linux.git/kernel/bpf, branch v6.10

bpf: Defer work in bpf_timer_cancel_and_free

2024-07-10T22:59:44+00:00

Currently, the same case as previous patch (two timer callbacks trying
to cancel each other) can be invoked through bpf_map_update_elem as
well, or more precisely, freeing map elements containing timers. Since
this relies on hrtimer_cancel as well, it is prone to the same deadlock
situation as the previous patch.

It would be sufficient to use hrtimer_try_to_cancel to fix this problem,
as the timer cannot be enqueued after async_cancel_and_free. Once
async_cancel_and_free has been done, the timer must be reinitialized
before it can be armed again. The callback running in parallel trying to
arm the timer will fail, and freeing bpf_hrtimer without waiting is
sufficient (given kfree_rcu), and bpf_timer_cb will return
HRTIMER_NORESTART, preventing the timer from being rearmed again.

However, there exists a UAF scenario where the callback arms the timer
before entering this function, such that if cancellation fails (due to
timer callback invoking this routine, or the target timer callback
running concurrently). In such a case, if the timer expiration is
significantly far in the future, the RCU grace period expiration
happening before it will free the bpf_hrtimer state and along with it
the struct hrtimer, that is enqueued.

Hence, it is clear cancellation needs to occur after
async_cancel_and_free, and yet it cannot be done inline due to deadlock
issues. We thus modify bpf_timer_cancel_and_free to defer work to the
global workqueue, adding a work_struct alongside rcu_head (both used at
_different_ points of time, so can share space).

Update existing code comments to reflect the new state of affairs.

Fixes: b00628b1c7d5 ("bpf: Introduce bpf timers.")
Signed-off-by: Kumar Kartikeya Dwivedi 
Link: https://lore.kernel.org/r/20240709185440.1104957-3-memxor@gmail.com
Signed-off-by: Alexei Starovoitov

bpf: Fail bpf_timer_cancel when callback is being cancelled

2024-07-10T22:59:44+00:00

Given a schedule:

timer1 cb			timer2 cb

bpf_timer_cancel(timer2);	bpf_timer_cancel(timer1);

Both bpf_timer_cancel calls would wait for the other callback to finish
executing, introducing a lockup.

Add an atomic_t count named 'cancelling' in bpf_hrtimer. This keeps
track of all in-flight cancellation requests for a given BPF timer.
Whenever cancelling a BPF timer, we must check if we have outstanding
cancellation requests, and if so, we must fail the operation with an
error (-EDEADLK) since cancellation is synchronous and waits for the
callback to finish executing. This implies that we can enter a deadlock
situation involving two or more timer callbacks executing in parallel
and attempting to cancel one another.

Note that we avoid incrementing the cancelling counter for the target
timer (the one being cancelled) if bpf_timer_cancel is not invoked from
a callback, to avoid spurious errors. The whole point of detecting
cur->cancelling and returning -EDEADLK is to not enter a busy wait loop
(which may or may not lead to a lockup). This does not apply in case the
caller is in a non-callback context, the other side can continue to
cancel as it sees fit without running into errors.

Background on prior attempts:

Earlier versions of this patch used a bool 'cancelling' bit and used the
following pattern under timer->lock to publish cancellation status.

lock(t->lock);
t->cancelling = true;
mb();
if (cur->cancelling)
	return -EDEADLK;
unlock(t->lock);
hrtimer_cancel(t->timer);
t->cancelling = false;

The store outside the critical section could overwrite a parallel
requests t->cancelling assignment to true, to ensure the parallely
executing callback observes its cancellation status.

It would be necessary to clear this cancelling bit once hrtimer_cancel
is done, but lack of serialization introduced races. Another option was
explored where bpf_timer_start would clear the bit when (re)starting the
timer under timer->lock. This would ensure serialized access to the
cancelling bit, but may allow it to be cleared before in-flight
hrtimer_cancel has finished executing, such that lockups can occur
again.

