In bpf_tcp_sendmsg() the sk_alloc_sg() may fail. In the case of
ENOMEM, it may also mean that we've partially filled the scatterlist
entries with pages. Later jumping to sk_stream_wait_memory()
we could further fail with an error for several reasons, however
we miss to call free_start_sg() if the local sk_msg_buff was used.

Fixes: 4f738adb ("bpf: create tcp_bpf_ulp allowing BPF to monitor socket TX/RX data")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

While working on bpf_tcp_sendmsg() code, I noticed that when a
sk->sk_err is set we error out with err = sk->sk_err. However
this is problematic since sk->sk_err is a positive error value
and therefore we will neither go into sk_stream_error() nor will
we report an error back to user space. I had this case with EPIPE
and user space was thinking sendmsg() succeeded since EPIPE is
a positive value, thinking we submitted 32 bytes. Fix it by
negating the sk->sk_err value.

Fixes: 4f738adb ("bpf: create tcp_bpf_ulp allowing BPF to monitor socket TX/RX data")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

In commit

  'bpf: bpf_compute_data uses incorrect cb structure' (8108a775)

we added the routine bpf_compute_data_end_sk_skb() to compute the
correct data_end values, but this has since been lost. In kernel
v4.14 this was correct and the above patch was applied in it
entirety. Then when v4.14 was merged into v4.15-rc1 net-next tree
we lost the piece that renamed bpf_compute_data_pointers to the
new function bpf_compute_data_end_sk_skb. This was done here,

e1ea2f98 ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net")

When it conflicted with the following rename patch,

6aaae2b6 ("bpf: rename bpf_compute_data_end into bpf_compute_data_pointers")

Finally, after a refactor I thought even the function
bpf_compute_data_end_sk_skb() was no longer needed and it was
erroneously removed.

However, we never reverted the sk_skb_convert_ctx_access() usage of
tcp_skb_cb which had been committed and survived the merge conflict.
Here we fix this by adding back the helper and *_data_end_sk_skb()
usage. Using the bpf_skc_data_end mapping is not correct because it
expects a qdisc_skb_cb object but at the sock layer this is not the
case. Even though it happens to work here because we don't overwrite
any data in-use at the socket layer and the cb structure is cleared
later this has potential to create some subtle issues. But, even
more concretely the filter.c access check uses tcp_skb_cb.

And by some act of chance though,

struct bpf_skb_data_end {
        struct qdisc_skb_cb        qdisc_cb;             /*     0    28 */

        /* XXX 4 bytes hole, try to pack */

        void *                     data_meta;            /*    32     8 */
        void *                     data_end;             /*    40     8 */

        /* size: 48, cachelines: 1, members: 3 */
        /* sum members: 44, holes: 1, sum holes: 4 */
        /* last cacheline: 48 bytes */
};

and then tcp_skb_cb,

struct tcp_skb_cb {
	[...]
                struct {
                        __u32      flags;                /*    24     4 */
                        struct sock * sk_redir;          /*    32     8 */
                        void *     data_end;             /*    40     8 */
                } bpf;                                   /*          24 */
        };

So when we use offset_of() to track down the byte offset we get 40 in
either case and everything continues to work. Fix this mess and use
correct structures its unclear how long this might actually work for
until someone moves the structs around.
Reported-by: NMartin KaFai Lau <kafai@fb.com>
Fixes: e1ea2f98 ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net")
Fixes: 6aaae2b6 ("bpf: rename bpf_compute_data_end into bpf_compute_data_pointers")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

Currently, when a sock is closed and the bpf_tcp_close() callback is
used we remove memory but do not free the skb. Call consume_skb() if
the skb is attached to the buffer.

Reported-by: syzbot+d464d2c20c717ef5a6a8@syzkaller.appspotmail.com
Fixes: 1aa12bdf ("bpf: sockmap, add sock close() hook to remove socks")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

After latest lock updates there is no longer anything preventing a
close and recvmsg call running in parallel. Additionally, we can
race update with close if we close a socket and simultaneously update
if via the BPF userspace API (note the cgroup ops are already run
with sock_lock held).

To resolve this take sock_lock in close and update paths.

