- 21 Jun, 2020 4 commits
-
-
Andrea Mayer authored
During the initialization phase of the VRF module, the callback for table to VRF device lookup is registered in l3mdev. Signed-off-by:
Andrea Mayer <andrea.mayer@uniroma2.it> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Andrea Mayer authored
Add net.vrf.strict_mode sysctl parameter. When net.vrf.strict_mode=0 (default) it is possible to associate multiple VRF devices to the same table. Conversely, when net.vrf.strict_mode=1 a table can be associated to a single VRF device. When switching from net.vrf.strict_mode=0 to net.vrf.strict_mode=1, a check is performed to verify that all tables have at most one VRF associated, otherwise the switch is not allowed. The net.vrf.strict_mode parameter is per network namespace. Signed-off-by:
-
Andrea Mayer authored
Add the data structures and the processing logic to keep track of the associations between VRF devices and routing tables. When a VRF is instantiated, it needs to refer to a given routing table. For each table, we explicitly keep track of the existing VRFs that refer to the table. Signed-off-by:
-
Andrea Mayer authored
Add infrastructure to l3mdev (the core code for Layer 3 master devices) in order to find out the corresponding VRF device for a given table id. Therefore, the l3mdev implementations: - can register a callback that returns the device index of the l3mdev associated with a given table id; - can offer the lookup function (table to VRF device). Signed-off-by:
-
- 20 Jun, 2020 13 commits
-
-
Gustavo A. R. Silva authored
Make use of the struct_size() helper instead of an open-coded version in order to avoid any potential type mistakes. This code was detected with the help of Coccinelle and, audited and fixed manually. Signed-off-by:
Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by:
-
Gustavo A. R. Silva authored
Make use of the struct_size() helper instead of an open-coded version in order to avoid any potential type mistakes. Also, remove unnecessary variable _size_. This code was detected with the help of Coccinelle and, audited and fixed manually. Signed-off-by:
-
David S. Miller authored
Russell King says: ==================== Clause 45 PHY probing improvements Last time this series was posted back in May, Florian reviewed the patches, which was the only feedback I received. I'm now posting them without the RFC tag. This series aims to improve the probing for Clause 45 PHYs. The first four patches clean up get_phy_device() and called functions, updating the kernel doc, adding information about the various error return values. We then provide better kerneldoc for get_phy_device(), describing what is going on, and more importantly what the various return codes mean. Patch 6 adds support for probing MMDs >= 8 to check for their presence. Patch 7 changes get_phy_c45_ids() to only set the returned devices_in_package if we successfully find a PHY. Patch 8 splits the use of "devices in package" from the "mmds present". Patch 9 expands our ID reading to cover the other MMDs. ==================== Signed-off-by: -
Russell King authored
Expand the device_ids[] array to allow all MMD IDs to be read rather than just the first 8 MMDs, but only read the ID if the MDIO_STAT2 register reports that a device really is present here for these new devices to maintain compatibility with our current behaviour. Note that only a limited number of devices have MDIO_STAT2. 88X3310 PHY vendor MMDs do are marked as present in the devices_in_package, but do not contain IEE 802.3 compatible register sets in their lower space. This avoids reading incorrect values as MMD identifiers. Reviewed-by:
Florian Fainelli <f.fainelli@gmail.com> Signed-off-by:
Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by:
-
Russell King authored
We have two competing requirements for the devices_in_package field. We want to use it as a bit array indicating which MMDs are present, but we also want to know if the Clause 22 registers are present. Since "devices in package" is a term used in the 802.3 specification, keep this as the as-specified values read from the PHY, and introduce a new member "mmds_present" to indicate which MMDs are actually present in the PHY, derived from the "devices in package" value. Reviewed-by:
-
Russell King authored