Thus, we choose an atomic counter to keep track of all outstanding
cancellation requests and use it to prevent lockups in case callbacks
attempt to cancel each other while executing in parallel.

Reported-by: Dohyun Kim 
Reported-by: Neel Natu 
Fixes: b00628b1c7d5 ("bpf: Introduce bpf timers.")
Signed-off-by: Kumar Kartikeya Dwivedi 
Link: https://lore.kernel.org/r/20240709185440.1104957-2-memxor@gmail.com
Signed-off-by: Alexei Starovoitov

bpf: fix order of args in call to bpf_map_kvcalloc

2024-07-10T22:31:19+00:00

The original function call passed size of smap->bucket before the number of
buckets which raises the error 'calloc-transposed-args' on compilation.

Vlastimil Babka added:

The order of parameters can be traced back all the way to 6ac99e8f23d4
("bpf: Introduce bpf sk local storage") accross several refactorings,
and that's why the commit is used as a Fixes: tag.

In v6.10-rc1, a different commit 2c321f3f70bc ("mm: change inlined
allocation helpers to account at the call site") however exposed the
order of args in a way that gcc-14 has enough visibility to start
warning about it, because (in !CONFIG_MEMCG case) bpf_map_kvcalloc is
then a macro alias for kvcalloc instead of a static inline wrapper.

To sum up the warning happens when the following conditions are all met:

- gcc-14 is used (didn't see it with gcc-13)
- commit 2c321f3f70bc is present
- CONFIG_MEMCG is not enabled in .config
- CONFIG_WERROR turns this from a compiler warning to error

Fixes: 6ac99e8f23d4 ("bpf: Introduce bpf sk local storage")
Reviewed-by: Andrii Nakryiko 
Tested-by: Christian Kujau 
Signed-off-by: Mohammad Shehar Yaar Tausif 
Signed-off-by: Vlastimil Babka 
Link: https://lore.kernel.org/r/20240710100521.15061-2-vbabka@suse.cz
Signed-off-by: Alexei Starovoitov

Merge tag 'asm-generic-fixes-6.10' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic

2024-06-27T17:53:52+00:00

Pull asm-generic fixes from Arnd Bergmann:
 "These are some bugfixes for system call ABI issues I found while
  working on a cleanup series. None of these are urgent since these bugs
  have gone unnoticed for many years, but I think we probably want to
  backport them all to stable kernels, so it makes sense to have the
  fixes included as early as possible.

  One more fix addresses a compile-time warning in kallsyms that was
  uncovered by a patch I did to enable additional warnings in 6.10. I
  had mistakenly thought that this fix was already merged through the
  module tree, but as Geert pointed out it was still missing"

* tag 'asm-generic-fixes-6.10' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic:
  kallsyms: rework symbol lookup return codes
  linux/syscalls.h: add missing __user annotations
  syscalls: mmap(): use unsigned offset type consistently
  s390: remove native mmap2() syscall
  hexagon: fix fadvise64_64 calling conventions
  csky, hexagon: fix broken sys_sync_file_range
  sh: rework sync_file_range ABI
  powerpc: restore some missing spu syscalls
  parisc: use generic sys_fanotify_mark implementation
  parisc: use correct compat recv/recvfrom syscalls
  sparc: fix compat recv/recvfrom syscalls
  sparc: fix old compat_sys_select()
  syscalls: fix compat_sys_io_pgetevents_time64 usage
  ftruncate: pass a signed offset

kallsyms: rework symbol lookup return codes

2024-06-27T15:43:40+00:00

Building with W=1 in some configurations produces a false positive
warning for kallsyms:

kernel/kallsyms.c: In function '__sprint_symbol.isra':
kernel/kallsyms.c:503:17: error: 'strcpy' source argument is the same as destination [-Werror=restrict]
  503 |                 strcpy(buffer, name);
      |                 ^~~~~~~~~~~~~~~~~~~~

This originally showed up while building with -O3, but later started
happening in other configurations as well, depending on inlining
decisions. The underlying issue is that the local 'name' variable is
always initialized to the be the same as 'buffer' in the called functions
that fill the buffer, which gcc notices while inlining, though it could
see that the address check always skips the copy.

