This is one of only two files that initialize a semaphore to a negative
value. We don't really need the two semaphores here at all, but can do
the same thing in more conventional and more effient way, by using a
single waitqueue and an atomic thread counter.

This gets us a little bit closer to eliminating classic semaphores from
the kernel. It also fixes a corner case where we fail to continue after
one of the threads fails to start up.

An alternative would be to use a split kthread_create()+wake_up_process()
and completely eliminate the separate synchronization.
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

So far phy_error() silently stops the PHY state machine. If the network
driver doesn't inform about a  MDIO error then the user may wonder why
his network is down. Let's print the stack trace to facilitate search
for the root cause of the error.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add helpers phy_is_started() and __phy_is_started() to avoid open-coded
checks whether PHY has been started. To make the check easier move
PHY_HALTED before PHY_UP in enum phy_state. Further improvements:

phy_start_aneg():
Return -EBUSY and print warning if function is called from a non-started
state (DOWN, READY, HALTED). Better check because function is exported
and drivers may use it incorrectly.

phy_interrupt():
Return IRQ_NONE also if state is DOWN or READY. We should never receive
an interrupt in one of these states, but better play safe.

phy_stop():
Just return and print a warning if PHY is in a non-started state.
This warning should help to identify drivers with unbalanced calls to
phy_start() / phy_stop().

phy_state_machine():
Schedule state machine run only if PHY is in a started state.
E.g. if state is READY we don't need the state machine, it will be
started by phy_start().

v2:
- don't use __func__ within phy_warn_state
v3:
- use WARN() instead of printing error message to facilitate debugging
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There's a workaround to restart the phylib state machine in case of a
MDIO access timeout. Seems it was introduced to deal with the
consequences of a too small MDIO timeout. See also commit message of
c3b084c2 ("net: fec: Adjust ENET MDIO timeouts") which increased
the timeout value later. Due to the later timeout value fix it seems
to be safe to remove the workaround.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There are two statements that are indented too much by one space each,
fix these.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Peng Li says:

====================
net: hns3: code optimizations & bugfixes for HNS3 driver

This patchset includes bugfixes and code optimizations for the HNS3
ethernet controller driver
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

When caculating the SSU buffer, it first allocate tx and
rx private buffer, then the remaining buffer is for rx
shared buffer. The remaining buffer size should be at
least bigger than or equal to the shared_std, which is the
minimum shared buffer size required by the driver, but
currently if the remaining buffer size is equal to the
shared_std, it returns failure, which causes SSU buffer
allocation failure problem.

This patch fixes this problem by rounding up shared_std before
checking the the remaining buffer size bigger than or equal to
the shared_std.

Fixes: 46a3df9f ("net: hns3: Add HNS3 Acceleration Engine & Compatibility Layer Support")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The hardware expects the buffer size set to SSU is aligned to
256 bytes, this patch aligns the buffer size to 256 byte using
roundup or rounddown function.
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds support of getting tx and dv buffer size through
firmware, because different version of hardware requires different
size of tx and dv buffer.

This patch also add dv_buf_size to tc' private buffer size even if
pfc is not enable for the tc.
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Driver calls phy_connect_direct and registers hclge_mac_adjust_link
to synchronize mac speed and duplex from phy. It is better to
synchronize mac speed and duplex from phy when phy link up.
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Our phy does not support 1000M/half, this patch removes 1000M/half from
PHY_SUPPORTED_FEATURES.
Signed-off-by: Fuyun Liang <liangfuyun1@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

coalesce param updates every 100 napi times, it may update a little
late if ping test after a high rate flow, may over napi poll is called
100 times as ping test sends packets every second.

This patch updates coalesce param every second, instead with every
100 napi times. It can not update the param 100% in time, but the
lag time is very short.
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In the hns3_nic_uninit_vector_data(), the procedure of uninitializing
the tqp_vector's IRQ has not set affinity_notify to NULL and changes
its init flag. This patch fixes it. And for simplificaton, local
variable tqp_vector is used instead of priv->tqp_vector[i].

Fixes: 424eb834 ("net: hns3: Unified HNS3 {VF|PF} Ethernet Driver for hip08 SoC")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When doing reset, it is unnecessary to get the hardware's default
configuration again, otherwise, the user's configuration will be
overwritten.

Fixes: 4ed340ab ("net: hns3: Add reset process in hclge_main")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When hclge_reset() completes successfully, it should update the
last_reset_time, set reset_fail_cnt to 0, and set reset_type of
hnae3_ae_dev to HNAE3_NONE_RESET.