Reported-by: syzbot+b680e42077a0d7c9a0c4@syzkaller.appspotmail.com
Fixes: e9db4ef6 ("bpf: sockhash fix omitted bucket lock in sock_close")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

This removes locking from readers of RCU hash table. Its not
necessary.

Fixes: 81110384 ("bpf: sockmap, add hash map support")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

The current code, in the error path of sock_hash_ctx_update_elem,
checks if the sock has a psock in the user data and if so decrements
the reference count of the psock. However, if the error happens early
in the error path we may have never incremented the psock reference
count and if the psock exists because the sock is in another map then
we may inadvertently decrement the reference count.

Fix this by making the error path only call smap_release_sock if the
error happens after the increment.

Reported-by: syzbot+d464d2c20c717ef5a6a8@syzkaller.appspotmail.com
Fixes: 81110384 ("bpf: sockmap, add hash map support")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

Add map_release_uref pointer to hashmap ops. This was dropped when
original sockhash code was ported into bpf-next before initial
commit.

Fixes: 81110384 ("bpf: sockmap, add hash map support")
Acked-by: NMartin KaFai Lau <kafai@fb.com>
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

First the sk_callback_lock() was being used to protect both the
sock callback hooks and the psock->maps list. This got overly
convoluted after the addition of sockhash (in sockmap it made
some sense because masp and callbacks were tightly coupled) so
lets split out a specific lock for maps and only use the callback
lock for its intended purpose. This fixes a couple cases where
we missed using maps lock when it was in fact needed. Also this
makes it easier to follow the code because now we can put the
locking closer to the actual code its serializing.

Next, in sock_hash_delete_elem() the pattern was as follows,

  sock_hash_delete_elem()
     [...]
     spin_lock(bucket_lock)
     l = lookup_elem_raw()
     if (l)
        hlist_del_rcu()
        write_lock(sk_callback_lock)
         .... destroy psock ...
        write_unlock(sk_callback_lock)
     spin_unlock(bucket_lock)

The ordering is necessary because we only know the {p}sock after
dereferencing the hash table which we can't do unless we have the
bucket lock held. Once we have the bucket lock and the psock element
it is deleted from the hashmap to ensure any other path doing a lookup
will fail. Finally, the refcnt is decremented and if zero the psock
is destroyed.

In parallel with the above (or free'ing the map) a tcp close event
may trigger tcp_close(). Which at the moment omits the bucket lock
altogether (oops!) where the flow looks like this,

  bpf_tcp_close()
     [...]
     write_lock(sk_callback_lock)
     for each psock->maps // list of maps this sock is part of
         hlist_del_rcu(ref_hash_node);
         .... destroy psock ...
     write_unlock(sk_callback_lock)

Obviously, and demonstrated by syzbot, this is broken because
we can have multiple threads deleting entries via hlist_del_rcu().

To fix this we might be tempted to wrap the hlist operation in a
bucket lock but that would create a lock inversion problem. In
summary to follow locking rules the psocks maps list needs the
sk_callback_lock (after this patch maps_lock) but we need the bucket
lock to do the hlist_del_rcu.

To resolve the lock inversion problem pop the head of the maps list
repeatedly and remove the reference until no more are left. If a
delete happens in parallel from the BPF API that is OK as well because
it will do a similar action, lookup the lock in the map/hash, delete
it from the map/hash, and dec the refcnt. We check for this case
before doing a destroy on the psock to ensure we don't have two
threads tearing down a psock. The new logic is as follows,

  bpf_tcp_close()
  e = psock_map_pop(psock->maps) // done with map lock
  bucket_lock() // lock hash list bucket
  l = lookup_elem_raw(head, hash, key, key_size);
  if (l) {
     //only get here if elmnt was not already removed
     hlist_del_rcu()
     ... destroy psock...
  }
  bucket_unlock()

And finally for all the above to work add missing locking around  map
operations per above. Then add RCU annotations and use
rcu_dereference/rcu_assign_pointer to manage values relying on RCU so
that the object is not free'd from sock_hash_free() while it is being
referenced in bpf_tcp_close().

