linux-2.6-microblaze.git
3 years agoskmsg: Use rcu work for destroying psock
Cong Wang [Wed, 31 Mar 2021 02:32:26 +0000 (19:32 -0700)]
skmsg: Use rcu work for destroying psock

The RCU callback sk_psock_destroy() only queues work psock->gc,
so we can just switch to rcu work to simplify the code.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20210331023237.41094-6-xiyou.wangcong@gmail.com
3 years agoskmsg: Avoid lock_sock() in sk_psock_backlog()
Cong Wang [Wed, 31 Mar 2021 02:32:25 +0000 (19:32 -0700)]
skmsg: Avoid lock_sock() in sk_psock_backlog()

We do not have to lock the sock to avoid losing sk_socket,
instead we can purge all the ingress queues when we close
the socket. Sending or receiving packets after orphaning
socket makes no sense.

We do purge these queues when psock refcnt reaches zero but
here we want to purge them explicitly in sock_map_close().
There are also some nasty race conditions on testing bit
SK_PSOCK_TX_ENABLED and queuing/canceling the psock work,
we can expand psock->ingress_lock a bit to protect them too.

As noticed by John, we still have to lock the psock->work,
because the same work item could be running concurrently on
different CPU's.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20210331023237.41094-5-xiyou.wangcong@gmail.com
3 years agonet: Introduce skb_send_sock() for sock_map
Cong Wang [Wed, 31 Mar 2021 02:32:24 +0000 (19:32 -0700)]
net: Introduce skb_send_sock() for sock_map

We only have skb_send_sock_locked() which requires callers
to use lock_sock(). Introduce a variant skb_send_sock()
which locks on its own, callers do not need to lock it
any more. This will save us from adding a ->sendmsg_locked
for each protocol.

To reuse the code, pass function pointers to __skb_send_sock()
and build skb_send_sock() and skb_send_sock_locked() on top.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Link: https://lore.kernel.org/bpf/20210331023237.41094-4-xiyou.wangcong@gmail.com
3 years agoskmsg: Introduce a spinlock to protect ingress_msg
Cong Wang [Wed, 31 Mar 2021 02:32:23 +0000 (19:32 -0700)]
skmsg: Introduce a spinlock to protect ingress_msg

Currently we rely on lock_sock to protect ingress_msg,
it is too big for this, we can actually just use a spinlock
to protect this list like protecting other skb queues.

__tcp_bpf_recvmsg() is still special because of peeking,
it still has to use lock_sock.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Jakub Sitnicki <jakub@cloudflare.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20210331023237.41094-3-xiyou.wangcong@gmail.com
3 years agoskmsg: Lock ingress_skb when purging
Cong Wang [Wed, 31 Mar 2021 02:32:22 +0000 (19:32 -0700)]
skmsg: Lock ingress_skb when purging

Currently we purge the ingress_skb queue only when psock
refcnt goes down to 0, so locking the queue is not necessary,
but in order to be called during ->close, we have to lock it
here.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Jakub Sitnicki <jakub@cloudflare.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20210331023237.41094-2-xiyou.wangcong@gmail.com
3 years agotools/resolve_btfids: Fix warnings
Stanislav Fomichev [Mon, 29 Mar 2021 22:31:43 +0000 (15:31 -0700)]
tools/resolve_btfids: Fix warnings

* make eprintf static, used only in main.c
* initialize ret in eprintf
* remove unused *tmp

v3:
* remove another err (Song Liu)

v2:
* remove unused 'int err = -1'

Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20210329223143.3659983-1-sdf@google.com
3 years agoselftests/bpf: Add an option for a debug shell in vmtest.sh
KP Singh [Tue, 23 Mar 2021 01:47:52 +0000 (01:47 +0000)]
selftests/bpf: Add an option for a debug shell in vmtest.sh

The newly introduced -s command line option starts an interactive shell.
If a command is specified, the shell is started after the command
finishes executing. It's useful to have a shell especially when
debugging failing tests or developing new tests.

Since the user may terminate the VM forcefully, an extra "sync" is added
after the execution of the command to persist any logs from the command
into the log file.

Signed-off-by: KP Singh <kpsingh@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210323014752.3198283-1-kpsingh@kernel.org
3 years agobpf: Remove redundant assignment of variable id
Colin Ian King [Fri, 26 Mar 2021 19:43:48 +0000 (19:43 +0000)]
bpf: Remove redundant assignment of variable id

The variable id is being assigned a value that is never read, the
assignment is redundant and can be removed.

Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20210326194348.623782-1-colin.king@canonical.com
3 years agobpf: Remove unused bpf_load_pointer
He Fengqing [Tue, 30 Mar 2021 02:48:43 +0000 (02:48 +0000)]
bpf: Remove unused bpf_load_pointer

Remove unused bpf_load_pointer function in filter.h. The last user of it has
been removed with 24dea04767e6 ("bpf, x32: remove ld_abs/ld_ind").

Signed-off-by: He Fengqing <hefengqing@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20210330024843.3479844-1-hefengqing@huawei.com
3 years agoMerge branch 'AF_XDP selftests improvements & bpf_link'
Alexei Starovoitov [Tue, 30 Mar 2021 16:24:40 +0000 (09:24 -0700)]
Merge branch 'AF_XDP selftests improvements & bpf_link'

Maciej Fijalkowski says:

====================

Changes since v4 (all in patch 6):
- do not close potentially invalid bpf_link fd (Toke)
- fix misspelling in label (Toke)
- mask out XDP_FLAGS_UPDATE_IF_NOEXIST and XDP_FLAGS_REPLACE explicitly when
  creating bpf_link (Toke)

Changes since v3:
- do not unload netlink-based XDP prog when updating map elem failed and
  current socket was not the creator of XDP resources (Toke)
- pull out code paths based on prog_id value within __xsk_setup_xdp_prog
  so that teardown in case of error at any point is more clear

Changes since v2:
- fix c&p failure in veth's get_channels implementation (Magnus)
- provide a backward compatibilty if bpf_link is not supported (Andrii)
- check for a link type while looking up existing bpf_links (Andrii)

Changes since v1:
- selftests improvements and test case for bpf_link persistence itself
- do not unload netlink-based prog when --force flag is set (John)
- simplify return semantics in xsk_link_lookup (John)

v4: https://lore.kernel.org/bpf/20210326230938.49998-1-maciej.fijalkowski@intel.com/
v3: https://lore.kernel.org/bpf/20210322205816.65159-1-maciej.fijalkowski@intel.com/
v2: https://lore.kernel.org/bpf/20210311152910.56760-1-maciej.fijalkowski@intel.com/
v1: https://lore.kernel.org/bpf/20210215154638.4627-1-maciej.fijalkowski@intel.com/

--------------------------------------------------

This set is another approach towards addressing the below issue:

// load xdp prog and xskmap and add entry to xskmap at idx 10
$ sudo ./xdpsock -i ens801f0 -t -q 10

// add entry to xskmap at idx 11
$ sudo ./xdpsock -i ens801f0 -t -q 11

terminate one of the processes and another one is unable to work due to
the fact that the XDP prog was unloaded from interface.

Previous attempt was, to put it mildly, a bit broken, as there was no
synchronization between updates to additional map, as Bjorn pointed out.
See https://lore.kernel.org/netdev/20190603131907.13395-5-maciej.fijalkowski@intel.com/

In the meantime bpf_link was introduced and it seems that it can address
the issue of refcounting the XDP prog on interface.

Although the bpf_link is the meat of the set, selftests improvements are a
bigger part of it. Overall, we've been able to reduce the complexity of xsk
selftests by removing a bunch of synchronization resources and
simplifying logic and structs.

Last but not least, for multiqueue veth working with AF-XDP, ethtool's
get_channels API needs to be implemented, so it's also included in that
set.

Note also that in order to make it work, a commit from bpf tree:
veth: store queue_mapping independently of XDP prog presence
https://lore.kernel.org/bpf/20210303152903.11172-1-maciej.fijalkowski@intel.com/

is needed.

Thanks,
Maciej

Björn Töpel (3):
  selftests: xsk: remove thread attribute
  selftests: xsk: Remove mutex and condition variable
  selftests: xsk: Remove unused defines
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
3 years agoselftests: xsk: Remove unused defines
Björn Töpel [Mon, 29 Mar 2021 22:43:16 +0000 (00:43 +0200)]
selftests: xsk: Remove unused defines

Remove two unused defines.

Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-18-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Remove mutex and condition variable
Björn Töpel [Mon, 29 Mar 2021 22:43:15 +0000 (00:43 +0200)]
selftests: xsk: Remove mutex and condition variable

The usage of the condition variable is broken, and overkill. Replace it
with a pthread barrier.

Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-17-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Remove thread attribute
Björn Töpel [Mon, 29 Mar 2021 22:43:14 +0000 (00:43 +0200)]
selftests: xsk: Remove thread attribute

There is really no reason to have a non-default thread stack
size. Remove that.

Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-16-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Implement bpf_link test
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:13 +0000 (00:43 +0200)]
selftests: xsk: Implement bpf_link test

Introduce a test that is supposed to verify the persistence of BPF
resources based on underlying bpf_link usage.

Test will:
1) create and bind two sockets on queue ids 0 and 1
2) run a traffic on queue ids 0
3) remove xsk sockets from queue 0 on both veth interfaces
4) run a traffic on queues ids 1

Running traffic successfully on qids 1 means that BPF resources were
not removed on step 3).

In order to make it work, change the command that creates veth pair to
have the 4 queue pairs by default.

Introduce the arrays of xsks and umems to ifobject struct but keep a
pointers to single entities, so rest of the logic around Rx/Tx can be
kept as-is.

For umem handling, double the size of mmapped space and split that
between the two sockets.

Rename also bidi_pass to a variable 'second_step' of a boolean type as
it's now used also for the test that is introduced here and it doesn't
have anything in common with bi-directional testing.

Drop opt_queue command line argument as it wasn't working before anyway.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-15-maciej.fijalkowski@intel.com
3 years agoveth: Implement ethtool's get_channels() callback
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:12 +0000 (00:43 +0200)]
veth: Implement ethtool's get_channels() callback

Libbpf's xsk part calls get_channels() API to retrieve the queue count
of the underlying driver so that XSKMAP is sized accordingly.

Implement that in veth so multi queue scenarios can work properly.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-14-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Remove sync_mutex_tx and atomic var
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:11 +0000 (00:43 +0200)]
selftests: xsk: Remove sync_mutex_tx and atomic var

Although thread_common_ops() are called in both Tx and Rx threads,
testapp_validate() will not spawn Tx thread until Rx thread signals that
it has finished its initialization via condition variable.

Therefore, locking in thread_common_ops is not needed and furthermore Tx
thread does not have to spin on atomic variable.

