16 __bpf_prog_run() interpreters for various stack sizes add .text
but not a lot comparing to run-time stack savings
text data bss dec hex filename
26350 10328 624 37302 91b6 kernel/bpf/core.o.before_split
25777 10328 624 36729 8f79 kernel/bpf/core.o.after_split
26970 10328 624 37922 9422 kernel/bpf/core.o.now
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
The next set of patches will take advantage of stack_depth tracking,
so make sure that the program that does bpf_tail_call() has
stack depth large enough for the callee.
We could have tracked the stack depth of the prog_array owner program
and only allow insertion of the programs with stack depth less
than the owner, but it will break existing applications.
Some of them have trivial root bpf program that only does
multiple bpf_tail_calls and at init time the prog array is empty.
In the future we may add a flag to do such tracking optionally,
but for now play simple and safe.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
teach verifier to track bpf program stack depth
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
free up BPF_JMP | BPF_CALL | BPF_X opcode to be used by actual
indirect call by register and use kernel internal opcode to
mark call instruction into bpf_tail_call() helper.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
trie_alloc() always needs to have BPF_F_NO_PREALLOC passed in via
attr->map_flags, since it does not support preallocation yet. We
check the flag, but we never copy the flag into trie->map.map_flags,
which is later on exposed into fdinfo and used by loaders such as
iproute2. Latter uses this in bpf_map_selfcheck_pinned() to test
whether a pinned map has the same spec as the one from the BPF obj
file and if not, bails out, which is currently the case for lpm
since it exposes always 0 as flags.
Also copy over flags in array_map_alloc() and stack_map_alloc().
They always have to be 0 right now, but we should make sure to not
miss to copy them over at a later point in time when we add actual
flags for them to use.
Fixes: b95a5c4db0 ("bpf: add a longest prefix match trie map implementation")
Reported-by: Jarno Rajahalme <jarno@covalent.io>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, after performing helper calls, we clear all caller saved
registers, that is r0 - r5 and fill r0 depending on struct bpf_func_proto
specification. The way we reset these regs can affect pruning decisions
in later paths, since we only reset register's imm to 0 and type to
NOT_INIT. However, we leave out clearing of other variables such as id,
min_value, max_value, etc, which can later on lead to pruning mismatches
due to stale data.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, when we enforce alignment tracking on direct packet access,
the verifier lets the following program pass despite doing a packet
write with unaligned access:
0: (61) r2 = *(u32 *)(r1 +76)
1: (61) r3 = *(u32 *)(r1 +80)
2: (61) r7 = *(u32 *)(r1 +8)
3: (bf) r0 = r2
4: (07) r0 += 14
5: (25) if r7 > 0x1 goto pc+4
R0=pkt(id=0,off=14,r=0) R1=ctx R2=pkt(id=0,off=0,r=0)
R3=pkt_end R7=inv,min_value=0,max_value=1 R10=fp
6: (2d) if r0 > r3 goto pc+1
R0=pkt(id=0,off=14,r=14) R1=ctx R2=pkt(id=0,off=0,r=14)
R3=pkt_end R7=inv,min_value=0,max_value=1 R10=fp
7: (63) *(u32 *)(r0 -4) = r0
8: (b7) r0 = 0
9: (95) exit
from 6 to 8:
R0=pkt(id=0,off=14,r=0) R1=ctx R2=pkt(id=0,off=0,r=0)
R3=pkt_end R7=inv,min_value=0,max_value=1 R10=fp
8: (b7) r0 = 0
9: (95) exit
from 5 to 10:
R0=pkt(id=0,off=14,r=0) R1=ctx R2=pkt(id=0,off=0,r=0)
R3=pkt_end R7=inv,min_value=2 R10=fp
10: (07) r0 += 1
11: (05) goto pc-6
6: safe <----- here, wrongly found safe
processed 15 insns
However, if we enforce a pruning mismatch by adding state into r8
which is then being mismatched in states_equal(), we find that for
the otherwise same program, the verifier detects a misaligned packet
access when actually walking that path:
0: (61) r2 = *(u32 *)(r1 +76)
1: (61) r3 = *(u32 *)(r1 +80)
2: (61) r7 = *(u32 *)(r1 +8)
3: (b7) r8 = 1
4: (bf) r0 = r2
5: (07) r0 += 14
6: (25) if r7 > 0x1 goto pc+4
R0=pkt(id=0,off=14,r=0) R1=ctx R2=pkt(id=0,off=0,r=0)
R3=pkt_end R7=inv,min_value=0,max_value=1
R8=imm1,min_value=1,max_value=1,min_align=1 R10=fp
7: (2d) if r0 > r3 goto pc+1
R0=pkt(id=0,off=14,r=14) R1=ctx R2=pkt(id=0,off=0,r=14)
R3=pkt_end R7=inv,min_value=0,max_value=1
R8=imm1,min_value=1,max_value=1,min_align=1 R10=fp
8: (63) *(u32 *)(r0 -4) = r0
9: (b7) r0 = 0
10: (95) exit
from 7 to 9:
R0=pkt(id=0,off=14,r=0) R1=ctx R2=pkt(id=0,off=0,r=0)
R3=pkt_end R7=inv,min_value=0,max_value=1
R8=imm1,min_value=1,max_value=1,min_align=1 R10=fp
9: (b7) r0 = 0
10: (95) exit
from 6 to 11:
R0=pkt(id=0,off=14,r=0) R1=ctx R2=pkt(id=0,off=0,r=0)
R3=pkt_end R7=inv,min_value=2
R8=imm1,min_value=1,max_value=1,min_align=1 R10=fp
11: (07) r0 += 1
12: (b7) r8 = 0
13: (05) goto pc-7 <----- mismatch due to r8
7: (2d) if r0 > r3 goto pc+1
R0=pkt(id=0,off=15,r=15) R1=ctx R2=pkt(id=0,off=0,r=15)
R3=pkt_end R7=inv,min_value=2
R8=imm0,min_value=0,max_value=0,min_align=2147483648 R10=fp
8: (63) *(u32 *)(r0 -4) = r0
misaligned packet access off 2+15+-4 size 4
The reason why we fail to see it in states_equal() is that the
third test in compare_ptrs_to_packet() ...
