In order to reach non-immediate remote node services that are
accessed through an intermediate node, the route to the remote
node needs to be saved.
E.g for a [node1 <=> node2 <=> node3] network
- node2 forwards node3 service to node1
- node1 must save node2 as route for reaching node3
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
A remote endpoint (immediate neighbors node) can forward services
from other nodes (non-immadiate), in that case ctrl packet node ID
(offering distant service) can differ from the qrtr source node
(forwarding the packet).
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The port ID for control messages was uncorrectly set with broadcast
node ID value, causing message to be dropped on remote side since
not passing packet filtering (cb->dst_port != QRTR_PORT_CTRL).
Fixes: d27e77a3de ("net: qrtr: Reset the node and port ID of broadcast messages")
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The referenced commit added in .config_intr() the part of code which upon
configuration of the IRQ state it also clears up any pending IRQ. If
there were actually pending IRQs, a read on the IRQ status register will
return something non zero. This should not result in the callback
returning an error.
Fix this by returning an error only when the result of the
phy_read_mmd() is negative.
Fixes: e11ef96d44 ("net: phy: aquantia: remove the use of .ack_interrupt()")
Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com>
Tested-by: Camelia Groza <camelia.groza@nxp.com>
Link: https://lore.kernel.org/r/20201109154601.3812574-1-ciorneiioana@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Alex Elder says:
====================
net: ipa: little fixes
This series adds a few small fixes to the IPA code.
The first patch appeared in a different form in June, and got some
pushback from David because he felt a problem that can be caught at
build time *should* be caught at build time.
https://lore.kernel.org/netdev/20200610195332.2612233-1-elder@linaro.org/
I agree with that, but in this case the "problem" was never actually
a problem. There's a little more explanation on the patch, but
basically now we remove the BUILD_BUG_ON() call entirely.
The second deletes a line of code that isn't needed.
The third converts a warning message to be a debug, as requested by
Stephen Boyd.
And the last one just gets rid of an error message that would be
output after a different message had already reported a problem.
====================
Link: https://lore.kernel.org/r/20201109165635.5449-1-elder@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
There is no need for gsi_modem_channel_halt() to report an error,
because gsi_generic_command() will already have done that if the
command times out. So get rid of the extra message.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When we determine from hardware what the size of IPA memory is
we compare it against what we learned about it from DT.
If DT defines a region that's larger than actual memory, we use the
smaller actual size and issue a warning.
If DT defines a smaller region than actual memory we issue a warning
too. But in this case the difference is harmless; so rather than
issuing a warning, just provide a debug message instead.
Reorder these checks so the one that matters more is done first.
Reported-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Delete a spurious line of code in ipa_hardware_config(). It reads a
register value then ignores the value, so is completely unnecessary.
Add a missing word in a comment.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The following call in ipa_validate_build() is erroneous:
BUILD_BUG_ON(sizeof(struct gsi_trans) > 128);
The fact is, it is not a bug for the size of a GSI transaction to be
bigger than 128 bytes. The correct operation of the driver is not
dependent on the size of this structure. The only consequence of
the transaction being large is that the amount of memory required
is larger.
The problem this was trying to flag is that a *slight* increase in
the size of this structure will have a disproportionate effect on
the amount of memory used. E.g. if the structure grew to 132 bytes
the memory requirement for the transaction arrays would be about
double.
With various debugging build flags enabled, the size grows to 160
bytes. But there's no reason to treat that as a build-time bug.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Eric Dumazet says:
====================
inet: prevent skb changes in udp{4|6}_lib_lookup_skb()
This came while reviewing Alexander Lobakin patch against UDP GRO:
We want to make sure skb wont be changed by these helpers
while it is owned by GRO stack.
====================
Link: https://lore.kernel.org/r/20201109231349.20946-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The skb is needed only to fetch the keys for the lookup.
Both functions are used from GRO stack, we do not want
accidental modification of the skb.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Alexander Lobakin <alobakin@pm.me>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
inet_sdif() does not modify the skb.
This will permit propagating the const qualifier in
udp{4|6}_lib_lookup_skb() functions.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Alexander Lobakin <alobakin@pm.me>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The left shift of u16 variable high is promoted to the type int and
then sign extended to a 64 bit u64 value. If the top bit of high is
set then the upper 32 bits of the result end up being set by the
sign extension. Fix this by explicitly casting the value in high to
a u64 before left shifting by 16 places.
Also, remove the initialisation of variable value to 0 at the start
of each loop iteration as the value is never read and hence the
assignment it is redundant.
