platform/kernel/linux-starfive.git
21 months agonet: systemport: Enable all RX descriptors for SYSTEMPORT Lite
Florian Fainelli [Fri, 7 Oct 2022 03:42:01 +0000 (20:42 -0700)]
net: systemport: Enable all RX descriptors for SYSTEMPORT Lite

The original commit that added support for the SYSTEMPORT Lite variant
halved the number of RX descriptors due to a confusion between the
number of descriptors and the number of descriptor words. There are 512
descriptor *words* which means 256 descriptors total.

Fixes: 44a4524c54af ("net: systemport: Add support for SYSTEMPORT Lite")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20221007034201.4126054-1-f.fainelli@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: prestera: span: do not unbind things things that were never bound
Maksym Glubokiy [Thu, 6 Oct 2022 19:06:00 +0000 (22:06 +0300)]
net: prestera: span: do not unbind things things that were never bound

Fixes: 13defa275eef ("net: marvell: prestera: Add matchall support")
Signed-off-by: Maksym Glubokiy <maksym.glubokiy@plvision.eu>
Link: https://lore.kernel.org/r/20221006190600.881740-1-maksym.glubokiy@plvision.eu
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoocteontx2-pf: mcs: fix possible memory leak in otx2_probe()
Yang Yingliang [Mon, 10 Oct 2022 03:39:45 +0000 (11:39 +0800)]
octeontx2-pf: mcs: fix possible memory leak in otx2_probe()

In error path after calling cn10k_mcs_init(), cn10k_mcs_free() need
be called to avoid memory leak.

Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months ago] ptp: ocp: remove symlink for second GNSS
Vadim Fedorenko [Mon, 10 Oct 2022 01:29:34 +0000 (04:29 +0300)]
] ptp: ocp: remove symlink for second GNSS

Destroy code doesn't remove symlink for ttyGNSS2 device introduced
earlier. Add cleanup code.

Fixes: 71d7e0850476 ("ptp: ocp: Add second GNSS device")
Signed-off-by: Vadim Fedorenko <vadfed@fb.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agoocteontx2-af: cn10k: mcs: Fix error return code in mcs_register_interrupts()
Yang Yingliang [Sun, 9 Oct 2022 01:51:26 +0000 (09:51 +0800)]
octeontx2-af: cn10k: mcs: Fix error return code in mcs_register_interrupts()

If alloc_mem() fails in mcs_register_interrupts(), it should return error
code.

Fixes: 6c635f78c474 ("octeontx2-af: cn10k: mcs: Handle MCS block interrupts")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agonet: dsa: fix wrong pointer passed to PTR_ERR() in dsa_port_phylink_create()
Yang Yingliang [Sat, 8 Oct 2022 08:39:42 +0000 (16:39 +0800)]
net: dsa: fix wrong pointer passed to PTR_ERR() in dsa_port_phylink_create()

Fix wrong pointer passed to PTR_ERR() in dsa_port_phylink_create() to print
error message.

Fixes: cf5ca4ddc37a ("net: dsa: don't leave dangling pointers in dp->pl when failing")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agoocteontx2-pf: mcs: fix missing unlock in some error paths
Yang Yingliang [Sat, 8 Oct 2022 08:26:50 +0000 (16:26 +0800)]
octeontx2-pf: mcs: fix missing unlock in some error paths

Add the missing unlock in some error paths.

Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agomacvlan: enforce a consistent minimal mtu
Eric Dumazet [Fri, 7 Oct 2022 22:57:43 +0000 (15:57 -0700)]
macvlan: enforce a consistent minimal mtu

macvlan should enforce a minimal mtu of 68, even at link creation.

This patch avoids the current behavior (which could lead to crashes
in ipv6 stack if the link is brought up)

$ ip link add macvlan1 link eno1 mtu 8 type macvlan  # This should fail !
$ ip link sh dev macvlan1
5: macvlan1@eno1: <BROADCAST,MULTICAST> mtu 8 qdisc noop
    state DOWN mode DEFAULT group default qlen 1000
    link/ether 02:47:6c:24:74:82 brd ff:ff:ff:ff:ff:ff
$ ip link set macvlan1 mtu 67
Error: mtu less than device minimum.
$ ip link set macvlan1 mtu 68
$ ip link set macvlan1 mtu 8
Error: mtu less than device minimum.

Fixes: 91572088e3fd ("net: use core MTU range checking in core net infra")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agomISDN: hfcpci: Fix use-after-free bug in hfcpci_softirq
Duoming Zhou [Sun, 9 Oct 2022 06:37:31 +0000 (14:37 +0800)]
mISDN: hfcpci: Fix use-after-free bug in hfcpci_softirq

The function hfcpci_softirq() is a timer handler. If it
is running, the timer_pending() will return 0 and the
del_timer_sync() in HFC_cleanup() will not be executed.
As a result, the use-after-free bug will happen. The
process is shown below:

    (cleanup routine)          |        (timer handler)
HFC_cleanup()                  | hfcpci_softirq()
 if (timer_pending(&hfc_tl))   |
   del_timer_sync()            |
 ...                           | ...
 pci_unregister_driver(hc)     |
  driver_unregister            |  driver_for_each_device
   bus_remove_driver           |   _hfcpci_softirq
    driver_detach              |   ...
     put_device(dev) //[1]FREE |
                               |    dev_get_drvdata(dev) //[2]USE

The device is deallocated is position [1] and used in
position [2].

Fix by removing the "timer_pending" check in HFC_cleanup(),
which makes sure that the hfcpci_softirq() have finished
before the resource is deallocated.

Fixes: 009fc857c5f6 ("mISDN: fix possible use-after-free in HFC_cleanup()")
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agonet: ethernet: bgmac: Remove -Warray-bounds exception
Kees Cook [Thu, 6 Oct 2022 19:20:53 +0000 (12:20 -0700)]
net: ethernet: bgmac: Remove -Warray-bounds exception

GCC-12 emits false positive -Warray-bounds warnings with
CONFIG_UBSAN_SHIFT (-fsanitize=shift). This is fixed in GCC 13[1],
and there is top-level Makefile logic to remove -Warray-bounds for
known-bad GCC versions staring with commit f0be87c42cbd ("gcc-12: disable
'-Warray-bounds' universally for now").

Remove the local work-around.

[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105679

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agonet: ethernet: mediatek: Remove -Warray-bounds exception
Kees Cook [Thu, 6 Oct 2022 19:20:52 +0000 (12:20 -0700)]
net: ethernet: mediatek: Remove -Warray-bounds exception

GCC-12 emits false positive -Warray-bounds warnings with
CONFIG_UBSAN_SHIFT (-fsanitize=shift). This is fixed in GCC 13[1],
and there is top-level Makefile logic to remove -Warray-bounds for
known-bad GCC versions staring with commit f0be87c42cbd ("gcc-12: disable
'-Warray-bounds' universally for now").

Remove the local work-around.

[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105679

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agoprestera: matchall: do not rollback if rule exists
Serhiy Boiko [Thu, 6 Oct 2022 19:04:09 +0000 (22:04 +0300)]
prestera: matchall: do not rollback if rule exists

If you try to create a 'mirror' ACL rule on a port that already has a
mirror rule, prestera_span_rule_add() will fail with EEXIST error.

This forces rollback procedure which destroys existing mirror rule on
hardware leaving it visible in linux.

Add an explicit check for EEXIST to prevent the deletion of the existing
rule but keep user seeing error message:

  $ tc filter add dev sw1p1 ... skip_sw action mirred egress mirror dev sw1p2
  $ tc filter add dev sw1p1 ... skip_sw action mirred egress mirror dev sw1p3
  RTNETLINK answers: File exists
  We have an error talking to the kernel

Fixes: 13defa275eef ("net: marvell: prestera: Add matchall support")
Signed-off-by: Serhiy Boiko <serhiy.boiko@plvision.eu>
Signed-off-by: Maksym Glubokiy <maksym.glubokiy@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agoipv4: Handle attempt to delete multipath route when fib_info contains an nh reference
David Ahern [Thu, 6 Oct 2022 16:48:49 +0000 (10:48 -0600)]
ipv4: Handle attempt to delete multipath route when fib_info contains an nh reference

Gwangun Jung reported a slab-out-of-bounds access in fib_nh_match:
    fib_nh_match+0xf98/0x1130 linux-6.0-rc7/net/ipv4/fib_semantics.c:961
    fib_table_delete+0x5f3/0xa40 linux-6.0-rc7/net/ipv4/fib_trie.c:1753
    inet_rtm_delroute+0x2b3/0x380 linux-6.0-rc7/net/ipv4/fib_frontend.c:874

Separate nexthop objects are mutually exclusive with the legacy
multipath spec. Fix fib_nh_match to return if the config for the
to be deleted route contains a multipath spec while the fib_info
is using a nexthop object.

Fixes: 493ced1ac47c ("ipv4: Allow routes to use nexthop objects")
Fixes: 6bf92d70e690 ("net: ipv4: fix route with nexthop object delete warning")
Reported-by: Gwangun Jung <exsociety@gmail.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Tested-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agonet: enetc: Remove duplicated include in enetc_qos.c
Yang Li [Thu, 6 Oct 2022 12:01:36 +0000 (20:01 +0800)]
net: enetc: Remove duplicated include in enetc_qos.c

net/pkt_sched.h is included twice in enetc_qos.c,
remove one of them.

Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=2334
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agoocteontx2-pf: mcs: remove unneeded semicolon
Yang Li [Thu, 6 Oct 2022 11:44:00 +0000 (19:44 +0800)]
octeontx2-pf: mcs: remove unneeded semicolon

Semicolon is not required after curly braces.

Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=2332
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agohv_netvsc: Fix race between VF offering and VF association message from host
Gaurav Kohli [Thu, 6 Oct 2022 05:52:59 +0000 (22:52 -0700)]
hv_netvsc: Fix race between VF offering and VF association message from host

During vm boot, there might be possibility that vf registration
call comes before the vf association from host to vm.

And this might break netvsc vf path, To prevent the same block
vf registration until vf bind message comes from host.

Cc: stable@vger.kernel.org
Fixes: 00d7ddba11436 ("hv_netvsc: pair VF based on serial number")
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Gaurav Kohli <gauravkohli@linux.microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agonet: ieee802154: return -EINVAL for unknown addr type
Alexander Aring [Thu, 6 Oct 2022 02:02:37 +0000 (22:02 -0400)]
net: ieee802154: return -EINVAL for unknown addr type

This patch adds handling to return -EINVAL for an unknown addr type. The
current behaviour is to return 0 as successful but the size of an
unknown addr type is not defined and should return an error like -EINVAL.

Fixes: 94160108a70c ("net/ieee802154: fix uninit value bug in dgram_sendmsg")
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agoMAINTAINERS: add Jan as SMC maintainer
Wenjia Zhang [Fri, 7 Oct 2022 06:54:36 +0000 (08:54 +0200)]
MAINTAINERS: add Jan as SMC maintainer

Add Jan as maintainer for Shared Memory Communications (SMC)
Sockets.

Acked-by: Jan Karcher <jaka@linux.ibm.com>
Acked-by: Alexandra Winter <wintera@linux.ibm.com>
Signed-off-by: Wenjia Zhang <wenjia@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
21 months agoMerge tag 'ieee802154-for-net-2022-10-05' of git://git.kernel.org/pub/scm/linux/kerne...
Jakub Kicinski [Thu, 6 Oct 2022 03:38:46 +0000 (20:38 -0700)]
Merge tag 'ieee802154-for-net-2022-10-05' of git://git./linux/kernel/git/sschmidt/wpan

Stefan Schmidt says:

====================
pull-request: ieee802154 for net 2022-10-05

Only two patches this time around. A revert from Alexander Aring to a patch
that hit net and the updated patch to fix the problem from Tetsuo Handa.

* tag 'ieee802154-for-net-2022-10-05' of git://git.kernel.org/pub/scm/linux/kernel/git/sschmidt/wpan:
  net/ieee802154: don't warn zero-sized raw_sendmsg()
  Revert "net/ieee802154: reject zero-sized raw_sendmsg()"
====================

Link: https://lore.kernel.org/r/20221005144508.787376-1-stefan@datenfreihafen.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: ethernet: adi: adin1110: Add check in netdev_event
Alexandru Tachici [Mon, 3 Oct 2022 11:16:36 +0000 (14:16 +0300)]
net: ethernet: adi: adin1110: Add check in netdev_event

Check whether this driver actually is the intended recipient of
upper change event.