if (old->off <= cur->off &&
old->off >= old->range && cur->off >= cur->range)
return true;
... will let the above pass. The situation we run into is that
old->off <= cur->off (14 <= 15), meaning that prior walked paths
went with smaller offset, which was later used in the packet
access after successful packet range check and found to be safe
already.
For example: Given is R0=pkt(id=0,off=0,r=0). Adding offset 14
as in above program to it, results in R0=pkt(id=0,off=14,r=0)
before the packet range test. Now, testing this against R3=pkt_end
with 'if r0 > r3 goto out' will transform R0 into R0=pkt(id=0,off=14,r=14)
for the case when we're within bounds. A write into the packet
at offset *(u32 *)(r0 -4), that is, 2 + 14 -4, is valid and
aligned (2 is for NET_IP_ALIGN). After processing this with
all fall-through paths, we later on check paths from branches.
When the above skb->mark test is true, then we jump near the
end of the program, perform r0 += 1, and jump back to the
'if r0 > r3 goto out' test we've visited earlier already. This
time, R0 is of type R0=pkt(id=0,off=15,r=0), and we'll prune
that part because this time we'll have a larger safe packet
range, and we already found that with off=14 all further insn
were already safe, so it's safe as well with a larger off.
However, the problem is that the subsequent write into the packet
with 2 + 15 -4 is then unaligned, and not caught by the alignment
tracking. Note that min_align, aux_off, and aux_off_align were
all 0 in this example.
Since we cannot tell at this time what kind of packet access was
performed in the prior walk and what minimal requirements it has
(we might do so in the future, but that requires more complexity),
fix it to disable this pruning case for strict alignment for now,
and let the verifier do check such paths instead. With that applied,
the test cases pass and reject the program due to misalignment.
Fixes: d117441674 ("bpf: Track alignment of register values in the verifier.")
Reference: http://patchwork.ozlabs.org/patch/761909/
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The assignmnet:
ip_align = strict ? 2 : NET_IP_ALIGN;
in compare_pkt_ptr_alignment() trips up Coverity because we can only
get to this code when strict is true, therefore ip_align will always
be 2 regardless of NET_IP_ALIGN's value.
So just assign directly to '2' and explain the situation in the
comment above.
Reported-by: "Gustavo A. R. Silva" <garsilva@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Current limits with regards to processing program paths do not
really reflect today's needs anymore due to programs becoming
more complex and verifier smarter, keeping track of more data
such as const ALU operations, alignment tracking, spilling of
PTR_TO_MAP_VALUE_ADJ registers, and other features allowing for
smarter matching of what LLVM generates.
This also comes with the side-effect that we result in fewer
opportunities to prune search states and thus often need to do
more work to prove safety than in the past due to different
register states and stack layout where we mismatch. Generally,
it's quite hard to determine what caused a sudden increase in
complexity, it could be caused by something as trivial as a
single branch somewhere at the beginning of the program where
LLVM assigned a stack slot that is marked differently throughout
other branches and thus causing a mismatch, where verifier
then needs to prove safety for the whole rest of the program.
Subsequently, programs with even less than half the insn size
limit can get rejected. We noticed that while some programs
load fine under pre 4.11, they get rejected due to hitting
limits on more recent kernels. We saw that in the vast majority
of cases (90+%) pruning failed due to register mismatches. In
case of stack mismatches, majority of cases failed due to
different stack slot types (invalid, spill, misc) rather than
differences in spilled registers.
This patch makes pruning more aggressive by also adding markers
that sit at conditional jumps as well. Currently, we only mark
jump targets for pruning. For example in direct packet access,
these are usually error paths where we bail out. We found that
adding these markers, it can reduce number of processed insns
by up to 30%. Another option is to ignore reg->id in probing
PTR_TO_MAP_VALUE_OR_NULL registers, which can help pruning
slightly as well by up to 7% observed complexity reduction as
stand-alone. Meaning, if a previous path with register type
PTR_TO_MAP_VALUE_OR_NULL for map X was found to be safe, then
in the current state a PTR_TO_MAP_VALUE_OR_NULL register for
the same map X must be safe as well. Last but not least the
patch also adds a scheduling point and bumps the current limit
for instructions to be processed to a more adequate value.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
We must accumulate into reg->aux_off rather than use a plain assignment.
