From 4a11678f683814df82fca9018d964771e02d7e6d Mon Sep 17 00:00:00 2001 From: Felix Maurer Date: Wed, 9 Feb 2022 16:55:26 +0100 Subject: [PATCH 1/8] bpf: Do not try bpf_msg_push_data with len 0 If bpf_msg_push_data() is called with len 0 (as it happens during selftests/bpf/test_sockmap), we do not need to do anything and can return early. Calling bpf_msg_push_data() with len 0 previously lead to a wrong ENOMEM error: we later called get_order(copy + len); if len was 0, copy + len was also often 0 and get_order() returned some undefined value (at the moment 52). alloc_pages() caught that and failed, but then bpf_msg_push_data() returned ENOMEM. This was wrong because we are most probably not out of memory and actually do not need any additional memory. Fixes: 6fff607e2f14b ("bpf: sk_msg program helper bpf_msg_push_data") Signed-off-by: Felix Maurer Signed-off-by: Daniel Borkmann Acked-by: Yonghong Song Acked-by: John Fastabend Link: https://lore.kernel.org/bpf/df69012695c7094ccb1943ca02b4920db3537466.1644421921.git.fmaurer@redhat.com --- net/core/filter.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/net/core/filter.c b/net/core/filter.c index 4603b7cd3cd17e..9eb785842258ab 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -2710,6 +2710,9 @@ BPF_CALL_4(bpf_msg_push_data, struct sk_msg *, msg, u32, start, if (unlikely(flags)) return -EINVAL; + if (unlikely(len == 0)) + return 0; + /* First find the starting scatterlist element */ i = msg->sg.start; do { From a8abb0c3dc1e28454851a00f8b7333d9695d566c Mon Sep 17 00:00:00 2001 From: Kumar Kartikeya Dwivedi Date: Wed, 9 Feb 2022 12:33:23 +0530 Subject: [PATCH 2/8] bpf: Fix crash due to incorrect copy_map_value When both bpf_spin_lock and bpf_timer are present in a BPF map value, copy_map_value needs to skirt both objects when copying a value into and out of the map. However, the current code does not set both s_off and t_off in copy_map_value, which leads to a crash when e.g. bpf_spin_lock is placed in map value with bpf_timer, as bpf_map_update_elem call will be able to overwrite the other timer object. When the issue is not fixed, an overwriting can produce the following splat: [root@(none) bpf]# ./test_progs -t timer_crash [ 15.930339] bpf_testmod: loading out-of-tree module taints kernel. [ 16.037849] ================================================================== [ 16.038458] BUG: KASAN: user-memory-access in __pv_queued_spin_lock_slowpath+0x32b/0x520 [ 16.038944] Write of size 8 at addr 0000000000043ec0 by task test_progs/325 [ 16.039399] [ 16.039514] CPU: 0 PID: 325 Comm: test_progs Tainted: G OE 5.16.0+ #278 [ 16.039983] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS ArchLinux 1.15.0-1 04/01/2014 [ 16.040485] Call Trace: [ 16.040645] [ 16.040805] dump_stack_lvl+0x59/0x73 [ 16.041069] ? __pv_queued_spin_lock_slowpath+0x32b/0x520 [ 16.041427] kasan_report.cold+0x116/0x11b [ 16.041673] ? __pv_queued_spin_lock_slowpath+0x32b/0x520 [ 16.042040] __pv_queued_spin_lock_slowpath+0x32b/0x520 [ 16.042328] ? memcpy+0x39/0x60 [ 16.042552] ? pv_hash+0xd0/0xd0 [ 16.042785] ? lockdep_hardirqs_off+0x95/0xd0 [ 16.043079] __bpf_spin_lock_irqsave+0xdf/0xf0 [ 16.043366] ? bpf_get_current_comm+0x50/0x50 [ 16.043608] ? jhash+0x11a/0x270 [ 16.043848] bpf_timer_cancel+0x34/0xe0 [ 16.044119] bpf_prog_c4ea1c0f7449940d_sys_enter+0x7c/0x81 [ 16.044500] bpf_trampoline_6442477838_0+0x36/0x1000 [ 16.044836] __x64_sys_nanosleep+0x5/0x140 [ 16.045119] do_syscall_64+0x59/0x80 [ 16.045377] ? lock_is_held_type+0xe4/0x140 [ 