- 22 5月, 2020 3 次提交
-
-
由 Björn Töpel 提交于
Continuing the path to support MEM_TYPE_XSK_BUFF_POOL, the AF_XDP zero-copy/sk_buff rx_bi rings are now separate. Functions to properly allocate the different rings are added as well. v3->v4: Made i40e_fd_handle_status() static. (kbuild test robot) v4->v5: Fix kdoc for i40e_clean_programming_status(). (Jakub) Signed-off-by: NBjörn Töpel <bjorn.topel@intel.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Cc: intel-wired-lan@lists.osuosl.org Link: https://lore.kernel.org/bpf/20200520192103.355233-8-bjorn.topel@gmail.com
-
由 Björn Töpel 提交于
As a first step to migrate i40e to the new MEM_TYPE_XSK_BUFF_POOL APIs, code that accesses the rx_bi (SW/shadow ring) is refactored to use an accessor function. Signed-off-by: NBjörn Töpel <bjorn.topel@intel.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Cc: intel-wired-lan@lists.osuosl.org Link: https://lore.kernel.org/bpf/20200520192103.355233-7-bjorn.topel@gmail.com
-
由 Magnus Karlsson 提交于
Move the AF_XDP zero-copy driver interface to its own include file called xdp_sock_drv.h. This, hopefully, will make it more clear for NIC driver implementors to know what functions to use for zero-copy support. v4->v5: Fix -Wmissing-prototypes by include header file. (Jakub) Signed-off-by: NMagnus Karlsson <magnus.karlsson@intel.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200520192103.355233-4-bjorn.topel@gmail.com
-
- 15 5月, 2020 6 次提交
-
-
由 Jesper Dangaard Brouer 提交于
Intel drivers implement native AF_XDP zerocopy in separate C-files, that have its own invocation of bpf_prog_run_xdp(). The setup of xdp_buff is also handled in separately from normal code path. This patch update XDP frame_sz for AF_XDP zerocopy drivers i40e, ice and ixgbe, as the code changes needed are very similar. Introduce a helper function xsk_umem_xdp_frame_sz() for calculating frame size. Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Acked-by: NBjörn Töpel <bjorn.topel@intel.com> Cc: intel-wired-lan@lists.osuosl.org Cc: Magnus Karlsson <magnus.karlsson@intel.com> Link: https://lore.kernel.org/bpf/158945347511.97035.8536753731329475655.stgit@firesoul
-
由 Jesper Dangaard Brouer 提交于
This driver uses different memory models depending on PAGE_SIZE at compile time. For PAGE_SIZE 4K it uses page splitting, meaning for normal MTU frame size is 2048 bytes (and headroom 192 bytes). For larger MTUs the driver still use page splitting, by allocating order-1 pages (8192 bytes) for RX frames. For PAGE_SIZE larger than 4K, driver instead advance its rx_buffer->page_offset with the frame size "truesize". For XDP frame size calculations, this mean that in PAGE_SIZE larger than 4K mode the frame_sz change on a per packet basis. For the page split 4K PAGE_SIZE mode, xdp.frame_sz is more constant and can be updated once outside the main NAPI loop. The default setting in the driver uses build_skb(), which provides the necessary headroom and tailroom for XDP-redirect in RX-frame (in both modes). There is one complication, which is legacy-rx mode (configurable via ethtool priv-flags). There are zero headroom in this mode, which is a requirement for XDP-redirect to work. The conversion to xdp_frame (convert_to_xdp_frame) will detect this insufficient space, and xdp_do_redirect() call will fail. This is deemed acceptable, as it allows other XDP actions to still work in legacy-mode. In legacy-mode + larger PAGE_SIZE due to lacking tailroom, we also accept that xdp_adjust_tail shrink doesn't work. Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Cc: intel-wired-lan@lists.osuosl.org Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Link: https://lore.kernel.org/bpf/158945347002.97035.328088795813704587.stgit@firesoul
-
由 Jesper Dangaard Brouer 提交于