Add a test for this situation to test_align.
Reported-by: Alexei Starovoitov <ast@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a new field, "prog_flags", and an initial flag value
BPF_F_STRICT_ALIGNMENT.
When set, the verifier will enforce strict pointer alignment
regardless of the setting of CONFIG_EFFICIENT_UNALIGNED_ACCESS.
The verifier, in this mode, will also use a fixed value of "2" in
place of NET_IP_ALIGN.
This facilitates test cases that will exercise and validate this part
of the verifier even when run on architectures where alignment doesn't
matter.
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
If log_level > 1, do a state dump every instruction and emit it in
a more compact way (without a leading newline).
This will facilitate more sophisticated test cases which inspect the
verifier log for register state.
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Currently if we add only constant values to pointers we can fully
validate the alignment, and properly check if we need to reject the
program on !CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS architectures.
However, once an unknown value is introduced we only allow byte sized
memory accesses which is too restrictive.
Add logic to track the known minimum alignment of register values,
and propagate this state into registers containing pointers.
The most common paradigm that makes use of this new logic is computing
the transport header using the IP header length field. For example:
struct ethhdr *ep = skb->data;
struct iphdr *iph = (struct iphdr *) (ep + 1);
struct tcphdr *th;
...
n = iph->ihl;
th = ((void *)iph + (n * 4));
port = th->dest;
The existing code will reject the load of th->dest because it cannot
validate that the alignment is at least 2 once "n * 4" is added the
the packet pointer.
In the new code, the register holding "n * 4" will have a reg->min_align
value of 4, because any value multiplied by 4 will be at least 4 byte
aligned. (actually, the eBPF code emitted by the compiler in this case
is most likely to use a shift left by 2, but the end result is identical)
At the critical addition:
th = ((void *)iph + (n * 4));
The register holding 'th' will start with reg->off value of 14. The
pointer addition will transform that reg into something that looks like:
reg->aux_off = 14
reg->aux_off_align = 4
Next, the verifier will look at the th->dest load, and it will see
a load offset of 2, and first check:
if (reg->aux_off_align % size)
which will pass because aux_off_align is 4. reg_off will be computed:
reg_off = reg->off;
...
reg_off += reg->aux_off;
plus we have off==2, and it will thus check:
if ((NET_IP_ALIGN + reg_off + off) % size != 0)
which evaluates to:
if ((NET_IP_ALIGN + 14 + 2) % size != 0)
On strict alignment architectures, NET_IP_ALIGN is 2, thus:
if ((2 + 14 + 2) % size != 0)
which passes.
These pointer transformations and checks work regardless of whether
the constant offset or the variable with known alignment is added
first to the pointer register.
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Pull networking fixes from David Miller:
1) Fix multiqueue in stmmac driver on PCI, from Andy Shevchenko.
2) cdc_ncm doesn't actually fully zero out the padding area is
allocates on TX, from Jim Baxter.
3) Don't leak map addresses in BPF verifier, from Daniel Borkmann.
4) If we randomize TCP timestamps, we have to do it everywhere
including SYN cookies. From Eric Dumazet.
5) Fix "ethtool -S" crash in aquantia driver, from Pavel Belous.
6) Fix allocation size for ntp filter bitmap in bnxt_en driver, from
Dan Carpenter.
7) Add missing memory allocation return value check to DSA loop driver,
from Christophe Jaillet.
8) Fix XDP leak on driver unload in qed driver, from Suddarsana Reddy
Kalluru.
9) Don't inherit MC list from parent inet connection sockets, another
syzkaller spotted gem. Fix from Eric Dumazet.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (43 commits)
dccp/tcp: do not inherit mc_list from parent
qede: Split PF/VF ndos.
qed: Correct doorbell configuration for !4Kb pages
qed: Tell QM the number of tasks
qed: Fix VF removal sequence
qede: Fix XDP memory leak on unload
net/mlx4_core: Reduce harmless SRIOV error message to debug level
net/mlx4_en: Avoid adding steering rules with invalid ring
net/mlx4_en: Change the error print to debug print
drivers: net: wimax: i2400m: i2400m-usb: Use time_after for time comparison
DECnet: Use container_of() for embedded struct
Revert "ipv4: restore rt->fi for reference counting"
net: mdio-mux: bcm-iproc: call mdiobus_free() in error path
net: ethernet: ti: cpsw: adjust cpsw fifos depth for fullduplex flow control
ipv6: reorder ip6_route_dev_notifier after ipv6_dev_notf
net: cdc_ncm: Fix TX zero padding
stmmac: pci: split out common_default_data() helper
stmmac: pci: RX queue routing configuration
stmmac: pci: TX and RX queue priority configuration
stmmac: pci: set default number of rx and tx queues
...
