- 02 Mar, 2019 7 commits
-
-
David S. Miller authored
Igor Russkikh says: ==================== net: aquantia: minor bug fixes after static analysis This patchset fixes minor errors and warnings found by smatch and kasan. Extra patch is to replace AQ_HW_WAIT_FOR with readx_poll_timeout to improve readability. V2: use readx_poll resubmitted to net-next since the changeset became quite big. ==================== Signed-off-by:David S. Miller <davem@davemloft.net>
-
Nikita Danilov authored
Replace some direct registers reads with better online functions. Signed-off-by:
Nikita Danilov <nikita.danilov@aquantia.com> Signed-off-by:
Igor Russkikh <igor.russkikh@aquantia.com> Signed-off-by:
-
Nikita Danilov authored
David noticed the original define was hiding 'err' variable reference. Thats confusing and counterintuitive. Andrew noted the whole macro could be replaced with standard readx_poll kernel macro. This makes code more readable. Signed-off-by:
-
Igor Russkikh authored
This is a real stack undercorruption found by kasan build. The issue did no harm normally because it only overflowed 2 bytes after `bitary` array which on most architectures were mapped into `err` local. Fixes: bab6de8f ("net: ethernet: aquantia: Atlantic A0 and B0 specific functions.") Signed-off-by:
-
Nikita Danilov authored
The overflow is detected by smatch: drivers/net/ethernet/aquantia/atlantic/aq_pci_func.c: 175 aq_pci_func_free_irqs() error: buffer overflow 'self->aq_vec' 8 <= 31 In reality msix_entry_mask always restricts number of iterations. Adding extra condition to make logic clear and smatch happy. Signed-off-by:
-
Nikita Danilov authored
drivers/net/ethernet/aquantia/atlantic/aq_nic.c: 991:1: warning: no newline at end of file Signed-off-by:
-
Nikita Danilov authored
Not careful array dereference caused analysis tools to think there could be memory overflow. There was actually no corruption because the array is two dimensional. drivers/net/ethernet/aquantia/atlantic/aq_ethtool.c: 140 aq_ethtool_get_strings() error: memcpy() '*aq_ethtool_stat_names' too small (32 vs 704) Signed-off-by:
-
- 01 Mar, 2019 12 commits
-
-
David S. Miller authored
Jakub Kicinski says: ==================== nfp: control processor DMA support and RJ45 This series starts with adding support for reporting twisted pair media type in ethtool. Remaining patches add support for using DMA with the control/service processor. Currently we always copy the command data into card's memory. DMA support allows us to have the NSP read the data from host memory by itself. Unfortunately, the FW loading and flashing cannot directly map the buffers for DMA because (a) the firmware ABI returns const buffers, and (b) the buffers may be vmalloc()ed in many mysterious/unmappable way. So just bite the bullet - allocate new host buffer for the command and copy. As Dirk explains, the NSP now supports updating all FWs at once which means the max flashing time grew significantly. He bumps the max wait to avoid timeouts. ==================== Signed-off-by: -
Dirk van der Merwe authored
The management firmware now supports being passed a bundle with multiple components to be stored in flash at once. This makes it easier to update all components to a known state with a single user command, however, this also has the potential to increase the time required to perform the update significantly. The management firmware only updates the components out of a bundle which are outdated, however, we need to make sure we can handle the absolute worst case where a CPLD update can take a long time to perform. We set a very conservative total timeout of 900s which already adds a contingency. Signed-off-by:
Dirk van der Merwe <dirk.vandermerwe@netronome.com> Reviewed-by:
Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by:
-
Jakub Kicinski authored
Newer versions of NSP can access host memory. Simplest access type requires all data to be in one contiguous area. Since we don't have the guarantee on where callers of the NSP ABI will allocate their buffers we allocate a bounce buffer and copy the data in and out. Signed-off-by:
-
Jakub Kicinski authored
DMA version of NSP communication is coming, move the code which copies data into the NFP buffer into a separate function. Signed-off-by:
-
Jakub Kicinski authored
NSP expresses the buffer size in MB and 4 kB blocks. For small buffers the kB part may make a difference, so count it in. Signed-off-by:
-
Jakub Kicinski authored
Add support for reporting twisted pair port type. Signed-off-by:
-
Marek Behún authored
The comphy driver for Armada 3700 by Miquèl Raynal (which is currently in linux-next) does not actually set comphy mode when phy_set_mode_ext is called. The mode is set at next call of phy_power_on. Update the driver to semantics similar to mvpp2: helper mvneta_comphy_init sets comphy mode and powers it on. When mode is to be changed in mvneta_mac_config, first power the comphy off, then call mvneta_comphy_init (which sets the mode to new one). Only do this when new mode is different from old mode. This should also work for Armada 38x, since in that comphy driver methods power_on and power_off are unimplemented. Signed-off-by:
Marek Behún <marek.behun@nic.cz> Signed-off-by:
-
David S. Miller authored