Also when hclgevf_reset() completes successfully, it should update
the last_reset_time, and set reset_type of hnae3_ae_dev to
HNAE3_NONE_RESET.
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

While doing DOWN, the calling of napi_disable() may not return, since the
napi_complete() in the hns3_nic_common_poll() will never be called when
HNS3_NIC_STATE_DOWN is set. So we need to call napi_complete() before
checking HNS3_NIC_STETE_DOWN.

Fixes: ff0699e0 ("net: hns3: stop napi polling when HNS3_NIC_STATE_DOWN is set")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In the hclgevf_pci_reset(), it only uninitialize and initialize
the msi, so if the initialization fails, hclgevf_uninit_hdev()
does not need to uninitialize the msi, but needs to uninitialize
the pci, otherwise it will cause pci resource not free.

Fixes: 862d969a ("net: hns3: do VF's pci re-initialization while PF doing FLR")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When hns3_get_vector_ring_chain() failed in the
hns3_nic_init_vector_data(), it should do the error handling instead
of return directly.

Also, cur_chain should be freed instead of chain and head->next should
be set to NULL in error handling of hns3_get_vector_ring_chain.

This patch fixes them.

Fixes: 73b907a0 ("net: hns3: bugfix for buffer not free problem during resetting")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Steffen Klassert says:

====================
pull request (net-next): ipsec-next 2018-12-18

1) Add xfrm policy selftest scripts.
   From Florian Westphal.

2) Split inexact policies into four different search list
   classes and use the rbtree infrastructure to store/lookup
   the policies. This is to improve the policy lookup
   performance after the flowcache removal.
   Patches from Florian Westphal.

3) Various coding style fixes, from Colin Ian King.

4) Fix policy lookup logic after adding the inexact policy
   search tree infrastructure. From Florian Westphal.

5) Remove a useless remove BUG_ON from xfrm6_dst_ifdown.
   From Li RongQing.

6) Use the correct policy direction for lookups on hash
   rebuilding. From Florian Westphal.

Please pull or let me know if there are problems.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Previously the identifier used for indirect block callback registry and
for block rule cb registry (when done via indirect blocks) was the pointer
to the netdev we were interested in receiving updates on. This worked fine
if a single app existed that registered one callback per netdev of
interest. However, if multiple cards are in place and, in turn, multiple
apps, then each app may register the same callback with the same
identifier to both the netdev's indirect block cb list and to a block's cb
list. This can lead to EEXIST errors and/or incorrect cb deletions.

Prevent this conflict by using the app pointer as the identifier for
netdev indirect block cb registry, allowing each app to register a unique
callback per netdev. For block cb registry, the same app may register
multiple cbs to the same block if using TC shared blocks. Instead of the
app, use the pointer to the allocated cb_priv data as the identifier here.
This means that there can be a unique block callback for each app/netdev
combo.

Fixes: 3166dd07 ("nfp: flower: offload tunnel decap rules via indirect TC blocks")
Reported-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This new firmware contains:
 * New packet traps for discarded packets
 * Secure firmware flash bug fix
 * Fence mechanism bug fix
 * TCAM RMA bug fix
Signed-off-by: Shalom Toledo <shalomt@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mfc6_cache is not needed by ip6mr_forward2 so drop it from the input
argument list.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mfc_cache is not needed by ipmr_queue_xmit so drop it from the input
argument list.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit d9fbc7f6 "net: tcp: prefer listeners bound to an address"
removes port-only listener lookups. This caused segfaults in DCCP
lookups because DCCP did not initialize the (addr,port) hashtable.

This patch adds said initialization.

The only non-trivial issue here is the size of the new hashtable.
It seemed reasonable to make it match the size of the port-only
hashtable (= INET_LHTABLE_SIZE) that was used previously. Other
parameters to inet_hashinfo2_init() match those used in TCP.

V2 changes: marked inet_hashinfo2_init as an exported symbol
so that DCCP compiles when configured as a module.

Tested: syzcaller issues fixed; the second patch in the patchset
        tests that DCCP lookups work correctly.

Fixes: d9fbc7f6 "net: tcp: prefer listeners bound to an address"
Reported-by: syzcaller <syzkaller@googlegroups.com>
Signed-off-by: Peter Oskolkov <posk@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Michael Chan says:

====================
bnxt_en: Updates for net-next.

Two main changes in this seris plus some miscellaneous changes.

1. Improvements and fixes for resource accounting which are required
for enabling SR-IOV and RDMA on the new 57500 chips.  Only SR-IOV
for 57500 chips is enabled in this series.