Pull misc vfs updates from Al Viro:
"Assorted bits and pieces from various people. No common topic in this
pile, sorry"
* 'work.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
fs/affs: add rename exchange
fs/affs: add rename2 to prepare multiple methods
Make stat/lstat/fstatat pass AT_NO_AUTOMOUNT to vfs_statx()
fs: don't set *REFERENCED on single use objects
fs: compat: Remove warning from COMPATIBLE_IOCTL
remove pointless extern of atime_need_update_rcu()
fs: completely ignore unknown open flags
fs: add a VALID_OPEN_FLAGS
fs: remove _submit_bh()
fs: constify tree_descr arrays passed to simple_fill_super()
fs: drop duplicate header percpu-rwsem.h
fs/affs: bugfix: Write files greater than page size on OFS
fs/affs: bugfix: enable writes on OFS disks
fs/affs: remove node generation check
fs/affs: import amigaffs.h
fs/affs: bugfix: make symbolic links work again
__vmalloc* allows users to provide gfp flags for the underlying
allocation. This API is quite popular
$ git grep "=[[:space:]]__vmalloc\|return[[:space:]]*__vmalloc" | wc -l
77
The only problem is that many people are not aware that they really want
to give __GFP_HIGHMEM along with other flags because there is really no
reason to consume precious lowmemory on CONFIG_HIGHMEM systems for pages
which are mapped to the kernel vmalloc space. About half of users don't
use this flag, though. This signals that we make the API unnecessarily
too complex.
This patch simply uses __GFP_HIGHMEM implicitly when allocating pages to
be mapped to the vmalloc space. Current users which add __GFP_HIGHMEM
are simplified and drop the flag.
Link: http://lkml.kernel.org/r/20170307141020.29107-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Matthew Wilcox <mawilcox@microsoft.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Cristopher Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The patch fixes two things at once:
1) It checks the env->allow_ptr_leaks and only prints the map address to
the log if we have the privileges to do so, otherwise it just dumps 0
as we would when kptr_restrict is enabled on %pK. Given the latter is
off by default and not every distro sets it, I don't want to rely on
this, hence the 0 by default for unprivileged.
2) Printing of ldimm64 in the verifier log is currently broken in that
we don't print the full immediate, but only the 32 bit part of the
first insn part for ldimm64. Thus, fix this up as well; it's okay to
access, since we verified all ldimm64 earlier already (including just
constants) through replace_map_fd_with_map_ptr().
Fixes: 1be7f75d16 ("bpf: enable non-root eBPF programs")
Fixes: cbd3570086 ("bpf: verifier (add ability to receive verification log)")
Reported-by: Jann Horn <jannh@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
llvm 4.0 and above generates the code like below:
....
440: (b7) r1 = 15
441: (05) goto pc+73
515: (79) r6 = *(u64 *)(r10 -152)
516: (bf) r7 = r10
517: (07) r7 += -112
518: (bf) r2 = r7
519: (0f) r2 += r1
520: (71) r1 = *(u8 *)(r8 +0)
521: (73) *(u8 *)(r2 +45) = r1
....
and the verifier complains "R2 invalid mem access 'inv'" for insn #521.
This is because verifier marks register r2 as unknown value after #519
where r2 is a stack pointer and r1 holds a constant value.
Teach verifier to recognize "stack_ptr + imm" and
"stack_ptr + reg with const val" as valid stack_ptr with new offset.
Signed-off-by: Yonghong Song <yhs@fb.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Hannes rightfully spotted that the bpf_lock doesn't need to be
irqsave variant. We never perform any such updates where this
would be necessary (neither right now nor in future), therefore
relax this further.
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
simple_fill_super() is passed an array of tree_descr structures which
describe the files to create in the filesystem's root directory. Since
these arrays are never modified intentionally, they should be 'const' so
that they are placed in .rodata and benefit from memory protection.
This patch updates the function signature and all users, and also
constifies tree_descr.name.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
When iterating through a map, we need to find a key that does not exist
in the map so map_get_next_key will give us the first key of the map.
This often requires a lot of guessing in production systems.
This patch makes map_get_next_key return the first key when the key
pointer in the parameter is NULL.
Signed-off-by: Teng Qin <qinteng@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that also the last in-tree user of the xdp_adjust_head bit has
been removed, we can remove the flag from struct bpf_prog altogether.
This, at the same time, also makes sure that any future driver for
XDP comes with bpf_xdp_adjust_head() support right away.
A rejection based on this flag would also mean that tail calls
couldn't be used with such driver as per c2002f9837 ("bpf: fix
checking xdp_adjust_head on tail calls") fix, thus lets not allow
for it in the first place.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
A function in kernel/bpf/syscall.c which got a bug fix in 'net'
was moved to kernel/bpf/verifier.c in 'net-next'.
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 17bedab272 ("bpf: xdp: Allow head adjustment in XDP prog")
added the xdp_adjust_head bit to the BPF prog in order to tell drivers
that the program that is to be attached requires support for the XDP
bpf_xdp_adjust_head() helper such that drivers not supporting this
helper can reject the program. There are also drivers that do support
the helper, but need to check for xdp_adjust_head bit in order to move
packet metadata prepended by the firmware away for making headroom.