Fixes: bc93e19d088b ("net: ethernet: adi: Add ADIN1110 support")
Signed-off-by: Alexandru Tachici <alexandru.tachici@analog.com>
Link: https://lore.kernel.org/r/20221003111636.54973-1-alexandru.tachici@analog.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agodocs: networking: phy: add missing space
Casper Andersson [Tue, 4 Oct 2022 07:32:42 +0000 (09:32 +0200)]
docs: networking: phy: add missing space

Missing space between "pins'" and "strength"

Signed-off-by: Casper Andersson <casper.casan@gmail.com>
Reviewed-by: Bagas Sanjaya <bagasdotme@gmail.com>
Link: https://lore.kernel.org/r/20221004073242.304425-1-casper.casan@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: pse-pd: PSE_REGULATOR should depend on REGULATOR
Geert Uytterhoeven [Tue, 4 Oct 2022 16:23:53 +0000 (18:23 +0200)]
net: pse-pd: PSE_REGULATOR should depend on REGULATOR

The Regulator based PSE controller driver relies on regulator support to
be enabled.  If regulator support is disabled, it will still compile
fine, but won't operate correctly.

Hence add a dependency on REGULATOR, to prevent asking the user about
this driver when configuring a kernel without regulator support.

Fixes: 66741b4e94ca7bb1 ("net: pse-pd: add regulator based PSE driver")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://lore.kernel.org/r/709caac8873ff2a8b72b92091429be7c1a939959.1664900558.git.geert+renesas@glider.be
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoRevert "net/sched: taprio: make qdisc_leaf() see the per-netdev-queue pfifo child...
Vladimir Oltean [Tue, 4 Oct 2022 22:01:00 +0000 (01:01 +0300)]
Revert "net/sched: taprio: make qdisc_leaf() see the per-netdev-queue pfifo child qdiscs"

taprio_attach() has this logic at the end, which should have been
removed with the blamed patch (which is now being reverted):

/* access to the child qdiscs is not needed in offload mode */
if (FULL_OFFLOAD_IS_ENABLED(q->flags)) {
kfree(q->qdiscs);
q->qdiscs = NULL;
}

because otherwise, we make use of q->qdiscs[] even after this array was
deallocated, namely in taprio_leaf(). Therefore, whenever one would try
to attach a valid child qdisc to a fully offloaded taprio root, one
would immediately dereference a NULL pointer.

$ tc qdisc replace dev eno0 handle 8001: parent root taprio \
num_tc 8 \
map 0 1 2 3 4 5 6 7 \
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 \
max-sdu 0 0 0 0 0 200 0 0 \
base-time 200 \
sched-entry S 80 20000 \
sched-entry S a0 20000 \
sched-entry S 5f 60000 \
flags 2
$ max_frame_size=1500
$ data_rate_kbps=20000
$ port_transmit_rate_kbps=1000000
$ idleslope=$data_rate_kbps
$ sendslope=$(($idleslope - $port_transmit_rate_kbps))
$ locredit=$(($max_frame_size * $sendslope / $port_transmit_rate_kbps))
$ hicredit=$(($max_frame_size * $idleslope / $port_transmit_rate_kbps))
$ tc qdisc replace dev eno0 parent 8001:7 cbs \
idleslope $idleslope \
sendslope $sendslope \
hicredit $hicredit \
locredit $locredit \
offload 0

Unable to handle kernel NULL pointer dereference at virtual address 0000000000000030
pc : taprio_leaf+0x28/0x40
lr : qdisc_leaf+0x3c/0x60
Call trace:
 taprio_leaf+0x28/0x40
 tc_modify_qdisc+0xf0/0x72c
 rtnetlink_rcv_msg+0x12c/0x390
 netlink_rcv_skb+0x5c/0x130
 rtnetlink_rcv+0x1c/0x2c

The solution is not as obvious as the problem. The code which deallocates
q->qdiscs[] is in fact copied and pasted from mqprio, which also
deallocates the array in mqprio_attach() and never uses it afterwards.

Therefore, the identical cleanup logic of priv->qdiscs[] that
mqprio_destroy() has is deceptive because it will never take place at
qdisc_destroy() time, but just at raw ops->destroy() time (otherwise
said, priv->qdiscs[] do not last for the entire lifetime of the mqprio
root), but rather, this is just the twisted way in which the Qdisc API
understands error path cleanup should be done (Qdisc_ops :: destroy() is
called even when Qdisc_ops :: init() never succeeded).

Side note, in fact this is also what the comment in mqprio_init() says:

/* pre-allocate qdisc, attachment can't fail */

Or reworded, mqprio's priv->qdiscs[] scheme is only meant to serve as
data passing between Qdisc_ops :: init() and Qdisc_ops :: attach().

[ this comment was also copied and pasted into the initial taprio
  commit, even though taprio_attach() came way later ]

The problem is that taprio also makes extensive use of the q->qdiscs[]
array in the software fast path (taprio_enqueue() and taprio_dequeue()),
but it does not keep a reference of its own on q->qdiscs[i] (you'd think
that since it creates these Qdiscs, it holds the reference, but nope,
this is not completely true).

To understand the difference between taprio_destroy() and mqprio_destroy()
one must look before commit 13511704f8d7 ("net: taprio offload: enforce
qdisc to netdev queue mapping"), because that just muddied the waters.

In the "original" taprio design, taprio always attached itself (the root
Qdisc) to all netdev TX queues, so that dev_qdisc_enqueue() would go
through taprio_enqueue().

It also called qdisc_refcount_inc() on itself for as many times as there
were netdev TX queues, in order to counter-balance what tc_get_qdisc()
does when destroying a Qdisc (simplified for brevity below):

if (n->nlmsg_type == RTM_DELQDISC)
err = qdisc_graft(dev, parent=NULL, new=NULL, q, extack);

qdisc_graft(where "new" is NULL so this deletes the Qdisc):

for (i = 0; i < num_q; i++) {
struct netdev_queue *dev_queue;

dev_queue = netdev_get_tx_queue(dev, i);

old = dev_graft_qdisc(dev_queue, new);
if (new && i > 0)
qdisc_refcount_inc(new);

