output, like so:
; <original C code> @ <filename>.bpf.c:<line>
If file name has directory names in it, they are stripped away. This
should be fine in practice as file names tend to be pretty unique with
C code anyways, and keeping log size smaller is always good.
In practice this might look something like below, where some code is
coming from application files, while others are from libbpf's usdt.bpf.h
header file:
; if (STROBEMETA_READ( @ strobemeta_probe.bpf.c:534
5592: (79) r1 = *(u64 *)(r10 -56) ; R1_w=mem_or_null(id=1589,sz=7680) R10=fp0
5593: (7b) *(u64 *)(r10 -56) = r1 ; R1_w=mem_or_null(id=1589,sz=7680) R10=fp0
5594: (79) r3 = *(u64 *)(r10 -8) ; R3_w=scalar() R10=fp0 fp-8=mmmmmmmm
...
170: (71) r1 = *(u8 *)(r8 +15) ; frame1: R1_w=scalar(...) R8_w=map_value(map=__bpf_usdt_spec,ks=4,vs=208)
171: (67) r1 <<= 56 ; frame1: R1_w=scalar(...)
172: (c7) r1 s>>= 56 ; frame1: R1_w=scalar(smin=smin32=-128,smax=smax32=127)
; val <<= arg_spec->arg_bitshift; @ usdt.bpf.h:183
173: (67) r1 <<= 32 ; frame1: R1_w=scalar(...)
174: (77) r1 >>= 32 ; frame1: R1_w=scalar(smin=0,smax=umax=0xffffffff,var_off=(0x0; 0xffffffff))
175: (79) r2 = *(u64 *)(r10 -8) ; frame1: R2_w=scalar() R10=fp0 fp-8=mmmmmmmm
176: (6f) r2 <<= r1 ; frame1: R1_w=scalar(smin=0,smax=umax=0xffffffff,var_off=(0x0; 0xffffffff)) R2_w=scalar()
177: (7b) *(u64 *)(r10 -8) = r2 ; frame1: R2_w=scalar(id=61) R10=fp0 fp-8_w=scalar(id=61)
; if (arg_spec->arg_signed) @ usdt.bpf.h:184
178: (bf) r3 = r2 ; frame1: R2_w=scalar(id=61) R3_w=scalar(id=61)
179: (7f) r3 >>= r1 ; frame1: R1_w=scalar(smin=0,smax=umax=0xffffffff,var_off=(0x0; 0xffffffff)) R3_w=scalar()
; if (arg_spec->arg_signed) @ usdt.bpf.h:184
180: (71) r4 = *(u8 *)(r8 +14)
181: safe
log_fixup tests needed a minor adjustment as verifier log output
increased a bit and that test is quite sensitive to such changes.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240212235944.2816107-1-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
For program types that don't have named context type name (e.g., BPF
iterator programs or tracepoint programs), ctx_tname will be a non-NULL
empty string. For such programs it shouldn't be possible to have
PTR_TO_CTX argument for global subprogs based on type name alone.
arg:ctx tag is the only way to have PTR_TO_CTX passed into global
subprog for such program types.
Fix this loophole, which currently would assume PTR_TO_CTX whenever
user uses a pointer to anonymous struct as an argument to their global
subprogs. This happens in practice with the following (quite common, in
practice) approach:
typedef struct { /* anonymous */
int x;
} my_type_t;
int my_subprog(my_type_t *arg) { ... }
User's intent is to have PTR_TO_MEM argument for `arg`, but verifier
will complain about expecting PTR_TO_CTX.
This fix also closes unintended s390x-specific KPROBE handling of
PTR_TO_CTX case. Selftest change is necessary to accommodate this.
Fixes: 91cc1a99740e ("bpf: Annotate context types")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240212233221.2575350-4-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
Expected canonical argument type for global function arguments
representing PTR_TO_CTX is `bpf_user_pt_regs_t *ctx`. This currently
works on s390x by accident because kernel resolves such typedef to
underlying struct (which is anonymous on s390x), and erroneously
accepting it as expected context type. We are fixing this problem next,
which would break s390x arch, so we need to handle `bpf_user_pt_regs_t`
case explicitly for KPROBE programs.
