- 16 Dec, 2021 19 commits
-
-
Tao Liu authored
Adding ethtool support for changing rx-coalesce-usec and tx-coalesce-usec when using the DQO queue format. Signed-off-by:
Tao Liu <xliutaox@google.com> Signed-off-by:
Jeroen de Borst <jeroendb@google.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Jordan Kim authored
Being able to see how many descriptors are in-use is helpful when diagnosing certain issues. Signed-off-by:
Jordan Kim <jrkim@google.com> Signed-off-by:
-
Catherine Sullivan authored
Add support for suspend, resume and shutdown. Signed-off-by:
Catherine Sullivan <csully@google.com> Signed-off-by:
David Awogbemila <awogbemila@google.com> Signed-off-by:
-
Willem de Bruijn authored
Allow drivers to pass metadata along with packet data to the device. Introduce a new metadata descriptor type * GVE_TXD_MTD This descriptor is optional. If present it immediate follows the packet descriptor and precedes the segment descriptor. This descriptor may be repeated. Multiple metadata descriptors may follow. There are no immediate uses for this, this is for future proofing. At present devices allow only 1 MTD descriptor. The lower four bits of the type_flags field encode GVE_TXD_MTD. The upper four bits of the type_flags field encodes a *sub*type. Introduce one such metadata descriptor subtype * GVE_MTD_SUBTYPE_PATH This shares path information with the device for network failure discovery and robust response: Linux derives ipv6 flowlabel and ECMP multipath from sk->sk_txhash, and updates this field on error with sk_rethink_txhash. Allow the host stack to do the same. Pass the tx_hash value if set. Also communicate whether the path hash is set, or more exactly, what its type is. Define two common types GVE_MTD_PATH_HASH_NONE GVE_MTD_PATH_HASH_L4 Concrete examples of error conditions that are resolved are mentioned in the commits that add sk_rethink_txhash calls. Such as commit 7788174e ("tcp: change IPv6 flow-label upon receiving spurious retransmission"). Experimental results mirror what the theory suggests: where IPv6 FlowLabel is included in path selection (e.g., LAG/ECMP), flowlabel rotation on TCP timeout avoids the vast majority of TCP disconnects that would otherwise have occurred during link failures in long-haul backbones, when an alternative path is available. Rotation can be applied to various bad connection signals, such as timeouts and spurious retransmissions. In aggregate, such flow level signals can help locate network issues. Define initial common states: GVE_MTD_PATH_STATE_DEFAULT GVE_MTD_PATH_STATE_TIMEOUT GVE_MTD_PATH_STATE_CONGESTION GVE_MTD_PATH_STATE_RETRANSMIT Signed-off-by:
Willem de Bruijn <willemb@google.com> Signed-off-by:
-
Catherine Sullivan authored
No longer needed after we introduced the barrier in gve_napi_poll. Signed-off-by:
-
Catherine Sullivan authored
The id field should be a u32 not a signed int. Signed-off-by:
-
Catherine Sullivan authored
Giving the device access to other kernel structs is not ideal. Move the indexes into their own array and just keep pointers to them in the ntfy block struct. Signed-off-by:
-
Jeroen de Borst authored
The legacy raw addressing device option was processed before the new RDA queue format option. This caused the supported features mask, which is provided only on the RDA queue format option, not to be set. This disabled jumbo-frame support when using raw adressing. Fixes: 255489f5 ("gve: Add a jumbo-frame device option") Signed-off-by:
-
David S. Miller authored
Russell King says: ==================== net: phylink: add PCS validation This series allows phylink to include the PCS in its validation step. There are two reasons to make this change: 1. Some of the network drivers that are making use of the split PCS support are already manually calling into their PCS drivers to perform validation. E.g. stmmac with xpcs. 2. Logically, some network drivers such as mvneta and mvpp2, the restriction we impose in the validate() callback is a property of the "PCS" block that we provide rather than the MAC. This series: 1. Gives phylink a mechanism to query the MAC driver which PCS is wishes to use for the PHY interface mode. This is necessary to allow the PCS to be involved in the validation step without making changes to the configuration. 2. Provide a pcs_validate() method that PCS can implement. This follows a similar model to the MAC's validate() callback, but with some minor differences due to observations from the various implementations. E.g. returning an error code for not-supported and the way the advertising bitmap is masked. 3. Convert mvpp2 and mvneta to this as examples of its use. Further Conversions are in the pipeline, including for stmmac+xpcs, as well as some DSA drivers. Note that DSA conversion to this is conditional upon all DSA drivers populating their supported_interfaces bitmap, since this is required before mac_select_pcs() can be used. Existing drivers that set a PCS in mac_prepare() or mac_config(), or shortly after phylink_create() will continue to work. However, it should be noted that mac_select_pcs() will be called during phylink_create(), and thus any PCS returned by mac_select_pcs() must be available by this time - or we drop the check in phylink_create(). v2: fix kerneldoc typo in patch 1. ==================== Signed-off-by: -