For these cases, the current check for xdp_adjust_head bit is insufficient
since there can be cases where the program itself does not use the
bpf_xdp_adjust_head() helper, but tail calls into another program that
uses bpf_xdp_adjust_head(). As such, the xdp_adjust_head bit is still
set to 0. Since the first program has no control over which program it
calls into, we need to assume that bpf_xdp_adjust_head() helper is used
upon tail calls. Thus, for the very same reasons in cb_access, set the
xdp_adjust_head bit to 1 when the main program uses tail calls.
Fixes: 17bedab272 ("bpf: xdp: Allow head adjustment in XDP prog")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Cc: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit ff936a04e5 ("bpf: fix cb access in socket filter programs")
added a fix for socket filter programs such that in i) AF_PACKET the
20 bytes of skb->cb[] area gets zeroed before use in order to not leak
data, and ii) socket filter programs attached to TCP/UDP sockets need
to save/restore these 20 bytes since they are also used by protocol
layers at that time.
The problem is that bpf_prog_run_save_cb() and bpf_prog_run_clear_cb()
only look at the actual attached program to determine whether to zero
or save/restore the skb->cb[] parts. There can be cases where the
actual attached program does not access the skb->cb[], but the program
tail calls into another program which does access this area. In such
a case, the zero or save/restore is currently not performed.
Since the programs we tail call into are unknown at verification time
and can dynamically change, we need to assume that whenever the attached
program performs a tail call, that later programs could access the
skb->cb[], and therefore we need to always set cb_access to 1.
Fixes: ff936a04e5 ("bpf: fix cb access in socket filter programs")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
After doing map_perf_test with a much bigger
BPF_F_NO_COMMON_LRU map, the perf report shows a
lot of time spent in rotating the inactive list (i.e.
__bpf_lru_list_rotate_inactive):
> map_perf_test 32 8 10000 1000000 | awk '{sum += $3}END{print sum}'
19644783 (19M/s)
> map_perf_test 32 8 10000000 10000000 | awk '{sum += $3}END{print sum}'
6283930 (6.28M/s)
By inactive, it usually means the element is not in cache. Hence,
there is a need to tune the PERCPU_NR_SCANS value.
This patch finds a better number of elements to
scan during each list rotation. The PERCPU_NR_SCANS (which
is defined the same as PERCPU_FREE_TARGET) decreases
from 16 elements to 4 elements. This change only
affects the BPF_F_NO_COMMON_LRU map.
The test_lru_dist does not show meaningful difference
between 16 and 4. Our production L4 load balancer which uses
the LRU map for conntrack-ing also shows little change in cache
hit rate. Since both benchmark and production data show no
cache-hit difference, PERCPU_NR_SCANS is lowered from 16 to 4.
We can consider making it configurable if we find a usecase
later that shows another value works better and/or use
a different rotation strategy.
After this change:
> map_perf_test 32 8 10000000 10000000 | awk '{sum += $3}END{print sum}'
9240324 (9.2M/s)
i.e. 6.28M/s -> 9.2M/s
The test_lru_dist has not shown meaningful difference:
> test_lru_dist zipf.100k.a1_01.out 4000 1:
nr_misses: 31575 (Before) vs 31566 (After)
> test_lru_dist zipf.100k.a0_01.out 40000 1
nr_misses: 67036 (Before) vs 67031 (After)
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts were simply overlapping changes. In the net/ipv4/route.c
case the code had simply moved around a little bit and the same fix
was made in both 'net' and 'net-next'.
In the net/sched/sch_generic.c case a fix in 'net' happened at
the same time that a new argument was added to qdisc_hash_add().
Signed-off-by: David S. Miller <davem@davemloft.net>
BPF helper functions access socket fields through skb->sk. This is not
set in ingress cgroup and socket filters. The association is only made
in skb_set_owner_r once the filter has accepted the packet. Sk is
available as socket lookup has taken place.
Temporarily set skb->sk to sk in these cases.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
There's no need to have struct bpf_map_type_list since
it just contains a list_head, the type, and the ops
pointer. Since the types are densely packed and not
actually dynamically registered, it's much easier and
smaller to have an array of type->ops pointer. Also
initialize this array statically to remove code needed
to initialize it.
In order to save duplicating the list, move it to the
types header file added by the previous patch and
include it in the same fashion.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
There's no need to have struct bpf_prog_type_list since
it just contains a list_head, the type, and the ops
pointer. Since the types are densely packed and not
actually dynamically registered, it's much easier and
smaller to have an array of type->ops pointer. Also
initialize this array statically to remove code needed
to initialize it.
In order to save duplicating the list, move it to a new
header file and include it in the places needing it.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
It took me quite some time to figure out how this was linked,
so in order to save the next person the effort of finding it
add a comment in __bpf_prog_run() that indicates what exactly
determines that a program can access the ctx == skb.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Mostly simple cases of overlapping changes (adding code nearby,
a function whose name changes, for example).
Signed-off-by: David S. Miller <davem@davemloft.net>
development and testing of networking bpf programs is quite cumbersome.