Note that this simplification wouldn't be possible if there would still
be a common worker thread.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-13-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Refactor teardown/bidi test cases and testapp_validate
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:10 +0000 (00:43 +0200)]
selftests: xsk: Refactor teardown/bidi test cases and testapp_validate

Currently, there is a testapp_sockets() that acts like a wrapper around
testapp_validate() and it is called for bidi and teardown test types.
Other test types call testapp_validate() directly.

Split testapp_sockets() onto two separate functions so a bunch of bidi
specific logic can be moved there and out of testapp_validate() itself.

Introduce function pointer to ifobject struct which will be used for
assigning the Rx/Tx function that is assigned to worker thread. Let's
also have a global ifobject Rx/Tx pointers so it's easier to swap the
vectors on a second run of a bi-directional test. Thread creation now is
easey to follow.

switching_notify variable is useless, info about vector switch can be
printed based on bidi_pass state.

Last but not least, init/destroy synchronization variables only once,
not per each test.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-12-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Remove Tx synchronization resources
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:09 +0000 (00:43 +0200)]
selftests: xsk: Remove Tx synchronization resources

Tx thread needs to be started after the Rx side is fully initialized so
that packets are not xmitted until xsk Rx socket is ready to be used.

It can be observed that atomic variable spinning_tx is not checked from
Rx side in any way, so thread_common_ops can be modified to only address
the spinning_rx. This means that spinning_tx can be removed altogheter.

signal_tx_condition is never utilized, so simply remove it.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-11-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Split worker thread
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:08 +0000 (00:43 +0200)]
selftests: xsk: Split worker thread

Let's a have a separate Tx/Rx worker threads instead of a one common
thread packed with Tx/Rx specific checks.

Move mmap for umem buffer space and a switch_namespace() call to
thread_common_ops.

This also allows for a bunch of simplifactions that are the subject of
the next commits. The final result will be a code base that is much
easier to follow.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-10-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Remove thread for netns switch
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:07 +0000 (00:43 +0200)]
selftests: xsk: Remove thread for netns switch

Currently, there is a dedicated thread for following remote ns operations:
- grabbing the ifindex of the interface moved to remote netns
- removing xdp prog from that interface

With bpf_link usage in place, this can be simply omitted, so remove
mentioned thread, as BPF resources will be managed by bpf_link itself,
so there's no further need for creating the thread that will switch to
remote netns and do the cleanup.

Keep most of the logic for switching the ns, though, but make
switch_namespace() return the fd so that it will be possible to close it
at the process termination time. Get rid of logic around making sure
that it's possible to switch ns in validate_interfaces().

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-9-maciej.fijalkowski@intel.com
3 years agosamples: bpf: Do not unload prog within xdpsock
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:06 +0000 (00:43 +0200)]
samples: bpf: Do not unload prog within xdpsock

With the introduction of bpf_link in xsk's libbpf part, there's no
further need for explicit unload of prog on xdpsock's termination. When
process dies, the bpf_link's refcount will be decremented and resources
will be unloaded/freed under the hood in case when there are no more
active users.

While at it, don't dump stats on error path.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-8-maciej.fijalkowski@intel.com
3 years agolibbpf: xsk: Use bpf_link
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:05 +0000 (00:43 +0200)]
libbpf: xsk: Use bpf_link

Currently, if there are multiple xdpsock instances running on a single
interface and in case one of the instances is terminated, the rest of
them are left in an inoperable state due to the fact of unloaded XDP
prog from interface.

Consider the scenario below:

// load xdp prog and xskmap and add entry to xskmap at idx 10
$ sudo ./xdpsock -i ens801f0 -t -q 10

// add entry to xskmap at idx 11
$ sudo ./xdpsock -i ens801f0 -t -q 11

terminate one of the processes and another one is unable to work due to
the fact that the XDP prog was unloaded from interface.

To address that, step away from setting bpf prog in favour of bpf_link.
This means that refcounting of BPF resources will be done automatically
by bpf_link itself.

Provide backward compatibility by checking if underlying system is
bpf_link capable. Do this by looking up/creating bpf_link on loopback
device. If it failed in any way, stick with netlink-based XDP prog.
therwise, use bpf_link-based logic.

When setting up BPF resources during xsk socket creation, check whether
bpf_link for a given ifindex already exists via set of calls to
bpf_link_get_next_id -> bpf_link_get_fd_by_id -> bpf_obj_get_info_by_fd
and comparing the ifindexes from bpf_link and xsk socket.

For case where resources exist but they are not AF_XDP related, bail out
and ask user to remove existing prog and then retry.

Lastly, do a bit of refactoring within __xsk_setup_xdp_prog and pull out
existing code branches based on prog_id value onto separate functions
that are responsible for resource initialization if prog_id was 0 and
for lookup existing resources for non-zero prog_id as that implies that
XDP program is present on the underlying net device. This in turn makes
it easier to follow, especially the teardown part of both branches.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-7-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Simplify frame traversal in dumping thread
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:04 +0000 (00:43 +0200)]
selftests: xsk: Simplify frame traversal in dumping thread

Store offsets to each layer in a separate variables rather than compute
them every single time.

Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-6-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Remove inline keyword from source file
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:03 +0000 (00:43 +0200)]
selftests: xsk: Remove inline keyword from source file

Follow the kernel coding style guidelines and let compiler do the
decision about inlining.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-5-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Remove unused function
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:02 +0000 (00:43 +0200)]
selftests: xsk: Remove unused function

Probably it was ported from xdpsock but is not used anywhere.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-4-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Remove struct ifaceconfigobj
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:01 +0000 (00:43 +0200)]
selftests: xsk: Remove struct ifaceconfigobj

ifaceconfigobj is not really useful, it is possible to keep the
functionality and simplify the code.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-3-maciej.fijalkowski@intel.com
3 years agoselftests: xsk: Don't call worker_pkt_dump() for stats test
Maciej Fijalkowski [Mon, 29 Mar 2021 22:43:00 +0000 (00:43 +0200)]
selftests: xsk: Don't call worker_pkt_dump() for stats test

For TEST_TYPE_STATS, worker_pkt_validate() that places frames onto
pkt_buf is not called. Therefore, when dump mode is set, don't call
worker_pkt_dump() for mentioned test type, so that it won't crash on
pkt_buf() access.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329224316.17793-2-maciej.fijalkowski@intel.com
3 years agolibbpf: Fix memory leak when emitting final btf_ext
Andrii Nakryiko [Sat, 27 Mar 2021 04:25:02 +0000 (21:25 -0700)]
libbpf: Fix memory leak when emitting final btf_ext

Free temporary allocated memory used to construct finalized .BTF.ext data.
Found by Coverity static analysis on libbpf's Github repo.

Fixes: 8fd27bf69b86 ("libbpf: Add BPF static linker BTF and BTF.ext support")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20210327042502.969745-1-andrii@kernel.org
3 years agoMerge branch 'bpf: Update doc about calling kernel function'
Alexei Starovoitov [Tue, 30 Mar 2021 14:31:01 +0000 (07:31 -0700)]
Merge branch 'bpf: Update doc about calling kernel function'

Martin KaFai says:

====================

This set updates the document about the bpf program calling kernel
function.  In particular, updates are regarding to the clang
requirement in selftests and kfunc-call not an ABI.
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
3 years agobpf: selftests: Update clang requirement in README.rst for testing kfunc call
Martin KaFai Lau [Tue, 30 Mar 2021 05:41:56 +0000 (22:41 -0700)]
bpf: selftests: Update clang requirement in README.rst for testing kfunc call

This patch updates the README.rst to specify the clang requirement
to compile the bpf selftests that call kernel function.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20210330054156.2933804-1-kafai@fb.com
3 years agobpf: Update bpf_design_QA.rst to clarify the kfunc call is not ABI
Martin KaFai Lau [Tue, 30 Mar 2021 05:41:50 +0000 (22:41 -0700)]
bpf: Update bpf_design_QA.rst to clarify the kfunc call is not ABI

This patch updates bpf_design_QA.rst to clarify that the kernel
function callable by bpf program is not an ABI.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20210330054150.2933542-1-kafai@fb.com
3 years agobpf: tcp: Limit calling some tcp cc functions to CONFIG_DYNAMIC_FTRACE
Martin KaFai Lau [Mon, 29 Mar 2021 22:13:57 +0000 (15:13 -0700)]
bpf: tcp: Limit calling some tcp cc functions to CONFIG_DYNAMIC_FTRACE

pahole currently only generates the btf_id for external function and
ftrace-able function.  Some functions in the bpf_tcp_ca_kfunc_ids
are static (e.g. cubictcp_init).  Thus, unless CONFIG_DYNAMIC_FTRACE
is set, btf_ids for those functions will not be generated and the
compilation fails during resolve_btfids.

This patch limits those functions to CONFIG_DYNAMIC_FTRACE.  I will
address the pahole generation in a followup and then remove the
CONFIG_DYNAMIC_FTRACE limitation.

Fixes: e78aea8b2170 ("bpf: tcp: Put some tcp cong functions in allowlist for bpf-tcp-cc")
Reported-by: Cong Wang <xiyou.wangcong@gmail.com>
Reported-by: Lorenz Bauer <lmb@cloudflare.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329221357.834438-1-kafai@fb.com
3 years agobpf: tcp: Fix an error in the bpf_tcp_ca_kfunc_ids list
Martin KaFai Lau [Mon, 29 Mar 2021 00:32:13 +0000 (17:32 -0700)]
bpf: tcp: Fix an error in the bpf_tcp_ca_kfunc_ids list

There is a typo in the bbr function, s/even/event/.
This patch fixes it.

Fixes: e78aea8b2170 ("bpf: tcp: Put some tcp cong functions in allowlist for bpf-tcp-cc")
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210329003213.2274210-1-kafai@fb.com
3 years agobpf: tcp: Remove comma which is causing build error
Atul Gopinathan [Sun, 28 Mar 2021 12:05:15 +0000 (17:35 +0530)]
bpf: tcp: Remove comma which is causing build error

Currently, building the bpf-next source with the CONFIG_BPF_SYSCALL
enabled is causing a compilation error:

"net/ipv4/bpf_tcp_ca.c:209:28: error: expected identifier or '(' before
',' token"

Fix this by removing an unnecessary comma.

Fixes: e78aea8b2170 ("bpf: tcp: Put some tcp cong functions in allowlist for bpf-tcp-cc")
Reported-by: syzbot+0b74d8ec3bf0cc4e4209@syzkaller.appspotmail.com
Signed-off-by: Atul Gopinathan <atulgopinathan@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210328120515.113895-1-atulgopinathan@gmail.com
3 years agoMerge branch 'bpf: Support calling kernel function'
Alexei Starovoitov [Sat, 27 Mar 2021 03:29:06 +0000 (20:29 -0700)]
Merge branch 'bpf: Support calling kernel function'

Martin KaFai says:

====================

This series adds support to allow bpf program calling kernel function.