Russell King (Oracle) authored
Convert mvneta to validate the autoneg state for 1000base-X in the pcs_validate() operation, rather than the MAC validate() operation. This allows us to switch the MAC validate() to use phylink_generic_validate(). Signed-off-by:
Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Signed-off-by:
-
Russell King authored
An initial stab at converting mvneta to PCS operations. There's a few FIXMEs to be solved. Signed-off-by:
-
Russell King authored
Convert mvneta to use the mac_prepare() and mac_finish() methods in preparation to converting mvneta to split-PCS support. Signed-off-by:
-
Russell King (Oracle) authored
Convert mvpp2 to validate the autoneg state for 1000base-X in the pcs_validate() operation, rather than the MAC validate() operation. This allows us to switch the MAC validate() to use phylink_generic_validate(). Signed-off-by:
-
Russell King (Oracle) authored
Use the mac_select_pcs() method to choose between the GMAC and XLG PCS implementations. Signed-off-by:
-
Russell King (Oracle) authored
Add a hook for PCS to validate the link parameters. This avoids MAC drivers having to have knowledge of their PCS in their validate() method, thereby allowing several MAC drivers to be simplfied. Signed-off-by:
-
Russell King (Oracle) authored
mac_select_pcs() allows us to have an explicit point to query which PCS the MAC wishes to use for a particular PHY interface mode, thereby allowing us to add support to validate the link settings with the PCS. Phylink will also use this to select the PCS to be used during a major configuration event without the MAC driver needing to call phylink_set_pcs(). Note that if mac_select_pcs() is present, the supported_interfaces bitmap must be filled in; this avoids mac_select_pcs() being called with PHY_INTERFACE_MODE_NA when we want to get support for all interface types. Phylink will return an error in phylink_create() unless this condition is satisfied. Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/nexDavid S. Miller authored
t-queue Tony Nguyen says: ==================== 100GbE Intel Wired LAN Driver Updates 2021-12-15 This series contains updates to ice driver only. Jake makes changes to flash update. This includes the following: * a new shadow-ram region similar to NVM region but for the device shadow RAM contents. This is distinct from NVM region because shadow RAM is built up during device init and may be different from the raw NVM flash data. * refactoring of the ice_flash_pldm_image to become the main flash update entry point. This is simpler than having both an ice_devlink_flash_update and an ice_flash_pldm_image. It will make additions like dry-run easier in the future. * reducing time to read Option ROM version information. * adding support for firmware activation via devlink reload, when possible. The major new work is the reload support, which allows activating firmware immediately without a reboot when possible. Reload support only supports firmware activation. Jesse improves transmit code: utilizing newer netif_tx* API, adding some prefetch calls, correcting expected conditions when calling ice_vsi_down(), and utilizing __netdev_tx_sent_queue() call. ==================== Signed-off-by: -
git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linuxDavid S. Miller authored
Saeed Mahameed says: ==================== mlx5-next branch 2021-12-15 Hi Dave, Jakub, Jason This pulls mlx5-next branch into net-next and rdma branches. All patches already reviewed on both rdma and netdev mailing lists. Please pull and let me know if there's any problem. 1) Add multiple FDB steering priorities [1] 2) Introduce HW bits needed to configure MAC list size of VF/SF. Required for ("net/mlx5: Memory optimizations") upcoming series [2]. [1] https://lore.kernel.org/netdev/20211201193621.9129-1-saeed@kernel.org/ [2] https://lore.kernel.org/lkml/20211208141722.13646-1-shayd@nvidia.com/ ==================== Signed-off-by: -
git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-nextJakub Kicinski authored
Pablo Neira Ayuso says: ==================== Netfilter updates for net-next The following patchset contains Netfilter updates for net-next, mostly rather small housekeeping patches: 1) Remove unused variable in IPVS, from GuoYong Zheng. 2) Use memset_after in conntrack, from Kees Cook. 3) Remove leftover function in nfnetlink_queue, from Florian Westphal. 4) Remove redundant test on bool in conntrack, from Bernard Zhao. 5) egress support for nft_fwd, from Lukas Wunner. 6) Make pppoe work for br_netfilter, from Florian Westphal. 7) Remove unused variable in conntrack resize routine, from luo penghao. * git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next: netfilter: conntrack: Remove useless assignment statements netfilter: bridge: add support for pppoe filtering netfilter: nft_fwd_netdev: Support egress hook netfilter: ctnetlink: remove useless type conversion to bool netfilter: nf_queue: remove leftover synchronize_rcu netfilter: conntrack: Use memset_startat() to zero struct nf_conn ipvs: remove unused variable for ip_vs_new_dest ==================== Link: https://lore.kernel.org/r/20211215234911.170741-1-pablo@netfilter.orgSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