Reported-by: syzbot+0ce137753c78f7b6acc1@syzkaller.appspotmail.com
Fixes: 81110384 ("bpf: sockmap, add hash map support")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

If a hashmap is free'd with open socks it removes the reference to
the hash entry from the psock. If that is the last reference to the
psock then it will also be free'd by the reference counting logic.
However the current logic that removes the hash reference from the
list of references is broken. In smap_list_remove() we first check
if the sockmap entry matches and then check if the hashmap entry
matches. But, the sockmap entry sill always match because its NULL in
this case which causes the first entry to be removed from the list.
If this is always the "right" entry (because the user adds/removes
entries in order) then everything is OK but otherwise a subsequent
bpf_tcp_close() may reference a free'd object.

To fix this create two list handlers one for sockmap and one for
sockhash.

Reported-by: syzbot+0ce137753c78f7b6acc1@syzkaller.appspotmail.com
Fixes: 81110384 ("bpf: sockmap, add hash map support")
Acked-by: NMartin KaFai Lau <kafai@fb.com>
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

This fixes a crash where we assign tcp_prot to IPv6 sockets instead
of tcpv6_prot.

Previously we overwrote the sk->prot field with tcp_prot even in the
AF_INET6 case. This patch ensures the correct tcp_prot and tcpv6_prot
are used.

Tested with 'netserver -6' and 'netperf -H [IPv6]' as well as
'netperf -H [IPv4]'. The ESTABLISHED check resolves the previously
crashing case here.

Fixes: 174a79ff ("bpf: sockmap with sk redirect support")
Reported-by: syzbot+5c063698bdbfac19f363@syzkaller.appspotmail.com
Acked-by: NMartin KaFai Lau <kafai@fb.com>
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NWei Wang <weiwan@google.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

If the kernel is compiled with CONFIG_CGROUP_BPF not enabled, it is not
possible to attach, detach or query IR BPF programs to /dev/lircN devices,
making them impossible to use. For embedded devices, it should be possible
to use IR decoding without cgroups or CONFIG_CGROUP_BPF enabled.

This change requires some refactoring, since bpf_prog_{attach,detach,query}
functions are now always compiled, but their code paths for cgroups need
moving out. Rather than a #ifdef CONFIG_CGROUP_BPF in kernel/bpf/syscall.c,
moving them to kernel/bpf/cgroup.c and kernel/bpf/sockmap.c does not
require #ifdefs since that is already conditionally compiled.

Fixes: f4364dcf ("media: rc: introduce BPF_PROG_LIRC_MODE2")
Signed-off-by: NSean Young <sean@mess.org>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Currently sk_msg programs only have access to the raw data. However,
it is often useful when building policies to have the policies specific
to the socket endpoint. This allows using the socket tuple as input
into filters, etc.

This patch adds ctx access to the sock fields.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Acked-by: NMartin KaFai Lau <kafai@fb.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

In the sockmap design BPF programs (SK_SKB_STREAM_PARSER,
SK_SKB_STREAM_VERDICT and SK_MSG_VERDICT) are attached to the sockmap
map type and when a sock is added to the map the programs are used by
the socket. However, sockmap updates from both userspace and BPF
programs can happen concurrently with the attach and detach of these
programs.

To resolve this we use the bpf_prog_inc_not_zero and a READ_ONCE()
primitive to ensure the program pointer is not refeched and
possibly NULL'd before the refcnt increment. This happens inside
a RCU critical section so although the pointer reference in the map
object may be NULL (by a concurrent detach operation) the reference
from READ_ONCE will not be free'd until after grace period. This
ensures the object returned by READ_ONCE() is valid through the
RCU criticl section and safe to use as long as we "know" it may
be free'd shortly.

Daniel spotted a case in the sock update API where instead of using
the READ_ONCE() program reference we used the pointer from the
original map, stab->bpf_{verdict|parse|txmsg}. The problem with this
is the logic checks the object returned from the READ_ONCE() is not
NULL and then tries to reference the object again but using the
above map pointer, which may have already been NULL'd by a parallel
detach operation. If this happened bpf_porg_inc_not_zero could
dereference a NULL pointer.

Fix this by using variable returned by READ_ONCE() that is checked
for NULL.