16.045670] ? irqentry_exit_to_user_mode+0xa/0x40 [ 16.046001] ? mark_held_locks+0x24/0x90 [ 16.046287] ? asm_exc_page_fault+0x1e/0x30 [ 16.046569] ? asm_exc_page_fault+0x8/0x30 [ 16.046851] ? lockdep_hardirqs_on+0x7e/0x100 [ 16.047137] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 16.047405] RIP: 0033:0x7f9e4831718d [ 16.047602] Code: b4 0c 00 0f 05 eb a9 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d b3 6c 0c 00 f7 d8 64 89 01 48 [ 16.048764] RSP: 002b:00007fff488086b8 EFLAGS: 00000206 ORIG_RAX: 0000000000000023 [ 16.049275] RAX: ffffffffffffffda RBX: 00007f9e48683740 RCX: 00007f9e4831718d [ 16.049747] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 00007fff488086d0 [ 16.050225] RBP: 00007fff488086f0 R08: 00007fff488085d7 R09: 00007f9e4cb594a0 [ 16.050648] R10: 0000000000000000 R11: 0000000000000206 R12: 00007f9e484cde30 [ 16.051124] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000 [ 16.051608] [ 16.051762] ================================================================== Fixes: 68134668c17f ("bpf: Add map side support for bpf timers.") Signed-off-by: Kumar Kartikeya Dwivedi Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/20220209070324.1093182-2-memxor@gmail.com --- include/linux/bpf.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/include/linux/bpf.h b/include/linux/bpf.h index fa517ae604ad2f..31a83449808b53 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -224,7 +224,8 @@ static inline void copy_map_value(struct bpf_map *map, void *dst, void *src) if (unlikely(map_value_has_spin_lock(map))) { s_off = map->spin_lock_off; s_sz = sizeof(struct bpf_spin_lock); - } else if (unlikely(map_value_has_timer(map))) { + } + if (unlikely(map_value_has_timer(map))) { t_off = map->timer_off; t_sz = sizeof(struct bpf_timer); } From a7e75016a0753c24d6c995bc02501ae35368e333 Mon Sep 17 00:00:00 2001 From: Kumar Kartikeya Dwivedi Date: Wed, 9 Feb 2022 12:33:24 +0530 Subject: [PATCH 3/8] selftests/bpf: Add test for bpf_timer overwriting crash Add a test that validates that timer value is not overwritten when doing a copy_map_value call in the kernel. Without the prior fix, this test triggers a crash. Signed-off-by: Kumar Kartikeya Dwivedi Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/20220209070324.1093182-3-memxor@gmail.com --- .../selftests/bpf/prog_tests/timer_crash.c | 32 +++++++++++ .../testing/selftests/bpf/progs/timer_crash.c | 54 +++++++++++++++++++ 2 files changed, 86 insertions(+) create mode 100644 tools/testing/selftests/bpf/prog_tests/timer_crash.c create mode 100644 tools/testing/selftests/bpf/progs/timer_crash.c diff --git a/tools/testing/selftests/bpf/prog_tests/timer_crash.c b/tools/testing/selftests/bpf/prog_tests/timer_crash.c new file mode 100644 index 00000000000000..f74b82305da8c8 --- /dev/null +++ b/tools/testing/selftests/bpf/prog_tests/timer_crash.c @@ -0,0 +1,32 @@ +// SPDX-License-Identifier: GPL-2.0 +#include +#include "timer_crash.skel.h" + +enum { + MODE_ARRAY, + MODE_HASH, +}; + +static void test_timer_crash_mode(int mode) +{ + struct timer_crash *skel; + + skel = timer_crash__open_and_load(); + if (!ASSERT_OK_PTR(skel, "timer_crash__open_and_load")) + return; + skel->bss->pid = getpid(); + skel->bss->crash_map = mode; + if (!ASSERT_OK(timer_crash__attach(skel), "timer_crash__attach")) + goto end; + usleep(1); +end: + timer_crash__destroy(skel); +} + +void test_timer_crash(void) +{ + if (test__start_subtest("array")) + test_timer_crash_mode(MODE_ARRAY); + if (test__start_subtest("hash")) + test_timer_crash_mode(MODE_HASH); +} diff --git a/tools/testing/selftests/bpf/progs/timer_crash.c b/tools/testing/selftests/bpf/progs/timer_crash.c new file mode 100644 index 00000000000000..f8f7944e70dae4 --- /dev/null +++ b/tools/testing/selftests/bpf/progs/timer_crash.c @@ -0,0 +1,54 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include +#include +#include + +struct map_elem { + struct bpf_timer timer; + struct bpf_spin_lock lock; +}; + +struct { + __uint(type, BPF_MAP_TYPE_ARRAY); + __uint(max_entries, 1); + __type(key, int); + __type(value, struct map_elem); +} amap SEC(".maps"); + +struct { + __uint(type, BPF_MAP_TYPE_HASH); + __uint(max_entries, 1); + __type(key, int); + __type(value, struct map_elem); +} hmap SEC(".maps"); + +int pid = 0; +int crash_map = 0; /* 0 for amap, 1 for hmap */ + +SEC("fentry/do_nanosleep") +int sys_enter(void *ctx) +{ + struct map_elem *e, value = {}; + void *map = crash_map ? (void *)&hmap : (void *)&amap; + + if (bpf_get_current_task_btf()->tgid != pid) + return 0; + + *(void **)&value = (void *)0xdeadcaf3; + + bpf_map_update_elem(map, &(int){0}, &value, 0); + /* For array map, doing bpf_map_update_elem will do a + * check_and_free_timer_in_array, which will trigger the crash if timer + * pointer was overwritten, for hmap we need to use bpf_timer_cancel. + */ + if (crash_map == 1) { + e = bpf_map_lookup_elem(map, &(int){0}); + if (!e) + return 0; + bpf_timer_cancel(&e->timer); + } + return 0; +} + +char _license[] SEC("license") = "GPL"; From 3bd916ee0ecbbdd902fc24845f2fef332b2a310c Mon Sep 17 00:00:00 2001 From: Yonghong Song Date: Fri, 11 Feb 2022 11:49:48 -0800 Subject: [PATCH 4/8] bpf: Emit bpf_timer in vmlinux BTF Currently the following code in check_and_init_map_value() *(struct bpf_timer *)(dst + map->timer_off) = (struct bpf_timer){}; can help generate bpf_timer definition in vmlinuxBTF. But the code above may not zero the whole structure due to anonymour members and that code will be replaced by memset in the subsequent patch and bpf_timer definition will disappear from vmlinuxBTF. Let us emit the type explicitly so bpf program can continue to use it from vmlinux.h. Signed-off-by: Yonghong Song Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/20220211194948.3141529-1-yhs@fb.com --- kernel/bpf/helpers.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c index 01cfdf40c838c0..55c084251fabf5 100644 --- a/kernel/bpf/helpers.c +++ b/kernel/bpf/helpers.c @@ -2,6 +2,7 @@ /* Copyright (c) 2011-2014 PLUMgrid, http://plumgrid.com */ #include +#include #include #include #include @@ -1075,6 +1076,7 @@ static enum hrtimer_restart bpf_timer_cb(struct hrtimer *hrtimer) void *key; u32 idx; + BTF_TYPE_EMIT(struct bpf_timer); callback_fn = rcu_dereference_check(t->callback_fn, rcu_read_lock_bh_held()); if (!callback_fn) goto out; From 5eaed6eedbe9612f642ad2b880f961d1c6c8ec2b Mon Sep 17 00:00:00 2001 From: Yonghong Song Date: Fri, 11 Feb 2022 11:49:53 -0800 Subject: [PATCH 5/8] bpf: Fix a bpf_timer initialization issue MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The patch in [1] intends to fix a bpf_timer related issue, but the fix caused existing 'timer' selftest to fail with hang or some random errors. After some debug, I found an issue with check_and_init_map_value() in the hashtab.c. More specifically, in hashtab.c, we have code l_new = bpf_map_kmalloc_node(&htab->map, ...) check_and_init_map_value(&htab->map, l_new...) Note that bpf_map_kmalloc_node() does not do initialization so l_new contains random value. The function check_and_init_map_value() intends