This driver uses different memory models depending on PAGE_SIZE at compile time. For PAGE_SIZE 4K it uses page splitting, meaning for normal MTU frame size is 2048 bytes (and headroom 192 bytes). For larger MTUs the driver still use page splitting, by allocating order-1 pages (8192 bytes) for RX frames. For PAGE_SIZE larger than 4K, driver instead advance its rx_buffer->page_offset with the frame size "truesize". For XDP frame size calculations, this mean that in PAGE_SIZE larger than 4K mode the frame_sz change on a per packet basis. For the page split 4K PAGE_SIZE mode, xdp.frame_sz is more constant and can be updated once outside the main NAPI loop. The default setting in the driver uses build_skb(), which provides the necessary headroom and tailroom for XDP-redirect in RX-frame (in both modes). There is one complication, which is legacy-rx mode (configurable via ethtool priv-flags). There are zero headroom in this mode, which is a requirement for XDP-redirect to work. The conversion to xdp_frame (convert_to_xdp_frame) will detect this insufficient space, and xdp_do_redirect() call will fail. This is deemed acceptable, as it allows other XDP actions to still work in legacy-mode. In legacy-mode + larger PAGE_SIZE due to lacking tailroom, we also accept that xdp_adjust_tail shrink doesn't work. Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Cc: intel-wired-lan@lists.osuosl.org Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Link: https://lore.kernel.org/bpf/158945346494.97035.12809400414566061815.stgit@firesoul
-
由 Jesper Dangaard Brouer 提交于
This patch mirrors the changes to ixgbe in previous patch. This VF driver doesn't support XDP_REDIRECT, but correct tailroom is still necessary for BPF-helper xdp_adjust_tail. In legacy-mode + larger PAGE_SIZE, due to lacking tailroom, we accept that xdp_adjust_tail shrink doesn't work. Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Cc: intel-wired-lan@lists.osuosl.org Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Link: https://lore.kernel.org/bpf/158945345984.97035.13518286183248025173.stgit@firesoul
-
由 Jesper Dangaard Brouer 提交于
This driver uses different memory models depending on PAGE_SIZE at compile time. For PAGE_SIZE 4K it uses page splitting, meaning for normal MTU frame size is 2048 bytes (and headroom 192 bytes). For larger MTUs the driver still use page splitting, by allocating order-1 pages (8192 bytes) for RX frames. For PAGE_SIZE larger than 4K, driver instead advance its rx_buffer->page_offset with the frame size "truesize". For XDP frame size calculations, this mean that in PAGE_SIZE larger than 4K mode the frame_sz change on a per packet basis. For the page split 4K PAGE_SIZE mode, xdp.frame_sz is more constant and can be updated once outside the main NAPI loop. The default setting in the driver uses build_skb(), which provides the necessary headroom and tailroom for XDP-redirect in RX-frame (in both modes). There is one complication, which is legacy-rx mode (configurable via ethtool priv-flags). There are zero headroom in this mode, which is a requirement for XDP-redirect to work. The conversion to xdp_frame (convert_to_xdp_frame) will detect this insufficient space, and xdp_do_redirect() call will fail. This is deemed acceptable, as it allows other XDP actions to still work in legacy-mode. In legacy-mode + larger PAGE_SIZE due to lacking tailroom, we also accept that xdp_adjust_tail shrink doesn't work. Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Cc: intel-wired-lan@lists.osuosl.org Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Link: https://lore.kernel.org/bpf/158945345455.97035.14334355929030628741.stgit@firesoul
-
由 Jesper Dangaard Brouer 提交于
The ixgbe driver have another memory model when compiled on archs with PAGE_SIZE above 4096 bytes. In this mode it doesn't split the page in two halves, but instead increment rx_buffer->page_offset by truesize of packet (which include headroom and tailroom for skb_shared_info). This is done correctly in ixgbe_build_skb(), but in ixgbe_rx_buffer_flip which is currently only called on XDP_TX and XDP_REDIRECT, it forgets to add the tailroom for skb_shared_info. This breaks XDP_REDIRECT, for veth and cpumap. Fix by adding size of skb_shared_info tailroom. Maintainers notice: This fix have been queued to Jeff. Fixes: 64530739 ("ixgbe: add initial support for xdp redirect") Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: NAlexei Starovoitov <ast@kernel.org> Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Link: https://lore.kernel.org/bpf/158945344946.97035.17031588499266605743.stgit@firesoul
-
- 01 5月, 2020 1 次提交
-
-
由 Wei Yongjun 提交于
Fix to return a error code from the error handling case instead of 0, as done elsewhere in this function. Fixes: 31ad4e4e ("ice: Allocate flow profile") Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
- 22 4月, 2020 2 次提交
-
-
由 Jason Yan 提交于