The use case included in this set is to allow bpf-tcp-cc to directly
call some tcp-cc helper functions (e.g. "tcp_cong_avoid_ai()").  Those
functions have already been used by some kernel tcp-cc implementations.

This set will also allow the bpf-tcp-cc program to directly call the
kernel tcp-cc implementation,  For example, a bpf_dctcp may only want to
implement its own dctcp_cwnd_event() and reuse other dctcp_*() directly
from the kernel tcp_dctcp.c instead of reimplementing (or
copy-and-pasting) them.

The tcp-cc kernel functions mentioned above will be white listed
for the struct_ops bpf-tcp-cc programs to use in a later patch.
The white listed functions are not bounded to a fixed ABI contract.
Those functions have already been used by the existing kernel tcp-cc.
If any of them has changed, both in-tree and out-of-tree kernel tcp-cc
implementations have to be changed.  The same goes for the struct_ops
bpf-tcp-cc programs which have to be adjusted accordingly.

Please see individual patch for details.

v2:
- Patch 2 in v1 is removed.  No need to support extern func in kernel.
  Changed libbpf to adjust the .ksyms datasec for extern func
  in patch 11. (Andrii)
- Name change: btf_check_func_arg_match() and btf_check_subprog_arg_match()
  in patch 2. (Andrii)
- Always set unreliable on any error in patch 2 since it does not
  matter. (Andrii)
- s/kern_func/kfunc/ and s/descriptor/desc/ in this set. (Andrii)
- Remove some unnecessary changes in disasm.h and disasm.c
  in patch 3.  In particular, no need to change the function
  signature in bpf_insn_revmap_call_t.  Also, removed the changes
  in print_bpf_insn().
- Fixed an issue in check_kfunc_call() when the calling kernel function
  returns a pointer in patch 3.  Added a selftest.
- Adjusted the verifier selftests due to the changes in the verifier log
  in patch 3.
- Fixed a comparison issue in kfunc_desc_cmp_by_imm() in patch 3. (Andrii)
- Name change: is_ldimm64_insn(),
  new helper: is_call_insn() in patch 10 (Andrii)
- Move btf_func_linkage() from btf.h to libbpf.c in patch 11. (Andrii)
- Fixed the linker error when CONFIG_BPF_SYSCALL is not defined.
  Moved the check_kfunc_call from filter.c to test_run.c in patch 14.
  (kernel test robot)
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
3 years agobpf: selftests: Add kfunc_call test
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:52 +0000 (18:52 -0700)]
bpf: selftests: Add kfunc_call test

This patch adds a few kernel function bpf_kfunc_call_test*() for the
selftest's test_run purpose.  They will be allowed for tc_cls prog.

The selftest calling the kernel function bpf_kfunc_call_test*()
is also added in this patch.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015252.1551395-1-kafai@fb.com
3 years agobpf: selftests: Bpf_cubic and bpf_dctcp calling kernel functions
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:46 +0000 (18:52 -0700)]
bpf: selftests: Bpf_cubic and bpf_dctcp calling kernel functions

This patch removes the bpf implementation of tcp_slow_start()
and tcp_cong_avoid_ai().  Instead, it directly uses the kernel
implementation.

It also replaces the bpf_cubic_undo_cwnd implementation by directly
calling tcp_reno_undo_cwnd().  bpf_dctcp also directly calls
tcp_reno_cong_avoid() instead.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015246.1551062-1-kafai@fb.com
3 years agobpf: selftests: Rename bictcp to bpf_cubic
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:40 +0000 (18:52 -0700)]
bpf: selftests: Rename bictcp to bpf_cubic

As a similar chanage in the kernel, this patch gives the proper
name to the bpf cubic.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015240.1550074-1-kafai@fb.com
3 years agolibbpf: Support extern kernel function
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:34 +0000 (18:52 -0700)]
libbpf: Support extern kernel function

This patch is to make libbpf able to handle the following extern
kernel function declaration and do the needed relocations before
loading the bpf program to the kernel.

extern int foo(struct sock *) __attribute__((section(".ksyms")))

In the collect extern phase, needed changes is made to
bpf_object__collect_externs() and find_extern_btf_id() to collect
extern function in ".ksyms" section.  The func in the BTF datasec also
needs to be replaced by an int var.  The idea is similar to the existing
handling in extern var.  In case the BTF may not have a var, a dummy ksym
var is added at the beginning of bpf_object__collect_externs()
if there is func under ksyms datasec.  It will also change the
func linkage from extern to global which the kernel can support.
It also assigns a param name if it does not have one.

In the collect relo phase, it will record the kernel function
call as RELO_EXTERN_FUNC.

bpf_object__resolve_ksym_func_btf_id() is added to find the func
btf_id of the running kernel.

During actual relocation, it will patch the BPF_CALL instruction with
src_reg = BPF_PSEUDO_FUNC_CALL and insn->imm set to the running
kernel func's btf_id.

The required LLVM patch: https://reviews.llvm.org/D93563

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015234.1548923-1-kafai@fb.com
3 years agolibbpf: Record extern sym relocation first
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:27 +0000 (18:52 -0700)]
libbpf: Record extern sym relocation first

This patch records the extern sym relocs first before recording
subprog relocs.  The later patch will have relocs for extern
kernel function call which is also using BPF_JMP | BPF_CALL.
It will be easier to handle the extern symbols first in
the later patch.

is_call_insn() helper is added.  The existing is_ldimm64() helper
is renamed to is_ldimm64_insn() for consistency.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015227.1548623-1-kafai@fb.com
3 years agolibbpf: Rename RELO_EXTERN to RELO_EXTERN_VAR
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:21 +0000 (18:52 -0700)]
libbpf: Rename RELO_EXTERN to RELO_EXTERN_VAR

This patch renames RELO_EXTERN to RELO_EXTERN_VAR.
It is to avoid the confusion with a later patch adding
RELO_EXTERN_FUNC.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015221.1547722-1-kafai@fb.com
3 years agolibbpf: Refactor codes for finding btf id of a kernel symbol
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:14 +0000 (18:52 -0700)]
libbpf: Refactor codes for finding btf id of a kernel symbol

This patch refactors code, that finds kernel btf_id by kind
and symbol name, to a new function find_ksym_btf_id().

It also adds a new helper __btf_kind_str() to return
a string by the numeric kind value.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015214.1547069-1-kafai@fb.com
3 years agolibbpf: Refactor bpf_object__resolve_ksyms_btf_id
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:07 +0000 (18:52 -0700)]
libbpf: Refactor bpf_object__resolve_ksyms_btf_id

This patch refactors most of the logic from
bpf_object__resolve_ksyms_btf_id() into a new function
bpf_object__resolve_ksym_var_btf_id().
It is to get ready for a later patch adding
bpf_object__resolve_ksym_func_btf_id() which resolves
a kernel function to the running kernel btf_id.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015207.1546749-1-kafai@fb.com
3 years agobpf: tcp: Put some tcp cong functions in allowlist for bpf-tcp-cc
Martin KaFai Lau [Thu, 25 Mar 2021 01:52:01 +0000 (18:52 -0700)]
bpf: tcp: Put some tcp cong functions in allowlist for bpf-tcp-cc

This patch puts some tcp cong helper functions, tcp_slow_start()
and tcp_cong_avoid_ai(), into the allowlist for the bpf-tcp-cc
program.

A few tcp cc implementation functions are also put into the
allowlist.  A potential use case is the bpf-tcp-cc implementation
may only want to override a subset of a tcp_congestion_ops.  For others,
the bpf-tcp-cc can directly call the kernel counter parts instead of
re-implementing (or copy-and-pasting) them to the bpf program.

They will only be available to the bpf-tcp-cc typed program.
The allowlist functions are not bounded to a fixed ABI contract.
When any of them has changed, the bpf-tcp-cc program has to be changed
like any in-tree/out-of-tree kernel tcp-cc implementations do also.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015201.1546345-1-kafai@fb.com
3 years agotcp: Rename bictcp function prefix to cubictcp
Martin KaFai Lau [Thu, 25 Mar 2021 01:51:55 +0000 (18:51 -0700)]
tcp: Rename bictcp function prefix to cubictcp

The cubic functions in tcp_cubic.c are using the bictcp prefix as
in tcp_bic.c.  This patch gives it the proper name cubictcp
because the later patch will allow the bpf prog to directly
call the cubictcp implementation.  Renaming them will avoid
the name collision when trying to find the intended
one to call during bpf prog load time.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015155.1545532-1-kafai@fb.com
3 years agobpf: Support kernel function call in x86-32
Martin KaFai Lau [Thu, 25 Mar 2021 01:51:49 +0000 (18:51 -0700)]
bpf: Support kernel function call in x86-32

This patch adds kernel function call support to the x86-32 bpf jit.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015149.1545267-1-kafai@fb.com
3 years agobpf: Support bpf program calling kernel function
Martin KaFai Lau [Thu, 25 Mar 2021 01:51:42 +0000 (18:51 -0700)]
bpf: Support bpf program calling kernel function

This patch adds support to BPF verifier to allow bpf program calling
kernel function directly.

The use case included in this set is to allow bpf-tcp-cc to directly
call some tcp-cc helper functions (e.g. "tcp_cong_avoid_ai()").  Those
functions have already been used by some kernel tcp-cc implementations.

This set will also allow the bpf-tcp-cc program to directly call the
kernel tcp-cc implementation,  For example, a bpf_dctcp may only want to
implement its own dctcp_cwnd_event() and reuse other dctcp_*() directly
from the kernel tcp_dctcp.c instead of reimplementing (or
copy-and-pasting) them.

The tcp-cc kernel functions mentioned above will be white listed
for the struct_ops bpf-tcp-cc programs to use in a later patch.
The white listed functions are not bounded to a fixed ABI contract.
Those functions have already been used by the existing kernel tcp-cc.
If any of them has changed, both in-tree and out-of-tree kernel tcp-cc
implementations have to be changed.  The same goes for the struct_ops
bpf-tcp-cc programs which have to be adjusted accordingly.

This patch is to make the required changes in the bpf verifier.

First change is in btf.c, it adds a case in "btf_check_func_arg_match()".
When the passed in "btf->kernel_btf == true", it means matching the
verifier regs' states with a kernel function.  This will handle the
PTR_TO_BTF_ID reg.  It also maps PTR_TO_SOCK_COMMON, PTR_TO_SOCKET,
and PTR_TO_TCP_SOCK to its kernel's btf_id.