Fixes: 2f857d04 ("bpf: sockmap, remove STRPARSER map_flags and add multi-map support")
Reported-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Acked-by: NMartin KaFai Lau <kafai@fb.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

If the user were to only attach one of the parse or verdict programs
then it is possible a subsequent sockmap update could incorrectly
decrement the refcnt on the program. This happens because in the
rollback logic, after an error, we have to decrement the program
reference count when its been incremented. However, we only increment
the program reference count if the user has both a verdict and a
parse program. The reason for this is because, at least at the
moment, both are required for any one to be meaningful. The problem
fixed here is in the rollback path we decrement the program refcnt
even if only one existing. But we never incremented the refcnt in
the first place creating an imbalance.

This patch fixes the error path to handle this case.

Fixes: 2f857d04 ("bpf: sockmap, remove STRPARSER map_flags and add multi-map support")
Reported-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Acked-by: NMartin KaFai Lau <kafai@fb.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

`e' is being freed twice.

Fix this by removing one of the kfree() calls.

Addresses-Coverity-ID: 1468983 ("Double free")
Fixes: 81110384 ("bpf: sockmap, add hash map support")
Signed-off-by: NGustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

There is a potential execution path in which variable err is
returned without being properly initialized previously.

Fix this by initializing variable err to 0.

Addresses-Coverity-ID: 1468964 ("Uninitialized scalar variable")
Fixes: e5cd3abc ("bpf: sockmap, refactor sockmap routines to work with hashmap")
Signed-off-by: NGustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

When an error happens in the update sockmap element logic also pass
the err up to the user.

Fixes: e5cd3abc ("bpf: sockmap, refactor sockmap routines to work with hashmap")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

syzbot reported a kernel warning below:
  WARNING: CPU: 0 PID: 4499 at mm/slab_common.c:996 kmalloc_slab+0x56/0x70 mm/slab_common.c:996
  Kernel panic - not syncing: panic_on_warn set ...

  CPU: 0 PID: 4499 Comm: syz-executor050 Not tainted 4.17.0-rc3+ #9
  Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
  Call Trace:
   __dump_stack lib/dump_stack.c:77 [inline]
   dump_stack+0x1b9/0x294 lib/dump_stack.c:113
   panic+0x22f/0x4de kernel/panic.c:184
   __warn.cold.8+0x163/0x1b3 kernel/panic.c:536
   report_bug+0x252/0x2d0 lib/bug.c:186
   fixup_bug arch/x86/kernel/traps.c:178 [inline]
   do_error_trap+0x1de/0x490 arch/x86/kernel/traps.c:296
   do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:315
   invalid_op+0x14/0x20 arch/x86/entry/entry_64.S:992
  RIP: 0010:kmalloc_slab+0x56/0x70 mm/slab_common.c:996
  RSP: 0018:ffff8801d907fc58 EFLAGS: 00010246
  RAX: 0000000000000000 RBX: ffff8801aeecb280 RCX: ffffffff8185ebd7
  RDX: 0000000000000000 RSI: 0000000000000000 RDI: 00000000ffffffe1
  RBP: ffff8801d907fc58 R08: ffff8801adb5e1c0 R09: ffffed0035a84700
  R10: ffffed0035a84700 R11: ffff8801ad423803 R12: ffff8801aeecb280
  R13: 00000000fffffff4 R14: ffff8801ad891a00 R15: 00000000014200c0
   __do_kmalloc mm/slab.c:3713 [inline]
   __kmalloc+0x25/0x760 mm/slab.c:3727
   kmalloc include/linux/slab.h:517 [inline]
   map_get_next_key+0x24a/0x640 kernel/bpf/syscall.c:858
   __do_sys_bpf kernel/bpf/syscall.c:2131 [inline]
   __se_sys_bpf kernel/bpf/syscall.c:2096 [inline]
   __x64_sys_bpf+0x354/0x4f0 kernel/bpf/syscall.c:2096
   do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
   entry_SYSCALL_64_after_hwframe+0x49/0xbe

The test case is against sock hashmap with a key size 0xffffffe1.
Such a large key size will cause the below code in function
sock_hash_alloc() overflowing and produces a smaller elem_size,
hence map creation will be successful.
    htab->elem_size = sizeof(struct htab_elem) +
                      round_up(htab->map.key_size, 8);

Later, when map_get_next_key is called and kernel tries
to allocate the key unsuccessfully, it will issue
the above warning.