qdisc_put(old);
~~~~~~~~~~~~~~
this decrements taprio's refcount once for each TX queue
}

notify_and_destroy(net, skb, n, classid,
   rtnl_dereference(dev->qdisc), new);
   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   and this finally decrements it to zero,
   making qdisc_put() call qdisc_destroy()

The q->qdiscs[] created using qdisc_create_dflt() (or their
replacements, if taprio_graft() was ever to get called) were then
privately freed by taprio_destroy().

This is still what is happening after commit 13511704f8d7 ("net: taprio
offload: enforce qdisc to netdev queue mapping"), but only for software
mode.

In full offload mode, the per-txq "qdisc_put(old)" calls from
qdisc_graft() now deallocate the child Qdiscs rather than decrement
taprio's refcount. So when notify_and_destroy(taprio) finally calls
taprio_destroy(), the difference is that the child Qdiscs were already
deallocated.

And this is exactly why the taprio_attach() comment "access to the child
qdiscs is not needed in offload mode" is deceptive too. Not only the
q->qdiscs[] array is not needed, but it is also necessary to get rid of
it as soon as possible, because otherwise, we will also call qdisc_put()
on the child Qdiscs in qdisc_destroy() -> taprio_destroy(), and this
will cause a nasty use-after-free/refcount-saturate/whatever.

In short, the problem is that since the blamed commit, taprio_leaf()
needs q->qdiscs[] to not be freed by taprio_attach(), while qdisc_destroy()
-> taprio_destroy() does need q->qdiscs[] to be freed by taprio_attach()
for full offload. Fixing one problem triggers the other.

All of this can be solved by making taprio keep its q->qdiscs[i] with a
refcount elevated at 2 (in offloaded mode where they are attached to the
netdev TX queues), both in taprio_attach() and in taprio_graft(). The
generic qdisc_graft() would just decrement the child qdiscs' refcounts
to 1, and taprio_destroy() would give them the final coup de grace.

However the rabbit hole of changes is getting quite deep, and the
complexity increases. The blamed commit was supposed to be a bug fix in
the first place, and the bug it addressed is not so significant so as to
justify further rework in stable trees. So I'd rather just revert it.
I don't know enough about multi-queue Qdisc design to make a proper
judgement right now regarding what is/isn't idiomatic use of Qdisc
concepts in taprio. I will try to study the problem more and come with a
different solution in net-next.

Fixes: 1461d212ab27 ("net/sched: taprio: make qdisc_leaf() see the per-netdev-queue pfifo child qdiscs")
Reported-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com>
Reported-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Link: https://lore.kernel.org/r/20221004220100.1650558-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/ieee802154: don't warn zero-sized raw_sendmsg()
Tetsuo Handa [Wed, 5 Oct 2022 01:47:50 +0000 (21:47 -0400)]
net/ieee802154: don't warn zero-sized raw_sendmsg()

syzbot is hitting skb_assert_len() warning at __dev_queue_xmit() [1],
for PF_IEEE802154 socket's zero-sized raw_sendmsg() request is hitting
__dev_queue_xmit() with skb->len == 0.

Since PF_IEEE802154 socket's zero-sized raw_sendmsg() request was
able to return 0, don't call __dev_queue_xmit() if packet length is 0.

  ----------
  #include <sys/socket.h>
  #include <netinet/in.h>

  int main(int argc, char *argv[])
  {
    struct sockaddr_in addr = { .sin_family = AF_INET, .sin_addr.s_addr = htonl(INADDR_LOOPBACK) };
    struct iovec iov = { };
    struct msghdr hdr = { .msg_name = &addr, .msg_namelen = sizeof(addr), .msg_iov = &iov, .msg_iovlen = 1 };
    sendmsg(socket(PF_IEEE802154, SOCK_RAW, 0), &hdr, 0);
    return 0;
  }
  ----------

Note that this might be a sign that commit fd1894224407c484 ("bpf: Don't
redirect packets with invalid pkt_len") should be reverted, for
skb->len == 0 was acceptable for at least PF_IEEE802154 socket.

Link: https://syzkaller.appspot.com/bug?extid=5ea725c25d06fb9114c4
Reported-by: syzbot <syzbot+5ea725c25d06fb9114c4@syzkaller.appspotmail.com>
Fixes: fd1894224407c484 ("bpf: Don't redirect packets with invalid pkt_len")
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20221005014750.3685555-2-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
21 months agoRevert "net/ieee802154: reject zero-sized raw_sendmsg()"
Alexander Aring [Wed, 5 Oct 2022 01:47:49 +0000 (21:47 -0400)]
Revert "net/ieee802154: reject zero-sized raw_sendmsg()"

This reverts commit 3a4d061c699bd3eedc80dc97a4b2a2e1af83c6f5.

There is a v2 which does return zero if zero length is given.

Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20221005014750.3685555-1-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
21 months agoMerge tag 'net-next-6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev...
Linus Torvalds [Tue, 4 Oct 2022 20:38:03 +0000 (13:38 -0700)]
Merge tag 'net-next-6.1' of git://git./linux/kernel/git/netdev/net-next

Pull networking updates from Jakub Kicinski:
 "Core:

   - Introduce and use a single page frag cache for allocating small skb
     heads, clawing back the 10-20% performance regression in UDP flood
     test from previous fixes.

   - Run packets which already went thru HW coalescing thru SW GRO. This
     significantly improves TCP segment coalescing and simplifies
     deployments as different workloads benefit from HW or SW GRO.

   - Shrink the size of the base zero-copy send structure.

   - Move TCP init under a new slow / sleepable version of DO_ONCE().

  BPF:

   - Add BPF-specific, any-context-safe memory allocator.

   - Add helpers/kfuncs for PKCS#7 signature verification from BPF
     programs.

   - Define a new map type and related helpers for user space -> kernel
     communication over a ring buffer (BPF_MAP_TYPE_USER_RINGBUF).

   - Allow targeting BPF iterators to loop through resources of one
     task/thread.

   - Add ability to call selected destructive functions. Expose
     crash_kexec() to allow BPF to trigger a kernel dump. Use
     CAP_SYS_BOOT check on the loading process to judge permissions.

   - Enable BPF to collect custom hierarchical cgroup stats efficiently
     by integrating with the rstat framework.

   - Support struct arguments for trampoline based programs. Only
     structs with size <= 16B and x86 are supported.

   - Invoke cgroup/connect{4,6} programs for unprivileged ICMP ping
     sockets (instead of just TCP and UDP sockets).

   - Add a helper for accessing CLOCK_TAI for time sensitive network
     related programs.

   - Support accessing network tunnel metadata's flags.

   - Make TCP SYN ACK RTO tunable by BPF programs with TCP Fast Open.

   - Add support for writing to Netfilter's nf_conn:mark.

  Protocols:

   - WiFi: more Extremely High Throughput (EHT) and Multi-Link Operation
     (MLO) work (802.11be, WiFi 7).

   - vsock: improve support for SO_RCVLOWAT.

   - SMC: support SO_REUSEPORT.

   - Netlink: define and document how to use netlink in a "modern" way.
     Support reporting missing attributes via extended ACK.

   - IPSec: support collect metadata mode for xfrm interfaces.

   - TCPv6: send consistent autoflowlabel in SYN_RECV state and RST
     packets.

   - TCP: introduce optional per-netns connection hash table to allow
     better isolation between namespaces (opt-in, at the cost of memory
     and cache pressure).

   - MPTCP: support TCP_FASTOPEN_CONNECT.

   - Add NEXT-C-SID support in Segment Routing (SRv6) End behavior.

   - Adjust IP_UNICAST_IF sockopt behavior for connected UDP sockets.

   - Open vSwitch:
      - Allow specifying ifindex of new interfaces.
      - Allow conntrack and metering in non-initial user namespace.

   - TLS: support the Korean ARIA-GCM crypto algorithm.

   - Remove DECnet support.

  Driver API:

   - Allow selecting the conduit interface used by each port in DSA
     switches, at runtime.

   - Ethernet Power Sourcing Equipment and Power Device support.

   - Add tc-taprio support for queueMaxSDU parameter, i.e. setting per
     traffic class max frame size for time-based packet schedules.

   - Support PHY rate matching - adapting between differing host-side
     and link-side speeds.

   - Introduce QUSGMII PHY mode and 1000BASE-KX interface mode.

   - Validate OF (device tree) nodes for DSA shared ports; make
     phylink-related properties mandatory on DSA and CPU ports.
     Enforcing more uniformity should allow transitioning to phylink.

   - Require that flash component name used during update matches one of
     the components for which version is reported by info_get().

   - Remove "weight" argument from driver-facing NAPI API as much as
     possible. It's one of those magic knobs which seemed like a good
     idea at the time but is too indirect to use in practice.

   - Support offload of TLS connections with 256 bit keys.

  New hardware / drivers:

   - Ethernet:
      - Microchip KSZ9896 6-port Gigabit Ethernet Switch
      - Renesas Ethernet AVB (EtherAVB-IF) Gen4 SoCs
      - Analog Devices ADIN1110 and ADIN2111 industrial single pair
        Ethernet (10BASE-T1L) MAC+PHY.
      - Rockchip RV1126 Gigabit Ethernet (a version of stmmac IP).

   - Ethernet SFPs / modules:
      - RollBall / Hilink / Turris 10G copper SFPs
      - HALNy GPON module

   - WiFi:
      - CYW43439 SDIO chipset (brcmfmac)
      - CYW89459 PCIe chipset (brcmfmac)
      - BCM4378 on Apple platforms (brcmfmac)

  Drivers:

   - CAN:
      - gs_usb: HW timestamp support

   - Ethernet PHYs:
      - lan8814: cable diagnostics

   - Ethernet NICs:
      - Intel (100G):
         - implement control of FCS/CRC stripping
         - port splitting via devlink
         - L2TPv3 filtering offload
      - nVidia/Mellanox:
         - tunnel offload for sub-functions
         - MACSec offload, w/ Extended packet number and replay window
           offload
         - significantly restructure, and optimize the AF_XDP support,
           align the behavior with other vendors
      - Huawei:
         - configuring DSCP map for traffic class selection
         - querying standard FEC statistics
         - querying SerDes lane number via ethtool
      - Marvell/Cavium:
         - egress priority flow control
         - MACSec offload
      - AMD/SolarFlare:
         - PTP over IPv6 and raw Ethernet
      - small / embedded:
         - ax88772: convert to phylink (to support SFP cages)
         - altera: tse: convert to phylink
         - ftgmac100: support fixed link
         - enetc: standard Ethtool counters
         - macb: ZynqMP SGMII dynamic configuration support
         - tsnep: support multi-queue and use page pool
         - lan743x: Rx IP & TCP checksum offload
         - igc: add xdp frags support to ndo_xdp_xmit

   - Ethernet high-speed switches:
      - Marvell (prestera):
         - support SPAN port features (traffic mirroring)
         - nexthop object offloading
      - Microchip (sparx5):
         - multicast forwarding offload
         - QoS queuing offload (tc-mqprio, tc-tbf, tc-ets)

   - Ethernet embedded switches:
      - Marvell (mv88e6xxx):
         - support RGMII cmode
      - NXP (felix):
         - standardized ethtool counters
      - Microchip (lan966x):
         - QoS queuing offload (tc-mqprio, tc-tbf, tc-cbs, tc-ets)
         - traffic policing and mirroring
         - link aggregation / bonding offload
         - QUSGMII PHY mode support

   - Qualcomm 802.11ax WiFi (ath11k):
      - cold boot calibration support on WCN6750
      - support to connect to a non-transmit MBSSID AP profile
      - enable remain-on-channel support on WCN6750
      - Wake-on-WLAN support for WCN6750
      - support to provide transmit power from firmware via nl80211
      - support to get power save duration for each client
      - spectral scan support for 160 MHz

   - MediaTek WiFi (mt76):
      - WiFi-to-Ethernet bridging offload for MT7986 chips

   - RealTek WiFi (rtw89):
      - P2P support"

* tag 'net-next-6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1864 commits)
  eth: pse: add missing static inlines
  once: rename _SLOW to _SLEEPABLE
  net: pse-pd: add regulator based PSE driver
  dt-bindings: net: pse-dt: add bindings for regulator based PoDL PSE controller
  ethtool: add interface to interact with Ethernet Power Equipment
  net: mdiobus: search for PSE nodes by parsing PHY nodes.
  net: mdiobus: fwnode_mdiobus_register_phy() rework error handling
  net: add framework to support Ethernet PSE and PDs devices
  dt-bindings: net: phy: add PoDL PSE property
  net: marvell: prestera: Propagate nh state from hw to kernel
  net: marvell: prestera: Add neighbour cache accounting
  net: marvell: prestera: add stub handler neighbour events
  net: marvell: prestera: Add heplers to interact with fib_notifier_info
  net: marvell: prestera: Add length macros for prestera_ip_addr
  net: marvell: prestera: add delayed wq and flush wq on deinit
  net: marvell: prestera: Add strict cleanup of fib arbiter
  net: marvell: prestera: Add cleanup of allocated fib_nodes
  net: marvell: prestera: Add router nexthops ABI
  eth: octeon: fix build after netif_napi_add() changes
  net/mlx5: E-Switch, Return EBUSY if can't get mode lock
  ...

21 months agoMerge tag 'landlock-6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mic...
Linus Torvalds [Tue, 4 Oct 2022 18:13:38 +0000 (11:13 -0700)]
Merge tag 'landlock-6.1-rc1' of git://git./linux/kernel/git/mic/linux

Pull landlock updates from Mickaël Salaün:
 "Improve user help for Landlock (documentation and sample)"

* tag 'landlock-6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux:
  landlock: Fix documentation style
  landlock: Slightly improve documentation and fix spelling
  samples/landlock: Print hints about ABI versions