In the later libbpf patch, the insn calling a kernel function will
look like:

insn->code == (BPF_JMP | BPF_CALL)
insn->src_reg == BPF_PSEUDO_KFUNC_CALL /* <- new in this patch */
insn->imm == func_btf_id /* btf_id of the running kernel */

[ For the future calling function-in-kernel-module support, an array
  of module btf_fds can be passed at the load time and insn->off
  can be used to index into this array. ]

At the early stage of verifier, the verifier will collect all kernel
function calls into "struct bpf_kfunc_desc".  Those
descriptors are stored in "prog->aux->kfunc_tab" and will
be available to the JIT.  Since this "add" operation is similar
to the current "add_subprog()" and looking for the same insn->code,
they are done together in the new "add_subprog_and_kfunc()".

In the "do_check()" stage, the new "check_kfunc_call()" is added
to verify the kernel function call instruction:
1. Ensure the kernel function can be used by a particular BPF_PROG_TYPE.
   A new bpf_verifier_ops "check_kfunc_call" is added to do that.
   The bpf-tcp-cc struct_ops program will implement this function in
   a later patch.
2. Call "btf_check_kfunc_args_match()" to ensure the regs can be
   used as the args of a kernel function.
3. Mark the regs' type, subreg_def, and zext_dst.

At the later do_misc_fixups() stage, the new fixup_kfunc_call()
will replace the insn->imm with the function address (relative
to __bpf_call_base).  If needed, the jit can find the btf_func_model
by calling the new bpf_jit_find_kfunc_model(prog, insn).
With the imm set to the function address, "bpftool prog dump xlated"
will be able to display the kernel function calls the same way as
it displays other bpf helper calls.

gpl_compatible program is required to call kernel function.

This feature currently requires JIT.

The verifier selftests are adjusted because of the changes in
the verbose log in add_subprog_and_kfunc().

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015142.1544736-1-kafai@fb.com
3 years agobpf: Refactor btf_check_func_arg_match
Martin KaFai Lau [Thu, 25 Mar 2021 01:51:36 +0000 (18:51 -0700)]
bpf: Refactor btf_check_func_arg_match

This patch moved the subprog specific logic from
btf_check_func_arg_match() to the new btf_check_subprog_arg_match().
The core logic is left in btf_check_func_arg_match() which
will be reused later to check the kernel function call.

The "if (!btf_type_is_ptr(t))" is checked first to improve the
indentation which will be useful for a later patch.

Some of the "btf_kind_str[]" usages is replaced with the shortcut
"btf_type_str(t)".

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015136.1544504-1-kafai@fb.com
3 years agobpf: Simplify freeing logic in linfo and jited_linfo
Martin KaFai Lau [Thu, 25 Mar 2021 01:51:30 +0000 (18:51 -0700)]
bpf: Simplify freeing logic in linfo and jited_linfo

This patch simplifies the linfo freeing logic by combining
"bpf_prog_free_jited_linfo()" and "bpf_prog_free_unused_jited_linfo()"
into the new "bpf_prog_jit_attempt_done()".
It is a prep work for the kernel function call support.  In a later
patch, freeing the kernel function call descriptors will also
be done in the "bpf_prog_jit_attempt_done()".

"bpf_prog_free_linfo()" is removed since it is only called by
"__bpf_prog_put_noref()".  The kvfree() are directly called
instead.

It also takes this chance to s/kcalloc/kvcalloc/ for the jited_linfo
allocation.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210325015130.1544323-1-kafai@fb.com
3 years agolibbpf: Preserve empty DATASEC BTFs during static linking
Andrii Nakryiko [Fri, 26 Mar 2021 04:30:36 +0000 (21:30 -0700)]
libbpf: Preserve empty DATASEC BTFs during static linking

Ensure that BPF static linker preserves all DATASEC BTF types, even if some of
them might not have any variable information at all. This may happen if the
compiler promotes local initialized variable contents into .rodata section and
there are no global or static functions in the program.

For example,

  $ cat t.c
  struct t { char a; char b; char c; };
  void bar(struct t*);
  void find() {
     struct t tmp = {1, 2, 3};
     bar(&tmp);
  }

  $ clang -target bpf -O2 -g -S t.c
         .long   104                             # BTF_KIND_DATASEC(id = 8)
         .long   251658240                       # 0xf000000
         .long   0

         .ascii  ".rodata"                       # string offset=104

  $ clang -target bpf -O2 -g -c t.c
  $ readelf -S t.o | grep data
     [ 4] .rodata           PROGBITS         0000000000000000  00000090

Fixes: 8fd27bf69b86 ("libbpf: Add BPF static linker BTF and BTF.ext support")
Reported-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://lore.kernel.org/bpf/20210326043036.3081011-1-andrii@kernel.org
3 years agobpf: struct sock is declared twice in bpf_sk_storage header
Wan Jiabing [Thu, 25 Mar 2021 07:06:02 +0000 (15:06 +0800)]
bpf: struct sock is declared twice in bpf_sk_storage header

struct sock has been declared twice, therefore remove the duplicate.

Signed-off-by: Wan Jiabing <wanjiabing@vivo.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20210325070602.858024-1-wanjiabing@vivo.com
3 years agobpf: Remove unused headers
Lu Wei [Wed, 24 Mar 2021 08:31:47 +0000 (16:31 +0800)]
bpf: Remove unused headers

The header <linux/version.h> is useless in sampleip_kern.c
and trace_event_kern.c, remove it.

Signed-off-by: Lu Wei <luwei32@huawei.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210324083147.149278-1-luwei32@huawei.com
3 years agolibbpf: Add bpf object kern_version attribute setter
Rafael David Tinoco [Tue, 23 Mar 2021 04:09:52 +0000 (01:09 -0300)]
libbpf: Add bpf object kern_version attribute setter

Unfortunately some distros don't have their kernel version defined
accurately in <linux/version.h> due to different long term support
reasons.

It is important to have a way to override the bpf kern_version
attribute during runtime: some old kernels might still check for
kern_version attribute during bpf_prog_load().

Signed-off-by: Rafael David Tinoco <rafaeldtinoco@ubuntu.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20210323040952.2118241-1-rafaeldtinoco@ubuntu.com
3 years agoMerge branch 'add support for batched ops in LPM trie'
Alexei Starovoitov [Fri, 26 Mar 2021 01:51:08 +0000 (18:51 -0700)]
Merge branch 'add support for batched ops in LPM trie'

Pedro Tammela says:

====================

The patch itself is straightforward thanks to the infrastructure that is
already in-place.

The tests follows the other '*_map_batch_ops' tests with minor tweaks.

v1 -> v2:
Fixes for checkpatch warnings
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
3 years agobpf: selftests: Add tests for batched ops in LPM trie maps
Pedro Tammela [Tue, 23 Mar 2021 02:50:54 +0000 (23:50 -0300)]
bpf: selftests: Add tests for batched ops in LPM trie maps

Uses the already existing infrastructure for testing batched ops.
The testing code is essentially the same, with minor tweaks for this use
case.

Suggested-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210323025058.315763-3-pctammela@gmail.com
3 years agobpf: Add support for batched ops in LPM trie maps
Pedro Tammela [Tue, 23 Mar 2021 02:50:53 +0000 (23:50 -0300)]
bpf: Add support for batched ops in LPM trie maps

Suggested-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210323025058.315763-2-pctammela@gmail.com
3 years agoselftests/bpf: Better error messages for ima_setup.sh failures
KP Singh [Mon, 22 Mar 2021 17:07:20 +0000 (17:07 +0000)]
selftests/bpf: Better error messages for ima_setup.sh failures

The current implementation uses the CHECK_FAIL macro which does not
provide useful error messages when the script fails. Use the CHECK macro
instead and provide more descriptive messages to aid debugging.

Signed-off-by: KP Singh <kpsingh@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210322170720.2926715-1-kpsingh@kernel.org
3 years agobpf: Fix NULL pointer dereference in bpf_get_local_storage() helper
Yonghong Song [Tue, 23 Mar 2021 05:51:46 +0000 (22:51 -0700)]
bpf: Fix NULL pointer dereference in bpf_get_local_storage() helper

Jiri Olsa reported a bug ([1]) in kernel where cgroup local
storage pointer may be NULL in bpf_get_local_storage() helper.
There are two issues uncovered by this bug:
  (1). kprobe or tracepoint prog incorrectly sets cgroup local storage
       before prog run,
  (2). due to change from preempt_disable to migrate_disable,
       preemption is possible and percpu storage might be overwritten
       by other tasks.

This issue (1) is fixed in [2]. This patch tried to address issue (2).
The following shows how things can go wrong:
  task 1:   bpf_cgroup_storage_set() for percpu local storage
         preemption happens
  task 2:   bpf_cgroup_storage_set() for percpu local storage
         preemption happens
  task 1:   run bpf program

task 1 will effectively use the percpu local storage setting by task 2
which will be either NULL or incorrect ones.

Instead of just one common local storage per cpu, this patch fixed
the issue by permitting 8 local storages per cpu and each local
storage is identified by a task_struct pointer. This way, we
allow at most 8 nested preemption between bpf_cgroup_storage_set()
and bpf_cgroup_storage_unset(). The percpu local storage slot
is released (calling bpf_cgroup_storage_unset()) by the same task
after bpf program finished running.
bpf_test_run() is also fixed to use the new bpf_cgroup_storage_set()
interface.

The patch is tested on top of [2] with reproducer in [1].
Without this patch, kernel will emit error in 2-3 minutes.
With this patch, after one hour, still no error.

 [1] https://lore.kernel.org/bpf/CAKH8qBuXCfUz=w8L+Fj74OaUpbosO29niYwTki7e3Ag044_aww@mail.gmail.com/T
 [2] https://lore.kernel.org/bpf/20210309185028.3763817-1-yhs@fb.com

Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Roman Gushchin <guro@fb.com>
Link: https://lore.kernel.org/bpf/20210323055146.3334476-1-yhs@fb.com
3 years agolibbpf: Constify few bpf_program getters
Andrii Nakryiko [Wed, 24 Mar 2021 17:29:41 +0000 (10:29 -0700)]
libbpf: Constify few bpf_program getters

bpf_program__get_type() and bpf_program__get_expected_attach_type() shouldn't
modify given bpf_program, so mark input parameter as const struct bpf_program.
This eliminates unnecessary compilation warnings or explicit casts in user
programs.

Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20210324172941.2609884-1-andrii@kernel.org
3 years agobpf: Fix typo 'accesible' into 'accessible'
Ricardo Ribalda [Thu, 18 Mar 2021 20:22:22 +0000 (21:22 +0100)]
bpf: Fix typo 'accesible' into 'accessible'

Trivial fix.

Signed-off-by: Ricardo Ribalda <ribalda@chromium.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20210318202223.164873-8-ribalda@chromium.org
3 years agobpf: Undo ptr_to_map_key alu sanitation for now
Daniel Borkmann [Thu, 25 Mar 2021 23:41:34 +0000 (00:41 +0100)]
bpf: Undo ptr_to_map_key alu sanitation for now

Remove PTR_TO_MAP_KEY for the time being from being sanitized on pointer ALU
through sanitize_ptr_alu() mainly for 3 reasons:

  1) It's currently unused and not available from unprivileged. However that by
     itself is not yet a strong reason to drop the code.