Only set the devices_in_package to a non-zero value if we find a valid value for this field, so we avoid leaving it set to e.g. 0x1fffffff. Reviewed-by:
-
Russell King authored
Add support for probing MMDs above 7 for a valid devices-in-package specifier, but only probe the vendor MMDs for this if they also report that there the device is present in status register 2. This avoids issues where the MMD is implemented, but does not provide IEEE 802.3 compliant registers (such as the MV88X3310 PHY.) Reviewed-by:
-
Russell King authored
Reword the get_phy_device() kerneldoc to be more explicit about how we probe for the PHY, and what the various return conditions signify. Reviewed-by:
-
Russell King authored
Move the ID check from get_phy_device() into get_phy_c22_id(), which simplifies get_phy_device(). The ID reading functions are now responsible for indicating whether they found a PHY or not via their return code - they must return -ENODEV when a PHY is not present. Reviewed-by:
-
Russell King authored
When we decide that a PHY is not present, we do not need to go through the hoops of setting *phy_id to 0xffffffff, and then return zero to make get_phy_device() fail - we can return -ENODEV which will have the same effect. Doing so means we no longer have to pass a pointer to phy_id in, and we can then clean up the clause 22 path in a similar way. Reviewed-by:
-
Russell King authored
Rearrange the code to read the PHY IDs, so we don't call get_phy_id() only to immediately call get_phy_c45_ids(). Move that logic into get_phy_device(), which results in better readability. Reviewed-by:
-
Russell King authored
Move the Cortina PHY workaround out of the "devices in package" loop; it doesn't need to be in there as the control flow will terminate the loop once we enter the workaround irrespective of the workaround's outcome. The workaround is triggered by the ID being mostly 1's, which will in any case terminate the loop. Signed-off-by:
-
Gustavo A. R. Silva authored
Make use of the struct_size() helper instead of an open-coded version in order to avoid any potential type mistakes. This code was detected with the help of Coccinelle and, audited and fixed manually. Signed-off-by:
-
- 19 Jun, 2020 23 commits
-
-
Gustavo A. R. Silva authored
Make use of the struct_size() helper instead of an open-coded version in order to avoid any potential type mistakes. This code was detected with the help of Coccinelle and, audited and fixed manually. Addresses-KSPP-ID: https://github.com/KSPP/linux/issues/83Signed-off-by:
Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by:
-
Gustavo A. R. Silva authored
Make use of the struct_size() helper instead of an open-coded version in order to avoid any potential type mistakes. This code was detected with the help of Coccinelle and, audited and fixed manually. Addresses-KSPP-ID: https://github.com/KSPP/linux/issues/83Signed-off-by:
-
Gustavo A. R. Silva authored
Make use of the struct_size() helper instead of an open-coded version in order to avoid any potential type mistakes. Also, remove unnecessary variable _size_. This code was detected with the help of Coccinelle and, audited and fixed manually. Addresses-KSPP-ID: https://github.com/KSPP/linux/issues/83Signed-off-by:
-
Eric Dumazet authored
If IPv6 is builtin, we do not need an expensive indirect call to reach icmp6_send(). v2: put inline keyword before the type to avoid sparse warnings. Signed-off-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
-
David S. Miller authored
Vishal Kulkarni says: ==================== cxgb4: add support for ethtool n-tuple filters Patch 1: Adds data structure to maintain list of filters and handles init/dinit of the same. Patch 2: Handles addition of filters via ETHTOOL_SRXCLSRLINS. Patch 3: Handles deletion of filtes via ETHTOOL_SRXCLSRLDEL. Patch 4: Handles viewing of added filters. Patch 5: Adds FLOW_ACTION_QUEUE support. ==================== Signed-off-by: -
Vishal Kulkarni authored
Add support for queue action to steer Rx traffic hitting the flows to specified Rxq. Signed-off-by:
Rahul Lakkireddy <rahul.lakkireddy@chelsio.com> Signed-off-by:
Vishal Kulkarni <vishal@chelsio.com> Signed-off-by:
-
Vishal Kulkarni authored
Add support to fetch the requested ethtool n-tuple filters by translating them from hardware spec to ethtool n-tuple spec. Signed-off-by:
-
Vishal Kulkarni authored