21 months agoMerge tag 'audit-pr-20221003' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoor...
Linus Torvalds [Tue, 4 Oct 2022 18:05:43 +0000 (11:05 -0700)]
Merge tag 'audit-pr-20221003' of git://git./linux/kernel/git/pcmoore/audit

Pull audit updates from Paul Moore:
 "Six audit patches for v6.1, most are pretty trivial, but a quick list
  of the highlights are below:

   - Only free the audit proctitle information on task exit. This allows
     us to cache the information and improve performance slightly.

   - Use the time_after() macro to do time comparisons instead of doing
     it directly and potentially causing ourselves problems when the
     timer wraps.

   - Convert an audit_context state comparison from a relative enum
     comparison, e.g. (x < y), to a not-equal comparison to ensure that
     we are not caught out at some unknown point in the future by an
     enum shuffle.

   - A handful of small cleanups such as tidying up comments and
     removing unused declarations"

* tag 'audit-pr-20221003' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/audit:
  audit: remove selinux_audit_rule_update() declaration
  audit: use time_after to compare time
  audit: free audit_proctitle only on task exit
  audit: explicitly check audit_context->context enum value
  audit: audit_context pid unused, context enum comment fix
  audit: fix repeated words in comments

21 months agoMerge tag 'x86_cleanups_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 4 Oct 2022 17:24:11 +0000 (10:24 -0700)]
Merge tag 'x86_cleanups_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 cleanups from Borislav Petkov:

 - The usual round of smaller fixes and cleanups all over the tree

* tag 'x86_cleanups_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/cpu: Include the header of init_ia32_feat_ctl()'s prototype
  x86/uaccess: Improve __try_cmpxchg64_user_asm() for x86_32
  x86: Fix various duplicate-word comment typos
  x86/boot: Remove superfluous type casting from arch/x86/boot/bitops.h

21 months agoMerge tag 'x86_cache_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 17:14:58 +0000 (10:14 -0700)]
Merge tag 'x86_cache_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 cache resource control updates from Borislav Petkov:

 - More work by James Morse to disentangle the resctrl filesystem
   generic code from the architectural one with the endgoal of plugging
   ARM's MPAM implementation into it too so that the user interface
   remains the same

 - Properly restore the MSR_MISC_FEATURE_CONTROL value instead of
   blindly overwriting it to 0

* tag 'x86_cache_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (22 commits)
  x86/resctrl: Make resctrl_arch_rmid_read() return values in bytes
  x86/resctrl: Add resctrl_rmid_realloc_limit to abstract x86's boot_cpu_data
  x86/resctrl: Rename and change the units of resctrl_cqm_threshold
  x86/resctrl: Move get_corrected_mbm_count() into resctrl_arch_rmid_read()
  x86/resctrl: Move mbm_overflow_count() into resctrl_arch_rmid_read()
  x86/resctrl: Pass the required parameters into resctrl_arch_rmid_read()
  x86/resctrl: Abstract __rmid_read()
  x86/resctrl: Allow per-rmid arch private storage to be reset
  x86/resctrl: Add per-rmid arch private storage for overflow and chunks
  x86/resctrl: Calculate bandwidth from the previous __mon_event_count() chunks
  x86/resctrl: Allow update_mba_bw() to update controls directly
  x86/resctrl: Remove architecture copy of mbps_val
  x86/resctrl: Switch over to the resctrl mbps_val list
  x86/resctrl: Create mba_sc configuration in the rdt_domain
  x86/resctrl: Abstract and use supports_mba_mbps()
  x86/resctrl: Remove set_mba_sc()s control array re-initialisation
  x86/resctrl: Add domain offline callback for resctrl work
  x86/resctrl: Group struct rdt_hw_domain cleanup
  x86/resctrl: Add domain online callback for resctrl work
  x86/resctrl: Merge mon_capable and mon_enabled
  ...

21 months agoMerge tag 'x86_microcode_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 4 Oct 2022 17:12:08 +0000 (10:12 -0700)]
Merge tag 'x86_microcode_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x75 microcode loader updates from Borislav Petkov:

 - Get rid of a single ksize() usage

 - By popular demand, print the previous microcode revision an update
   was done over

 - Remove more code related to the now gone MICROCODE_OLD_INTERFACE

 - Document the problems stemming from microcode late loading

* tag 'x86_microcode_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/microcode/AMD: Track patch allocation size explicitly
  x86/microcode: Print previous version of microcode after reload
  x86/microcode: Remove ->request_microcode_user()
  x86/microcode: Document the whole late loading problem

21 months agoMerge tag 'x86_paravirt_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 4 Oct 2022 17:03:40 +0000 (10:03 -0700)]
Merge tag 'x86_paravirt_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 paravirt fix from Borislav Petkov:

 - Ensure paravirt patching site descriptors are aligned properly so
   that code can do proper arithmetic with their addresses

* tag 'x86_paravirt_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/paravirt: Ensure proper alignment

21 months agoMerge tag 'x86_misc_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 17:00:27 +0000 (10:00 -0700)]
Merge tag 'x86_misc_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull misc x86 fixes from Borislav Petkov:

 - Drop misleading "RIP" from the opcodes dumping message

 - Correct APM entry's Konfig help text

* tag 'x86_misc_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/dumpstack: Don't mention RIP in "Code: "
  x86/Kconfig: Specify idle=poll instead of no-hlt

21 months agoMerge tag 'x86_asm_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 16:49:50 +0000 (09:49 -0700)]
Merge tag 'x86_asm_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 asm update from Borislav Petkov:

 - Use the __builtin_ffs/ctzl() compiler builtins for the constant
   argument case in the kernel's optimized ffs()/ffz() helpers in order
   to make use of the compiler's constant folding optmization passes.

* tag 'x86_asm_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/asm/bitops: Use __builtin_ctzl() to evaluate constant expressions
  x86/asm/bitops: Use __builtin_ffs() to evaluate constant expressions

21 months agoMerge tag 'x86_core_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 16:46:22 +0000 (09:46 -0700)]
Merge tag 'x86_core_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 core fixes from Borislav Petkov:

 - Make sure an INT3 is slapped after every unconditional retpoline JMP
   as both vendors suggest

 - Clean up pciserial a bit

* tag 'x86_core_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86,retpoline: Be sure to emit INT3 after JMP *%\reg
  x86/earlyprintk: Clean up pciserial

21 months agoMerge tag 'x86_apic_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 16:37:02 +0000 (09:37 -0700)]
Merge tag 'x86_apic_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 APIC update from Borislav Petkov:

 - Add support for locking the APIC in X2APIC mode to prevent SGX
   enclave leaks

* tag 'x86_apic_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/apic: Don't disable x2APIC if locked

21 months agoMerge tag 'ras_core_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 16:33:12 +0000 (09:33 -0700)]
Merge tag 'ras_core_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 RAS updates from Borislav Petkov:

 - Fix the APEI MCE callback handler to consult the hardware about the
   granularity of the memory error instead of hard-coding it

 - Offline memory pages on Intel machines after 2 errors reported per
   page

* tag 'ras_core_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/mce: Retrieve poison range from hardware
  RAS/CEC: Reduce offline page threshold for Intel systems

21 months agoMerge tag 'x86_cpu_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 16:21:30 +0000 (09:21 -0700)]
Merge tag 'x86_cpu_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 cpu updates from Borislav Petkov:

 - Print the CPU number at segfault time.

   The number printed is not always accurate (preemption is enabled at
   that time) but the print string contains "likely" and after a lot of
   back'n'forth on this, this was the consensus that was reached. See
   thread at [1].

 - After a *lot* of testing and polishing, finally the clear_user()
   improvements to inline REP; STOSB by default

Link: https://lore.kernel.org/r/5d62c1d0-7425-d5bb-ecb5-1dc3b4d7d245@intel.com
* tag 'x86_cpu_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/mm: Print likely CPU at segfault time
  x86/clear_user: Make it faster

21 months agoMerge tag 'x86_sgx_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 16:17:44 +0000 (09:17 -0700)]
Merge tag 'x86_sgx_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 SGX update from Borislav Petkov:

 - Improve the documentation of a couple of SGX functions handling
   backing storage

* tag 'x86_sgx_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/sgx: Improve comments for sgx_encl_lookup/alloc_backing()

21 months agoMerge tag 'x86_timers_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 4 Oct 2022 16:13:21 +0000 (09:13 -0700)]
Merge tag 'x86_timers_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 RTC cleanups from Borislav Petkov:

 - Cleanup x86/rtc.c and delete duplicated functionality in favor of
   using the respective functionality from the RTC library

* tag 'x86_timers_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/rtc: Rename mach_set_rtc_mmss() to mach_set_cmos_time()
  x86/rtc: Rewrite & simplify mach_get_cmos_time() by deleting duplicated functionality

21 months agoMerge tag 'x86_platform_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 4 Oct 2022 16:06:35 +0000 (09:06 -0700)]
Merge tag 'x86_platform_for_v6.1_rc1' of git://git./linux/kernel/git/tip/tip

Pull x86 platform update from Borislav Petkov:
 "A single x86/platform improvement when the kernel is running as an
  ACRN guest:

   - Get TSC and CPU frequency from CPUID leaf 0x40000010 when the
     kernel is running as a guest on the ACRN hypervisor"

* tag 'x86_platform_for_v6.1_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/acrn: Set up timekeeping

21 months agoMerge tag 'edac_updates_for_v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 15:58:02 +0000 (08:58 -0700)]
Merge tag 'edac_updates_for_v6.1' of git://git./linux/kernel/git/ras/ras

Pull EDAC updates from Borislav Petkov:

 - Add support for Skylake-S CPUs to ie31200_edac

 - Improve error decoding speed of the Intel drivers by avoiding the
   ACPI facilities but doing decoding in the driver itself

 - Other misc improvements to the Intel drivers

 - The usual cleanups and fixlets all over EDAC land

* tag 'edac_updates_for_v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
  EDAC/i7300: Correct the i7300_exit() function name in comment
  x86/sb_edac: Add row column translation for Broadwell
  EDAC/i10nm: Print an extra register set of retry_rd_err_log
  EDAC/i10nm: Retrieve and print retry_rd_err_log registers for HBM
  EDAC/skx_common: Add ChipSelect ADXL component
  EDAC/ppc_4xx: Reorder symbols to get rid of a few forward declarations
  EDAC: Remove obsolete declarations in edac_module.h
  EDAC/i10nm: Add driver decoder for Ice Lake and Tremont CPUs
  EDAC/skx_common: Make output format similar
  EDAC/skx_common: Use driver decoder first
  EDAC/mc: Drop duplicated dimm->nr_pages debug printout
  EDAC/mc: Replace spaces with tabs in memtype flags definition
  EDAC/wq: Remove unneeded flush_workqueue()
  EDAC/ie31200: Add Skylake-S support

21 months agoMerge branches 'edac-drivers' and 'edac-misc' into edac-updates-for-v6.1
Borislav Petkov [Tue, 4 Oct 2022 08:00:25 +0000 (10:00 +0200)]
Merge branches 'edac-drivers' and 'edac-misc' into edac-updates-for-v6.1

Combine all queued EDAC changes for submission into v6.1:

* edac-drivers:
  EDAC/ie31200: Add Skylake-S support

* edac-misc:
  EDAC/i7300: Correct the i7300_exit() function name in comment
  x86/sb_edac: Add row column translation for Broadwell
  EDAC/i10nm: Print an extra register set of retry_rd_err_log
  EDAC/i10nm: Retrieve and print retry_rd_err_log registers for HBM
  EDAC/skx_common: Add ChipSelect ADXL component
  EDAC/ppc_4xx: Reorder symbols to get rid of a few forward declarations
  EDAC: Remove obsolete declarations in edac_module.h
  EDAC/i10nm: Add driver decoder for Ice Lake and Tremont CPUs
  EDAC/skx_common: Make output format similar
  EDAC/skx_common: Use driver decoder first
  EDAC/mc: Drop duplicated dimm->nr_pages debug printout
  EDAC/mc: Replace spaces with tabs in memtype flags definition
  EDAC/wq: Remove unneeded flush_workqueue()

Signed-off-by: Borislav Petkov <bp@suse.de>
21 months agoeth: pse: add missing static inlines
Jakub Kicinski [Tue, 4 Oct 2022 04:03:27 +0000 (21:03 -0700)]
eth: pse: add missing static inlines

build bot reports missing 'static inline' qualifiers in the header.

Reported-by: kernel test robot <lkp@intel.com>
Fixes: 18ff0bcda6d1 ("ethtool: add interface to interact with Ethernet Power Equipment")
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://lore.kernel.org/r/20221004040327.2034878-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoMerge tag 'statx-dioalign-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 4 Oct 2022 03:33:41 +0000 (20:33 -0700)]
Merge tag 'statx-dioalign-for-linus' of git://git./linux/kernel/git/ebiggers/linux

Pull STATX_DIOALIGN support from Eric Biggers:
 "Make statx() support reporting direct I/O (DIO) alignment information.

  This provides a generic interface for userspace programs to determine