  2) Commit 69c087ba6225 ("bpf: Add bpf_for_each_map_elem() helper") implemented
     the sanitation not fully correct in that unlike stack or map_value pointer
     it doesn't probe whether the access to the map key /after/ the simulated ALU
     operation is still in bounds. This means that the generated mask can truncate
     the offset in the non-speculative domain whereas it should only truncate in
     the speculative domain. The verifier should instead reject such program as
     we do for other types.

  3) Given the recent fixes from f232326f6966 ("bpf: Prohibit alu ops for pointer
     types not defining ptr_limit"), 10d2bb2e6b1d ("bpf: Fix off-by-one for area
     size in creating mask to left"), b5871dca250c ("bpf: Simplify alu_limit masking
     for pointer arithmetic") as well as 1b1597e64e1a ("bpf: Add sanity check for
     upper ptr_limit") the code changed quite a bit and the merge in efd13b71a3fa
     broke the PTR_TO_MAP_KEY case due to an incorrect merge conflict.

Remove the relevant pieces for the time being and we can rework the PTR_TO_MAP_KEY
case once everything settles.

Fixes: efd13b71a3fa ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net")
Fixes: 69c087ba6225 ("bpf: Add bpf_for_each_map_elem() helper")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
3 years agoMerge https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
David S. Miller [Thu, 25 Mar 2021 23:30:46 +0000 (16:30 -0700)]
Merge https://git./linux/kernel/git/bpf/bpf-next

Alexei Starovoitov says:

====================
pull-request: bpf-next 2021-03-24

The following pull-request contains BPF updates for your *net-next* tree.

We've added 37 non-merge commits during the last 15 day(s) which contain
a total of 65 files changed, 3200 insertions(+), 738 deletions(-).

The main changes are:

1) Static linking of multiple BPF ELF files, from Andrii.

2) Move drop error path to devmap for XDP_REDIRECT, from Lorenzo.

3) Spelling fixes from various folks.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
David S. Miller [Thu, 25 Mar 2021 22:31:22 +0000 (15:31 -0700)]
Merge git://git./linux/kernel/git/netdev/net

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Thu, 25 Mar 2021 18:43:43 +0000 (11:43 -0700)]
Merge branch 'akpm' (patches from Andrew)

Merge misc fixes from Andrew Morton:
 "14 patches.

  Subsystems affected by this patch series: mm (hugetlb, kasan, gup,
  selftests, z3fold, kfence, memblock, and highmem), squashfs, ia64,
  gcov, and mailmap"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  mailmap: update Andrey Konovalov's email address
  mm/highmem: fix CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP
  mm: memblock: fix section mismatch warning again
  kfence: make compatible with kmemleak
  gcov: fix clang-11+ support
  ia64: fix format strings for err_inject
  ia64: mca: allocate early mca with GFP_ATOMIC
  squashfs: fix xattr id and id lookup sanity checks
  squashfs: fix inode lookup sanity checks
  z3fold: prevent reclaim/free race for headless pages
  selftests/vm: fix out-of-tree build
  mm/mmu_notifiers: ensure range_end() is paired with range_start()
  kasan: fix per-page tags for non-page_alloc pages
  hugetlb_cgroup: fix imbalanced css_get and css_put pair for shared mappings

3 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma
Linus Torvalds [Thu, 25 Mar 2021 18:23:35 +0000 (11:23 -0700)]
Merge tag 'for-linus' of git://git./linux/kernel/git/rdma/rdma

Pull rdma fixes from Jason Gunthorpe:
 "Not much going on, just some small bug fixes:

   - Typo causing a regression in mlx5 devx

   - Regression in the recent hns rework causing the HW to get out of
     sync

   - Long-standing cxgb4 adaptor crash when destroying cm ids"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
  RDMA/cxgb4: Fix adapter LE hash errors while destroying ipv6 listening server
  RDMA/hns: Fix bug during CMDQ initialization
  RDMA/mlx5: Fix typo in destroy_mkey inbox

3 years agoMerge tag 'mfd-fixes-5.12' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd
Linus Torvalds [Thu, 25 Mar 2021 18:11:45 +0000 (11:11 -0700)]
Merge tag 'mfd-fixes-5.12' of git://git./linux/kernel/git/lee/mfd

Pull mfs fix from Lee Jones:
 "Unconstify editable placeholder structures"

* tag 'mfd-fixes-5.12' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd:
  mfd: intel_quark_i2c_gpio: Revert "Constify static struct resources"

3 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Thu, 25 Mar 2021 18:07:40 +0000 (11:07 -0700)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Will Deacon:
 "Minor fixes all over, ranging from typos to tests to errata
  workarounds:

   - Fix possible memory hotplug failure with KASLR

   - Fix FFR value in SVE kselftest

   - Fix backtraces reported in /proc/$pid/stack

   - Disable broken CnP implementation on NVIDIA Carmel

   - Typo fixes and ACPI documentation clarification

   - Fix some W=1 warnings"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: kernel: disable CNP on Carmel
  arm64/process.c: fix Wmissing-prototypes build warnings
  kselftest/arm64: sve: Do not use non-canonical FFR register value
  arm64: mm: correct the inside linear map range during hotplug check
  arm64: kdump: update ppos when reading elfcorehdr
  arm64: cpuinfo: Fix a typo
  Documentation: arm64/acpi : clarify arm64 support of IBFT
  arm64: stacktrace: don't trace arch_stack_walk()
  arm64: csum: cast to the proper type

3 years agomailmap: update the email address for Chris Chiu
Chris Chiu [Thu, 25 Mar 2021 14:04:19 +0000 (22:04 +0800)]
mailmap: update the email address for Chris Chiu

Redirect my older email addresses in the git logs.

Signed-off-by: Chris Chiu <chris.chiu@canonical.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomailmap: update Andrey Konovalov's email address
Andrey Konovalov [Thu, 25 Mar 2021 04:37:56 +0000 (21:37 -0700)]
mailmap: update Andrey Konovalov's email address

Use my personal email, the @google.com one will stop functioning soon.

Link: https://lkml.kernel.org/r/ead0e9c32a2f70e0bde6f63b3b9470e0ef13d2ee.1616107969.git.andreyknvl@google.com
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomm/highmem: fix CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP
Ira Weiny [Thu, 25 Mar 2021 04:37:53 +0000 (21:37 -0700)]
mm/highmem: fix CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP

The kernel test robot found that __kmap_local_sched_out() was not
correctly skipping the guard pages when DEBUG_KMAP_LOCAL_FORCE_MAP was
set.[1] This was due to DEBUG_HIGHMEM check being used.

Change the configuration check to be correct.

[1] https://lore.kernel.org/lkml/20210304083825.GB17830@xsang-OptiPlex-9020/

Link: https://lkml.kernel.org/r/20210318230657.1497881-1-ira.weiny@intel.com
Fixes: 0e91a0c6984c ("mm/highmem: Provide CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP")
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reported-by: kernel test robot <oliver.sang@intel.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Oliver Sang <oliver.sang@intel.com>
Cc: Chaitanya Kulkarni <Chaitanya.Kulkarni@wdc.com>
Cc: David Sterba <dsterba@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomm: memblock: fix section mismatch warning again
Mike Rapoport [Thu, 25 Mar 2021 04:37:50 +0000 (21:37 -0700)]
mm: memblock: fix section mismatch warning again

Commit 34dc2efb39a2 ("memblock: fix section mismatch warning") marked
memblock_bottom_up() and memblock_set_bottom_up() as __init, but they
could be referenced from non-init functions like
memblock_find_in_range_node() on architectures that enable
CONFIG_ARCH_KEEP_MEMBLOCK.

For such builds kernel test robot reports:

   WARNING: modpost: vmlinux.o(.text+0x74fea4): Section mismatch in reference from the function memblock_find_in_range_node() to the function .init.text:memblock_bottom_up()
   The function memblock_find_in_range_node() references the function __init memblock_bottom_up().
   This is often because memblock_find_in_range_node lacks a __init  annotation or the annotation of memblock_bottom_up is wrong.

Replace __init annotations with __init_memblock annotations so that the
appropriate section will be selected depending on
CONFIG_ARCH_KEEP_MEMBLOCK.

Link: https://lore.kernel.org/lkml/202103160133.UzhgY0wt-lkp@intel.com
Link: https://lkml.kernel.org/r/20210316171347.14084-1-rppt@kernel.org
Fixes: 34dc2efb39a2 ("memblock: fix section mismatch warning")
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agokfence: make compatible with kmemleak
Marco Elver [Thu, 25 Mar 2021 04:37:47 +0000 (21:37 -0700)]
kfence: make compatible with kmemleak

Because memblock allocations are registered with kmemleak, the KFENCE
pool was seen by kmemleak as one large object.  Later allocations
through kfence_alloc() that were registered with kmemleak via
slab_post_alloc_hook() would then overlap and trigger a warning.
Therefore, once the pool is initialized, we can remove (free) it from
kmemleak again, since it should be treated as allocator-internal and be
seen as "free memory".

The second problem is that kmemleak is passed the rounded size, and not
the originally requested size, which is also the size of KFENCE objects.
To avoid kmemleak scanning past the end of an object and trigger a
KFENCE out-of-bounds error, fix the size if it is a KFENCE object.

For simplicity, to avoid a call to kfence_ksize() in
slab_post_alloc_hook() (and avoid new IS_ENABLED(CONFIG_DEBUG_KMEMLEAK)
guard), just call kfence_ksize() in mm/kmemleak.c:create_object().

Link: https://lkml.kernel.org/r/20210317084740.3099921-1-elver@google.com
Signed-off-by: Marco Elver <elver@google.com>
Reported-by: Luis Henriques <lhenriques@suse.de>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: Luis Henriques <lhenriques@suse.de>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Jann Horn <jannh@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agogcov: fix clang-11+ support
Nick Desaulniers [Thu, 25 Mar 2021 04:37:44 +0000 (21:37 -0700)]
gcov: fix clang-11+ support

LLVM changed the expected function signatures for llvm_gcda_start_file()
and llvm_gcda_emit_function() in the clang-11 release.  Users of
clang-11 or newer may have noticed their kernels failing to boot due to
a panic when enabling CONFIG_GCOV_KERNEL=y +CONFIG_GCOV_PROFILE_ALL=y.
Fix up the function signatures so calling these functions doesn't panic
the kernel.