- 15 Dec, 2021 21 commits
-
-
luo penghao authored
The old_size assignment here will not be used anymore The clang_analyzer complains as follows: Value stored to 'old_size' is never read Reported-by:
Zeal Robot <zealci@zte.com.cn> Signed-off-by:
luo penghao <luo.penghao@zte.com.cn> Signed-off-by:
Pablo Neira Ayuso <pablo@netfilter.org>
-
Shay Drory authored
Downstream patch will use this bit in order to know whether the device supports changing of max_uc_list. Signed-off-by:
Shay Drory <shayd@nvidia.com> Reviewed-by:
Moshe Shemesh <moshe@nvidia.com> Signed-off-by:
Saeed Mahameed <saeedm@nvidia.com>
-
Jesse Brandeburg authored
The kernel gained a new interface for drivers to use to combine tail bump (doorbell) and BQL updates, attempt to use those new interfaces. Signed-off-by:
Jesse Brandeburg <jesse.brandeburg@intel.com> Tested-by:
Gurucharan G <gurucharanx.g@intel.com> Signed-off-by:
Tony Nguyen <anthony.l.nguyen@intel.com>
-
Jesse Brandeburg authored
The driver had comments to the effect of: This flag should be set before calling this function. While reviewing code it was found that there were several violations of this policy, which could introduce hard to find bugs or races. Fix the violations of the "VSI DOWN state must be set before calling ice_down" and make checking the state into code with a WARN_ON. Signed-off-by:
-
Jesse Brandeburg authored
The kernel provides some prefetch mechanisms to speed up commonly cold cache line accesses during receive processing. Since these are software structures it helps to have these strategically placed prefetches. Be careful to call BQL prefetch complete only for non XDP queues. Co-developed-by:
Piotr Raczynski <piotr.raczynski@intel.com> Signed-off-by:
-
Jesse Brandeburg authored
Use the netif_tx_* API from netdevice.h which has simpler parameters. Signed-off-by:
-
Jacob Keller authored
The ice hardware contains an embedded chip with firmware which can be updated using devlink flash. The firmware which runs on this chip is referred to as the Embedded Management Processor firmware (EMP firmware). Activating the new firmware image currently requires that the system be rebooted. This is not ideal as rebooting the system can cause unwanted downtime. In practical terms, activating the firmware does not always require a full system reboot. In many cases it is possible to activate the EMP firmware immediately. There are a couple of different scenarios to cover. * The EMP firmware itself can be reloaded by issuing a special update to the device called an Embedded Management Processor reset (EMP reset). This reset causes the device to reset and reload the EMP firmware. * PCI configuration changes are only reloaded after a cold PCIe reset. Unfortunately there is no generic way to trigger this for a PCIe device without a system reboot. When performing a flash update, firmware is capable of responding with some information about the specific update requirements. The driver updates the flash by programming a secondary inactive bank with the contents of the new image, and then issuing a command to request to switch the active bank starting from the next load. The response to the final command for updating the inactive NVM flash bank includes an indication of the minimum reset required to fully update the device. This can be one of the following: * A full power on is required * A cold PCIe reset is required * An EMP reset is required The response to the command to switch flash banks includes an indication of whether or not the firmware will allow an EMP reset request. For most updates, an EMP reset is sufficient to load the new EMP firmware without issues. In some cases, this reset is not sufficient because the PCI configuration space has changed. When this could cause incompatibility with the new EMP image, the firmware is capable of rejecting the EMP reset request. Add logic to ice_fw_update.c to handle the response data flash update AdminQ commands. For the reset level, issue a devlink status notification informing the user of how to complete the update with a simple suggestion like "Activate new firmware by rebooting the system". Cache the status of whether or not firmware will restrict the EMP reset for use in implementing devlink reload. Implement support for devlink reload with the "fw_activate" flag. This allows user space to request the firmware be activated immediately. For the .reload_down handler, we will issue a request for the EMP reset using the appropriate firmware AdminQ command. If we know that the firmware will not allow an EMP reset, simply exit with a suitable netlink extended ACK message indicating that the EMP reset is not available. For the .reload_up handler, simply wait until