to zero the bpf_spin_lock and bpf_timer if they exist in the map. But I found bpf_spin_lock is zero'ed but bpf_timer is not zero'ed. With [1], later copy_map_value() skips copying of bpf_spin_lock and bpf_timer. The non-zero bpf_timer caused random failures for 'timer' selftest. Without [1], for both bpf_spin_lock and bpf_timer case, bpf_timer will be zero'ed, so 'timer' self test is okay. For check_and_init_map_value(), why bpf_spin_lock is zero'ed properly while bpf_timer not. In bpf uapi header, we have struct bpf_spin_lock { __u32 val; }; struct bpf_timer { __u64 :64; __u64 :64; } __attribute__((aligned(8))); The initialization code: *(struct bpf_spin_lock *)(dst + map->spin_lock_off) = (struct bpf_spin_lock){}; *(struct bpf_timer *)(dst + map->timer_off) = (struct bpf_timer){}; It appears the compiler has no obligation to initialize anonymous fields. For example, let us use clang with bpf target as below: $ cat t.c struct bpf_timer { unsigned long long :64; }; struct bpf_timer2 { unsigned long long a; }; void test(struct bpf_timer *t) { *t = (struct bpf_timer){}; } void test2(struct bpf_timer2 *t) { *t = (struct bpf_timer2){}; } $ clang -target bpf -O2 -c -g t.c $ llvm-objdump -d t.o ... 0000000000000000 : 0: 95 00 00 00 00 00 00 00 exit 0000000000000008 : 1: b7 02 00 00 00 00 00 00 r2 = 0 2: 7b 21 00 00 00 00 00 00 *(u64 *)(r1 + 0) = r2 3: 95 00 00 00 00 00 00 00 exit gcc11.2 does not have the above issue. But from INTERNATIONAL STANDARD ©ISO/IEC ISO/IEC 9899:201x Programming languages — C http://www.open-std.org/Jtc1/sc22/wg14/www/docs/n1547.pdf page 157: Except where explicitly stated otherwise, for the purposes of this subclause unnamed members of objects of structure and union type do not participate in initialization. Unnamed members of structure objects have indeterminate value even after initialization. To fix the problem, let use memset for bpf_timer case in check_and_init_map_value(). For consistency, memset is also used for bpf_spin_lock case. [1] https://lore.kernel.org/bpf/20220209070324.1093182-2-memxor@gmail.com/ Fixes: 68134668c17f3 ("bpf: Add map side support for bpf timers.") Signed-off-by: Yonghong Song Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/20220211194953.3142152-1-yhs@fb.com --- include/linux/bpf.h | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 31a83449808b53..d0ad379d1e62a7 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -209,11 +209,9 @@ static inline bool map_value_has_timer(const struct bpf_map *map) static inline void check_and_init_map_value(struct bpf_map *map, void *dst) { if (unlikely(map_value_has_spin_lock(map))) - *(struct bpf_spin_lock *)(dst + map->spin_lock_off) = - (struct bpf_spin_lock){}; + memset(dst + map->spin_lock_off, 0, sizeof(struct bpf_spin_lock)); if (unlikely(map_value_has_timer(map))) - *(struct bpf_timer *)(dst + map->timer_off) = - (struct bpf_timer){}; + memset(dst + map->timer_off, 0, sizeof(struct bpf_timer)); } /* copy everything but bpf_spin_lock and bpf_timer. There could be one of each. */ From 61d06f01f9710b327a53492e5add9f972eb909b3 Mon Sep 17 00:00:00 2001 From: Felix Maurer Date: Fri, 11 Feb 2022 18:43:36 +0100 Subject: [PATCH 6/8] selftests: bpf: Check bpf_msg_push_data return value bpf_msg_push_data may return a non-zero value to indicate an error. The return value should be checked to prevent undetected errors. To indicate an error, the BPF programs now perform a different action than their intended one to make the userspace test program notice the error, i.e., the