  whether a file supports DIO, and if so with what alignment
  restrictions. Specifically, STATX_DIOALIGN works on block devices, and
  on regular files when their containing filesystem has implemented
  support.

  An interface like this has been requested for years, since the
  conditions for when DIO is supported in Linux have gotten increasingly
  complex over time. Today, DIO support and alignment requirements can
  be affected by various filesystem features such as multi-device
  support, data journalling, inline data, encryption, verity,
  compression, checkpoint disabling, log-structured mode, etc.

  Further complicating things, Linux v6.0 relaxed the traditional rule
  of DIO needing to be aligned to the block device's logical block size;
  now user buffers (but not file offsets) only need to be aligned to the
  DMA alignment.

  The approach of uplifting the XFS specific ioctl XFS_IOC_DIOINFO was
  discarded in favor of creating a clean new interface with statx().

  For more information, see the individual commits and the man page
  update[1]"

Link: https://lore.kernel.org/r/20220722074229.148925-1-ebiggers@kernel.org
* tag 'statx-dioalign-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux:
  xfs: support STATX_DIOALIGN
  f2fs: support STATX_DIOALIGN
  f2fs: simplify f2fs_force_buffered_io()
  f2fs: move f2fs_force_buffered_io() into file.c
  ext4: support STATX_DIOALIGN
  fscrypt: change fscrypt_dio_supported() to prepare for STATX_DIOALIGN
  vfs: support STATX_DIOALIGN on block devices
  statx: add direct I/O alignment information

21 months agoMerge tag 'fsverity-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt
Linus Torvalds [Tue, 4 Oct 2022 03:27:34 +0000 (20:27 -0700)]
Merge tag 'fsverity-for-linus' of git://git./fs/fscrypt/fscrypt

Pull fsverity updates from Eric Biggers:
 "Minor changes to convert uses of kmap() to kmap_local_page()"

* tag 'fsverity-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt:
  fs-verity: use kmap_local_page() instead of kmap()
  fs-verity: use memcpy_from_page()

21 months agoMerge tag 'fscrypt-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt
Linus Torvalds [Tue, 4 Oct 2022 03:18:34 +0000 (20:18 -0700)]
Merge tag 'fscrypt-for-linus' of git://git./fs/fscrypt/fscrypt

Pull fscrypt updates from Eric Biggers:
 "This release contains some implementation changes, but no new
  features:

   - Rework the implementation of the fscrypt filesystem-level keyring
     to not be as tightly coupled to the keyrings subsystem. This
     resolves several issues.

   - Eliminate most direct uses of struct request_queue from fs/crypto/,
     since struct request_queue is considered to be a block layer
     implementation detail.

   - Stop using the PG_error flag to track decryption failures. This is
     a prerequisite for freeing up PG_error for other uses"

* tag 'fscrypt-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt:
  fscrypt: work on block_devices instead of request_queues
  fscrypt: stop holding extra request_queue references
  fscrypt: stop using keyrings subsystem for fscrypt_master_key
  fscrypt: stop using PG_error to track error status
  fscrypt: remove fscrypt_set_test_dummy_encryption()

21 months agoMerge tag 'dlm-6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/linux-dlm
Linus Torvalds [Tue, 4 Oct 2022 03:11:59 +0000 (20:11 -0700)]
Merge tag 'dlm-6.1' of git://git./linux/kernel/git/teigland/linux-dlm

Pull dlm updates from David Teigland:

 - Fix a couple races found with a new torture test

 - Improve errors when api functions are used incorrectly

 - Improve tracing for lock requests from user space

 - Fix use after free in recently added tracing cod.

 - Small internal code cleanups

* tag 'dlm-6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/linux-dlm:
  fs: dlm: fix possible use after free if tracing
  fs: dlm: const void resource name parameter
  fs: dlm: LSFL_CB_DELAY only for kernel lockspaces
  fs: dlm: remove DLM_LSFL_FS from uapi
  fs: dlm: trace user space callbacks
  fs: dlm: change ls_clear_proc_locks to spinlock
  fs: dlm: remove dlm_del_ast prototype
  fs: dlm: handle rcom in else if branch
  fs: dlm: allow lockspaces have zero lvblen
  fs: dlm: fix invalid derefence of sb_lvbptr
  fs: dlm: handle -EINVAL as log_error()
  fs: dlm: use __func__ for function name
  fs: dlm: handle -EBUSY first in unlock validation
  fs: dlm: handle -EBUSY first in lock arg validation
  fs: dlm: fix race between test_bit() and queue_work()
  fs: dlm: fix race in lowcomms

21 months agoMerge tag 'nfsd-6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux
Linus Torvalds [Tue, 4 Oct 2022 03:07:15 +0000 (20:07 -0700)]
Merge tag 'nfsd-6.1' of git://git./linux/kernel/git/cel/linux

Pull nfsd updates from Chuck Lever:
 "This release is mostly bug fixes, clean-ups, and optimizations.

  One notable set of fixes addresses a subtle buffer overflow issue that
  occurs if a small RPC Call message arrives in an oversized RPC record.
  This is only possible on a framed RPC transport such as TCP.

  Because NFSD shares the receive and send buffers in one set of pages,
  an oversized RPC record steals pages from the send buffer that will be
  used to construct the RPC Reply message. NFSD must not assume that a
  full-sized buffer is always available to it; otherwise, it will walk
  off the end of the send buffer while constructing its reply.

  In this release, we also introduce the ability for the server to wait
  a moment for clients to return delegations before it responds with
  NFS4ERR_DELAY. This saves a retransmit and a network round- trip when
  a delegation recall is needed. This work will be built upon in future
  releases.

  The NFS server adds another shrinker to its collection. Because
  courtesy clients can linger for quite some time, they might be
  freeable when the server host comes under memory pressure. A new
  shrinker has been added that releases courtesy client resources during
  low memory scenarios.

  Lastly, of note: the maximum number of operations per NFSv4 COMPOUND
  that NFSD can handle is increased from 16 to 50. There are NFSv4
  client implementations that need more than 16 to successfully perform
  a mount operation that uses a pathname with many components"

* tag 'nfsd-6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux: (53 commits)
  nfsd: extra checks when freeing delegation stateids
  nfsd: make nfsd4_run_cb a bool return function
  nfsd: fix comments about spinlock handling with delegations
  nfsd: only fill out return pointer on success in nfsd4_lookup_stateid
  NFSD: fix use-after-free on source server when doing inter-server copy
  NFSD: Cap rsize_bop result based on send buffer size
  NFSD: Rename the fields in copy_stateid_t
  nfsd: use DEFINE_SHOW_ATTRIBUTE to define nfsd_file_cache_stats_fops
  nfsd: use DEFINE_SHOW_ATTRIBUTE to define nfsd_reply_cache_stats_fops
  nfsd: use DEFINE_SHOW_ATTRIBUTE to define client_info_fops
  nfsd: use DEFINE_SHOW_ATTRIBUTE to define export_features_fops and supported_enctypes_fops
  nfsd: use DEFINE_PROC_SHOW_ATTRIBUTE to define nfsd_proc_ops
  NFSD: Pack struct nfsd4_compoundres
  NFSD: Remove unused nfsd4_compoundargs::cachetype field
  NFSD: Remove "inline" directives on op_rsize_bop helpers
  NFSD: Clean up nfs4svc_encode_compoundres()
  SUNRPC: Fix typo in xdr_buf_subsegment's kdoc comment
  NFSD: Clean up WRITE arg decoders
  NFSD: Use xdr_inline_decode() to decode NFSv3 symlinks
  NFSD: Refactor common code out of dirlist helpers
  ...

21 months agoMerge tag 'erofs-for-6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang...
Linus Torvalds [Tue, 4 Oct 2022 03:01:40 +0000 (20:01 -0700)]
Merge tag 'erofs-for-6.1-rc1' of git://git./linux/kernel/git/xiang/erofs

Pull erofs updates from Gao Xiang:
 "In this cycle, for container use cases, fscache-based shared domain is
  introduced [1] so that data blobs in the same domain will be storage
  deduplicated and it will also be used for page cache sharing later.

  Also, a special packed inode is now introduced to record inode
  fragments which keep the tail part of files by Yue Hu [2]. You can
  keep arbitary length or (at will) the whole file as a fragment and
  then fragments can be optionally compressed in the packed inode
  together and even deduplicated for smaller image sizes.

  In addition to that, global compressed data deduplication by sharing
  partial-referenced pclusters is also supported in this cycle.

  Summary:

   - Introduce fscache-based domain to share blobs between images

   - Support recording fragments in a special packed inode

   - Support partial-referenced pclusters for global compressed data
     deduplication

   - Fix an order >= MAX_ORDER warning due to crafted negative i_size

   - Several cleanups"

Link: https://lore.kernel.org/r/20220916085940.89392-1-zhujia.zj@bytedance.com
Link: https://lore.kernel.org/r/cover.1663065968.git.huyue2@coolpad.com
* tag 'erofs-for-6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
  erofs: clean up erofs_iget()
  erofs: clean up unnecessary code and comments
  erofs: fold in z_erofs_reload_indexes()
  erofs: introduce partial-referenced pclusters
  erofs: support on-disk compressed fragments data
  erofs: support interlaced uncompressed data for compressed files
  erofs: clean up .read_folio() and .readahead() in fscache mode
  erofs: introduce 'domain_id' mount option
  erofs: Support sharing cookies in the same domain
  erofs: introduce a pseudo mnt to manage shared cookies
  erofs: introduce fscache-based domain
  erofs: code clean up for fscache
  erofs: use kill_anon_super() to kill super in fscache mode
  erofs: fix order >= MAX_ORDER warning due to crafted negative i_size

21 months agoMerge tag 'fs.vfsuid.fat.v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs...
Linus Torvalds [Tue, 4 Oct 2022 02:54:29 +0000 (19:54 -0700)]
Merge tag 'fs.vfsuid.fat.v6.1' of git://git./linux/kernel/git/vfs/idmapping

Pull fatfs vfsuid conversion from Christian Brauner:
 "Last cycle we introduced the new vfs{g,u}id_t types that we had agreed
  on. The most important parts of the vfs have been converted but there
  are a few more places we need to switch before we can remove the old
  helpers completely.

  This cycle we converted all filesystems that called idmapped mount
  helpers directly. The affected filesystems are f2fs, fat, fuse, ksmbd,
  overlayfs, and xfs. We've sent patches for all of them. Looking at
  -next f2fs, ksmbd, overlayfs, and xfs have all picked up these patches
  and they should land in mainline during the v6.1 merge window.

  So all filesystems that have a separate tree should send the vfsuid
  conversion themselves. Onle the fat conversion is going through this
  generic fs trees because there is no fat tree.

  In order to change time settings on an inode fat checks that the
  caller either is the owner of the inode or the inode's group is in the
  caller's group list. If fat is on an idmapped mount we compare whether
  the inode mapped into the mount is equivalent to the caller's fsuid.
  If it isn't we compare whether the inode's group mapped into the mount
  is in the caller's group list.

  We now use the new vfsuid based helpers for that"

* tag 'fs.vfsuid.fat.v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping:
  fat: port to vfs{g,u}id_t and associated helpers

21 months agoMerge tag 'fs.acl.rework.prep.v6.1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 4 Oct 2022 02:48:54 +0000 (19:48 -0700)]
Merge tag 'fs.acl.rework.prep.v6.1' of git://git./linux/kernel/git/vfs/idmapping

Pull vfs acl updates from Christian Brauner:
 "These are general fixes and preparatory changes related to the ongoing
  posix acl rework. The actual rework where we build a type safe posix
  acl api wasn't ready for this merge window but we're hopeful for the
  next merge window.

  General fixes:

   - Some filesystems like 9p and cifs have to implement custom posix
     acl handlers because they require access to the dentry in order to
     set and get posix acls while the set and get inode operations
     currently don't. But the ntfs3 filesystem has no such requirement
     and thus implemented custom posix acl xattr handlers when it really
     didn't have to. So this pr contains patch that just implements set
     and get inode operations for ntfs3 and switches it to rely on the
     generic posix acl xattr handlers. (We would've appreciated reviews
     from the ntfs3 maintainers but we didn't get any. But hey, if we
     really broke it we'll fix it. But fstests for ntfs3 said it's
     fine.)

   - The posix_acl_fix_xattr_common() helper has been adapted so it can
     be used by a few more callers and avoiding open-coding the same
     checks over and over.

  Other than the two general fixes this series introduces a new helper
  vfs_set_acl_prepare(). The reason for this helper is so that we can
  mitigate one of the source that change {g,u}id values directly in the
  uapi struct. With the vfs_set_acl_prepare() helper we can move the
  idmapped mount fixup into the generic posix acl set handler.

  The advantage of this is that it allows us to remove the
  posix_acl_setxattr_idmapped_mnt() helper which so far we had to call
  in vfs_setxattr() to account for idmapped mounts. While semantically
  correct the problem with this approach was that we had to keep the
  value parameter of the generic vfs_setxattr() call as non-const. This
  is rectified in this series.

  Ultimately, we will get rid of all the extreme kludges and type
  unsafety once we have merged the posix api - hopefully during the next
  merge window - built solely around get and set inode operations. Which
  incidentally will also improve handling of posix acls in security and