the driver has finished resetting. Logic to handle processing of an EMP reset already exists in the driver as part of its reset and rebuild flows. Implement support for the devlink reload interface with the "fw_activate" action. This allows userspace to request activation of firmware without a reboot. Note that support for indicating the required reset and EMP reset restriction is not supported on old versions of firmware. The driver can determine if the two features are supported by checking the device capabilities report. I confirmed support has existed since at least version 5.5.2 as reported by the 'fw.mgmt' version. Support to issue the EMP reset request has existed in all version of the EMP firmware for the ice hardware. Check the device capabilities report to determine whether or not the indications are reported by the running firmware. If the reset requirement indication is not supported, always assume a full power on is necessary. If the reset restriction capability is not supported, always assume the EMP reset is available. Users can verify if the EMP reset has activated the firmware by using the devlink info report to check that the 'running' firmware version has updated. For example a user might do the following: # Check current version $ devlink dev info # Update the device $ devlink dev flash pci/0000:af:00.0 file firmware.bin # Confirm stored version updated $ devlink dev info # Reload to activate new firmware $ devlink dev reload pci/0000:af:00.0 action fw_activate # Confirm running version updated $ devlink dev info Finally, this change does *not* implement basic driver-only reload support. I did look into trying to do this. However, it requires significant refactor of how the ice driver probes and loads everything. The ice driver probe and allocation flows were not designed with such a reload in mind. Refactoring the flow to support this is beyond the scope of this change. Signed-off-by:
Jacob Keller <jacob.e.keller@intel.com> Tested-by:
-
Jacob Keller authored
During probe and device reset, the ice driver reads some data from the NVM image as part of ice_init_nvm. Part of this data includes a section of the Option ROM which contains version information. The function ice_get_orom_civd_data is used to locate the '$CIV' data section of the Option ROM. Timing of ice_probe and ice_rebuild indicate that the ice_get_orom_civd_data function takes about 10 seconds to finish executing. The function locates the section by scanning the Option ROM every 512 bytes. This requires a significant number of NVM read accesses, since the Option ROM bank is 500KB. In the worst case it would take about 1000 reads. Worse, all PFs serialize this operation during reload because of acquiring the NVM semaphore. The CIVD section is located at the end of the Option ROM image data. Unfortunately, the driver has no easy method to determine the offset manually. Practical experiments have shown that the data could be at a variety of locations, so simply reversing the scanning order is not sufficient to reduce the overall read time. Instead, copy the entire contents of the Option ROM into memory. This allows reading the data using 4Kb pages instead of 512 bytes at a time. This reduces the total number of firmware commands by a factor of 8. In addition, reading the whole section together at once allows better indication to firmware of when we're "done". Re-write ice_get_orom_civd_data to allocate virtual memory to store the Option ROM data. Copy the entire OptionROM contents at once using ice_read_flash_module. Finally, use this memory copy to scan for the '$CIV' section. This change significantly reduces the time to read the Option ROM CIVD section from ~10 seconds down to ~1 second. This has a significant impact on the total time to complete a driver rebuild or probe. Signed-off-by:
-
Jacob Keller authored
The ice_devlink_flash_update function performs a few upfront checks and then calls ice_flash_pldm_image. Most if these checks make more sense in the context of code within ice_flash_pldm_image. Merge ice_devlink_flash_update and ice_flash_pldm_image into one function, placing it in ice_fw_update.c Since this is still the entry point for devlink, call the function ice_devlink_flash_update instead of ice_flash_pldm_image. This leaves a single function which handles the devlink parameters and then initiates a PLDM update. With this change, the ice_devlink_flash_update function in ice_fw_update.c becomes the main entry point for flash update. It elimintes some unnecessary boiler plate code between the two previous functions. The ultimate motivation for this is that it eases supporting a dry run with the PLDM library in a future change. Signed-off-by:
-
Jacob Keller authored