Addresses-Coverity: ("Unintended sign extension")
Fixes: e4b27ebc78 ("net: dsa: Add DSA driver for Hirschmann Hellcreek switches")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de>
Link: https://lore.kernel.org/r/20201109124008.2079873-1-colin.king@canonical.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Similar to commit fda55eca5a
("net: introduce skb_transport_header_was_set()"), avoid resetting
transport offsets that were already set by GRO layer. This not only
mirrors the behavior of __netif_receive_skb_core(), but also makes
sense when it comes to UDP GSO fraglists forwarding: transport offset
of such skbs is set only once by GRO receive callback and remains
untouched and correct up to the xmitting driver in 1:1 case, but
becomes junk after untagging in ingress VLAN case and breaks UDP
GSO offload. This does not happen after this change, and all types
of forwarding of UDP GSO fraglists work as expected.
Since v1 [1]:
- keep the code 1:1 with __netif_receive_skb_core() (Jakub).
[1] https://lore.kernel.org/netdev/zYurwsZRN7BkqSoikWQLVqHyxz18h4LhHU4NFa2Vw@cp4-web-038.plabs.ch
Signed-off-by: Alexander Lobakin <alobakin@pm.me>
Link: https://lore.kernel.org/r/7JgIkgEztzt0W6ZtC9V9Cnk5qfkrUFYcpN871syCi8@cp4-web-040.plabs.ch
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Heiner Kallweit says:
====================
net: add and use dev_get_tstats64
It's a frequent pattern to use netdev->stats for the less frequently
accessed counters and per-cpu counters for the frequently accessed
counters (rx/tx bytes/packets). Add a default ndo_get_stats64()
implementation for this use case. Subsequently switch more drivers
to use this pattern.
v2:
- add patches for replacing ip_tunnel_get_stats64
Requested additional migrations will come in a separate series.
v3:
- add atomic_long_t member rx_frame_errors in patch 3 for making
counter updates atomic
====================
Link: https://lore.kernel.org/r/99273e2f-c218-cd19-916e-9161d8ad8c56@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
After having migrated all users remove ip_tunnel_get_stats64().
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Replace ip_tunnel_get_stats64() with the new identical core function
dev_get_tstats64().
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Replace ip_tunnel_get_stats64() with the new identical core function
dev_get_tstats64().
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Replace ip_tunnel_get_stats64() with the new identical core function
dev_get_tstats64().
Reviewed-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Replace ip_tunnel_get_stats64() with the new identical core function
dev_get_tstats64().
Acked-by: Harald Welte <laforge@gnumonks.org>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Replace ip_tunnel_get_stats64() with the new identical core function
dev_get_tstats64().
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Switch ip6_tunnel to the standard statistics pattern:
- use dev->stats for the less frequently accessed counters
- use dev->tstats for the frequently accessed counters
An additional benefit is that we now have 64bit statistics also on
32bit systems.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Switch tun to the standard statistics pattern:
- use netdev->stats for the less frequently accessed counters
- use netdev->tstats for the frequently accessed per-cpu counters
v3:
- add atomic_long_t member rx_frame_errors for making counter updates
atomic
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Use netdev->tstats instead of a member of dsa_slave_priv for storing
a pointer to the per-cpu counters. This allows us to use core
functionality for statistics handling.
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
It's a frequent pattern to use netdev->stats for the less frequently
accessed counters and per-cpu counters for the frequently accessed
counters (rx/tx bytes/packets). Add a default ndo_get_stats64()
implementation for this use case.
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add an additional control test that verifies:
-specifying two different max_num_members values fails
-specifying max_num_members > PACKET_FANOUT_MAX fails
In datapath tests, set max_num_members to PACKET_FANOUT_MAX.
Signed-off-by: Tanner Love <tannerlove@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
One use case of PACKET_FANOUT is lockless reception with one socket
per CPU. 256 is a practical limit on increasingly many machines.
Increase PACKET_FANOUT_MAX to 64K. Expand setsockopt PACKET_FANOUT to
take an extra argument max_num_members. Also explicitly define a
fanout_args struct, instead of implicitly casting to an integer. This
documents the API and simplifies the control flow.
If max_num_members is not specified or is set to 0, then 256 is used,
same as before.
Signed-off-by: Tanner Love <tannerlove@google.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When udp_memory_allocated is at the limit, __udp_enqueue_schedule_skb
will return a -ENOBUFS, and skb will be dropped in __udp_queue_rcv_skb
without any counters being done. It's hard to find out what happened
once this happen.