programs supposed to pass/redirect drop, the program supposed to drop passes. Fixes: 84fbfe026acaa ("bpf: test_sockmap add options to use msg_push_data") Signed-off-by: Felix Maurer Signed-off-by: Alexei Starovoitov Acked-by: John Fastabend Link: https://lore.kernel.org/bpf/89f767bb44005d6b4dd1f42038c438f76b3ebfad.1644601294.git.fmaurer@redhat.com --- .../selftests/bpf/progs/test_sockmap_kern.h | 26 +++++++++++++------ 1 file changed, 18 insertions(+), 8 deletions(-) diff --git a/tools/testing/selftests/bpf/progs/test_sockmap_kern.h b/tools/testing/selftests/bpf/progs/test_sockmap_kern.h index 2966564b8497aa..6c85b00f27b2e5 100644 --- a/tools/testing/selftests/bpf/progs/test_sockmap_kern.h +++ b/tools/testing/selftests/bpf/progs/test_sockmap_kern.h @@ -235,7 +235,7 @@ SEC("sk_msg1") int bpf_prog4(struct sk_msg_md *msg) { int *bytes, zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5; - int *start, *end, *start_push, *end_push, *start_pop, *pop; + int *start, *end, *start_push, *end_push, *start_pop, *pop, err = 0; bytes = bpf_map_lookup_elem(&sock_apply_bytes, &zero); if (bytes) @@ -249,8 +249,11 @@ int bpf_prog4(struct sk_msg_md *msg) bpf_msg_pull_data(msg, *start, *end, 0); start_push = bpf_map_lookup_elem(&sock_bytes, &two); end_push = bpf_map_lookup_elem(&sock_bytes, &three); - if (start_push && end_push) - bpf_msg_push_data(msg, *start_push, *end_push, 0); + if (start_push && end_push) { + err = bpf_msg_push_data(msg, *start_push, *end_push, 0); + if (err) + return SK_DROP; + } start_pop = bpf_map_lookup_elem(&sock_bytes, &four); pop = bpf_map_lookup_elem(&sock_bytes, &five); if (start_pop && pop) @@ -263,6 +266,7 @@ int bpf_prog6(struct sk_msg_md *msg) { int zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5, key = 0; int *bytes, *start, *end, *start_push, *end_push, *start_pop, *pop, *f; + int err = 0; __u64 flags = 0; bytes = bpf_map_lookup_elem(&sock_apply_bytes, &zero); @@ -279,8 +283,11 @@ int bpf_prog6(struct sk_msg_md *msg) start_push = bpf_map_lookup_elem(&sock_bytes, &two); end_push = bpf_map_lookup_elem(&sock_bytes, &three); - if (start_push && end_push) - bpf_msg_push_data(msg, *start_push, *end_push, 0); + if (start_push && end_push) { + err = bpf_msg_push_data(msg, *start_push, *end_push, 0); + if (err) + return SK_DROP; + } start_pop = bpf_map_lookup_elem(&sock_bytes, &four); pop = bpf_map_lookup_elem(&sock_bytes, &five); @@ -338,7 +345,7 @@ SEC("sk_msg5") int bpf_prog10(struct sk_msg_md *msg) { int *bytes, *start, *end, *start_push, *end_push, *start_pop, *pop; - int zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5; + int zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5, err = 0; bytes = bpf_map_lookup_elem(&sock_apply_bytes, &zero); if (bytes) @@ -352,8 +359,11 @@ int bpf_prog10(struct sk_msg_md *msg) bpf_msg_pull_data(msg, *start, *end, 0); start_push = bpf_map_lookup_elem(&sock_bytes, &two); end_push = bpf_map_lookup_elem(&sock_bytes, &three); - if (start_push && end_push) - bpf_msg_push_data(msg, *start_push, *end_push, 0); + if (start_push && end_push) { + err = bpf_msg_push_data(msg, *start_push, *end_push, 0); + if (err) + return SK_PASS; + } start_pop = bpf_map_lookup_elem(&sock_bytes, &four); pop = bpf_map_lookup_elem(&sock_bytes, &five); if (start_pop && pop) From 45ce4b4f9009102cd9f581196d480a59208690c1 Mon Sep 17 00:00:00 2001 From: Kumar Kartikeya Dwivedi Date: Thu, 17 Feb 2022 01:49:43 +0530 Subject: [PATCH 7/8] bpf: Fix crash due to out of bounds access into reg2btf_ids. When commit e6ac2450d6de ("bpf: Support