The '==' expression itself is bool, no need to convert it to bool again. This fixes the following coccicheck warning: drivers/net/ethernet/intel/e1000/e1000_main.c:1479:44-49: WARNING: conversion to bool not needed here Signed-off-by: NJason Yan <yanaijie@huawei.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Jason Yan 提交于
The '==' expression itself is bool, no need to convert it to bool again. This fixes the following coccicheck warning: drivers/net/ethernet/intel/i40e/i40e_main.c:1614:52-57: WARNING: conversion to bool not needed here drivers/net/ethernet/intel/i40e/i40e_main.c:11439:52-57: WARNING: conversion to bool not needed here Signed-off-by: NJason Yan <yanaijie@huawei.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
- 20 4月, 2020 14 次提交
-
-
由 Andre Guedes 提交于
This patch adds log messages to functions related to the MAC address filtering code to ease debugging. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
This patch does a code refactoring in igc_del_mac_filter() so it uses the new helper igc_find_mac_filter() and improves the comment about the special handling when deleting the default filter. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
The helper igc_mac_entry_can_be_used() implementation is a bit convoluted since it does two different things: find a not-in-use slot in mac_table or find an in-use slot where the address and address type match. This patch does a code refactoring and break it up into two helper functions. With this patch we might traverse mac_table twice in some situations, but this is not harmful performance-wise (mac_table has only 16 entries and adding mac filters is not hot-path), and it improves igc_add_mac_ filter() readability considerably. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
With the previous two patches, igc_add_mac_steering_filter() and igc_del_mac_steering_filter() became a pointless wrapper of igc_add_mac_filter() and igc_del_mac_filter(). This patch removes these wrappers and update callers to call igc_add_mac_filter() and igc_del_mac_filter() directly. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
The IGC_MAC_STATE_QUEUE_STEERING bit in mac_table[i].state is utilized to indicate that frames matching the filter are assigned to mac_table[i].queue. This bit is not strictly necessary since we can convey the same information as follows: queue == -1 means queue assignment is disabled, otherwise it is enabled. In addition to make the code simpler, this change fixes some awkward situations where we pass a complete misleading 'queue' value such as in igc_uc_sync(). So this patch removes IGC_MAC_STATE_QUEUE_STEERING and also takes the opportunity to improve the igc_add_mac_filter documentation. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
igc_add_mac_filter() doesn't allow us to have more than one entry with the same address and address type in adapter->mac_table so checking if 'queue' matches in igc_del_mac_filter() isn't necessary. This patch removes that check. This patch also takes the opportunity to improve the igc_del_mac_filter documentation and remove comment which is not applicable to this I225 controller. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
igc_add_mac_filter() doesn't allow filters with invalid MAC address to be added to adapter->mac_table so, in igc_del_mac_filter(), we can early return if MAC address is invalid. No need to traverse the table. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
Current igc_rar_set_index() implementation is a bit convoluted so this patch does some code refactoring to improve it. The helper igc_rar_set_index() is about writing MAC filter settings into hardware registers. Logic such as address validation belongs to functions upper in the call chain such as igc_set_mac() and igc_add_mac_filter(). So this patch moves the is_valid_ether_addr() call to igc_add_mac_filter(). No need to touch igc_set_mac() since it already checks it. The variables 'rar_low' and 'rar_high' represent the value in registers RAL and RAH so we rename them to 'ral' and 'rah', respectively, to match the registers names. To make it explicit, filter settings are passed as arguments to the function instead of reading them from adapter->mac_table "under the hood". Also, the function was renamed to igc_set_mac_filter_hw to make it more clear what it does. Finally, the patch removes some wrfl() calls and comments not needed. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