Link: https://reviews.llvm.org/rGcdd683b516d147925212724b09ec6fb792a40041
Link: https://reviews.llvm.org/rG13a633b438b6500ecad9e4f936ebadf3411d0f44
Link: https://lkml.kernel.org/r/20210312224132.3413602-2-ndesaulniers@google.com
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Reported-by: Prasad Sodagudi <psodagud@quicinc.com>
Suggested-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Fangrui Song <maskray@google.com>
Tested-by: Nathan Chancellor <nathan@kernel.org>
Acked-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Cc: <stable@vger.kernel.org> [5.4+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoia64: fix format strings for err_inject
Sergei Trofimovich [Thu, 25 Mar 2021 04:37:41 +0000 (21:37 -0700)]
ia64: fix format strings for err_inject

Fix warning with %lx / u64 mismatch:

  arch/ia64/kernel/err_inject.c: In function 'show_resources':
  arch/ia64/kernel/err_inject.c:62:22: warning:
    format '%lx' expects argument of type 'long unsigned int',
    but argument 3 has type 'u64' {aka 'long long unsigned int'}
     62 |  return sprintf(buf, "%lx", name[cpu]);   \
        |                      ^~~~~~~

Link: https://lkml.kernel.org/r/20210313104312.1548232-1-slyfox@gentoo.org
Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoia64: mca: allocate early mca with GFP_ATOMIC
Sergei Trofimovich [Thu, 25 Mar 2021 04:37:38 +0000 (21:37 -0700)]
ia64: mca: allocate early mca with GFP_ATOMIC

The sleep warning happens at early boot right at secondary CPU
activation bootup:

    smp: Bringing up secondary CPUs ...
    BUG: sleeping function called from invalid context at mm/page_alloc.c:4942
    in_atomic(): 0, irqs_disabled(): 1, non_block: 0, pid: 0, name: swapper/1
    CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.12.0-rc2-00007-g79e228d0b611-dirty #99
    ..
    Call Trace:
      show_stack+0x90/0xc0
      dump_stack+0x150/0x1c0
      ___might_sleep+0x1c0/0x2a0
      __might_sleep+0xa0/0x160
      __alloc_pages_nodemask+0x1a0/0x600
      alloc_page_interleave+0x30/0x1c0
      alloc_pages_current+0x2c0/0x340
      __get_free_pages+0x30/0xa0
      ia64_mca_cpu_init+0x2d0/0x3a0
      cpu_init+0x8b0/0x1440
      start_secondary+0x60/0x700
      start_ap+0x750/0x780
    Fixed BSP b0 value from CPU 1

As I understand interrupts are not enabled yet and system has a lot of
memory.  There is little chance to sleep and switch to GFP_ATOMIC should
be a no-op.

Link: https://lkml.kernel.org/r/20210315085045.204414-1-slyfox@gentoo.org
Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agosquashfs: fix xattr id and id lookup sanity checks
Phillip Lougher [Thu, 25 Mar 2021 04:37:35 +0000 (21:37 -0700)]
squashfs: fix xattr id and id lookup sanity checks

The checks for maximum metadata block size is missing
SQUASHFS_BLOCK_OFFSET (the two byte length count).

Link: https://lkml.kernel.org/r/2069685113.2081245.1614583677427@webmail.123-reg.co.uk
Fixes: f37aa4c7366e23f ("squashfs: add more sanity checks in id lookup")
Signed-off-by: Phillip Lougher <phillip@squashfs.org.uk>
Cc: Sean Nyekjaer <sean@geanix.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agosquashfs: fix inode lookup sanity checks
Sean Nyekjaer [Thu, 25 Mar 2021 04:37:32 +0000 (21:37 -0700)]
squashfs: fix inode lookup sanity checks

When mouting a squashfs image created without inode compression it fails
with: "unable to read inode lookup table"

It turns out that the BLOCK_OFFSET is missing when checking the
SQUASHFS_METADATA_SIZE agaist the actual size.

Link: https://lkml.kernel.org/r/20210226092903.1473545-1-sean@geanix.com
Fixes: eabac19e40c0 ("squashfs: add more sanity checks in inode lookup")
Signed-off-by: Sean Nyekjaer <sean@geanix.com>
Acked-by: Phillip Lougher <phillip@squashfs.org.uk>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoz3fold: prevent reclaim/free race for headless pages
Thomas Hebb [Thu, 25 Mar 2021 04:37:29 +0000 (21:37 -0700)]
z3fold: prevent reclaim/free race for headless pages

Commit ca0246bb97c2 ("z3fold: fix possible reclaim races") introduced
the PAGE_CLAIMED flag "to avoid racing on a z3fold 'headless' page
release." By atomically testing and setting the bit in each of
z3fold_free() and z3fold_reclaim_page(), a double-free was avoided.

However, commit dcf5aedb24f8 ("z3fold: stricter locking and more careful
reclaim") appears to have unintentionally broken this behavior by moving
the PAGE_CLAIMED check in z3fold_reclaim_page() to after the page lock
gets taken, which only happens for non-headless pages.  For headless
pages, the check is now skipped entirely and races can occur again.

I have observed such a race on my system:

    page:00000000ffbd76b7 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x165316
    flags: 0x2ffff0000000000()
    raw: 02ffff0000000000 ffffea0004535f48 ffff8881d553a170 0000000000000000
    raw: 0000000000000000 0000000000000011 00000000ffffffff 0000000000000000
    page dumped because: VM_BUG_ON_PAGE(page_ref_count(page) == 0)
    ------------[ cut here ]------------
    kernel BUG at include/linux/mm.h:707!
    invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI
    CPU: 2 PID: 291928 Comm: kworker/2:0 Tainted: G    B             5.10.7-arch1-1-kasan #1
    Hardware name: Gigabyte Technology Co., Ltd. H97N-WIFI/H97N-WIFI, BIOS F9b 03/03/2016
    Workqueue: zswap-shrink shrink_worker
    RIP: 0010:__free_pages+0x10a/0x130
    Code: c1 e7 06 48 01 ef 45 85 e4 74 d1 44 89 e6 31 d2 41 83 ec 01 e8 e7 b0 ff ff eb da 48 c7 c6 e0 32 91 88 48 89 ef e8 a6 89 f8 ff <0f> 0b 4c 89 e7 e8 fc 79 07 00 e9 33 ff ff ff 48 89 ef e8 ff 79 07
    RSP: 0000:ffff88819a2ffb98 EFLAGS: 00010296
    RAX: 0000000000000000 RBX: ffffea000594c5a8 RCX: 0000000000000000
    RDX: 1ffffd4000b298b7 RSI: 0000000000000000 RDI: ffffea000594c5b8
    RBP: ffffea000594c580 R08: 000000000000003e R09: ffff8881d5520bbb
    R10: ffffed103aaa4177 R11: 0000000000000001 R12: ffffea000594c5b4
    R13: 0000000000000000 R14: ffff888165316000 R15: ffffea000594c588
    FS:  0000000000000000(0000) GS:ffff8881d5500000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 00007f7c8c3654d8 CR3: 0000000103f42004 CR4: 00000000001706e0
    Call Trace:
     z3fold_zpool_shrink+0x9b6/0x1240
     shrink_worker+0x35/0x90
     process_one_work+0x70c/0x1210
     worker_thread+0x539/0x1200
     kthread+0x330/0x400
     ret_from_fork+0x22/0x30
    Modules linked in: rfcomm ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter ccm algif_aead des_generic libdes ecb algif_skcipher cmac bnep md4 algif_hash af_alg vfat fat intel_rapl_msr intel_rapl_common x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel iwlmvm hid_logitech_hidpp kvm at24 mac80211 snd_hda_codec_realtek iTCO_wdt snd_hda_codec_generic intel_pmc_bxt snd_hda_codec_hdmi ledtrig_audio iTCO_vendor_support mei_wdt mei_hdcp snd_hda_intel snd_intel_dspcfg libarc4 soundwire_intel irqbypass iwlwifi soundwire_generic_allocation rapl soundwire_cadence intel_cstate snd_hda_codec intel_uncore btusb joydev mousedev snd_usb_audio pcspkr btrtl uvcvideo nouveau btbcm i2c_i801 btintel snd_hda_core videobuf2_vmalloc i2c_smbus snd_usbmidi_lib videobuf2_memops bluetooth snd_hwdep soundwire_bus snd_soc_rt5640 videobuf2_v4l2 cfg80211 snd_soc_rl6231 videobuf2_common snd_rawmidi lpc_ich alx videodev mdio snd_seq_device snd_soc_core mc ecdh_generic mxm_wmi mei_me
     hid_logitech_dj wmi snd_compress e1000e ac97_bus mei ttm rfkill snd_pcm_dmaengine ecc snd_pcm snd_timer snd soundcore mac_hid acpi_pad pkcs8_key_parser it87 hwmon_vid crypto_user fuse ip_tables x_tables ext4 crc32c_generic crc16 mbcache jbd2 dm_crypt cbc encrypted_keys trusted tpm rng_core usbhid dm_mod crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel aesni_intel crypto_simd cryptd glue_helper xhci_pci xhci_pci_renesas i915 video intel_gtt i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops cec drm agpgart
    ---[ end trace 126d646fc3dc0ad8 ]---

To fix the issue, re-add the earlier test and set in the case where we
have a headless page.

Link: https://lkml.kernel.org/r/c8106dbe6d8390b290cd1d7f873a2942e805349e.1615452048.git.tommyhebb@gmail.com
Fixes: dcf5aedb24f8 ("z3fold: stricter locking and more careful reclaim")
Signed-off-by: Thomas Hebb <tommyhebb@gmail.com>
Reviewed-by: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: Jongseok Kim <ks77sj@gmail.com>
Cc: Snild Dolkow <snild@sony.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoselftests/vm: fix out-of-tree build
Rong Chen [Thu, 25 Mar 2021 04:37:26 +0000 (21:37 -0700)]
selftests/vm: fix out-of-tree build

When building out-of-tree, attempting to make target from $(OUTPUT) directory:

  make[1]: *** No rule to make target '$(OUTPUT)/protection_keys.c', needed by '$(OUTPUT)/protection_keys_32'.

Link: https://lkml.kernel.org/r/20210315094700.522753-1-rong.a.chen@intel.com
Signed-off-by: Rong Chen <rong.a.chen@intel.com>
Reported-by: kernel test robot <lkp@intel.com>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomm/mmu_notifiers: ensure range_end() is paired with range_start()
Sean Christopherson [Thu, 25 Mar 2021 04:37:23 +0000 (21:37 -0700)]
mm/mmu_notifiers: ensure range_end() is paired with range_start()

If one or more notifiers fails .invalidate_range_start(), invoke
.invalidate_range_end() for "all" notifiers.  If there are multiple
notifiers, those that did not fail are expecting _start() and _end() to
be paired, e.g.  KVM's mmu_notifier_count would become imbalanced.
Disallow notifiers that can fail _start() from implementing _end() so
that it's unnecessary to either track which notifiers rejected _start(),
or had already succeeded prior to a failed _start().