Add support to delete ethtool n-tuple filter. Fetch the appropriate filter region (HPFILTER, HASH, NORMAL) in which the filter exists, and delete it from the respective region, accordingly. Signed-off-by:
-
Vishal Kulkarni authored
Add support to parse and insert ethtool n-tuple filters. Translate n-tuple spec to flow spec and use the existing tc-flower offload infra to insert ethtool n-tuple filters. Signed-off-by:
-
Vishal Kulkarni authored
Allocate and manage resources required for ethtool n-tuple filters. Also fetch the HASH filter region size and calculate nhash entries. Signed-off-by:
-
Po Liu authored
This patch adds a drop frames counter to tc flower offloading. Reporting h/w dropped frames is necessary for some actions. Some actions like police action and the coming introduced stream gate action would produce dropped frames which is necessary for user. Status update shows how many filtered packets increasing and how many dropped in those packets. v2: Changes - Update commit comments suggest by Jiri Pirko. Signed-off-by:
Po Liu <Po.Liu@nxp.com> Reviewed-by:
Simon Horman <simon.horman@netronome.com> Reviewed-by:
Vlad Buslov <vladbu@mellanox.com> Signed-off-by:
-
David S. Miller authored
Vishal Kulkarni says: ==================== cxgb4: add support to read/write flash This series of patches adds support to read/write different binary images of serial flash present in Chelsio terminator. V2 changes: Patch 1: No change Patch 2: No change Patch 3: Fix 4 compilation warnings reported by C=1, W=1 flags Patch 4: No change Patch 5: No change ==================== Signed-off-by: -
Vishal Kulkarni authored
This patch adds support to dump flash memory via ethtool --get-dump Signed-off-by:
-
Vishal Kulkarni authored
Update set_flash to flash boot cfg image to flash region Signed-off-by:
-
Vishal Kulkarni authored
Update set_flash to flash boot image to flash region Signed-off-by:
-
Vishal Kulkarni authored
Update set_flash to flash PHY image to flash region Signed-off-by:
-
Vishal Kulkarni authored
Chelsio adapter contains different flash regions and each region is used by different binary files. This patch adds support to flash images like PHY firmware, boot and boot config using ethtool -f N. The N value mapping is as follows. N = 0 : Parse image and decide which region to flash N = 1 : Firmware N = 2 : PHY firmware N = 3 : boot image N = 4 : boot cfg Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>" Signed-off-by: -
David S. Miller authored
Eric Dumazet says: ==================== net: tso: expand to UDP support With QUIC getting more attention these days, it is worth implementing UDP direct segmentation, the same we did for TCP. Drivers will need to advertize NETIF_F_GSO_UDP_L4 so that GSO stack does not do the (more expensive) segmentation. Note the two first patches are stable candidates, after tests confirm they do not add regressions. v2: addressed Jakub feedback : 1) Added a prep patch for octeontx2-af 2) calls tso_start() earlier in otx2_sq_append_tso() ==================== Signed-off-by: -
Eric Dumazet authored
Note that like TCP, we do not support additional encapsulations, and that checksums must be offloaded to the NIC. Signed-off-by:
-
Eric Dumazet authored
Add tlen field into struct tso_t, and change tso_start() to return skb_transport_offset(skb) + tso->tlen This removes from callers the need to use tcp_hdrlen(skb) and will ease UDP segmentation offload addition. v2: calls tso_start() earlier in otx2_sq_append_tso() [Jakub] Signed-off-by:
-
Eric Dumazet authored
skb argument of tso_count_descs(), tso_build_hdr() and tso_build_data() can be const. Signed-off-by:
-
Eric Dumazet authored
size field can be an int, no need for size_t Removes a 32bit hole on 64bit kernels. And align fields for better readability. Signed-off-by:
-
Eric Dumazet authored
Transport header size could be 60 bytes, and network header size can also be 60 bytes. Add the Ethernet header and we are above 128 bytes. Since drivers using net/core/tso.c usually allocates one DMA coherent piece of memory per TX queue, this patch might cause issues if a driver was using too many slots. For 1024 slots, we would need 256 KB of physically contiguous memory instead of 128 KB. Alternative fix would be to add checks in the fast path, but this involves more work in all drivers using net/core/tso.c. Fixes: f9cbe9a5 ("net: define the TSO header size in net/tso.h") Signed-off-by:
-