Despite availability of user space bpf interpreters the kernel is
the ultimate authority and execution environment.
Current test frameworks for TC include creation of netns, veth,
qdiscs and use of various packet generators just to test functionality
of a bpf program. XDP testing is even more complicated, since
qemu needs to be started with gro/gso disabled and precise queue
configuration, transferring of xdp program from host into guest,
attaching to virtio/eth0 and generating traffic from the host
while capturing the results from the guest.
Moreover analyzing performance bottlenecks in XDP program is
impossible in virtio environment, since cost of running the program
is tiny comparing to the overhead of virtio packet processing,
so performance testing can only be done on physical nic
with another server generating traffic.
Furthermore ongoing changes to user space control plane of production
applications cannot be run on the test servers leaving bpf programs
stubbed out for testing.
Last but not least, the upstream llvm changes are validated by the bpf
backend testsuite which has no ability to test the code generated.
To improve this situation introduce BPF_PROG_TEST_RUN command
to test and performance benchmark bpf programs.
Joint work with Daniel Borkmann.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, the verifier doesn't reject unaligned access for map_value_adj
register types. Commit 484611357c ("bpf: allow access into map value
arrays") added logic to check_ptr_alignment() extending it from PTR_TO_PACKET
to also PTR_TO_MAP_VALUE_ADJ, but for PTR_TO_MAP_VALUE_ADJ no enforcement
is in place, because reg->id for PTR_TO_MAP_VALUE_ADJ reg types is never
non-zero, meaning, we can cause BPF_H/_W/_DW-based unaligned access for
architectures not supporting efficient unaligned access, and thus worst
case could raise exceptions on some archs that are unable to correct the
unaligned access or perform a different memory access to the actual
requested one and such.
i) Unaligned load with !CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS
on r0 (map_value_adj):
0: (bf) r2 = r10
1: (07) r2 += -8
2: (7a) *(u64 *)(r2 +0) = 0
3: (18) r1 = 0x42533a00
5: (85) call bpf_map_lookup_elem#1
6: (15) if r0 == 0x0 goto pc+11
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
7: (61) r1 = *(u32 *)(r0 +0)
8: (35) if r1 >= 0xb goto pc+9
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R1=inv,min_value=0,max_value=10 R10=fp
9: (07) r0 += 3
10: (79) r7 = *(u64 *)(r0 +0)
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R1=inv,min_value=0,max_value=10 R10=fp
11: (79) r7 = *(u64 *)(r0 +2)
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R1=inv,min_value=0,max_value=10 R7=inv R10=fp
[...]
ii) Unaligned store with !CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS
on r0 (map_value_adj):
0: (bf) r2 = r10
1: (07) r2 += -8
2: (7a) *(u64 *)(r2 +0) = 0
3: (18) r1 = 0x4df16a00
5: (85) call bpf_map_lookup_elem#1
6: (15) if r0 == 0x0 goto pc+19
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
7: (07) r0 += 3
8: (7a) *(u64 *)(r0 +0) = 42
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R10=fp
9: (7a) *(u64 *)(r0 +2) = 43
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R10=fp
10: (7a) *(u64 *)(r0 -2) = 44
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R10=fp
[...]
For the PTR_TO_PACKET type, reg->id is initially zero when skb->data
was fetched, it later receives a reg->id from env->id_gen generator
once another register with UNKNOWN_VALUE type was added to it via
check_packet_ptr_add(). The purpose of this reg->id is twofold: i) it
is used in find_good_pkt_pointers() for setting the allowed access
range for regs with PTR_TO_PACKET of same id once verifier matched
on data/data_end tests, and ii) for check_ptr_alignment() to determine
that when not having efficient unaligned access and register with
UNKNOWN_VALUE was added to PTR_TO_PACKET, that we're only allowed
to access the content bytewise due to unknown unalignment. reg->id
was never intended for PTR_TO_MAP_VALUE{,_ADJ} types and thus is
always zero, the only marking is in PTR_TO_MAP_VALUE_OR_NULL that
was added after 484611357c via 57a09bf0a4 ("bpf: Detect identical
PTR_TO_MAP_VALUE_OR_NULL registers"). Above tests will fail for
non-root environment due to prohibited pointer arithmetic.
The fix splits register-type specific checks into their own helper
instead of keeping them combined, so we don't run into a similar
issue in future once we extend check_ptr_alignment() further and
forget to add reg->type checks for some of the checks.