2. New statistics counters and improvements to keep the basic
counters and port counters during IFDOWN.

3. Msic. small changes for ETS, returning proper error codes
when flashing NVRAM, and a link speed related fix for ethtool
loopback selftest.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

With autoneg enabled, PHY loopback test fails. To disable autoneg,
driver needs to send a valid forced speed to FW. FW is not sending
async event for invalid speeds. To fix this, query forced speeds
and send the correct speed when disabling autoneg mode.
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Port statistics which include RDMA counters are useful even when the
netdevice is down.  Do not free the port statistics DMA buffers
when the netdevice is down.  This is keep the snapshot of the port
statistics and counters will just continue counting when the
netdevice goes back up.

Split the bnxt_free_stats() function into 2 functions.  The port
statistics buffers will only be freed when the netdevice is
removed.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

With the current driver, the statistics reported by .ndo_get_stats64()
are reset when the device goes down.  Store a snapshot of the
rtnl_link_stats64 before shutdown.  This snapshot is added to the
current counters in .ndo_get_stats64() so that the counters will not
get reset when the device is down.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently firmware specific errors are returned directly in flash_device
and reset ethtool hooks. Modify it to return linux standard errors
to userspace when flashing operations fail.
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently, the code allows ETS bandwidth weight 0 to be set on unused TCs.
We should not set any DCB parameters on unused TCs at all.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Display the CoS counters as additional priority counters by looking up
the priority to CoS queue mapping.  If the TX extended port statistics
block size returned by firmware is big enough to cover the CoS counters,
then we will display the new priority counters.  We call firmware to get
the up-to-date pri2cos mapping to convert the CoS counters to
priority counters.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There are some minor differences when assigning VF resources on the
new chips.  The MSIX (NQ) resource has to be assigned and ring group
is not needed on the new chips.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When bringing up a device, the code checks to see if the number of
MSIX has changed.  pci_disable_msix() should be called first before
changing the number of reserved NQs/CMPL rings.  This ensures that
the MSIX vectors associated with the NQs/CMPL rings are still
properly mapped when pci_disable_msix() masks the vectors.

This patch will prevent errors when RDMA support is added for the new
57500 chips.  When the RDMA driver shuts down, the number of NQs is
decreased and we must use the new sequence to prevent MSIX errors.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnxt_en requires same number of stat_ctxs as CP rings but RDMA
requires only 1 stat_ctx.  Also add a new parameter resv_stat_ctxs
to better keep track of stat_ctxs reserved including resources used
by RDMA.  Add a stat_ctxs parameter to all the relevant resource
reservation functions so we can reserve the correct number of
stat_ctxs.

Prior to this patch, we were not reserving the extra stat_ctx for
RDMA and RDMA would not work on the new 57500 chips.
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Calling bnxt_set_max_func_stat_ctxs() to modify max stat_ctxs requested
or freed by the RDMA driver is wrong. After introducing reservation of
resources recently, the driver has to keep track of all stat_ctxs
including the ones used by the RDMA driver.  This will provide a better
foundation for accurate accounting of the stat_ctxs.
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

For bnxt_en driver, stat_ctxs created will always be same as
cp_nr_rings. Remove extra variable that duplicates the value.
Also introduce bnxt_get_avail_stat_ctxs_for_en() helper to get
available stat_ctxs and bnxt_get_ulp_stat_ctxs() helper to return
number of stat_ctxs used by RDMA.
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The available CP rings are calculated differently on the new 57500
chips, so add this helper to do this calculation correctly.  The
VFs will be assigned these available CP rings.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The PF has a pool of NQs and MSIX vectors assigned to it based on
NVRAM configurations.  The number of usable MSIX vectors on the PF
is the minimum of the NQs and MSIX vectors.  Any excess NQs without
associated MSIX may be used for the VFs, so we need to store this
max_nqs value.  max_nqs minus the NQs used by the PF will be the
available NQs for the VFs.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Handling exceptions for direct UDP encapsulation in GUE (that is,
UDP-in-UDP) leads to unbounded recursion in the GUE exception handler,
syzbot reported.

While draft-ietf-intarea-gue-06 doesn't explicitly forbid direct
encapsulation of UDP in GUE, it probably doesn't make sense to set up GUE
this way, and it's currently not even possible to configure this.

Skip exception handling if the GUE proto/ctype field is set to the UDP
protocol number. Should we need to handle exceptions for UDP-in-GUE one
day, we might need to either explicitly set a bound for recursion, or
implement a special iterative handling for these cases.

Reported-and-tested-by: syzbot+43f6755d1c2e62743468@syzkaller.appspotmail.com
Fixes: b8a51b38 ("fou, fou6: ICMP error handlers for FoU and GUE")
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This allows to control carrier from /sys/class/net/ethX/carrier
for Fixed PHYs.
Signed-off-by: Joakim Tjernlund <joakim.tjernlund@infinera.com>
Signed-off-by: David S. Miller <davem@davemloft.net>