Fixes: 91cc1a99740e ("bpf: Annotate context types")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240212233221.2575350-3-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
Return result of btf_get_prog_ctx_type() is never used and callers only
check NULL vs non-NULL case to determine if given type matches expected
PTR_TO_CTX type. So rename function to `btf_is_prog_ctx_type()` and
return a simple true/false. We'll use this simpler interface to handle
kprobe program type's special typedef case in the next patch.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240212233221.2575350-2-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
Originally, this patch removed a redundant check in
BPF_CGROUP_RUN_PROG_INET_EGRESS, as the check was already being done in
the function it called, __cgroup_bpf_run_filter_skb. For v2, it was
reccomended that I remove the check from __cgroup_bpf_run_filter_skb,
and add the checks to the other macro that calls that function,
BPF_CGROUP_RUN_PROG_INET_INGRESS.
To sum it up, checking that the socket exists and that it is a full
socket is now part of both macros BPF_CGROUP_RUN_PROG_INET_EGRESS and
BPF_CGROUP_RUN_PROG_INET_INGRESS, and it is no longer part of the
function they call, __cgroup_bpf_run_filter_skb.
v3->v4: Fixed weird merge conflict.
v2->v3: Sent to bpf-next instead of generic patch
v1->v2: Addressed feedback about where check should be removed.
Signed-off-by: Oliver Crumrine <ozlinuxc@gmail.com>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/r/7lv62yiyvmj5a7eozv2iznglpkydkdfancgmbhiptrgvgan5sy@3fl3onchgdz3
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
|
|
Collect argument information from the type information of stub functions to
mark arguments of BPF struct_ops programs with PTR_MAYBE_NULL if they are
nullable. A nullable argument is annotated by suffixing "__nullable" at
the argument name of stub function.
For nullable arguments, this patch sets a struct bpf_ctx_arg_aux to label
their reg_type with PTR_TO_BTF_ID | PTR_TRUSTED | PTR_MAYBE_NULL. This
makes the verifier to check programs and ensure that they properly check
the pointer. The programs should check if the pointer is null before
accessing the pointed memory.
The implementer of a struct_ops type should annotate the arguments that can
be null. The implementer should define a stub function (empty) as a
placeholder for each defined operator. The name of a stub function should
be in the pattern "<st_op_type>__<operator name>". For example, for
test_maybe_null of struct bpf_testmod_ops, it's stub function name should
be "bpf_testmod_ops__test_maybe_null". You mark an argument nullable by
suffixing the argument name with "__nullable" at the stub function.
Since we already has stub functions for kCFI, we just reuse these stub
functions with the naming convention mentioned earlier. These stub
functions with the naming convention is only required if there are nullable
arguments to annotate. For functions having not nullable arguments, stub
functions are not necessary for the purpose of this patch.
This patch will prepare a list of struct bpf_ctx_arg_aux, aka arg_info, for
each member field of a struct_ops type. "arg_info" will be assigned to
"prog->aux->ctx_arg_info" of BPF struct_ops programs in
check_struct_ops_btf_id() so that it can be used by btf_ctx_access() later
to set reg_type properly for the verifier.
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240209023750.1153905-4-thinker.li@gmail.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
|
|
Move __kfunc_param_match_suffix() to btf.c and rename it as
btf_param_match_suffix(). It can be reused by bpf_struct_ops later.
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240209023750.1153905-3-thinker.li@gmail.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
|
|
Enable the providers to use types defined in a module instead of in the
kernel (btf_vmlinux).
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240209023750.1153905-2-thinker.li@gmail.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
|
|
Compiling with CONFIG_BPF_SYSCALL & !CONFIG_BPF_JIT throws the below
warning:
"WARN: resolve_btfids: unresolved symbol bpf_cpumask"
Fix it by adding the appropriate #ifdef.
Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Stanislav Fomichev <sdf@google.com>
Acked-by: David Vernet <void@manifault.com>
Link: https://lore.kernel.org/bpf/20240208100115.602172-1-hbathini@linux.ibm.com
|
|
Currently tracing is supposed not to allow for bpf_spin_{lock,unlock}()
helper calls. This is to prevent deadlock for the following cases:
- there is a prog (prog-A) calling bpf_spin_{lock,unlock}().
- there is a tracing program (prog-B), e.g., fentry, attached
to bpf_spin_lock() and/or bpf_spin_unlock().
- prog-B calls bpf_spin_{lock,unlock}().
For such a case, when prog-A calls bpf_spin_{lock,unlock}(),
a deadlock will happen.