Claudiu Manoil says: ==================== enetc: Add mdio support and device tree nodes This is the missing part to enable PCI probing of the ENETC ethernet ports on the LS1028A SoC and external traffic on the LS1028A RDB board. It's one of the first items on the TODO list for the recently merged ENETC ethernet driver. v3: Add DT bindings doc for ENETC connections v4: none ==================== Signed-off-by: -
Claudiu Manoil authored
Define connection bindings (external PHY connections and internal links) for the ENETC on-chip ethernet controllers. Signed-off-by:
Claudiu Manoil <claudiu.manoil@nxp.com> Reviewed-by:
Rob Herring <robh@kernel.org> Signed-off-by:
-
Claudiu Manoil authored
Each ENETC PF has its own MDIO interface, the corresponding MDIO registers are mapped in the ENETC's Port register block. The current patch adds a driver for these PF level MDIO buses, so that each PF can manage directly its own external link. Signed-off-by:
Alex Marginean <alexandru.marginean@nxp.com> Signed-off-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
Claudiu Manoil authored
The LS1028A RDB board features an Atheros PHY connected over SGMII to the ENETC PF0 (or Port0). ENETC Port1 (PF1) has no external connection on this board, so it can be disabled for now. Signed-off-by:
-
Claudiu Manoil authored
The LS1028A SoC features a PCI Integrated Endpoint Root Complex (IERC) defining several integrated PCI devices, including the ENETC ethernet controller integrated endpoints (IEPs). The IERC implements ECAM (Enhanced Configuration Access Mechanism) to provide access to the PCIe config space of the IEPs. This means the the IEPs (including ENETC) do not support the standard PCIe BARs, instead the Enhanced Allocation (EA) capability structures in the ECAM space are used to fix the base addresses in the system, and the PCI subsystem uses these structures for device enumeration and discovery. The "ranges" entries contain basic information from these EA capabily structures required by the kernel for device enumeration. The current patch also enables the first 2 ENETC PFs (Physiscal Functions) and the associated VFs (Virtual Functions), 2 VFs for each PF. Each of these ENETC PFs has an external ethernet port on the LS1028A SoC. Signed-off-by:
-
- 28 Feb, 2019 21 commits
-
-
David Ahern authored
'ip' can switch network namespaces internally and then run a given command relative to that namespace without the need to fork and exec another ip instance. Update all references of the form: ip netns exec "$testns" ip ... to ip -netns "$testns" ... Signed-off-by:David Ahern <dsahern@gmail.com> Signed-off-by:
-
David S. Miller authored
Julian Wiedmann says: ==================== s390/qeth: updates 2019-02-28 please apply one more qeth patch series for net-next. This eliminates some of the quirks in our reset code, and slims down the internal state machine. ==================== Signed-off-by: -
Julian Wiedmann authored
Now that qeth always uses dev_close() to shutdown the interface, we can trust the locking and remove some custom state checks. qeth_l?_stop_card() is no longer called for a card in UP state, so remove the checks there too. This basically makes the UP state obsolete, so rip out the whole thing (except for the sysfs-visible string). Signed-off-by:
Julian Wiedmann <jwi@linux.ibm.com> Signed-off-by:
-
Julian Wiedmann authored
It makes no difference whether we 1. manually disarm the HW trap and call the offline code with recovery_mode == 1, or 2. call the offline code with recovery_mode == 0, and let it disarm the HW trap for us. So consolidate the two code paths in the suspend callback. Signed-off-by:
-
Julian Wiedmann authored
The qeth-wide workqueue is now only used by a single caller to schedule close_dev work. Just put it on a system queue instead. Signed-off-by:
-
Julian Wiedmann authored
The recovery code already runs in a kthread, we don't have to defer the offlining further. Signed-off-by:
-
Julian Wiedmann authored
smatch complains that __qeth_l3_set_offline() first accesses card->dev, and then later checks whether the pointer is valid. Since commit d3d1b205 ("s390/qeth: allocate netdevice early"), the pointer is _always_ valid - that patch merely missed to remove this one check. Signed-off-by:
-
Julian Wiedmann authored
When resetting an interface ("recovery"), qeth currently attempts to elide the call to dev_close(). We initially only call .ndo_close to quiesce the data path, and then offline & online the ccwgroup device. If the reset succeeded, a call to .ndo_open then resumes the data path along with some internal setup (dev_addr validation, RX modeset) that dev_open() would have usually triggered. dev_close() only gets called (via the close_dev worker) if the reset action fails. It's unclear whether this was initially done due to locking concerns, or rather to execute the reset transparently. Either way, temporarily closing the interface without dev_close() is fragile, and means we're susceptible to various races and unexpected behaviour. For instance: - Bypassing dev_deactivate_many() means that the qdiscs are not set to __QDISC_STATE_DEACTIVATED. Consequently any intermittent TX completion can wake up the txq, resulting in calls to .ndo_start_xmit while the data path is down. We have custom state checking to detect this case and drop such packets. - Because the IFF_UP flag doesn't reflect the interface's actual state during a reset, we have custom state checking in .ndo_open and .ndo_close to guard against invalid calls. - Considering that the reset might take a considerable amount of time (in particular if an IO fails and we end up waiting for its timeout), we _do_ want NETDEV_GOING_DOWN and NETDEV_DOWN events so that components like bonding, team, bridge, macvlan, vlan, ... can take appropriate action. Signed-off-by: -
Julian Wiedmann authored
In its attempt to run only the minimal amount of tear down steps, qeth_l2_stop_card() fails to reset the "is dev_addr registered?" flag in some rare scenarios. But a future change to the tear down sequence would cause us to _always_ hit this issue, so patch it up before that code lands. Fix it by unconditionally clearing the flag bit. This also allows us to remove the additional cleanup step in qeth_dev_layer2_store(). Signed-off-by:
-
Julian Wiedmann authored
When setting a L2 qeth device online, enable the HW trap as soon as the control plane is available. This allows us to catch any error that occurs during the very first commands. In the same spirit, the offline code should disable the HW trap as the very first step of its processing. Signed-off-by:
-
Julian Wiedmann authored
The offline code uses a specific RECOVER state to indicate that the interface should be brought up when a qeth device is set online again. Rather than having a specific card-state for this, just put it in an internal flag bit and set the state to DOWN. When working with the card's state transitions, this reduces the complexity quite a bit. Signed-off-by:
-
Ursula Braun authored
Without hardware pnetid support there must currently be a pnet table configured to determine the IB device port to be used for SMC RDMA traffic. This patch enables a setup without pnet table, if the used handshake interface belongs already to a RoCE port. Signed-off-by:
Ursula Braun <ubraun@linux.ibm.com> Signed-off-by:
-
Leslie Monis authored
As per RFC 8033, it is sufficient for the drop probability decay factor to have a value of (1 - 1/64) instead of 98%. This avoids the need to do slow division. Suggested-by:
David Laight <David.Laight@aculab.com> Signed-off-by:
Leslie Monis <lesliemonis@gmail.com> Signed-off-by:
-
Arjun Vynipadath authored
If we are not able to reach firmware, enter debugging mode that will help us to get adapter logs. Signed-off-by:
Arjun Vynipadath <arjun@chelsio.com> Signed-off-by:
Vishal Kulkarni <vishal@chelsio.com> Signed-off-by:
-
Arjun Vynipadath authored
T6 adapters support outer UDP checksum offload for encapsulated packets, hence enabling netdev feature flag NETIF_F_GSO_UDP_TUNNEL_CSUM. Signed-off-by:
-
Arjun Vynipadath authored
GRO is done by cxgb4/cxgb4vf. Hence set NETIF_F_GRO flag for both cxgb4/cxgb4vf. Cleaned up VLAN netdev features in cxgb4vf. Also fixed NETIF_F_HIGHDMA being set unconditionally for vlan netdev features. Signed-off-by:
-
Eli Britstein authored
The csum calculation is different for IPv4/6. For VLAN packets, tc_skb_protocol returns the VLAN protocol rather than the packet's one (e.g. IPv4/6), so csum is not calculated. Furthermore, VLAN may not be stripped so csum is not calculated in this case too. Calculate the csum for those cases. Fixes: d8b9605d ("net: sched: fix skb->protocol use in case of accelerated vlan path") Signed-off-by:
Eli Britstein <elibr@mellanox.com> Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Signed-off-by:
-
Gustavo A. R. Silva authored
One of the more common cases of allocation size calculations is finding the size of a structure that has a zero-sized array at the end, along with memory for some number of elements for that array. For example: struct foo { int stuff; struct boo entry[]; }; instance = devm_kzalloc(dev, sizeof(struct foo) + sizeof(struct boo) * count, GFP_KERNEL); Instead of leaving these open-coded and prone to type mistakes, we can now use the new struct_size() helper: instance = devm_kzalloc(dev, struct_size(instance, entry, count), GFP_KERNEL); This code was detected with the help of Coccinelle. Signed-off-by:Gustavo A. R. Silva <gustavo@embeddedor.com> Signed-off-by:
-
David S. Miller authored
Maxime Chevallier says: ==================== net: phy: marvell10g: Clean .get_features by using C45 helpers Recent work on C45 helpers by Heiner made the genphy_c45_pma_read_abilities function generic enough to use as a default .get_featutes implementation. This series removes the remaining redundant code in mv3310_get_features(), and makes the 2110 PHY use genphy_c45_pma_read_abilities() directly, since it doesn't have the issue with the wrong abilities being reported. ==================== Signed-off-by: -
Maxime Chevallier authored
Contrary to the 3310, the 2110 PHY correctly reports it's 2.5G/5G abilities. We can therefore use the genphy_c45_pma_read_abilities helper to build the list of features. Signed-off-by:
Maxime Chevallier <maxime.chevallier@bootlin.com> Signed-off-by:
-
Maxime Chevallier authored
The genphy_c45_pma_read_abilities helper now sets the Autoneg ability in phydev->supported according to what the AN MMD reports. We therefore don't need to manually do that in mv3310_get_features(). Signed-off-by:
Heiner Kallweit <hkallweit1@gmail.com> Signed-off-by:
-