In case igc_del_mac_filter() returns error, that error is masked since the functions always return 0 (success). This patch fixes igc_uc_unsync() so it returns whatever value igc_del_mac_filter() returns (0 on success, negative number on error). Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
In case of success, igc_add_mac_filter() returns the index in adapter->mac_table where the requested filter was added. This information, however, is not used by any caller of that function. In fact, callers have extra code just to handle this returning index as 0 (success). So this patch changes the function to return 0 on success instead, and cleans up the extra code. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
The IGC_MAC_STATE_SRC_ADDR flags is not supported by igc_add_mac_ filter() so this patch adds a check for it and returns -ENOTSUPP in case it is set. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
This patch does a code refactoring in the MAC address filtering logic to get rid of some duplicate code. IGC driver has two functions to add MAC address filters that are pretty much the same: igc_add_mac_filter() and igc_add_mac_filter_flags(). The only difference is that the latter allows the callee to specify the 'flags' parameter while the former has it hard coded as zero. The same rationale applies to filter deletion counterparts. So this patch refactors igc_add_mac_filter() and igc_del_mac_filter() so they handle the 'flags' parameters, removes the _flags() functions, and fixes callees accordingly. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Vitaly Lifshits 提交于
Added a fix to S0ix entry and exit flows for TGP and above MAC types, to the case when the Ethernet cable is connected and the link is up. With that the system is able to reach SLP_S0 when going to freeze power state. Signed-off-by: NVitaly Lifshits <vitaly.lifshits@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Sasha Neftin 提交于
Add new device IDs for the next step of i225 Signed-off-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
- 18 4月, 2020 14 次提交
-
-
由 Andre Guedes 提交于
This patch fixes a bug when the user adds the first MAC address filter via ethtool NFC mechanism. When the first MAC address filter is added, it overwrites the default MAC address filter configured at RAL[0] and RAH[0]. As consequence, frames addressed to the interface MAC address are not sent to host anymore. This patch fixes the bug by calling igc_set_default_mac_filter() during adapter init so the position 0 of adapter->mac_table[] is assigned to the default MAC address. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
IGC driver has no support for tc-flower filters so this patch removes some leftover code, probably copied from IGB driver by mistake. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
The support for ethtool Network Flow Classification (NFC) queue redirection based on destination MAC address is currently broken in IGC. For instance, if we add the following rule, matching frames aren't enqueued on the expected rx queue. $ ethtool -N IFNAME flow-type ether dst 3c:fd:fe:9e:7f:71 queue 2 The issue here is due to the fact that igc_rar_set_index() is missing code to enable the queue selection feature from Receive Address High (RAH) register. This patch adds the missing code and fixes the issue. Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Acked-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Sasha Neftin 提交于
i225 device support copper mode only PHY signal detect indication for copper fiber switch not applicable to i225 part Signed-off-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Sasha Neftin 提交于
This assignment of the feature NETIF_F_HW_TC occurs prior to the initial setup of the local hw_features variable. This ensures that NETIF_F_HW_TC are marked as user changeable, and also enables it by default when the driver loads. Signed-off-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Sasha Neftin 提交于
IGC_START_ITR has beed defined twice This patch come to fix it Signed-off-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Andre Guedes 提交于