The related source codes are below in kernel/bpf/helpers.c:
notrace BPF_CALL_1(bpf_spin_lock, struct bpf_spin_lock *, lock)
notrace BPF_CALL_1(bpf_spin_unlock, struct bpf_spin_lock *, lock)
notrace is supposed to prevent fentry prog from attaching to
bpf_spin_{lock,unlock}().
But actually this is not the case and fentry prog can successfully
attached to bpf_spin_lock(). Siddharth Chintamaneni reported
the issue in [1]. The following is the macro definition for
above BPF_CALL_1:
#define BPF_CALL_x(x, name, ...) \
static __always_inline \
u64 ____##name(__BPF_MAP(x, __BPF_DECL_ARGS, __BPF_V, __VA_ARGS__)); \
typedef u64 (*btf_##name)(__BPF_MAP(x, __BPF_DECL_ARGS, __BPF_V, __VA_ARGS__)); \
u64 name(__BPF_REG(x, __BPF_DECL_REGS, __BPF_N, __VA_ARGS__)); \
u64 name(__BPF_REG(x, __BPF_DECL_REGS, __BPF_N, __VA_ARGS__)) \
{ \
return ((btf_##name)____##name)(__BPF_MAP(x,__BPF_CAST,__BPF_N,__VA_ARGS__));\
} \
static __always_inline \
u64 ____##name(__BPF_MAP(x, __BPF_DECL_ARGS, __BPF_V, __VA_ARGS__))
#define BPF_CALL_1(name, ...) BPF_CALL_x(1, name, __VA_ARGS__)
The notrace attribute is actually applied to the static always_inline function
____bpf_spin_{lock,unlock}(). The actual callback function
bpf_spin_{lock,unlock}() is not marked with notrace, hence
allowing fentry prog to attach to two helpers, and this
may cause the above mentioned deadlock. Siddharth Chintamaneni
actually has a reproducer in [2].
To fix the issue, a new macro NOTRACE_BPF_CALL_1 is introduced which
will add notrace attribute to the original function instead of
the hidden always_inline function and this fixed the problem.
[1] https://lore.kernel.org/bpf/CAE5sdEigPnoGrzN8WU7Tx-h-iFuMZgW06qp0KHWtpvoXxf1OAQ@mail.gmail.com/
[2] https://lore.kernel.org/bpf/CAE5sdEg6yUc_Jz50AnUXEEUh6O73yQ1Z6NV2srJnef0ZrQkZew@mail.gmail.com/
Fixes: d83525ca62cf ("bpf: introduce bpf_spin_lock")
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/bpf/20240207070102.335167-1-yonghong.song@linux.dev
|
|
Since 20d59ee55172 ("libbpf: add bpf_core_cast() macro"), libbpf is now
exporting a const arg version of bpf_rdonly_cast(). This causes the
following conflicting type error when generating kfunc prototypes from
BTF:
In file included from skeleton/pid_iter.bpf.c:5:
/home/dxu/dev/linux/tools/bpf/bpftool/bootstrap/libbpf/include/bpf/bpf_core_read.h:297:14: error: conflicting types for 'bpf_rdonly_cast'
extern void *bpf_rdonly_cast(const void *obj__ign, __u32 btf_id__k) __ksym __weak;
^
./vmlinux.h:135625:14: note: previous declaration is here
extern void *bpf_rdonly_cast(void *obj__ign, u32 btf_id__k) __weak __ksym;
This is b/c the kernel defines bpf_rdonly_cast() with non-const arg.
Since const arg is more permissive and thus backwards compatible, we
change the kernel definition as well to avoid conflicting type errors.
Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/bpf/dfd3823f11ffd2d4c838e961d61ec9ae8a646773.1707080349.git.dxu@dxuuu.xyz
|
|
In various performance profiles of kernels with BPF programs attached,
bpf_local_storage_lookup() appears as a significant portion of CPU
cycles spent. To enable the compiler generate more optimal code, turn
bpf_local_storage_lookup() into a static inline function, where only the
cache insertion code path is outlined
Notably, outlining cache insertion helps avoid bloating callers by
duplicating setting up calls to raw_spin_{lock,unlock}_irqsave() (on
architectures which do not inline spin_lock/unlock, such as x86), which
would cause the compiler produce worse code by deciding to outline
otherwise inlinable functions. The call overhead is neutral, because we
make 2 calls either way: either calling raw_spin_lock_irqsave() and
raw_spin_unlock_irqsave(); or call __bpf_local_storage_insert_cache(),
which calls raw_spin_lock_irqsave(), followed by a tail-call to
raw_spin_unlock_irqsave() where the compiler can perform TCO and (in
optimized uninstrumented builds) turns it into a plain jump. The call to
__bpf_local_storage_insert_cache() can be elided entirely if
cacheit_lockit is a false constant expression.