Fixes: 484611357c ("bpf: allow access into map value arrays")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
While looking into map_value_adj, I noticed that alu operations
directly on the map_value() resp. map_value_adj() register (any
alu operation on a map_value() register will turn it into a
map_value_adj() typed register) are not sufficiently protected
against some of the operations. Two non-exhaustive examples are
provided that the verifier needs to reject:
i) BPF_AND on r0 (map_value_adj):
0: (bf) r2 = r10
1: (07) r2 += -8
2: (7a) *(u64 *)(r2 +0) = 0
3: (18) r1 = 0xbf842a00
5: (85) call bpf_map_lookup_elem#1
6: (15) if r0 == 0x0 goto pc+2
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
7: (57) r0 &= 8
8: (7a) *(u64 *)(r0 +0) = 22
R0=map_value_adj(ks=8,vs=48,id=0),min_value=0,max_value=8 R10=fp
9: (95) exit
from 6 to 9: R0=inv,min_value=0,max_value=0 R10=fp
9: (95) exit
processed 10 insns
ii) BPF_ADD in 32 bit mode on r0 (map_value_adj):
0: (bf) r2 = r10
1: (07) r2 += -8
2: (7a) *(u64 *)(r2 +0) = 0
3: (18) r1 = 0xc24eee00
5: (85) call bpf_map_lookup_elem#1
6: (15) if r0 == 0x0 goto pc+2
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
7: (04) (u32) r0 += (u32) 0
8: (7a) *(u64 *)(r0 +0) = 22
R0=map_value_adj(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
9: (95) exit
from 6 to 9: R0=inv,min_value=0,max_value=0 R10=fp
9: (95) exit
processed 10 insns
Issue is, while min_value / max_value boundaries for the access
are adjusted appropriately, we change the pointer value in a way
that cannot be sufficiently tracked anymore from its origin.
Operations like BPF_{AND,OR,DIV,MUL,etc} on a destination register
that is PTR_TO_MAP_VALUE{,_ADJ} was probably unintended, in fact,
all the test cases coming with 484611357c ("bpf: allow access
into map value arrays") perform BPF_ADD only on the destination
register that is PTR_TO_MAP_VALUE_ADJ.
Only for UNKNOWN_VALUE register types such operations make sense,
f.e. with unknown memory content fetched initially from a constant
offset from the map value memory into a register. That register is
then later tested against lower / upper bounds, so that the verifier
can then do the tracking of min_value / max_value, and properly
check once that UNKNOWN_VALUE register is added to the destination
register with type PTR_TO_MAP_VALUE{,_ADJ}. This is also what the
original use-case is solving. Note, tracking on what is being
added is done through adjust_reg_min_max_vals() and later access
to the map value enforced with these boundaries and the given offset
from the insn through check_map_access_adj().
Tests will fail for non-root environment due to prohibited pointer
arithmetic, in particular in check_alu_op(), we bail out on the
is_pointer_value() check on the dst_reg (which is false in root
case as we allow for pointer arithmetic via env->allow_ptr_leaks).
Similarly to PTR_TO_PACKET, one way to fix it is to restrict the
allowed operations on PTR_TO_MAP_VALUE{,_ADJ} registers to 64 bit
mode BPF_ADD. The test_verifier suite runs fine after the patch
and it also rejects mentioned test cases.
Fixes: 484611357c ("bpf: allow access into map value arrays")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
llvm can optimize the 'if (ptr > data_end)' checks to be in the order
slightly different than the original C code which will confuse verifier.
Like:
if (ptr + 16 > data_end)
return TC_ACT_SHOT;
// may be followed by
if (ptr + 14 > data_end)
return TC_ACT_SHOT;
while llvm can see that 'ptr' is valid for all 16 bytes,
the verifier could not.
Fix verifier logic to account for such case and add a test.
Reported-by: Huapeng Zhou <hzhou@fb.com>
Fixes: 969bf05eb3 ("bpf: direct packet access")
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/broadcom/genet/bcmmii.c
drivers/net/hyperv/netvsc.c
kernel/bpf/hashtab.c
Almost entirely overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds hash of maps support (hashmap->bpf_map).
BPF_MAP_TYPE_HASH_OF_MAPS is added.
A map-in-map contains a pointer to another map and lets call
this pointer 'inner_map_ptr'.
Notes on deleting inner_map_ptr from a hash map:
1. For BPF_F_NO_PREALLOC map-in-map, when deleting
an inner_map_ptr, the htab_elem itself will go through
a rcu grace period and the inner_map_ptr resides
in the htab_elem.
2. For pre-allocated htab_elem (!BPF_F_NO_PREALLOC),
when deleting an inner_map_ptr, the htab_elem may
get reused immediately. This situation is similar
to the existing prealloc-ated use cases.
However, the bpf_map_fd_put_ptr() calls bpf_map_put() which calls
inner_map->ops->map_free(inner_map) which will go
through a rcu grace period (i.e. all bpf_map's map_free
currently goes through a rcu grace period). Hence,
the inner_map_ptr is still safe for the rcu reader side.
This patch also includes BPF_MAP_TYPE_HASH_OF_MAPS to the
check_map_prealloc() in the verifier. preallocation is a
must for BPF_PROG_TYPE_PERF_EVENT. Hence, even we don't expect
heavy updates to map-in-map, enforcing BPF_F_NO_PREALLOC for map-in-map
is impossible without disallowing BPF_PROG_TYPE_PERF_EVENT from using
map-in-map first.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds a few helper funcs to enable map-in-map
support (i.e. outer_map->inner_map). The first outer_map type
BPF_MAP_TYPE_ARRAY_OF_MAPS is also added in this patch.
The next patch will introduce a hash of maps type.
Any bpf map type can be acted as an inner_map. The exception
is BPF_MAP_TYPE_PROG_ARRAY because the extra level of
indirection makes it harder to verify the owner_prog_type
and owner_jited.
Multi-level map-in-map is not supported (i.e. map->map is ok
but not map->map->map).
When adding an inner_map to an outer_map, it currently checks the
map_type, key_size, value_size, map_flags, max_entries and ops.
The verifier also uses those map's properties to do static analysis.
map_flags is needed because we need to ensure BPF_PROG_TYPE_PERF_EVENT
is using a preallocated hashtab for the inner_hash also. ops and
max_entries are needed to generate inlined map-lookup instructions.
For simplicity reason, a simple '==' test is used for both map_flags
and max_entries. The equality of ops is implied by the equality of
map_type.
During outer_map creation time, an inner_map_fd is needed to create an
outer_map. However, the inner_map_fd's life time does not depend on the
outer_map. The inner_map_fd is merely used to initialize
the inner_map_meta of the outer_map.
Also, for the outer_map:
* It allows element update and delete from syscall
* It allows element lookup from bpf_prog
The above is similar to the current fd_array pattern.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix in verifier:
For the same bpf_map_lookup_elem() instruction (i.e. "call 1"),
a broken case is "a different type of map could be used for the
same lookup instruction". For example, an array in one case and a
hashmap in another. We have to resort to the old dynamic call behavior
in this case. The fix is to check for collision on insn_aux->map_ptr.
If there is collision, don't inline the map lookup.
Please see the "do_reg_lookup()" in test_map_in_map_kern.c in the later
patch for how-to trigger the above case.
Simplifications on array_map_gen_lookup():
1. Calculate elem_size from map->value_size. It removes the
need for 'struct bpf_array' which makes the later map-in-map
implementation easier.
2. Remove the 'elem_size == 1' test
Fixes: 81ed18ab30 ("bpf: add helper inlining infra and optimize map_array lookup")
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
In both kmalloc and prealloc mode the bpf_map_update_elem() is using
per-cpu extra_elems to do atomic update when the map is full.
There are two issues with it. The logic can be misused, since it allows
max_entries+num_cpus elements to be present in the map. And alloc_extra_elems()
at map creation time can fail percpu alloc for large map values with a warn:
WARNING: CPU: 3 PID: 2752 at ../mm/percpu.c:892 pcpu_alloc+0x119/0xa60
illegal size (32824) or align (8) for percpu allocation
The fixes for both of these issues are different for kmalloc and prealloc modes.
For prealloc mode allocate extra num_possible_cpus elements and store
their pointers into extra_elems array instead of actual elements.
Hence we can use these hidden(spare) elements not only when the map is full
but during bpf_map_update_elem() that replaces existing element too.
That also improves performance, since pcpu_freelist_pop/push is avoided.
Unfortunately this approach cannot be used for kmalloc mode which needs
to kfree elements after rcu grace period. Therefore switch it back to normal
kmalloc even when full and old element exists like it was prior to
commit 6c90598174 ("bpf: pre-allocate hash map elements").
Add tests to check for over max_entries and large map values.
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Fixes: 6c90598174 ("bpf: pre-allocate hash map elements")
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Optimize:
bpf_call
bpf_map_lookup_elem
map->ops->map_lookup_elem
htab_map_lookup_elem
__htab_map_lookup_elem
into:
bpf_call
__htab_map_lookup_elem
to improve performance of JITed programs.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Optimize bpf_call -> bpf_map_lookup_elem() -> array_map_lookup_elem()
into a sequence of bpf instructions.
When JIT is on the sequence of bpf instructions is the sequence
of native cpu instructions with significantly faster performance
than indirect call and two function's prologue/epilogue.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
convert_ctx_accesses() replaces single bpf instruction with a set of
instructions. Adjust corresponding insn_aux_data while patching.
It's needed to make sure subsequent 'for(all insn)' loops
have matching insn and insn_aux_data.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
reduce indent and make it iterate over instructions similar to
convert_ctx_accesses(). Also convert hard BUG_ON into soft verifier error.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
no functional change.
move fixup_bpf_calls() to verifier.c
it's being refactored in the next patch
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
when all map elements are pre-allocated one cpu can delete and reuse htab_elem
while another cpu is still walking the hlist. In such case the lookup may
miss the element. Convert hlist to hlist_nulls to avoid such scenario.
When bucket lock is taken there is no need to take such precautions,
so only convert map_lookup and map_get_next to nulls.
The race window is extremely small and only reproducible with explicit
udelay() inside lookup_nulls_elem_raw()
Similar to hlist add hlist_nulls_for_each_entry_safe() and
hlist_nulls_entry_safe() helpers.
Fixes: 6c90598174 ("bpf: pre-allocate hash map elements")
Reported-by: Jonathan Perry <jonperry@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
when htab_elem is removed from the bucket list the htab_elem.hash_node.next
field should not be overridden too early otherwise we have a tiny race window
between lookup and delete.
The bug was discovered by manual code analysis and reproducible
only with explicit udelay() in lookup_elem_raw().
Fixes: 6c90598174 ("bpf: pre-allocate hash map elements")
Reported-by: Jonathan Perry <jonperry@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