The ice_devlink_flash_update function performs a few checks and then calls ice_flash_pldm_image. One of these checks is to call ice_check_for_pending_update. This function checks if the device has a pending update, and cancels it if so. This is necessary to allow a new flash update to proceed. We want to refactor the ice code to eliminate ice_devlink_flash_update, moving its checks into ice_flash_pldm_image. To do this, ice_check_for_pending_update will become static, and only called by ice_flash_pldm_image. To make this change easier to review, first just move the function up within the ice_fw_update.c file. While at it, note that the function has a misleading name. Its primary action is to cancel a pending update. Using the verb "check" does not imply this. Rename it to ice_cancel_pending_update. Signed-off-by:
-
Jacob Keller authored
We have a region for reading the contents of the NVM flash as a snapshot. This region does not allow reading the Shadow RAM, as it always passes the FLASH_ONLY bit to the low level firmware interface. Add a separate shadow-ram region which will allow snapshot of the current contents of the Shadow RAM. This data is built from the NVM contents but is distinct as the device builds up the Shadow RAM during initialization, so being able to snapshot its contents can be useful when attempting to debug flash related issues. Fix the comment description of the nvm-flash region which incorrectly stated that it filled the shadow-ram region, and add a comment explaining that the nvm-flash region does not actually read the Shadow RAM. Signed-off-by:
-
Jakub Kicinski authored
Commit 0976b888 ("ethtool: fix null-ptr-deref on ref tracker") made the write to req_info.dev conditional, but as Eric points out in a different follow up the structure is often allocated on the stack and not kzalloc()'d so seems safer to always write the dev, in case it's garbage on input. Signed-off-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
-
Eric Dumazet authored
Most notable changes are in af_packet, tipc ones are trivial. Signed-off-by:
-
David S. Miller authored
Ido Schimmel says: ==================== mlxsw: Add support for VxLAN with IPv6 underlay So far, mlxsw only supported VxLAN with IPv4 underlay. This patchset extends mlxsw to also support VxLAN with IPv6 underlay. The main difference is related to the way IPv6 addresses are handled by the device. See patch #1 for a detailed explanation. Patch #1 creates a common hash table to store the mapping from IPv6 addresses to KVDL indexes. This table is useful for both IP-in-IP and VxLAN tunnels with an IPv6 underlay. Patch #2 converts the IP-in-IP code to use the new hash table. Patches #3-#6 are preparations. Patch #7 finally adds support for VxLAN with IPv6 underlay. Patch #8 removes a test case that checked that VxLAN configurations with IPv6 underlay are vetoed by the driver. A follow-up patchset will add forwarding selftests. ==================== Signed-off-by: -
Amit Cohen authored
Currently, there is a test case to verify that VxLAN with IPv6 underlay is forbidden. Remove this test case as support for VxLAN with IPv6 underlay was added by the previous patch. Signed-off-by:
Amit Cohen <amcohen@nvidia.com> Signed-off-by:
Ido Schimmel <idosch@nvidia.com> Signed-off-by:
-
Amit Cohen authored
Currently, mlxsw driver supports VxLAN with IPv4 underlay only. Add support for IPv6 underlay. The main differences are: * Learning is not supported for IPv6 FDB entries, use static entries and do not allow 'learning' flag for IPv6 VxLAN. * IPv6 addresses for FDB entries should be saved as part of KVDL. Use the new API to allocate and release entries for IPv6 addresses. * Spectrum ASICs do not fill UDP checksum, while in software IPv6 UDP packets with checksum zero are dropped. Force the relevant flags which allow the VxLAN device to generate UDP packets with zero checksum and also receive them. Signed-off-by:
-
Amit Cohen authored
FDB entries that perform VxLAN encapsulation with an IPv6 underlay hold a reference on a resource. Namely, the KVDL entry where the IPv6 underlay destination IP is stored. When such an FDB entry is deleted, it needs to drop the reference from the corresponding KVDL entry. To that end, maintain a hash table that maps an FDB entry (i.e., {MAC, FID}) to the IPv6 address used by it. Signed-off-by: -
Amit Cohen authored
Add a function to fill IPv6 unicast FDB entries. Use the common function for common fields. Unlike IPv4 entries, the underlay IP address is not filled in the register payload, but instead a pointer to KVDL is used. Signed-off-by:
-
Amit Cohen authored
Currently, the function which adds/removes unicast tunnel FDB entries is shared between IPv4 and IPv6, while for IPv6 it warns because there is no support for it. The code for IPv6 will be more complicated because it needs to allocate/release a KVDL pointer for the underlay IPv6 address. As a preparation for IPv6 underlay support, split the code according to address family. Signed-off-by:
-
Amit Cohen authored
As part of 'can_offload' checks, there is a check of VxLAN flags. The supported flags for IPv6 VxLAN will be different from the existing flags because of some limitations. As preparation for IPv6 underlay support, make this check per address family. Signed-off-by:
-
Amit Cohen authored
Use the common hash table introduced by the previous patch instead of the IP-in-IP specific implementation. Signed-off-by:
-