Similar to hashtab, ensure the key size is at most
MAX_BPF_STACK for a successful map creation.

Fixes: 81110384 ("bpf: sockmap, add hash map support")
Reported-by: syzbot+e4566d29080e7f3460ff@syzkaller.appspotmail.com
Signed-off-by: NYonghong Song <yhs@fb.com>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Sockmap is currently backed by an array and enforces keys to be
four bytes. This works well for many use cases and was originally
modeled after devmap which also uses four bytes keys. However,
this has become limiting in larger use cases where a hash would
be more appropriate. For example users may want to use the 5-tuple
of the socket as the lookup key.

To support this add hash support.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

This patch only refactors the existing sockmap code. This will allow
much of the psock initialization code path and bpf helper codes to
work for both sockmap bpf map types that are backed by an array, the
currently supported type, and the new hash backed bpf map type
sockhash.

Most the fallout comes from three changes,

  - Pushing bpf programs into an independent structure so we
    can use it from the htab struct in the next patch.
  - Generalizing helpers to use void *key instead of the hardcoded
    u32.
  - Instead of passing map/key through the metadata we now do
    the lookup inline. This avoids storing the key in the metadata
    which will be useful when keys can be longer than 4 bytes. We
    rename the sk pointers to sk_redir at this point as well to
    avoid any confusion between the current sk pointer and the
    redirect pointer sk_redir.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

When a redirect failure happens we release the buffers in-flight
without calling a sk_mem_uncharge(), the uncharge is called before
dropping the sock lock for the redirecte, however we missed updating
the ring start index. When no apply actions are in progress this
is OK because we uncharge the entire buffer before the redirect.
But, when we have apply logic running its possible that only a
portion of the buffer is being redirected. In this case we only
do memory accounting for the buffer slice being redirected and
expect to be able to loop over the BPF program again and/or if
a sock is closed uncharge the memory at sock destruct time.

With an invalid start index however the program logic looks at
the start pointer index, checks the length, and when seeing the
length is zero (from the initial release and failure to update
the pointer) aborts without uncharging/releasing the remaining
memory.

The fix for this is simply to update the start index. To avoid
fixing this error in two locations we do a small refactor and
remove one case where it is open-coded. Then fix it in the
single function.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

When an error occurs during a redirect we have two cases that need
to be handled (i) we have a cork'ed buffer (ii) we have a normal
sendmsg buffer.

In the cork'ed buffer case we don't currently support recovering from
errors in a redirect action. So the buffer is released and the error
should _not_ be pushed back to the caller of sendmsg/sendpage. The
rationale here is the user will get an error that relates to old
data that may have been sent by some arbitrary thread on that sock.
Instead we simple consume the data and tell the user that the data
has been consumed. We may add proper error recovery in the future.
However, this patch fixes a bug where the bytes outstanding counter
sg_size was not zeroed. This could result in a case where if the user
has both a cork'ed action and apply action in progress we may
incorrectly call into the BPF program when the user expected an
old verdict to be applied via the apply action. I don't have a use
case where using apply and cork at the same time is valid but we
never explicitly reject it because it should work fine. This patch
ensures the sg_size is zeroed so we don't have this case.

In the normal sendmsg buffer case (no cork data) we also do not
zero sg_size. Again this can confuse the apply logic when the logic
calls into the BPF program when the BPF programmer expected the old
verdict to remain. So ensure we set sg_size to zero here as well. And
additionally to keep the psock state in-sync with the sk_msg_buff
release all the memory as well. Previously we did this before
returning to the user but this left a gap where psock and sk_msg_buff
states were out of sync which seems fragile. No additional overhead
is taken here except for a call to check the length and realize its
already been freed. This is in the error path as well so in my
opinion lets have robust code over optimized error paths.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

When the call to do_tcp_sendpage() fails to send the complete block
requested we either retry if only a partial send was completed or
abort if we receive a error less than or equal to zero. Before
returning though we must update the scatterlist length/offset to
account for any partial send completed.