bpf program calling kernel function") added kfunc support, it defined reg2btf_ids as a cheap way to translate the verifier reg type to the appropriate btf_vmlinux BTF ID, however commit c25b2ae13603 ("bpf: Replace PTR_TO_XXX_OR_NULL with PTR_TO_XXX | PTR_MAYBE_NULL") moved the __BPF_REG_TYPE_MAX from the last member of bpf_reg_type enum to after the base register types, and defined other variants using type flag composition. However, now, the direct usage of reg->type to index into reg2btf_ids may no longer fall into __BPF_REG_TYPE_MAX range, and hence lead to out of bounds access and kernel crash on dereference of bad pointer. Fixes: c25b2ae13603 ("bpf: Replace PTR_TO_XXX_OR_NULL with PTR_TO_XXX | PTR_MAYBE_NULL") Signed-off-by: Kumar Kartikeya Dwivedi Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/20220216201943.624869-1-memxor@gmail.com --- kernel/bpf/btf.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c index e16dafeb24504a..3e23b3fa79ff61 100644 --- a/kernel/bpf/btf.c +++ b/kernel/bpf/btf.c @@ -5688,7 +5688,8 @@ static int btf_check_func_arg_match(struct bpf_verifier_env *env, } if (check_ptr_off_reg(env, reg, regno)) return -EINVAL; - } else if (is_kfunc && (reg->type == PTR_TO_BTF_ID || reg2btf_ids[reg->type])) { + } else if (is_kfunc && (reg->type == PTR_TO_BTF_ID || + (reg2btf_ids[base_type(reg->type)] && !type_flag(reg->type)))) { const struct btf_type *reg_ref_t; const struct btf *reg_btf; const char *reg_ref_tname; @@ -5706,7 +5707,7 @@ static int btf_check_func_arg_match(struct bpf_verifier_env *env, reg_ref_id = reg->btf_id; } else { reg_btf = btf_vmlinux; - reg_ref_id = *reg2btf_ids[reg->type]; + reg_ref_id = *reg2btf_ids[base_type(reg->type)]; } reg_ref_t = btf_type_skip_modifiers(reg_btf, reg_ref_id, From 75134f16e7dd0007aa474b281935c5f42e79f2c8 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 17 Feb 2022 10:19:02 -0800 Subject: [PATCH 8/8] bpf: Add schedule points in batch ops syzbot reported various soft lockups caused by bpf batch operations. INFO: task kworker/1:1:27 blocked for more than 140 seconds. INFO: task hung in rcu_barrier Nothing prevents batch ops to process huge amount of data, we need to add schedule points in them. Note that maybe_wait_bpf_programs(map) calls from generic_map_delete_batch() can be factorized by moving the call after the loop. This will be done later in -next tree once we get this fix merged, unless there is strong opinion doing this optimization sooner. Fixes: aa2e93b8e58e ("bpf: Add generic support for update and delete batch ops") Fixes: cb4d03ab499d ("bpf: Add generic support for lookup batch op") Reported-by: syzbot Signed-off-by: Eric Dumazet Signed-off-by: Alexei Starovoitov Reviewed-by: Stanislav Fomichev Acked-by: Brian Vazquez Link: https://lore.kernel.org/bpf/20220217181902.808742-1-eric.dumazet@gmail.com --- kernel/bpf/syscall.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index fa4505f9b6119b..ca70fe6fba3879 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -1355,6 +1355,7 @@ int generic_map_delete_batch(struct bpf_map *map, maybe_wait_bpf_programs(map); if (err) break; + cond_resched(); } if (copy_to_user(&uattr->batch.count, &cp, sizeof(cp))) err = -EFAULT; @@ -1412,6 +1413,7 @@ int generic_map_update_batch(struct bpf_map *map, if (err) break; + cond_resched(); } if (copy_to_user(&uattr->batch.count, &cp, sizeof(cp))) @@ -1509,6 +1511,7 @@ int generic_map_lookup_batch(struct bpf_map *map, swap(prev_key, key); retry = MAP_LOOKUP_RETRIES; cp++; + cond_resched(); } if (err == -EFAULT)