  especially in integrity modesl. While this will come with temporarily
  having two inode operation for posix acls that is nothing compared to
  the problems we have right now and so well worth it. We'll end up with
  something that we can actually reason about instead of needing to
  write novels to explain what's going on"

* tag 'fs.acl.rework.prep.v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping:
  xattr: always us is_posix_acl_xattr() helper
  acl: fix the comments of posix_acl_xattr_set
  xattr: constify value argument in vfs_setxattr()
  ovl: use vfs_set_acl_prepare()
  acl: move idmapping handling into posix_acl_xattr_set()
  acl: add vfs_set_acl_prepare()
  acl: return EOPNOTSUPP in posix_acl_fix_xattr_common()
  ntfs3: rework xattr handlers and switch to POSIX ACL VFS helpers

21 months agoMerge tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Tue, 4 Oct 2022 01:03:58 +0000 (18:03 -0700)]
Merge tag 'pull-fixes' of git://git./linux/kernel/git/viro/vfs

Pull coredump fix from Al Viro:
 "Brown paper bag bug fix for the coredumping fix late in the 6.0
  release cycle"

* tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  [brown paperbag] fix coredump breakage

21 months agoMerge tag 'lsm-pr-20221003' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/lsm
Linus Torvalds [Tue, 4 Oct 2022 00:51:52 +0000 (17:51 -0700)]
Merge tag 'lsm-pr-20221003' of git://git./linux/kernel/git/pcmoore/lsm

Pull LSM updates from Paul Moore:
 "Seven patches for the LSM layer and we've got a mix of trivial and
  significant patches. Highlights below, starting with the smaller bits
  first so they don't get lost in the discussion of the larger items:

   - Remove some redundant NULL pointer checks in the common LSM audit
     code.

   - Ratelimit the lockdown LSM's access denial messages.

     With this change there is a chance that the last visible lockdown
     message on the console is outdated/old, but it does help preserve
     the initial series of lockdown denials that started the denial
     message flood and my gut feeling is that these might be the more
     valuable messages.

   - Open userfaultfds as readonly instead of read/write.

     While this code obviously lives outside the LSM, it does have a
     noticeable impact on the LSMs with Ondrej explaining the situation
     in the commit description. It is worth noting that this patch
     languished on the VFS list for over a year without any comments
     (objections or otherwise) so I took the liberty of pulling it into
     the LSM tree after giving fair notice. It has been in linux-next
     since the end of August without any noticeable problems.

   - Add a LSM hook for user namespace creation, with implementations
     for both the BPF LSM and SELinux.

     Even though the changes are fairly small, this is the bulk of the
     diffstat as we are also including BPF LSM selftests for the new
     hook.

     It's also the most contentious of the changes in this pull request
     with Eric Biederman NACK'ing the LSM hook multiple times during its
     development and discussion upstream. While I've never taken NACK's
     lightly, I'm sending these patches to you because it is my belief
     that they are of good quality, satisfy a long-standing need of
     users and distros, and are in keeping with the existing nature of
     the LSM layer and the Linux Kernel as a whole.

     The patches in implement a LSM hook for user namespace creation
     that allows for a granular approach, configurable at runtime, which
     enables both monitoring and control of user namespaces. The general
     consensus has been that this is far preferable to the other
     solutions that have been adopted downstream including outright
     removal from the kernel, disabling via system wide sysctls, or
     various other out-of-tree mechanisms that users have been forced to
     adopt since we haven't been able to provide them an upstream
     solution for their requests. Eric has been steadfast in his
     objections to this LSM hook, explaining that any restrictions on
     the user namespace could have significant impact on userspace.
     While there is the possibility of impacting userspace, it is
     important to note that this solution only impacts userspace when it
     is requested based on the runtime configuration supplied by the
     distro/admin/user. Frederick (the pathset author), the LSM/security
     community, and myself have tried to work with Eric during
     development of this patchset to find a mutually acceptable
     solution, but Eric's approach and unwillingness to engage in a
     meaningful way have made this impossible. I have CC'd Eric directly
     on this pull request so he has a chance to provide his side of the
     story; there have been no objections outside of Eric's"

* tag 'lsm-pr-20221003' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/lsm:
  lockdown: ratelimit denial messages
  userfaultfd: open userfaultfds with O_RDONLY
  selinux: Implement userns_create hook
  selftests/bpf: Add tests verifying bpf lsm userns_create hook
  bpf-lsm: Make bpf_lsm_userns_create() sleepable
  security, lsm: Introduce security_create_user_ns()
  lsm: clean up redundant NULL pointer check

21 months agoMerge tag 'selinux-pr-20221003' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 4 Oct 2022 00:45:15 +0000 (17:45 -0700)]
Merge tag 'selinux-pr-20221003' of git://git./linux/kernel/git/pcmoore/selinux

Pull SELinux updates from Paul Moore:
 "Six SELinux patches, all are simple and easily understood, but a list
  of the highlights is below:

   - Use 'grep -E' instead of 'egrep' in the SELinux policy install
     script.

     Fun fact, this seems to be GregKH's *second* dedicated SELinux
     patch since we transitioned to git (ignoring merges, the SPDX
     stuff, and a trivial fs reference removal when lustre was yanked);
     the first was back in 2011 when selinuxfs was placed in
     /sys/fs/selinux. Oh, the memories ...

   - Convert the SELinux policy boolean values to use signed integer
     types throughout the SELinux kernel code.

     Prior to this we were using a mix of signed and unsigned integers
     which was probably okay in this particular case, but it is
     definitely not a good idea in general.

   - Remove a reference to the SELinux runtime disable functionality in
     /etc/selinux/config as we are in the process of deprecating that.

     See [1] for more background on this if you missed the previous
     notes on the deprecation.

   - Minor cleanups: remove unneeded variables and function parameter
     constification"

Link: https://github.com/SELinuxProject/selinux-kernel/wiki/DEPRECATE-runtime-disable
* tag 'selinux-pr-20221003' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux:
  selinux: remove runtime disable message in the install_policy.sh script
  selinux: use "grep -E" instead of "egrep"
  selinux: remove the unneeded result variable
  selinux: declare read-only parameters const
  selinux: use int arrays for boolean values
  selinux: remove an unneeded variable in sel_make_class_dir_entries()

21 months agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Jakub Kicinski [Tue, 4 Oct 2022 00:44:18 +0000 (17:44 -0700)]
Merge git://git./linux/kernel/git/netdev/net

Merge in the left-over fixes before the net-next pull-request.

Conflicts:

drivers/net/ethernet/mediatek/mtk_ppe.c
  ae3ed15da588 ("net: ethernet: mtk_eth_soc: fix state in __mtk_foe_entry_clear")
  9d8cb4c096ab ("net: ethernet: mtk_eth_soc: add foe_entry_size to mtk_eth_soc")
https://lore.kernel.org/all/6cb6893b-4921-a068-4c30-1109795110bb@tessares.net/

kernel/bpf/helpers.c
  8addbfc7b308 ("bpf: Gate dynptr API behind CAP_BPF")
  5679ff2f138f ("bpf: Move bpf_loop and bpf_for_each_map_elem under CAP_BPF")
  8a67f2de9b1d ("bpf: expose bpf_strtol and bpf_strtoul to all program types")
https://lore.kernel.org/all/20221003201957.13149-1-daniel@iogearbox.net/

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoMerge tag 'integrity-v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/zohar...
Linus Torvalds [Tue, 4 Oct 2022 00:42:12 +0000 (17:42 -0700)]
Merge tag 'integrity-v6.1' of git://git./linux/kernel/git/zohar/linux-integrity

Pull integrity updates from Mimi Zohar:
 "Just two bug fixes"

* tag 'integrity-v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/zohar/linux-integrity:
  efi: Correct Macmini DMI match in uefi cert quirk
  ima: fix blocking of security.ima xattrs of unsupported algorithms

21 months agoMerge tag 'Smack-for-6.1' of https://github.com/cschaufler/smack-next
Linus Torvalds [Tue, 4 Oct 2022 00:38:09 +0000 (17:38 -0700)]
Merge tag 'Smack-for-6.1' of https://github.com/cschaufler/smack-next

Pull smack updates from Casey Schaufler:
 "Two minor code clean-ups: one removes constants left over from the old
  mount API, while the other gets rid of an unneeded variable.

  The other change fixes a flaw in handling IPv6 labeling"

* tag 'Smack-for-6.1' of https://github.com/cschaufler/smack-next:
  smack: cleanup obsolete mount option flags
  smack: lsm: remove the unneeded result variable
  SMACK: Add sk_clone_security LSM hook

21 months agoonce: rename _SLOW to _SLEEPABLE
Jason A. Donenfeld [Mon, 3 Oct 2022 18:14:13 +0000 (20:14 +0200)]
once: rename _SLOW to _SLEEPABLE

The _SLOW designation wasn't really descriptive of anything. This is
meant to be called from process context when it's possible to sleep. So
name this more aptly _SLEEPABLE, which better fits its intended use.

Fixes: 62c07983bef9 ("once: add DO_ONCE_SLOW() for sleepable contexts")
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20221003181413.1221968-1-Jason@zx2c4.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoMerge branch 'add-generic-pse-support'
Jakub Kicinski [Tue, 4 Oct 2022 00:34:06 +0000 (17:34 -0700)]
Merge branch 'add-generic-pse-support'

Oleksij Rempel says:

====================
add generic PSE support

Add generic support for the Ethernet Power Sourcing Equipment.
====================

Link: https://lore.kernel.org/r/20221003065202.3889095-1-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: pse-pd: add regulator based PSE driver
Oleksij Rempel [Mon, 3 Oct 2022 06:52:02 +0000 (08:52 +0200)]
net: pse-pd: add regulator based PSE driver

Add generic, regulator based PSE driver to support simple Power Sourcing
Equipment without automatic classification support.

This driver was tested on 10Bast-T1L switch with regulator based PoDL PSE.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agodt-bindings: net: pse-dt: add bindings for regulator based PoDL PSE controller
Oleksij Rempel [Mon, 3 Oct 2022 06:52:01 +0000 (08:52 +0200)]
dt-bindings: net: pse-dt: add bindings for regulator based PoDL PSE controller

Add bindings for the regulator based Ethernet PoDL PSE controller and
generic bindings for all PSE controllers.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoethtool: add interface to interact with Ethernet Power Equipment
Oleksij Rempel [Mon, 3 Oct 2022 06:52:00 +0000 (08:52 +0200)]
ethtool: add interface to interact with Ethernet Power Equipment

Add interface to support Power Sourcing Equipment. At current step it
provides generic way to address all variants of PSE devices as defined
in IEEE 802.3-2018 but support only objects specified for IEEE 802.3-2018 104.4
PoDL Power Sourcing Equipment (PSE).

Currently supported and mandatory objects are:
IEEE 802.3-2018 30.15.1.1.3 aPoDLPSEPowerDetectionStatus
IEEE 802.3-2018 30.15.1.1.2 aPoDLPSEAdminState
IEEE 802.3-2018 30.15.1.2.1 acPoDLPSEAdminControl

This is minimal interface needed to control PSE on each separate
ethernet port but it provides not all mandatory objects specified in
IEEE 802.3-2018.

Since "PoDL PSE" and "PSE" have similar names, but some different values
I decide to not merge them and keep separate naming schema. This should
allow as to be as close to IEEE 802.3 spec as possible and avoid name
conflicts in the future.

This implementation is connected to PHYs instead of MACs because PSE
auto classification can potentially interfere with PHY auto negotiation.
So, may be some extra PHY related initialization will be needed.

With WIP version of ethtools interaction with PSE capable link looks
as following:

$ ip l
...
5: t1l1@eth0: <BROADCAST,MULTICAST> ..
...

$ ethtool --show-pse t1l1
PSE attributs for t1l1:
PoDL PSE Admin State: disabled
PoDL PSE Power Detection Status: disabled

$ ethtool --set-pse t1l1 podl-pse-admin-control enable
$ ethtool --show-pse t1l1
PSE attributs for t1l1:
PoDL PSE Admin State: enabled
PoDL PSE Power Detection Status: delivering power

Signed-off-by: kernel test robot <lkp@intel.com>
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Bagas Sanjaya <bagasdotme@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: mdiobus: search for PSE nodes by parsing PHY nodes.
Oleksij Rempel [Mon, 3 Oct 2022 06:51:59 +0000 (08:51 +0200)]
net: mdiobus: search for PSE nodes by parsing PHY nodes.

Some PHYs can be linked with PSE (Power Sourcing Equipment), so search
for related nodes and attach it to the phydev.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: mdiobus: fwnode_mdiobus_register_phy() rework error handling
Oleksij Rempel [Mon, 3 Oct 2022 06:51:58 +0000 (08:51 +0200)]
net: mdiobus: fwnode_mdiobus_register_phy() rework error handling

Rework error handling as preparation for PSE patch. This patch should
make it easier to extend this function.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: add framework to support Ethernet PSE and PDs devices
Oleksij Rempel [Mon, 3 Oct 2022 06:51:57 +0000 (08:51 +0200)]
net: add framework to support Ethernet PSE and PDs devices

This framework was create with intention to provide support for Ethernet PSE
(Power Sourcing Equipment) and PDs (Powered Device).

At current step this patch implements generic PSE support for PoDL (Power over
Data Lines 802.3bu) specification with reserving name space for PD devices as
well.

This framework can be extended to support 802.3af and 802.3at "Power via the
Media Dependent Interface" (or PoE/Power over Ethernet)

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agodt-bindings: net: phy: add PoDL PSE property
Oleksij Rempel [Mon, 3 Oct 2022 06:51:56 +0000 (08:51 +0200)]
dt-bindings: net: phy: add PoDL PSE property

Add property to reference node representing a PoDL Power Sourcing Equipment.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months ago[brown paperbag] fix coredump breakage
Al Viro [Tue, 4 Oct 2022 00:26:08 +0000 (20:26 -0400)]
[brown paperbag] fix coredump breakage

Let me count the ways in which I'd screwed up:

* when emitting a page, handling of gaps in coredump should happen
before fetching the current file position.
* fix for a problem that occurs on rather uncommon setups (and hadn't
been observed in the wild) had been sent very late in the cycle.
* ... with badly insufficient testing, introducing an easily
reproducible breakage.  Without giving it time to soak in -next.

Fucked-up-by: Al Viro <viro@zeniv.linux.org.uk>
Reported-by: "J. R. Okajima" <hooanon05g@gmail.com>
Tested-by: "J. R. Okajima" <hooanon05g@gmail.com>
Fixes: 06bbaa6dc53c "[coredump] don't use __kernel_write() on kmap_local_page()"
Cc: stable@kernel.org # v6.0-only
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
21 months agoMerge tag 'hardening-v6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees...
Linus Torvalds [Tue, 4 Oct 2022 00:24:22 +0000 (17:24 -0700)]
Merge tag 'hardening-v6.1-rc1' of git://git./linux/kernel/git/kees/linux

Pull kernel hardening updates from Kees Cook:
 "Most of the collected changes here are fixes across the tree for
  various hardening features (details noted below).

  The most notable new feature here is the addition of the memcpy()
  overflow warning (under CONFIG_FORTIFY_SOURCE), which is the next step
  on the path to killing the common class of "trivially detectable"
  buffer overflow conditions (i.e. on arrays with sizes known at compile
  time) that have resulted in many exploitable vulnerabilities over the
  years (e.g. BleedingTooth).

  This feature is expected to still have some undiscovered false
  positives. It's been in -next for a full development cycle and all the
  reported false positives have been fixed in their respective trees.
  All the known-bad code patterns we could find with Coccinelle are also
  either fixed in their respective trees or in flight.

  The commit message in commit 54d9469bc515 ("fortify: Add run-time WARN
  for cross-field memcpy()") for the feature has extensive details, but
  I'll repeat here that this is a warning _only_, and is not intended to
  actually block overflows (yet). The many patches fixing array sizes
  and struct members have been landing for several years now, and we're
  finally able to turn this on to find any remaining stragglers.

  Summary:

  Various fixes across several hardening areas:

   - loadpin: Fix verity target enforcement (Matthias Kaehlcke).

   - zero-call-used-regs: Add missing clobbers in paravirt (Bill
     Wendling).

   - CFI: clean up sparc function pointer type mismatches (Bart Van
     Assche).

   - Clang: Adjust compiler flag detection for various Clang changes
     (Sami Tolvanen, Kees Cook).

   - fortify: Fix warnings in arch-specific code in sh, ARM, and xen.

  Improvements to existing features:

   - testing: improve overflow KUnit test, introduce fortify KUnit test,
     add more coverage to LKDTM tests (Bart Van Assche, Kees Cook).

   - overflow: Relax overflow type checking for wider utility.

  New features:

   - string: Introduce strtomem() and strtomem_pad() to fill a gap in
     strncpy() replacement needs.

   - um: Enable FORTIFY_SOURCE support.

   - fortify: Enable run-time struct member memcpy() overflow warning"

* tag 'hardening-v6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux: (27 commits)
  Makefile.extrawarn: Move -Wcast-function-type-strict to W=1
  hardening: Remove Clang's enable flag for -ftrivial-auto-var-init=zero
  sparc: Unbreak the build
  x86/paravirt: add extra clobbers with ZERO_CALL_USED_REGS enabled
  x86/paravirt: clean up typos and grammaros
  fortify: Convert to struct vs member helpers
  fortify: Explicitly check bounds are compile-time constants
  x86/entry: Work around Clang __bdos() bug
  ARM: decompressor: Include .data.rel.ro.local
  fortify: Adjust KUnit test for modular build
  sh: machvec: Use char[] for section boundaries
  kunit/memcpy: Avoid pathological compile-time string size
  lib: Improve the is_signed_type() kunit test
  LoadPin: Require file with verity root digests to have a header
  dm: verity-loadpin: Only trust verity targets with enforcement
  LoadPin: Fix Kconfig doc about format of file with verity digests
  um: Enable FORTIFY_SOURCE
  lkdtm: Update tests for memcpy() run-time warnings
  fortify: Add run-time WARN for cross-field memcpy()
  fortify: Use SIZE_MAX instead of (size_t)-1
  ...

21 months agoMerge branch 'net-marvell-prestera-add-nexthop-routes-offloading'
Jakub Kicinski [Tue, 4 Oct 2022 00:14:56 +0000 (17:14 -0700)]
Merge branch 'net-marvell-prestera-add-nexthop-routes-offloading'

Yevhen Orlov says:

====================
net: marvell: prestera: add nexthop routes offloading

Add support for nexthop routes for Marvell Prestera driver.
Subscribe on NEIGH_UPDATE events.

Add features:
 - Support connected route adding
   e.g.: "ip address add 1.1.1.1/24 dev sw1p1"
   e.g.: "ip route add 6.6.6/24 dev sw1p1"
 - Support nexthop route adding
   e.g.: "ip route add 5.5.5/24 via 1.1.1.2"
 - Support ECMP route adding
   e.g.: "ip route add 5.5.5/24 nexthop via 1.1.1.2 nexthop via 1.1.1.3"
 - Support "offload" and "trap" flags per each nexthop
 - Support "offload" flag for neighbours

Limitations:
 - Only "local" and "main" tables supported
 - Only generic interfaces supported for router (no bridges or vlans)

Flags meaning:
  ip route add 5.5.5/24 nexthop via 2.2.2.2 nexthop via 2.2.2.3
  ip route show
  ...
  5.5.5.0/24 rt_offload
        nexthop via 2.2.2.2 dev sw1p31 weight 1 trap
        nexthop via 2.2.2.3 dev sw1p31 weight 1 trap
  ...
  # When you just add route - lpm entry became occupied
  # in HW ("rt_offload" flag), but related to nexthops neighbours
  # still not resolved ("trap" flag).
  #
  # After some time...
  ip route show
  ...
  5.5.5.0/24 rt_offload
        nexthop via 2.2.2.2 dev sw1p31 weight 1 offload
        nexthop via 2.2.2.3 dev sw1p31 weight 1 offload
  ...
  # You will see, that appropriate neighbours was resolved and nexthop
  # entries occupied in HW too ("offload" flag)

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Changes for v2:
* Add more reviewers in CC
* Check if route nexthop or direct with fib_nh_gw_family instead of fib_nh_scope
  This is needed after,
  747c14307214 ("ip: fix dflt addr selection for connected nexthop"),
  because direct route is now with the same scope as nexthop (RT_SCOPE_LINK)

Changes for v3:
* Resolve "unused functions" warnings, after
  patch ("net: marvell: prestera: Add heplers to interact ... "), and before
  patch ("net: marvell: prestera: Add neighbour cache accounting")

Changes for v4:
* Rebase to the latest master to resolve patch applying issues

Changes for v5:
* Repack structures to prevent holes
* Remove unused variables
* Fix misspeling issues

Changes for v6:
* Rebase on top of master
* Fix smatch warnings

Changes for v7:
* Rebase on top of master
* Refactor: use "fib_lookup" instead of "fib_new_table"+"fib_table_lookup",
  according to Paolo Abeni suggestion
* Refactor: use "rhashtable_free_and_destroy" instead of rhashtable
  walk, according to Paolo Abeni suggestion
====================

Link: https://lore.kernel.org/r/20221001093417.22388-1-yevhen.orlov@plvision.eu
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: Propagate nh state from hw to kernel
Yevhen Orlov [Sat, 1 Oct 2022 09:34:17 +0000 (12:34 +0300)]
net: marvell: prestera: Propagate nh state from hw to kernel

We poll nexthops in HW and call for each active nexthop appropriate
neighbour.

Also we provide implicity neighbour resolving.
For example, user have added nexthop route:
  # ip route add 5.5.5.5 via 1.1.1.2
But neighbour 1.1.1.2 doesn't exist. In this case we will try to call
neigh_event_send, even if there is no traffic.
This is useful, when you have add route, which will be used after some
time but with a lot of traffic (burst). So, we has prepared, offloaded
route in advance.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: Add neighbour cache accounting
Yevhen Orlov [Sat, 1 Oct 2022 09:34:16 +0000 (12:34 +0300)]
net: marvell: prestera: Add neighbour cache accounting

Move forward and use new PRESTERA_FIB_TYPE_UC_NH to provide basic
nexthop routes support.
Provide deinitialization sequence for all created router objects.

Limitations:
- Only "local" and "main" tables supported
- Only generic interfaces supported for router (no bridges or vlans)

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: add stub handler neighbour events
Yevhen Orlov [Sat, 1 Oct 2022 09:34:15 +0000 (12:34 +0300)]
net: marvell: prestera: add stub handler neighbour events

Actual handler will be added in next patches

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: Add heplers to interact with fib_notifier_info
Yevhen Orlov [Sat, 1 Oct 2022 09:34:14 +0000 (12:34 +0300)]
net: marvell: prestera: Add heplers to interact with fib_notifier_info

This will be used to implement nexthops related logic in next patches.
Also try to keep ipv4/6 abstraction to be able to reuse helpers for ipv6
in the future.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: Add length macros for prestera_ip_addr
Yevhen Orlov [Sat, 1 Oct 2022 09:34:13 +0000 (12:34 +0300)]
net: marvell: prestera: Add length macros for prestera_ip_addr

Add macros to determine IP address length (internal driver types).
This will be used in next patches for nexthops logic.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: add delayed wq and flush wq on deinit
Yevhen Orlov [Sat, 1 Oct 2022 09:34:12 +0000 (12:34 +0300)]
net: marvell: prestera: add delayed wq and flush wq on deinit

Flushing workqueues ensures, that no more pending works, related to just
unregistered or deinitialized notifiers. After that we can free memory.

Delayed wq will be used for neighbours in next patches.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: Add strict cleanup of fib arbiter
Yevhen Orlov [Sat, 1 Oct 2022 09:34:11 +0000 (12:34 +0300)]
net: marvell: prestera: Add strict cleanup of fib arbiter

This will, ensure, that there is no more, preciously allocated fib_cache
entries left after deinit.
Will be used to free allocated resources of nexthop routes, that points
to "not our" port (e.g. eth0).

Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: Add cleanup of allocated fib_nodes
Yevhen Orlov [Sat, 1 Oct 2022 09:34:10 +0000 (12:34 +0300)]
net: marvell: prestera: Add cleanup of allocated fib_nodes

Do explicity cleanup on router_hw_fini, to ensure, that all allocated
objects cleaned. This will be used in cases,
when upper layer (cache) is not mapped to router_hw layer.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: marvell: prestera: Add router nexthops ABI
Yevhen Orlov [Sat, 1 Oct 2022 09:34:09 +0000 (12:34 +0300)]
net: marvell: prestera: Add router nexthops ABI

- Add functions to allocate/delete/set nexthop group
  - NOTE: non-ECMP nexthop is nexthop group with allocated size = 1
- Add function to read state of HW nh (if packets going through it)

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoMerge tag 'kcfi-v6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux
Linus Torvalds [Tue, 4 Oct 2022 00:11:07 +0000 (17:11 -0700)]
Merge tag 'kcfi-v6.1-rc1' of git://git./linux/kernel/git/kees/linux

Pull kcfi updates from Kees Cook:
 "This replaces the prior support for Clang's standard Control Flow
  Integrity (CFI) instrumentation, which has required a lot of special
  conditions (e.g. LTO) and work-arounds.

  The new implementation ("Kernel CFI") is specific to C, directly
  designed for the Linux kernel, and takes advantage of architectural
  features like x86's IBT. This series retains arm64 support and adds
  x86 support.

  GCC support is expected in the future[1], and additional "generic"
  architectural support is expected soon[2].

  Summary:

   - treewide: Remove old CFI support details

   - arm64: Replace Clang CFI support with Clang KCFI support

   - x86: Introduce Clang KCFI support"

Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107048
Link: https://github.com/samitolvanen/llvm-project/commits/kcfi_generic
* tag 'kcfi-v6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux: (22 commits)
  x86: Add support for CONFIG_CFI_CLANG
  x86/purgatory: Disable CFI
  x86: Add types to indirectly called assembly functions
  x86/tools/relocs: Ignore __kcfi_typeid_ relocations
  kallsyms: Drop CONFIG_CFI_CLANG workarounds
  objtool: Disable CFI warnings
  objtool: Preserve special st_shndx indexes in elf_update_symbol
  treewide: Drop __cficanonical
  treewide: Drop WARN_ON_FUNCTION_MISMATCH
  treewide: Drop function_nocfi
  init: Drop __nocfi from __init
  arm64: Drop unneeded __nocfi attributes
  arm64: Add CFI error handling
  arm64: Add types to indirect called assembly functions
  psci: Fix the function type for psci_initcall_t
  lkdtm: Emit an indirect call for CFI tests
  cfi: Add type helper macros
  cfi: Switch to -fsanitize=kcfi
  cfi: Drop __CFI_ADDRESSABLE
  cfi: Remove CONFIG_CFI_CLANG_SHADOW
  ...