Before this patch we did this at the end of the retry loop, but
this was buggy when used while applying a verdict to fewer bytes
than in the scatterlist. When the scatterlist length was being set
we forgot to account for the apply logic reducing the size variable.
So the result was we chopped off some bytes in the scatterlist without
doing proper cleanup on them. This results in a WARNING when the
sock is tore down because the bytes have previously been charged to
the socket but are never uncharged.

The simple fix is to simply do the accounting inside the retry loop
subtracting from the absolute scatterlist values rather than trying
to accumulate the totals and subtract at the end.
Reported-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

In the case where the socket memory boundary is hit the redirect
path returns an ENOMEM error. However, before checking for this
condition the redirect scatterlist buffer is setup with a valid
page and length. This is never unwound so when the buffers are
released latter in the error path we do a put_page() and clear
the scatterlist fields. But, because the initial error happens
before completing the scatterlist buffer we end up with both the
original buffer and the redirect buffer pointing to the same page
resulting in duplicate put_page() calls.

To fix this simply move the initial configuration of the redirect
scatterlist buffer below the sock memory check.

Found this while running TCP_STREAM test with netperf using Cilium.

Fixes: fa246693 ("bpf: sockmap, BPF_F_INGRESS flag for BPF_SK_SKB_STREAM_VERDICT")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

In the recvmsg handler we need to add a wait event to support the
blocking use cases. Without this we return zero and may confuse
user applications. In the wait event any data received on the
sk either via sk_receive_queue or the psock ingress list will
wake up the sock.

Fixes: fa246693 ("bpf: sockmap, BPF_F_INGRESS flag for BPF_SK_SKB_STREAM_VERDICT")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Relying on map_release hook to decrement the reference counts when a
map is removed only works if the map is not being pinned. In the
pinned case the ref is decremented immediately and the BPF programs
released. After this BPF programs may not be in-use which is not
what the user would expect.

This patch moves the release logic into bpf_map_put_uref() and brings
sockmap in-line with how a similar case is handled in prog array maps.

Fixes: 3d9e9526 ("bpf: sockmap, fix leaking maps with attached but not detached progs")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Remove dead code that bails on `attr->value_size > KMALLOC_MAX_SIZE` - the
previous check already bails on `attr->value_size != 4`.
Signed-off-by: NJann Horn <jannh@google.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

It is possible to have multiple ULP tcp_release call paths in flight
if a sock is closed and simultaneously being removed from the sockmap
control path. The result would be setting the sk_prot to the saved
values on the first iteration and then on the second iteration setting
the value to NULL.

This patch resolves this by ensuring we only reset the sk_prot pointer
if we have a valid saved state to set.

Fixes: 4f738adb ("bpf: create tcp_bpf_ulp allowing BPF to monitor socket TX/RX data")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

If a socket with pending cork data is closed we do not return the
memory to the socket until the garbage collector free's the psock
structure. The garbage collector though can run after the sock has
completed its close operation. If this ordering happens the sock code
will through a WARN_ON because there is still outstanding memory
accounted to the sock.

To resolve this ensure we return memory to the sock when a socket
is closed.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Fixes: 91843d54 ("bpf: sockmap, add msg_cork_bytes() helper")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

When CONFIG_DEBUG_SG is set, sg->sg_magic is initialized in
sg_init_table() and it is verified in sg api while navigating. We hit
BUG_ON when magic check is failed.

In functions sg_tcp_sendpage and sg_tcp_sendmsg, the struct containing
the scatterlist is already zeroed out. So to avoid extra memset, we
use sg_init_marker() to initialize sg_magic.

Fixed following things:
- In bpf_tcp_sendpage: initialize sg using sg_init_marker
- In bpf_tcp_sendmsg: Replace sg_init_table with sg_init_marker
- In bpf_tcp_push: Replace memset with sg_init_table where consumed
  sg entry needs to be re-initialized.
Signed-off-by: NPrashant Bhole <bhole_prashant_q7@lab.ntt.co.jp>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Add support for the BPF_F_INGRESS flag in skb redirect helper. To
do this convert skb into a scatterlist and push into ingress queue.
This is the same logic that is used in the sk_msg redirect helper
so it should feel familiar.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Add support for the BPF_F_INGRESS flag in sk_msg redirect helper.
To do this add a scatterlist ring for receiving socks to check
before calling into regular recvmsg call path. Additionally, because
the poll wakeup logic only checked the skb recv queue we need to
add a hook in TCP stack (similar to write side) so that we have
a way to wake up polling socks when a scatterlist is redirected
to that sock.