The calling conventions here are rather unusual, as all of the internal
lookup functions (bpf_address_lookup, ftrace_mod_address_lookup,
ftrace_func_address_lookup, module_address_lookup and
kallsyms_lookup_buildid) already use the provided buffer and either return
the address of that buffer to indicate success, or NULL for failure,
but the callers are written to also expect an arbitrary other buffer
to be returned.

Rework the calling conventions to return the length of the filled buffer
instead of its address, which is simpler and easier to follow as well
as avoiding the warning. Leave only the kallsyms_lookup() calling conventions
unchanged, since that is called from 16 different functions and
adapting this would be a much bigger change.

Link: https://lore.kernel.org/lkml/20200107214042.855757-1-arnd@arndb.de/
Link: https://lore.kernel.org/lkml/20240326130647.7bfb1d92@gandalf.local.home/
Tested-by: Geert Uytterhoeven 
Reviewed-by: Luis Chamberlain 
Acked-by: Steven Rostedt (Google) 
Signed-off-by: Arnd Bergmann

bpf: Fix may_goto with negative offset.

2024-06-24T11:44:02+00:00

Zac's syzbot crafted a bpf prog that exposed two bugs in may_goto.
The 1st bug is the way may_goto is patched. When offset is negative
it should be patched differently.
The 2nd bug is in the verifier:
when current state may_goto_depth is equal to visited state may_goto_depth
it means there is an actual infinite loop. It's not correct to prune
exploration of the program at this point.
Note, that this check doesn't limit the program to only one may_goto insn,
since 2nd and any further may_goto will increment may_goto_depth only
in the queued state pushed for future exploration. The current state
will have may_goto_depth == 0 regardless of number of may_goto insns
and the verifier has to explore the program until bpf_exit.

Fixes: 011832b97b31 ("bpf: Introduce may_goto instruction")
Reported-by: Zac Ecob 
Signed-off-by: Alexei Starovoitov 
Signed-off-by: Andrii Nakryiko 
Signed-off-by: Daniel Borkmann 
Acked-by: Eduard Zingerman 
Closes: https://lore.kernel.org/bpf/CAADnVQL-15aNp04-cyHRn47Yv61NXfYyhopyZtUyxNojUZUXpA@mail.gmail.com/
Link: https://lore.kernel.org/bpf/20240619235355.85031-1-alexei.starovoitov@gmail.com

bpf: Fix overrunning reservations in ringbuf

2024-06-21T20:04:21+00:00

The BPF ring buffer internally is implemented as a power-of-2 sized circular
buffer, with two logical and ever-increasing counters: consumer_pos is the
consumer counter to show which logical position the consumer consumed the
data, and producer_pos which is the producer counter denoting the amount of
data reserved by all producers.

Each time a record is reserved, the producer that "owns" the record will
successfully advance producer counter. In user space each time a record is
read, the consumer of the data advanced the consumer counter once it finished
processing. Both counters are stored in separate pages so that from user
space, the producer counter is read-only and the consumer counter is read-write.

One aspect that simplifies and thus speeds up the implementation of both
producers and consumers is how the data area is mapped twice contiguously
back-to-back in the virtual memory, allowing to not take any special measures
for samples that have to wrap around at the end of the circular buffer data
area, because the next page after the last data page would be first data page
again, and thus the sample will still appear completely contiguous in virtual
memory.