21 months agoeth: octeon: fix build after netif_napi_add() changes
Jakub Kicinski [Sun, 2 Oct 2022 17:56:50 +0000 (10:56 -0700)]
eth: octeon: fix build after netif_napi_add() changes

Guenter reports I missed a netif_napi_add() call
in one of the platform-specific drivers:

drivers/net/ethernet/cavium/octeon/octeon_mgmt.c: In function 'octeon_mgmt_probe':
drivers/net/ethernet/cavium/octeon/octeon_mgmt.c:1399:9: error: too many arguments to function 'netif_napi_add'
 1399 |         netif_napi_add(netdev, &p->napi, octeon_mgmt_napi_poll,
      |         ^~~~~~~~~~~~~~

Reported-by: Guenter Roeck <linux@roeck-us.net>
Fixes: b48b89f9c189 ("net: drop the weight argument from netif_napi_add")
Link: https://lore.kernel.org/r/20221002175650.1491124-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoMerge tag 'execve-v6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees...
Linus Torvalds [Mon, 3 Oct 2022 23:56:40 +0000 (16:56 -0700)]
Merge tag 'execve-v6.1-rc1' of git://git./linux/kernel/git/kees/linux

Pull execve updates from Kees Cook:
 "This removes a.out support globally; it has been disabled for a while
  now.

   - Remove a.out implementation globally (Eric W. Biederman)

   - Remove unused linux_binprm::taso member (Lukas Bulwahn)"

* tag 'execve-v6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  binfmt: remove taso from linux_binprm struct
  a.out: Remove the a.out implementation

21 months agoMerge branch 'mlx5-xsk-updates-part4-and-more'
Jakub Kicinski [Mon, 3 Oct 2022 23:55:33 +0000 (16:55 -0700)]
Merge branch 'mlx5-xsk-updates-part4-and-more'

Saeed Mahameed says:

====================
mlx5 xsk updates part4 and more

1) Final part of xsk improvements,
in this series Maxim continues to improve xsk implementation
 a) XSK Busy polling support
 b) Use KLM to avoid Frame overrun in unaligned mode
 c) Optimize unaligned more for certain frame sizes
 d) Other straight forward minor optimizations.

part 1: https://lore.kernel.org/netdev/20220927203611.244301-1-saeed@kernel.org/
part 2: https://lore.kernel.org/netdev/20220929072156.93299-1-saeed@kernel.org/
part 3: https://lore.kernel.org/netdev/20220930162903.62262-1-saeed@kernel.org/

2) Oversize packets firmware counter, from Gal.

3) Set default grace period for health reporters based on function type

4) Some minor E-Switch improvements
====================

Link: https://lore.kernel.org/r/20221002045632.291612-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5: E-Switch, Return EBUSY if can't get mode lock
Jianbo Liu [Sun, 2 Oct 2022 04:56:32 +0000 (21:56 -0700)]
net/mlx5: E-Switch, Return EBUSY if can't get mode lock

It is to avoid tc retrying during device mode change.

Signed-off-by: Jianbo Liu <jianbol@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5: E-switch, Don't update group if qos is not enabled
Chris Mi [Sun, 2 Oct 2022 04:56:31 +0000 (21:56 -0700)]
net/mlx5: E-switch, Don't update group if qos is not enabled

Currently, qos group will be updated and qos will be enabled when
unregistering devlink port. Actually no need to update group if qos
is not enabled.

Add a check to prevent unnecessary enabling and disabling qos for
every port.

Signed-off-by: Chris Mi <cmi@nvidia.com>
Reviewed-by: Dmytro Linkin <dlinkin@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5: E-Switch, Allow offloading fwd dest flow table with vport
Roi Dayan [Sun, 2 Oct 2022 04:56:30 +0000 (21:56 -0700)]
net/mlx5: E-Switch, Allow offloading fwd dest flow table with vport

Before this commit a fwd dest flow table resulted in ignoring vport dests
which is incorrect and is supported.
With this commit the dests can be a mix of flow table and vport dests.
There is still a limitation that there cannot be more than one flow table dest.

Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5: Set default grace period based on function type
Maher Sanalla [Sun, 2 Oct 2022 04:56:29 +0000 (21:56 -0700)]
net/mlx5: Set default grace period based on function type

Currently, driver sets the same grace period for fw fatal health reporter
to any type of function.

Since the lower level functions are more vulnerable to fw fatal errors as a
result of parent function closure/reload, set a smaller grace period for
the lower level functions, as follows:

1. For ECPF: 180 seconds.
2. For PF: 60 seconds.
3. For VF/SF: 30 seconds.

Signed-off-by: Maher Sanalla <msanalla@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5: Start health poll at earlier stage of driver load
Moshe Shemesh [Sun, 2 Oct 2022 04:56:28 +0000 (21:56 -0700)]
net/mlx5: Start health poll at earlier stage of driver load

Start health poll at earlier stage, so if fw fatal issue occurred before
or during initialization commands such as init_hca or set_hca_cap the
poll health can detect and indicate that the driver is already in error
state.

Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: Expose rx_oversize_pkts_buffer counter
Gal Pressman [Sun, 2 Oct 2022 04:56:27 +0000 (21:56 -0700)]
net/mlx5e: Expose rx_oversize_pkts_buffer counter

Add the rx_oversize_pkts_buffer counter to ethtool statistics.
This counter exposes the number of dropped received packets due to
length which arrived to RQ and exceed software buffer size allocated by
the device for incoming traffic. It might imply that the device MTU is
larger than the software buffers size.

Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: xsk: Optimize for unaligned mode with 3072-byte frames
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:26 +0000 (21:56 -0700)]
net/mlx5e: xsk: Optimize for unaligned mode with 3072-byte frames

When XSK frame size is 3072 (or another power of two multiplied by 3),
KLM mechanism for NIC virtual memory page mapping can be optimized by
replacing it with KSM.

Before this change, two KLM entries were needed to map an XSK frame that
is not a power of two: one entry maps the UMEM memory up to the frame
length, the other maps the rest of the stride to the garbage page.

When the frame length divided by 3 is a power of two, it can be mapped
using 3 KSM entries, and the fourth will map the rest of the stride to
the garbage page. All 4 KSM entries are of the same size, which allows
for a much faster lookup.

Frame size 3072 is useful in certain use cases, because it allows
packing 4 frames into 3 pages. Generally speaking, other frame sizes
equal to PAGE_SIZE minus a power of two can be optimized in a similar
way, but it will require many more KSMs per frame, which slows down UMRs
a little bit, but more importantly may hit the limit for the maximum
number of KSM entries.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: xsk: Print a warning in slow configurations
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:25 +0000 (21:56 -0700)]
net/mlx5e: xsk: Print a warning in slow configurations

On striding RQ, when the XSK frame size doesn't match the MKey page
size, KLM is used for memory mappings, which is a slower mechanism than
MTT or KSM. It may happen in two cases:

1. Frame size is not a power of two (only possible in the unaligned mode
of XSK).

2. Frame size is 2048 bytes, and the firmware doesn't support MKey pages
smaller than 4096 bytes.

Depending on the case, print a warning and recommend to disable striding
RQ or upgrade the firmware.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: xsk: Use KLM to protect frame overrun in unaligned mode
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:24 +0000 (21:56 -0700)]
net/mlx5e: xsk: Use KLM to protect frame overrun in unaligned mode

XSK RQs support striding RQ linear mode, but the stride size may be
bigger than the XSK frame size, because:

1. The stride size must be a power of two.

2. The stride size must be equal to the UMR page size. Each XSK frame is
treated as a separate page, because they aren't necessarily adjacent in
physical memory, so the driver can't put more than one stride per page.

3. The minimal MTT page size is 4096 on older firmware.

That means that if XSK frame size is 2048 or not a power of two, the
strides may be bigger than XSK frames. Normally, it's not a problem if
the hardware enforces the MTU. However, traffic between vports skips the
hardware MTU check, and oversized packets may be received.

If an oversized packet is bigger than the XSK frame but not bigger than
the stride, it will cause overwriting of the adjacent UMEM region. If
the packet takes more than one stride, they can be recycled for reuse,
so it's not a problem when the XSK frame size matches the stride size.

Work around the above issue by leveraging KLM to make a more
fine-grained mapping. The beginning of each stride is mapped to the
frame memory, and the padding up to the closest power of two is mapped
to the overflow page that doesn't belong to UMEM. This way, application
data corruption won't happen upon receiving packets bigger than MTU.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: Improve MTT/KSM alignment
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:23 +0000 (21:56 -0700)]
net/mlx5e: Improve MTT/KSM alignment

Make mlx5e_mpwrq_mtts_per_wqe take into account that KSM requires
smaller alignment than MTT.

Ensure that there is always an even amount of MTTs in a UMR WQE, so that
complete octwords are formed, and no garbage is mapped.

Drop extra alignment in MLX5_MTT_OCTW that may cause setting too big
ucseg->xlt_octowords, also leading to mapping garbage.

Generalize some calculations by introducing the MLX5_OCTWORD constant.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: xsk: Use umr_mode to calculate striding RQ parameters
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:22 +0000 (21:56 -0700)]
net/mlx5e: xsk: Use umr_mode to calculate striding RQ parameters

Instead of passing the unaligned flag, pass an enum that indicates the
UMR mode. The next commit will add the third mode (KLM for certain
configurations of XSK), which will be added to this enum instead of
adding another bool flag everywhere.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: xsk: Improve need_wakeup logic
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:21 +0000 (21:56 -0700)]
net/mlx5e: xsk: Improve need_wakeup logic

XSK need_wakeup mechanism allows the driver to stop busy waiting for
buffers when the fill ring is empty, yield to the application and signal
it that the driver needs to be waken up after the application refills
the fill ring.

Add protection against the race condition on the RX (refill) side: if
the application refills buffers after xskrq->post_wqes is called, but
before mlx5e_xsk_update_rx_wakeup, NAPI will exit, skipping taking these
buffers to the hardware WQ, and the application won't wake it up again.

Optimize the whole need_wakeup logic, removing unneeded flows, to
compensate for this new check.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: xsk: Include XSK skb_from_cqe callbacks in INDIRECT_CALL
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:20 +0000 (21:56 -0700)]
net/mlx5e: xsk: Include XSK skb_from_cqe callbacks in INDIRECT_CALL

XSK is a performance-critical data path. To avoid an indirect function
call with a retpoline, include XSK callbacks in the INDIRECT_CALL macro,
so that they are called directly in XSK flows.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: xsk: Set napi_id to support busy polling
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:19 +0000 (21:56 -0700)]
net/mlx5e: xsk: Set napi_id to support busy polling

xdp_rxq_info_reg should get the actual napi_id, not 0, in order to
support socket busy polling properly.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/mlx5e: xsk: Flush RQ on XSK activation to save memory
Maxim Mikityanskiy [Sun, 2 Oct 2022 04:56:18 +0000 (21:56 -0700)]
net/mlx5e: xsk: Flush RQ on XSK activation to save memory

The regular RQ remains open after opening an XSK socket, in order to
guarantee that closing the XSK socket never fails due to an error when
reopening the regular RQ.

To save memory, the regular RQ can be deactivated and flushed, releasing
all pages, when an XSK socket is open.

Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: mvpp2: fix mvpp2 debugfs leak
Russell King (Oracle) [Mon, 3 Oct 2022 16:19:27 +0000 (17:19 +0100)]
net: mvpp2: fix mvpp2 debugfs leak

When mvpp2 is unloaded, the driver specific debugfs directory is not
removed, which technically leads to a memory leak. However, this
directory is only created when the first device is probed, so the
hardware is present. Removing the module is only something a developer
would to when e.g. testing out changes, so the module would be
reloaded. So this memory leak is minor.

The original attempt in commit fe2c9c61f668 ("net: mvpp2: debugfs: fix
memory leak when using debugfs_lookup()") that was labelled as a memory
leak fix was not, it fixed a refcount leak, but in doing so created a
problem when the module is reloaded - the directory already exists, but
mvpp2_root is NULL, so we lose all debugfs entries. This fix has been
reverted.

This is the alternative fix, where we remove the offending directory
whenever the driver is unloaded.

Fixes: 21da57a23125 ("net: mvpp2: add a debugfs interface for the Header Parser")
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Marcin Wojtas <mw@semihalf.com>
Link: https://lore.kernel.org/r/E1ofOAB-00CzkG-UO@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet: ipa: update copyrights
Alex Elder [Fri, 30 Sep 2022 22:45:49 +0000 (17:45 -0500)]
net: ipa: update copyrights

Some source files state copyright dates that are earlier than the
last modification of the file.  Change the copyright year to 2022 in
all such cases.

Signed-off-by: Alex Elder <elder@linaro.org>
Link: https://lore.kernel.org/r/20220930224549.3503434-1-elder@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>