After this all that is needed is for the redirect helper to
push the scatterlist into the psock receive queue.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

This implements a BPF ULP layer to allow policy enforcement and
monitoring at the socket layer. In order to support this a new
program type BPF_PROG_TYPE_SK_MSG is used to run the policy at
the sendmsg/sendpage hook. To attach the policy to sockets a
sockmap is used with a new program attach type BPF_SK_MSG_VERDICT.

Similar to previous sockmap usages when a sock is added to a
sockmap, via a map update, if the map contains a BPF_SK_MSG_VERDICT
program type attached then the BPF ULP layer is created on the
socket and the attached BPF_PROG_TYPE_SK_MSG program is run for
every msg in sendmsg case and page/offset in sendpage case.

BPF_PROG_TYPE_SK_MSG Semantics/API:

BPF_PROG_TYPE_SK_MSG supports only two return codes SK_PASS and
SK_DROP. Returning SK_DROP free's the copied data in the sendmsg
case and in the sendpage case leaves the data untouched. Both cases
return -EACESS to the user. Returning SK_PASS will allow the msg to
be sent.

In the sendmsg case data is copied into kernel space buffers before
running the BPF program. The kernel space buffers are stored in a
scatterlist object where each element is a kernel memory buffer.
Some effort is made to coalesce data from the sendmsg call here.
For example a sendmsg call with many one byte iov entries will
likely be pushed into a single entry. The BPF program is run with
data pointers (start/end) pointing to the first sg element.

In the sendpage case data is not copied. We opt not to copy the
data by default here, because the BPF infrastructure does not
know what bytes will be needed nor when they will be needed. So
copying all bytes may be wasteful. Because of this the initial
start/end data pointers are (0,0). Meaning no data can be read or
written. This avoids reading data that may be modified by the
user. A new helper is added later in this series if reading and
writing the data is needed. The helper call will do a copy by
default so that the page is exclusively owned by the BPF call.

The verdict from the BPF_PROG_TYPE_SK_MSG applies to the entire msg
in the sendmsg() case and the entire page/offset in the sendpage case.
This avoids ambiguity on how to handle mixed return codes in the
sendmsg case. Again a helper is added later in the series if
a verdict needs to apply to multiple system calls and/or only
a subpart of the currently being processed message.

The helper msg_redirect_map() can be used to select the socket to
send the data on. This is used similar to existing redirect use
cases. This allows policy to redirect msgs.

Pseudo code simple example:

The basic logic to attach a program to a socket is as follows,

  // load the programs
  bpf_prog_load(SOCKMAP_TCP_MSG_PROG, BPF_PROG_TYPE_SK_MSG,
		&obj, &msg_prog);

  // lookup the sockmap
  bpf_map_msg = bpf_object__find_map_by_name(obj, "my_sock_map");

  // get fd for sockmap
  map_fd_msg = bpf_map__fd(bpf_map_msg);

  // attach program to sockmap
  bpf_prog_attach(msg_prog, map_fd_msg, BPF_SK_MSG_VERDICT, 0);

Adding sockets to the map is done in the normal way,

  // Add a socket 'fd' to sockmap at location 'i'
  bpf_map_update_elem(map_fd_msg, &i, fd, BPF_ANY);

After the above any socket attached to "my_sock_map", in this case
'fd', will run the BPF msg verdict program (msg_prog) on every
sendmsg and sendpage system call.

For a complete example see BPF selftests or sockmap samples.

Implementation notes:

It seemed the simplest, to me at least, to use a refcnt to ensure
psock is not lost across the sendmsg copy into the sg, the bpf program
running on the data in sg_data, and the final pass to the TCP stack.
Some performance testing may show a better method to do this and avoid
the refcnt cost, but for now use the simpler method.

Another item that will come after basic support is in place is
supporting MSG_MORE flag. At the moment we call sendpages even if
the MSG_MORE flag is set. An enhancement would be to collect the
pages into a larger scatterlist and pass down the stack. Notice that
bpf_tcp_sendmsg() could support this with some additional state saved
across sendmsg calls. I built the code to support this without having
to do refactoring work. Other features TBD include ZEROCOPY and the
TCP_RECV_QUEUE/TCP_NO_QUEUE support. This will follow initial series
shortly.

Future work could improve size limits on the scatterlist rings used
here. Currently, we use MAX_SKB_FRAGS simply because this was being
used already in the TLS case. Future work could extend the kernel sk
APIs to tune this depending on workload. This is a trade-off
between memory usage and throughput performance.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

The sockmap refcnt up until now has been wrapped in the
sk_callback_lock(). So its not actually needed any locking of its
own. The counter itself tracks the lifetime of the psock object.
Sockets in a sockmap have a lifetime that is independent of the
map they are part of. This is possible because a single socket may
be in multiple maps. When this happens we can only release the
psock data associated with the socket when the refcnt reaches
zero. There are three possible delete sock reference decrement
paths first through the normal sockmap process, the user deletes
the socket from the map. Second the map is removed and all sockets
in the map are removed, delete path is similar to case 1. The third
case is an asyncronous socket event such as a closing the socket. The
last case handles removing sockets that are no longer available.
For completeness, although inc does not pose any problems in this
patch series, the inc case only happens when a psock is added to a
map.

Next we plan to add another socket prog type to handle policy and
monitoring on the TX path. When we do this however we will need to
keep a reference count open across the sendmsg/sendpage call and
holding the sk_callback_lock() here (on every send) seems less than
ideal, also it may sleep in cases where we hit memory pressure.
Instead of dealing with these issues in some clever way simply make
the reference counting a refcnt_t type and do proper atomic ops.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

In case user program provides silly parameters, we want
a map_alloc() handler to return an error, not a NULL pointer,
otherwise we crash later in find_and_alloc_map()

Fixes: 1aa12bdf ("bpf: sockmap, add sock close() hook to remove socks")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: Nsyzbot <syzkaller@googlegroups.com>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

When a program is attached to a map we increment the program refcnt
to ensure that the program is not removed while it is potentially
being referenced from sockmap side. However, if this same program
also references the map (this is a reasonably common pattern in
my programs) then the verifier will also increment the maps refcnt
from the verifier. This is to ensure the map doesn't get garbage
collected while the program has a reference to it.

So we are left in a state where the map holds the refcnt on the
program stopping it from being removed and releasing the map refcnt.
And vice versa the program holds a refcnt on the map stopping it
from releasing the refcnt on the prog.

All this is fine as long as users detach the program while the
map fd is still around. But, if the user omits this detach command
we are left with a dangling map we can no longer release.

To resolve this when the map fd is released decrement the program
references and remove any reference from the map to the program.
This fixes the issue with possibly dangling map and creates a
user side API constraint. That is, the map fd must be held open
for programs to be attached to a map.

Fixes: 174a79ff ("bpf: sockmap with sk redirect support")
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

The selftests test_maps program was leaving dangling BPF sockmap
programs around because not all psock elements were removed from
the map. The elements in turn hold a reference on the BPF program
they are attached to causing BPF programs to stay open even after
test_maps has completed.

The original intent was that sk_state_change() would be called
when TCP socks went through TCP_CLOSE state. However, because
socks may be in SOCK_DEAD state or the sock may be a listening
socket the event is not always triggered.

To resolve this use the ULP infrastructure and register our own
proto close() handler. This fixes the above case.

Fixes: 174a79ff ("bpf: sockmap with sk redirect support")
Reported-by: NPrashant Bhole <bhole_prashant_q7@lab.ntt.co.jp>
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

All map types reimplement the field-by-field copy of union bpf_attr
members into struct bpf_map.  Add a helper to perform this operation.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NQuentin Monnet <quentin.monnet@netronome.com>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Add psock NULL check to handle a racing sock event that can get the
sk_callback_lock before this case but after xchg happens causing the
refcnt to hit zero and sock user data (psock) to be null and queued
for garbage collection.

Also add a comment in the code because this is a bit subtle and
not obvious in my opinion.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>