Note, the existing behavior of calling _start() on all notifiers even
after a previous notifier failed _start() was an unintented "feature".
Make it canon now that the behavior is depended on for correctness.

As of today, the bug is likely benign:

  1. The only caller of the non-blocking notifier is OOM kill.
  2. The only notifiers that can fail _start() are the i915 and Nouveau
     drivers.
  3. The only notifiers that utilize _end() are the SGI UV GRU driver
     and KVM.
  4. The GRU driver will never coincide with the i195/Nouveau drivers.
  5. An imbalanced kvm->mmu_notifier_count only causes soft lockup in the
     _guest_, and the guest is already doomed due to being an OOM victim.

Fix the bug now to play nice with future usage, e.g.  KVM has a
potential use case for blocking memslot updates in KVM while an
invalidation is in-progress, and failure to unblock would result in said
updates being blocked indefinitely and hanging.

Found by inspection.  Verified by adding a second notifier in KVM that
periodically returns -EAGAIN on non-blockable ranges, triggering OOM,
and observing that KVM exits with an elevated notifier count.

Link: https://lkml.kernel.org/r/20210311180057.1582638-1-seanjc@google.com
Fixes: 93065ac753e4 ("mm, oom: distinguish blockable mode for mmu notifiers")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Suggested-by: Jason Gunthorpe <jgg@ziepe.ca>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ben Gardon <bgardon@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Jérôme Glisse" <jglisse@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Dimitri Sivanich <dimitri.sivanich@hpe.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agokasan: fix per-page tags for non-page_alloc pages
Andrey Konovalov [Thu, 25 Mar 2021 04:37:20 +0000 (21:37 -0700)]
kasan: fix per-page tags for non-page_alloc pages

To allow performing tag checks on page_alloc addresses obtained via
page_address(), tag-based KASAN modes store tags for page_alloc
allocations in page->flags.

Currently, the default tag value stored in page->flags is 0x00.
Therefore, page_address() returns a 0x00ffff...  address for pages that
were not allocated via page_alloc.

This might cause problems.  A particular case we encountered is a
conflict with KFENCE.  If a KFENCE-allocated slab object is being freed
via kfree(page_address(page) + offset), the address passed to kfree()
will get tagged with 0x00 (as slab pages keep the default per-page
tags).  This leads to is_kfence_address() check failing, and a KFENCE
object ending up in normal slab freelist, which causes memory
corruptions.

This patch changes the way KASAN stores tag in page-flags: they are now
stored xor'ed with 0xff.  This way, KASAN doesn't need to initialize
per-page flags for every created page, which might be slow.

With this change, page_address() returns natively-tagged (with 0xff)
pointers for pages that didn't have tags set explicitly.

This patch fixes the encountered conflict with KFENCE and prevents more
similar issues that can occur in the future.

Link: https://lkml.kernel.org/r/1a41abb11c51b264511d9e71c303bb16d5cb367b.1615475452.git.andreyknvl@google.com
Fixes: 2813b9c02962 ("kasan, mm, arm64: tag non slab memory allocated via pagealloc")
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Reviewed-by: Marco Elver <elver@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Peter Collingbourne <pcc@google.com>
Cc: Evgenii Stepanov <eugenis@google.com>
Cc: Branislav Rankov <Branislav.Rankov@arm.com>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agohugetlb_cgroup: fix imbalanced css_get and css_put pair for shared mappings
Miaohe Lin [Thu, 25 Mar 2021 04:37:17 +0000 (21:37 -0700)]
hugetlb_cgroup: fix imbalanced css_get and css_put pair for shared mappings

The current implementation of hugetlb_cgroup for shared mappings could
have different behavior.  Consider the following two scenarios:

 1.Assume initial css reference count of hugetlb_cgroup is 1:
  1.1 Call hugetlb_reserve_pages with from = 1, to = 2. So css reference
      count is 2 associated with 1 file_region.
  1.2 Call hugetlb_reserve_pages with from = 2, to = 3. So css reference
      count is 3 associated with 2 file_region.
  1.3 coalesce_file_region will coalesce these two file_regions into
      one. So css reference count is 3 associated with 1 file_region
      now.

 2.Assume initial css reference count of hugetlb_cgroup is 1 again:
  2.1 Call hugetlb_reserve_pages with from = 1, to = 3. So css reference
      count is 2 associated with 1 file_region.

Therefore, we might have one file_region while holding one or more css
reference counts. This inconsistency could lead to imbalanced css_get()
and css_put() pair. If we do css_put one by one (i.g. hole punch case),
scenario 2 would put one more css reference. If we do css_put all
together (i.g. truncate case), scenario 1 will leak one css reference.

The imbalanced css_get() and css_put() pair would result in a non-zero
reference when we try to destroy the hugetlb cgroup. The hugetlb cgroup
directory is removed __but__ associated resource is not freed. This
might result in OOM or can not create a new hugetlb cgroup in a busy
workload ultimately.

In order to fix this, we have to make sure that one file_region must
hold exactly one css reference. So in coalesce_file_region case, we
should release one css reference before coalescence. Also only put css
reference when the entire file_region is removed.

The last thing to note is that the caller of region_add() will only hold
one reference to h_cg->css for the whole contiguous reservation region.
But this area might be scattered when there are already some
file_regions reside in it. As a result, many file_regions may share only
one h_cg->css reference. In order to ensure that one file_region must
hold exactly one css reference, we should do css_get() for each
file_region and release the reference held by caller when they are done.

[linmiaohe@huawei.com: fix imbalanced css_get and css_put pair for shared mappings]
Link: https://lkml.kernel.org/r/20210316023002.53921-1-linmiaohe@huawei.com
Link: https://lkml.kernel.org/r/20210301120540.37076-1-linmiaohe@huawei.com
Fixes: 075a61d07a8e ("hugetlb_cgroup: add accounting for shared mappings")
Reported-by: kernel test robot <lkp@intel.com> (auto build test ERROR)
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mina Almasry <almasrymina@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoRDMA/cxgb4: Fix adapter LE hash errors while destroying ipv6 listening server
Potnuri Bharat Teja [Wed, 24 Mar 2021 19:04:53 +0000 (00:34 +0530)]
RDMA/cxgb4: Fix adapter LE hash errors while destroying ipv6 listening server

Not setting the ipv6 bit while destroying ipv6 listening servers may
result in potential fatal adapter errors due to lookup engine memory hash
errors. Therefore always set ipv6 field while destroying ipv6 listening
servers.

Fixes: 830662f6f032 ("RDMA/cxgb4: Add support for active and passive open connection with IPv6 address")
Link: https://lore.kernel.org/r/20210324190453.8171-1-bharat@chelsio.com
Signed-off-by: Potnuri Bharat Teja <bharat@chelsio.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
3 years agoarm64: kernel: disable CNP on Carmel
Rich Wiley [Wed, 24 Mar 2021 00:28:09 +0000 (17:28 -0700)]
arm64: kernel: disable CNP on Carmel

On NVIDIA Carmel cores, CNP behaves differently than it does on standard
ARM cores. On Carmel, if two cores have CNP enabled and share an L2 TLB
entry created by core0 for a specific ASID, a non-shareable TLBI from
core1 may still see the shared entry. On standard ARM cores, that TLBI
will invalidate the shared entry as well.

This causes issues with patchsets that attempt to do local TLBIs based
on cpumasks instead of broadcast TLBIs. Avoid these issues by disabling
CNP support for NVIDIA Carmel cores.

Signed-off-by: Rich Wiley <rwiley@nvidia.com>
Link: https://lore.kernel.org/r/20210324002809.30271-1-rwiley@nvidia.com
[will: Fix pre-existing whitespace issue]
Signed-off-by: Will Deacon <will@kernel.org>
3 years agoarm64/process.c: fix Wmissing-prototypes build warnings
Maninder Singh [Wed, 24 Mar 2021 06:54:58 +0000 (12:24 +0530)]
arm64/process.c: fix Wmissing-prototypes build warnings

Fix GCC warnings reported when building with "-Wmissing-prototypes":

  arch/arm64/kernel/process.c:261:6: warning: no previous prototype for '__show_regs' [-Wmissing-prototypes]
      261 | void __show_regs(struct pt_regs *regs)
          |      ^~~~~~~~~~~
  arch/arm64/kernel/process.c:307:6: warning: no previous prototype for '__show_regs_alloc_free' [-Wmissing-prototypes]
      307 | void __show_regs_alloc_free(struct pt_regs *regs)
          |      ^~~~~~~~~~~~~~~~~~~~~~
  arch/arm64/kernel/process.c:365:5: warning: no previous prototype for 'arch_dup_task_struct' [-Wmissing-prototypes]
      365 | int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
          |     ^~~~~~~~~~~~~~~~~~~~
  arch/arm64/kernel/process.c:546:41: warning: no previous prototype for '__switch_to' [-Wmissing-prototypes]
      546 | __notrace_funcgraph struct task_struct *__switch_to(struct task_struct *prev,
          |                                         ^~~~~~~~~~~
  arch/arm64/kernel/process.c:710:25: warning: no previous prototype for 'arm64_preempt_schedule_irq' [-Wmissing-prototypes]
      710 | asmlinkage void __sched arm64_preempt_schedule_irq(void)
          |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~

Link: https://lore.kernel.org/lkml/202103192250.AennsfXM-lkp@intel.com
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Maninder Singh <maninder1.s@samsung.com>
Link: https://lore.kernel.org/r/1616568899-986-1-git-send-email-maninder1.s@samsung.com
Signed-off-by: Will Deacon <will@kernel.org>
3 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 25 Mar 2021 01:16:04 +0000 (18:16 -0700)]
Merge git://git./linux/kernel/git/netdev/net

Pull networking fixes from David Miller:
 "Various fixes, all over:

   1) Fix overflow in ptp_qoriq_adjfine(), from Yangbo Lu.

   2) Always store the rx queue mapping in veth, from Maciej
      Fijalkowski.

   3) Don't allow vmlinux btf in map_create, from Alexei Starovoitov.

   4) Fix memory leak in octeontx2-af from Colin Ian King.

   5) Use kvalloc in bpf x86 JIT for storing jit'd addresses, from
      Yonghong Song.

   6) Fix tx ptp stats in mlx5, from Aya Levin.

   7) Check correct ip version in tun decap, fropm Roi Dayan.

   8) Fix rate calculation in mlx5 E-Switch code, from arav Pandit.

   9) Work item memork leak in mlx5, from Shay Drory.

  10) Fix ip6ip6 tunnel crash with bpf, from Daniel Borkmann.

  11) Lack of preemptrion awareness in macvlan, from Eric Dumazet.

  12) Fix data race in pxa168_eth, from Pavel Andrianov.

  13) Range validate stab in red_check_params(), from Eric Dumazet.

  14) Inherit vlan filtering setting properly in b53 driver, from
      Florian Fainelli.

  15) Fix rtnl locking in igc driver, from Sasha Neftin.

  16) Pause handling fixes in igc driver, from Muhammad Husaini
      Zulkifli.

  17) Missing rtnl locking in e1000_reset_task, from Vitaly Lifshits.

  18) Use after free in qlcnic, from Lv Yunlong.

  19) fix crash in fritzpci mISDN, from Tong Zhang.

  20) Premature rx buffer reuse in igb, from Li RongQing.

  21) Missing termination of ip[a driver message handler arrays, from
      Alex Elder.

  22) Fix race between "x25_close" and "x25_xmit"/"x25_rx" in hdlc_x25
      driver, from Xie He.

  23) Use after free in c_can_pci_remove(), from Tong Zhang.

  24) Uninitialized variable use in nl80211, from Jarod Wilson.

  25) Off by one size calc in bpf verifier, from Piotr Krysiuk.

  26) Use delayed work instead of deferrable for flowtable GC, from
      Yinjun Zhang.

  27) Fix infinite loop in NPC unmap of octeontx2 driver, from
      Hariprasad Kelam.

  28) Fix being unable to change MTU of dwmac-sun8i devices due to lack
      of fifo sizes, from Corentin Labbe.

  29) DMA use after free in r8169 with WoL, fom Heiner Kallweit.

  30) Mismatched prototypes in isdn-capi, from Arnd Bergmann.

  31) Fix psample UAPI breakage, from Ido Schimmel"

* git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (171 commits)
  psample: Fix user API breakage
  math: Export mul_u64_u64_div_u64
  ch_ktls: fix enum-conversion warning
  octeontx2-af: Fix memory leak of object buf
  ptp_qoriq: fix overflow in ptp_qoriq_adjfine() u64 calcalation
  net: bridge: don't notify switchdev for local FDB addresses
  net/sched: act_ct: clear post_ct if doing ct_clear
  net: dsa: don't assign an error value to tag_ops
  isdn: capi: fix mismatched prototypes
  net/mlx5: SF, do not use ecpu bit for vhca state processing
  net/mlx5e: Fix division by 0 in mlx5e_select_queue
  net/mlx5e: Fix error path for ethtool set-priv-flag
  net/mlx5e: Offload tuple rewrite for non-CT flows
  net/mlx5e: Allow to match on MPLS parameters only for MPLS over UDP
  net/mlx5: Add back multicast stats for uplink representor
  net: ipconfig: ic_dev can be NULL in ic_close_devs
  MAINTAINERS: Combine "QLOGIC QLGE 10Gb ETHERNET DRIVER" sections into one
  docs: networking: Fix a typo
  r8169: fix DMA being used after buffer free if WoL is enabled
  net: ipa: fix init header command validation
  ...

3 years agohinic: avoid gcc -Wrestrict warning
Arnd Bergmann [Wed, 24 Mar 2021 13:07:22 +0000 (14:07 +0100)]
hinic: avoid gcc -Wrestrict warning

With extra warnings enabled, gcc complains that snprintf should not
take the same buffer as source and destination:

drivers/net/ethernet/huawei/hinic/hinic_ethtool.c: In function 'hinic_set_settings_to_hw':
drivers/net/ethernet/huawei/hinic/hinic_ethtool.c:480:9: error: 'snprintf' argument 4 overlaps destination object 'set_link_str' [-Werror=restrict]
  480 |   err = snprintf(set_link_str, SET_LINK_STR_MAX_LEN,
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  481 |           "%sspeed %d ", set_link_str, speed);
      |           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/net/ethernet/huawei/hinic/hinic_ethtool.c:464:7: note: destination object referenced by 'restrict'-qualified argument 1 was declared here
  464 |  char set_link_str[SET_LINK_STR_MAX_LEN] = {0};

Rewrite this to avoid the nested sprintf and instead use separate
buffers, which is simpler.

Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: stmmac: support FPE link partner hand-shaking procedure
Ong Boon Leong [Wed, 24 Mar 2021 09:07:42 +0000 (17:07 +0800)]
net: stmmac: support FPE link partner hand-shaking procedure

In order to discover whether remote station supports frame preemption,
local station sends verify mPacket and expects response mPacket in
return from the remote station.

So, we add the functions to send and handle event when verify mPacket
and response mPacket are exchanged between the networked stations.

The mechanism to handle different FPE states between local and remote
station (link partner) is implemented using workqueue which starts a
task each time there is some sign of verify & response mPacket exchange
as check in FPE IRQ event. The task retries couple of times to try to
spot the states that both stations are ready to enter FPE ON. This allows
different end points to enable FPE at different time and verify-response
mPacket can happen asynchronously. Ultimately, the task will only turn
FPE ON when local station have both exchange response in both directions.

Thanks to Voon Weifeng for implementing the core functions for detecting
FPE events and send mPacket and phylink related change.

Signed-off-by: Ong Boon Leong <boon.leong.ong@intel.com>
Co-developed-by: Voon Weifeng <weifeng.voon@intel.com>
Signed-off-by: Voon Weifeng <weifeng.voon@intel.com>
Co-developed-by: Tan Tee Min <tee.min.tan@intel.com>
Signed-off-by: Tan Tee Min <tee.min.tan@intel.com>
Co-developed-by: Mohammad Athari Bin Ismail <mohammad.athari.ismail@intel.com>
Signed-off-by: Mohammad Athari Bin Ismail <mohammad.athari.ismail@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years ago6lowpan: Fix some typos in nhc_udp.c
Wang Hai [Wed, 24 Mar 2021 06:22:24 +0000 (14:22 +0800)]
6lowpan: Fix some typos in nhc_udp.c

s/Orignal/Original/
s/infered/inferred/

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wang Hai <wanghai38@huawei.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet/packet: Fix a typo in af_packet.c
Wang Hai [Wed, 24 Mar 2021 06:19:31 +0000 (14:19 +0800)]
net/packet: Fix a typo in af_packet.c

s/sequencially/sequentially/

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wang Hai <wanghai38@huawei.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet/tls: Fix a typo in tls_device.c
Wang Hai [Wed, 24 Mar 2021 06:16:22 +0000 (14:16 +0800)]
net/tls: Fix a typo in tls_device.c

s/beggining/beginning/

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wang Hai <wanghai38@huawei.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoSimplify the code by using module_platform_driver macro
Zhichao Cai [Wed, 24 Mar 2021 02:30:47 +0000 (10:30 +0800)]
Simplify the code by using module_platform_driver macro

for ftmac100

Signed-off-by: Zhichao Cai <caizhichao@yulong.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'ipa-versions-and-registers'
David S. Miller [Wed, 24 Mar 2021 23:52:48 +0000 (16:52 -0700)]
Merge branch 'ipa-versions-and-registers'

Alex Elder says:

====================
net: ipa: versions and registers

Version 2 of this series adds kernel-doc descriptions for all
members of the ipa_version enumerated type in patch 2.

The original description of the series is below.

-Alex

This series is sort of a mix of things, generally related to
updating IPA versions and register definitions.

The first patch fixes some version-related tests throughout the code
so the conditions are valid for IPA versions other than the two that
are currently supported.  Support for additional versions is
forthcoming, and this is a preparatory step.

The second patch adds to the set of defined IPA versions, to include
all versions between 3.0 and 4.11.

The next defines an endpoint initialization register that was
previously not being configured.  We now initialize that register
(so that NAT is explicitly disabled) on all AP endpoints.

The fourth adds support for an extra bit in a field in a register,
which is present starting at IPA v4.5.

The last two are sort of standalone.  One just moves a function
definition and makes it private.  The other increases the number of
GSI channels and events supported by the driver, sufficient for IPA
v4.5.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ipa: increase channels and events
Alex Elder [Wed, 24 Mar 2021 13:15:28 +0000 (08:15 -0500)]
net: ipa: increase channels and events

Increase the maximum number of channels and event rings supported by
the driver, to allow the maximum available on the SDX55.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ipa: move ipa_aggr_granularity_val()
Alex Elder [Wed, 24 Mar 2021 13:15:27 +0000 (08:15 -0500)]
net: ipa: move ipa_aggr_granularity_val()

We only use ipa_aggr_granularity_val() inside "ipa_main.c", so it
doesn't really need to be defined in a header file.  It makes some
sense to be grouped with the register definitions, but it is unlike
the other inline functions now defined in "ipa_reg.h".  So move it
into "ipa_main.c" where it's used.  TIMER_FREQUENCY is used only
by that function, so move that definition as well.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ipa: limit local processing context address
Alex Elder [Wed, 24 Mar 2021 13:15:26 +0000 (08:15 -0500)]
net: ipa: limit local processing context address

Not all of the bits of the LOCAL_PKT_PROC_CNTXT register are valid.
Until IPA v4.5, there are 17 valid bits (though the bottom three
must be zero).  Starting with IPA v4.5, 18 bits are valid.

Introduce proc_cntxt_base_addr_encoded() to encode the base address
for use in the register using only the valid bits.

Shorten the name of the register (omit "_BASE") to avoid the need to
wrap the line in the one place it's used.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ipa: define the ENDP_INIT_NAT register
Alex Elder [Wed, 24 Mar 2021 13:15:25 +0000 (08:15 -0500)]
net: ipa: define the ENDP_INIT_NAT register

Define the ENDP_INIT_NAT register for setting up the NAT
configuration for an endpoint.  We aren't using NAT at this
time, so explicitly set the type to BYPASS for all endpoints.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ipa: update version definitions
Alex Elder [Wed, 24 Mar 2021 13:15:24 +0000 (08:15 -0500)]
net: ipa: update version definitions

Add IPA version definitions for all IPA v3.x and v4.x.  Fix the GSI
version associated with IPA version 4.1.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ipa: reduce IPA version assumptions
Alex Elder [Wed, 24 Mar 2021 13:15:23 +0000 (08:15 -0500)]
net: ipa: reduce IPA version assumptions

Modify conditional tests throughout the IPA code so they do not
assume that IPA v3.5.1 is the oldest version supported.  Also remove
assumptions that IPA v4.5 is the newest version of IPA supported.

Augment versions in comments with "+", to be clearer that the
comment applies to a version and subsequent versions.  (E.g.,
"present for IPA v4.2+" instead of just "present for v4.2".)

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agotcp_metrics: tcpm_hash_bucket is strictly local
Eric Dumazet [Wed, 24 Mar 2021 22:01:38 +0000 (15:01 -0700)]
tcp_metrics: tcpm_hash_bucket is strictly local

After commit 098a697b497e ("tcp_metrics: Use a single hash table
for all network namespaces."), tcpm_hash_bucket is local to
net/ipv4/tcp_metrics.c

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>