So we introduce a UDP_MIB_MEMERRORS to do this job. Well, this change
looks friendly to the existing users, such as netstat:
$ netstat -u -s
Udp:
0 packets received
639 packets to unknown port received.
158689 packet receive errors
180022 packets sent
RcvbufErrors: 20930
MemErrors: 137759
UdpLite:
IpExt:
InOctets: 257426235
OutOctets: 257460598
InNoECTPkts: 181177
v2:
- Fix some alignment problems
Signed-off-by: Menglong Dong <dong.menglong@zte.com.cn>
Link: https://lore.kernel.org/r/1604627354-43207-1-git-send-email-dong.menglong@zte.com.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Alex Elder says:
====================
net: ipa: constrain GSI interrupts
The goal of this series is to more tightly control when GSI
interrupts are enabled. This is a long-ish series, so I'll
describe it in parts.
The first patch is actually unrelated... I forgot to include
it in my previous series (which exposed the GSI layer to the
IPA version). It is a trivial comments-only update patch.
The second patch defers registering the GSI interrupt handler
until *after* all of the resources that handler touches have
been initialized. In practice, we don't see this interrupt
that early, but this precludes an obvious problem.
The next two patches are simple changes. The first just
trivially renames a field. The second switches from using
constant mask values to using an enumerated type of bit
positions to represent each GSI interrupt type.
The rest implement the "real work." First, all interrupts
are disabled at initialization time. Next, we keep track of
a bitmask of enabled GSI interrupt types, updating it each
time we enable or disable one of them. From there we have
a set of patches that one-by-one enable each interrupt type
only during the period it is required. This includes allowing
a channel to generate IEOB interrupts only when it has been
enabled. And finally, the last patch simplifies some code
now that all GSI interrupt types are handled uniformly.
====================
Link: https://lore.kernel.org/r/20201105181407.8006-1-elder@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Now that all of the GSI interrupts are handled uniformly,
change gsi_irq_type_update() so it takes a value. Have the
function assign that value to the cached mask of enabled GSI
IRQ types before writing it to hardware.
Note that gsi_irq_teardown() will only be called after
gsi_irq_disable(), so it's not necessary for the former
to disable all IRQ types. Get rid of that.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Most GSI general errors are unrecoverable without a full reset.
Despite that, we want to receive these errors so we can at least
report what happened before whatever undefined behavior ensues.
Explicitly disable all such interrupts in gsi_irq_setup(), then
enable those we want in gsi_irq_enable(). List the interrupt types
we are interested in (everything but breakpoint) explicitly rather
than using GSI_CNTXT_GSI_IRQ_ALL, and remove that symbol's
definition.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
It is possible for other execution environments (EEs, like the modem)
to request changes to local (AP) channel or event ring state. We do
not support this feature.
In gsi_irq_setup(), explicitly zero the mask that defines which
channels are permitted to generate inter-EE channel state change
interrupts. Do the same for the event ring mask.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
A GSI channel must be started in order to use it to perform a
transfer data (or command) transaction. And the only time we'll see
an IEOB interrupt is if we send a transaction to a started channel.
Therefore we do not need to have the IEOB interrupt type enabled
until at least one channel has been started. And once the last
started channel has been stopped, we can disable the IEOB interrupt
type again.
We already enable the IEOB interrupt for a particular channel only
when it is started. Extend that by having the IEOB interrupt *type*
be enabled only when at least one channel is in STARTED state.
Disallow all channels from triggering the IEOB interrupt in
gsi_irq_setup(). We only enable an channel's interrupt when
needed, so there is no longer any need to zero the channel mask
in gsi_irq_disable().
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The completion of a generic EE GSI command is signaled by a global
interrupt of type GP_INT1. The only other used type for a global
interrupt is a hardware error report.
First, disallow all global interrupt types in gsi_irq_setup(). We
want to know about hardware errors, so re-enable the interrupt type
in gsi_irq_enable(), to allow hardware errors to be reported.
Disable that interrupt type again in gsi_irq_disable().
We only issue generic EE commands one at a time, and there's no
reason to keep the completion interrupt enabled when no generic
EE command is pending. We furthermore have no need to enable the
GP_INT2 or GP_INT3 interrupt types (which aren't used).
The change in gsi_irq_enable() makes GSI_CNTXT_GLOB_IRQ_ALL unused,
so get rid of it. Have gsi_generic_command() enable the GP_INT1
interrupt type (in addition to the ERROR_INT type) only while a
generic command is pending.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