This patch fixes some register overwriting when dumping registers via ethtool. We have a total of 16 RAL registers, starting at offset 139. So RAH offset should be 139 + 16 = 155, not 145. As result some RAL registers are overwritten. Likewise, RAH registers are also overwritten by TDBAL, TDBAH, TDLEN, and TDH registers. To fix this bug while preserving the ABI, this patch re-writes RAL and RAH registers at the end of 'regs_buff' and bumps regs->version. It also removes some pointless comments in the middle of igc_set_regs(). Signed-off-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Sasha Neftin 提交于
Move igc_adapter and igc_ring structures up to avoid forward declaration It is not necessary to forward declare these structures Signed-off-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Sasha Neftin 提交于
We support only copper mode Not applicable for i225 parts Signed-off-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Sasha Neftin 提交于
Formally Destination bit should be kept reserved to support legacy drivers and ignore on write/read operation Not applicable for i225 parts Signed-off-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Kees Cook 提交于
Variables declared in a switch statement before any case statements cannot be automatically initialized with compiler instrumentation (as they are not part of any execution flow). With GCC's proposed automatic stack variable initialization feature, this triggers a warning (and they don't get initialized). Clang's automatic stack variable initialization (via CONFIG_INIT_STACK_ALL=y) doesn't throw a warning, but it also doesn't initialize such variables[1]. Note that these warnings (or silent skipping) happen before the dead-store elimination optimization phase, so even when the automatic initializations are later elided in favor of direct initializations, the warnings remain. To avoid these problems, move such variables into the "case" where they're used or lift them up into the main function body. drivers/net/ethernet/intel/e1000/e1000_main.c: In function ‘e1000_xmit_frame’: drivers/net/ethernet/intel/e1000/e1000_main.c:3143:18: warning: statement will never be executed [-Wswitch-unreachable] 3143 | unsigned int pull_size; | ^~~~~~~~~ [1] https://bugs.llvm.org/show_bug.cgi?id=44916Signed-off-by: NKees Cook <keescook@chromium.org> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Vinicius Costa Gomes 提交于
This adds support for ETF offloading for the i225 controller. For i225, the LaunchTime feature is almost a subset of the Qbv feature. The main change from the i210 is that the launchtime of each packet is specified as an offset applied to the BASET register. BASET is automatically incremented each cycle. For i225, the approach chosen is to re-use most of the setup used for taprio offloading. With a few changes: - The more or less obvious one is that when ETF is enabled, we should set add the expected launchtime to the (advanced) transmit descriptor; - The less obvious, is that when taprio offloading is not enabled, we add a dummy schedule (all queues are open all the time, with a cycle time of 1 second). Signed-off-by: NVinicius Costa Gomes <vinicius.gomes@intel.com> Reviewed-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Vinicius Costa Gomes 提交于
Adds support for translating taprio schedules into i225 cycles. This will allow schedules to run in the hardware, making the schedules enforcement more precise and saving CPU time. Right now, the only simple schedules are allowed, complex schedules are rejected. "simple" in this context are schedules that each HW queue is opened and closed only once in each cycle. Changing schedules is still not supported as well. Signed-off-by: NVinicius Costa Gomes <vinicius.gomes@intel.com> Reviewed-by: NAndre Guedes <andre.guedes@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-
由 Sasha Neftin 提交于
Partial generic segmentation offload is a hybrid between TSO and GSO. What is effectively does is take advantage of certain traits of TCP and tunnels so that instead of having to rewrite the packet headers for each segment only in the inner-most transport header and possible the outer-most network header need to be updated. This allows devices that do not support tunnel offload or tunnels offloads with checksum to still make use of segmentation. Signed-off-by: NSasha Neftin <sasha.neftin@intel.com> Tested-by: NAaron Brown <aaron.f.brown@intel.com> Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
-