Each record has a struct bpf_ringbuf_hdr { u32 len; u32 pg_off; } header for
book-keeping the length and offset, and is inaccessible to the BPF program.
Helpers like bpf_ringbuf_reserve() return `(void *)hdr + BPF_RINGBUF_HDR_SZ`
for the BPF program to use. Bing-Jhong and Muhammad reported that it is however
possible to make a second allocated memory chunk overlapping with the first
chunk and as a result, the BPF program is now able to edit first chunk's
header.

For example, consider the creation of a BPF_MAP_TYPE_RINGBUF map with size
of 0x4000. Next, the consumer_pos is modified to 0x3000 /before/ a call to
bpf_ringbuf_reserve() is made. This will allocate a chunk A, which is in
[0x0,0x3008], and the BPF program is able to edit [0x8,0x3008]. Now, lets
allocate a chunk B with size 0x3000. This will succeed because consumer_pos
was edited ahead of time to pass the `new_prod_pos - cons_pos > rb->mask`
check. Chunk B will be in range [0x3008,0x6010], and the BPF program is able
to edit [0x3010,0x6010]. Due to the ring buffer memory layout mentioned
earlier, the ranges [0x0,0x4000] and [0x4000,0x8000] point to the same data
pages. This means that chunk B at [0x4000,0x4008] is chunk A's header.
bpf_ringbuf_submit() / bpf_ringbuf_discard() use the header's pg_off to then
locate the bpf_ringbuf itself via bpf_ringbuf_restore_from_rec(). Once chunk
B modified chunk A's header, then bpf_ringbuf_commit() refers to the wrong
page and could cause a crash.

Fix it by calculating the oldest pending_pos and check whether the range
from the oldest outstanding record to the newest would span beyond the ring
buffer size. If that is the case, then reject the request. We've tested with
the ring buffer benchmark in BPF selftests (./benchs/run_bench_ringbufs.sh)
before/after the fix and while it seems a bit slower on some benchmarks, it
is still not significantly enough to matter.

Fixes: 457f44363a88 ("bpf: Implement BPF ring buffer and verifier support for it")
Reported-by: Bing-Jhong Billy Jheng 
Reported-by: Muhammad Ramdhan 
Co-developed-by: Bing-Jhong Billy Jheng 
Co-developed-by: Andrii Nakryiko 
Signed-off-by: Bing-Jhong Billy Jheng 
Signed-off-by: Andrii Nakryiko 
Signed-off-by: Daniel Borkmann 
Signed-off-by: Andrii Nakryiko 
Link: https://lore.kernel.org/bpf/20240621140828.18238-1-daniel@iogearbox.net

bpf: Fix the corner case with may_goto and jump to the 1st insn.

2024-06-21T18:18:40+00:00

When the following program is processed by the verifier:
L1: may_goto L2
    goto L1
L2: w0 = 0
    exit

the may_goto insn is first converted to:
L1: r11 = *(u64 *)(r10 -8)
    if r11 == 0x0 goto L2
    r11 -= 1
    *(u64 *)(r10 -8) = r11
    goto L1
L2: w0 = 0
    exit

then later as the last step the verifier inserts:
  *(u64 *)(r10 -8) = BPF_MAX_LOOPS
as the first insn of the program to initialize loop count.

When the first insn happens to be a branch target of some jmp the
bpf_patch_insn_data() logic will produce:
L1: *(u64 *)(r10 -8) = BPF_MAX_LOOPS
    r11 = *(u64 *)(r10 -8)
    if r11 == 0x0 goto L2
    r11 -= 1
    *(u64 *)(r10 -8) = r11
    goto L1
L2: w0 = 0
    exit

because instruction patching adjusts all jmps and calls, but for this
particular corner case it's incorrect and the L1 label should be one
instruction down, like:
    *(u64 *)(r10 -8) = BPF_MAX_LOOPS
L1: r11 = *(u64 *)(r10 -8)
    if r11 == 0x0 goto L2
    r11 -= 1
    *(u64 *)(r10 -8) = r11
    goto L1
L2: w0 = 0
    exit

and that's what this patch is fixing.
After bpf_patch_insn_data() call adjust_jmp_off() to adjust all jmps
that point to newly insert BPF_ST insn to point to insn after.