Based on results from './benchs/run_bench_local_storage.sh' (21 trials,
reboot between each trial; x86 defconfig + BPF, clang 16) this produces
improvements in throughput and latency in the majority of cases, with an
average (geomean) improvement of 8%:
+---- Hashmap Control --------------------
|
| + num keys: 10
| : <before> | <after>
| +-+ hashmap (control) sequential get +----------------------+----------------------
| +- hits throughput | 14.789 M ops/s | 14.745 M ops/s ( ~ )
| +- hits latency | 67.679 ns/op | 67.879 ns/op ( ~ )
| +- important_hits throughput | 14.789 M ops/s | 14.745 M ops/s ( ~ )
|
| + num keys: 1000
| : <before> | <after>
| +-+ hashmap (control) sequential get +----------------------+----------------------
| +- hits throughput | 12.233 M ops/s | 12.170 M ops/s ( ~ )
| +- hits latency | 81.754 ns/op | 82.185 ns/op ( ~ )
| +- important_hits throughput | 12.233 M ops/s | 12.170 M ops/s ( ~ )
|
| + num keys: 10000
| : <before> | <after>
| +-+ hashmap (control) sequential get +----------------------+----------------------
| +- hits throughput | 7.220 M ops/s | 7.204 M ops/s ( ~ )
| +- hits latency | 138.522 ns/op | 138.842 ns/op ( ~ )
| +- important_hits throughput | 7.220 M ops/s | 7.204 M ops/s ( ~ )
|
| + num keys: 100000
| : <before> | <after>
| +-+ hashmap (control) sequential get +----------------------+----------------------
| +- hits throughput | 5.061 M ops/s | 5.165 M ops/s (+2.1%)
| +- hits latency | 198.483 ns/op | 194.270 ns/op (-2.1%)
| +- important_hits throughput | 5.061 M ops/s | 5.165 M ops/s (+2.1%)
|
| + num keys: 4194304
| : <before> | <after>
| +-+ hashmap (control) sequential get +----------------------+----------------------
| +- hits throughput | 2.864 M ops/s | 2.882 M ops/s ( ~ )
| +- hits latency | 365.220 ns/op | 361.418 ns/op (-1.0%)
| +- important_hits throughput | 2.864 M ops/s | 2.882 M ops/s ( ~ )
|
+---- Local Storage ----------------------
|
| + num_maps: 1
| : <before> | <after>
| +-+ local_storage cache sequential get +----------------------+----------------------
| +- hits throughput | 33.005 M ops/s | 39.068 M ops/s (+18.4%)
| +- hits latency | 30.300 ns/op | 25.598 ns/op (-15.5%)
| +- important_hits throughput | 33.005 M ops/s | 39.068 M ops/s (+18.4%)
| :
| : <before> | <after>
| +-+ local_storage cache interleaved get +----------------------+----------------------
| +- hits throughput | 37.151 M ops/s | 44.926 M ops/s (+20.9%)
| +- hits latency | 26.919 ns/op | 22.259 ns/op (-17.3%)
| +- important_hits throughput | 37.151 M ops/s | 44.926 M ops/s (+20.9%)
|
| + num_maps: 10
| : <before> | <after>
| +-+ local_storage cache sequential get +----------------------+----------------------
| +- hits throughput | 32.288 M ops/s | 38.099 M ops/s (+18.0%)
| +- hits latency | 30.972 ns/op | 26.248 ns/op (-15.3%)
| +- important_hits throughput | 3.229 M ops/s | 3.810 M ops/s (+18.0%)
| :
| : <before> | <after>
| +-+ local_storage cache interleaved get +----------------------+----------------------
| +- hits throughput | 34.473 M ops/s | 41.145 M ops/s (+19.4%)
| +- hits latency | 29.010 ns/op | 24.307 ns/op (-16.2%)
| +- important_hits throughput | 12.312 M ops/s | 14.695 M ops/s (+19.4%)
|
| + num_maps: 16
| : <before> |