Note that bpf_patch_insn_data() cannot easily be changed to accommodate
this logic, since jumps that point before or after a sequence of patched
instructions have to be adjusted with the full length of the patch.

Conceptually it's somewhat similar to "insert" of instructions between other
instructions with weird semantics. Like "insert" before 1st insn would require
adjustment of CALL insns to point to newly inserted 1st insn, but not an
adjustment JMP insns that point to 1st, yet still adjusting JMP insns that
cross over 1st insn (point to insn before or insn after), hence use simple
adjust_jmp_off() logic to fix this corner case. Ideally bpf_patch_insn_data()
would have an auxiliary info to say where 'the start of newly inserted patch
is', but it would be too complex for backport.

Fixes: 011832b97b31 ("bpf: Introduce may_goto instruction")
Reported-by: Zac Ecob 
Signed-off-by: Alexei Starovoitov 
Signed-off-by: Daniel Borkmann 
Acked-by: Eduard Zingerman 
Closes: https://lore.kernel.org/bpf/CAADnVQJ_WWx8w4b=6Gc2EpzAjgv+6A0ridnMz2TvS2egj4r3Gw@mail.gmail.com/
Link: https://lore.kernel.org/bpf/20240619011859.79334-1-alexei.starovoitov@gmail.com

bpf: Fix remap of arena.

2024-06-18T15:19:46+00:00

The bpf arena logic didn't account for mremap operation. Add a refcnt for
multiple mmap events to prevent use-after-free in arena_vm_close.

Fixes: 317460317a02 ("bpf: Introduce bpf_arena.")
Reported-by: Pengfei Xu 
Signed-off-by: Alexei Starovoitov 
Signed-off-by: Daniel Borkmann 
Reviewed-by: Barret Rhoden 
Tested-by: Pengfei Xu 
Closes: https://lore.kernel.org/bpf/Zmuw29IhgyPNKnIM@xpf.sh.intel.com
Link: https://lore.kernel.org/bpf/20240617171812.76634-1-alexei.starovoitov@gmail.com

bpf: Add missed var_off setting in coerce_subreg_to_size_sx()

2024-06-17T17:45:46+00:00

In coerce_subreg_to_size_sx(), for the case where upper
sign extension bits are the same for smax32 and smin32
values, we missed to setup properly. This is especially
problematic if both smax32 and smin32's sign extension
bits are 1.

The following is a simple example illustrating the inconsistent
verifier states due to missed var_off:

  0: (85) call bpf_get_prandom_u32#7    ; R0_w=scalar()
  1: (bf) r3 = r0                       ; R0_w=scalar(id=1) R3_w=scalar(id=1)
  2: (57) r3 &= 15                      ; R3_w=scalar(smin=smin32=0,smax=umax=smax32=umax32=15,var_off=(0x0; 0xf))
  3: (47) r3 |= 128                     ; R3_w=scalar(smin=umin=smin32=umin32=128,smax=umax=smax32=umax32=143,var_off=(0x80; 0xf))
  4: (bc) w7 = (s8)w3
  REG INVARIANTS VIOLATION (alu): range bounds violation u64=[0xffffff80, 0x8f] s64=[0xffffff80, 0x8f]
    u32=[0xffffff80, 0x8f] s32=[0x80, 0xffffff8f] var_off=(0x80, 0xf)

The var_off=(0x80, 0xf) is not correct, and the correct one should
be var_off=(0xffffff80; 0xf) since from insn 3, we know that at
insn 4, the sign extension bits will be 1. This patch fixed this
issue by setting var_off properly.

Fixes: 8100928c8814 ("bpf: Support new sign-extension mov insns")
Signed-off-by: Yonghong Song 
Link: https://lore.kernel.org/r/20240615174632.3995278-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov