- 28 Jun, 2019 40 commits
-
-
Vedang Patel authored
When the taprio qdisc is running in "txtime offload" mode, it will set the launchtime value (in skb->tstamp) for all the packets which do not have the SO_TXTIME socket option. But, the TCP packets already have this value set and it indicates the earliest departure time represented in CLOCK_MONOTONIC clock. We need to respect the timestamp set by the TCP subsystem. So, convert this time to the clock which taprio is using and ensure that the packet is not transmitted before the deadline set by TCP. Signed-off-by:
Vedang Patel <vedang.patel@intel.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Vedang Patel authored
Later in this series we will need to transform from CLOCK_MONOTONIC (used in TCP) to the clock reference used in TAPRIO. Signed-off-by:
Vinicius Costa Gomes <vinicius.gomes@intel.com> Signed-off-by:
-
Vedang Patel authored
Currently, we are seeing non-critical packets being transmitted outside of their timeslice. We can confirm that the packets are being dequeued at the right time. So, the delay is induced in the hardware side. The most likely reason is the hardware queues are starving the lower priority queues. In order to improve the performance of taprio, we will be making use of the txtime feature provided by the ETF qdisc. For all the packets which do not have the SO_TXTIME option set, taprio will set the transmit timestamp (set in skb->tstamp) in this mode. TAPrio Qdisc will ensure that the transmit time for the packet is set to when the gate is open. If SO_TXTIME is set, the TAPrio qdisc will validate whether the timestamp (in skb->tstamp) occurs when the gate corresponding to skb's traffic class is open. Following two parameters added to support this mode: - flags: used to enable txtime-assist mode. Will also be used to enable other modes (like hardware offloading) later. - txtime-delay: This indicates the minimum time it will take for the packet to hit the wire. This is useful in determining whether we can transmit the packet in the remaining time if the gate corresponding to the packet is currently open. An example configuration for enabling txtime-assist: tc qdisc replace dev eth0 parent root handle 100 taprio \\ num_tc 3 \\ map 2 2 1 0 2 2 2 2 2 2 2 2 2 2 2 2 \\ queues 1@0 1@0 1@0 \\ base-time 1558653424279842568 \\ sched-entry S 01 300000 \\ sched-entry S 02 300000 \\ sched-entry S 04 400000 \\ flags 0x1 \\ txtime-delay 40000 \\ clockid CLOCK_TAI tc qdisc replace dev $IFACE parent 100:1 etf skip_sock_check \\ offload delta 200000 clockid CLOCK_TAI Note that all the traffic classes are mapped to the same queue. This is only possible in taprio when txtime-assist is enabled. Also, note that the ETF Qdisc is enabled with offload mode set. In this mode, if the packet's traffic class is open and the complete packet can be transmitted, taprio will try to transmit the packet immediately. This will be done by setting skb->tstamp to current_time + the time delta indicated in the txtime-delay parameter. This parameter indicates the time taken (in software) for packet to reach the network adapter. If the packet cannot be transmitted in the current interval or if the packet's traffic is not currently transmitting, the skb->tstamp is set to the next available timestamp value. This is tracked in the next_launchtime parameter in the struct sched_entry. The behaviour w.r.t admin and oper schedules is not changed from what is present in software mode. The transmit time is already known in advance. So, we do not need the HR timers to advance the schedule and wakeup the dequeue side of taprio. So, HR timer won't be run when this mode is enabled. Signed-off-by: -
Vedang Patel authored
Remove inline directive from length_to_duration(). We will let the compiler make the decisions. Signed-off-by:
-
Vedang Patel authored
cycle time for a particular schedule is calculated only when it is first installed. So, it makes sense to just calculate it once right after the 'cycle_time' parameter has been parsed and store it in cycle_time. Signed-off-by:
-
Vedang Patel authored
Currently, etf expects a socket with SO_TXTIME option set for each packet it encounters. So, it will drop all other packets. But, in the future commits we are planning to add functionality where tstamp value will be set by another qdisc. Also, some packets which are generated from within the kernel (e.g. ICMP packets) do not have any socket associated with them. So, this commit adds support for skip_sock_check. When this option is set, etf will skip checking for a socket and other associated options for all skbs. Signed-off-by:
-
Vedang Patel authored
The BIT() macro isn't exported as part of the UAPI interface. So, the compile-test to ensure they are self contained fails. So, use _BITUL() instead. Signed-off-by:
-
Vedang Patel authored
If a packet which is utilizing the launchtime feature (via SO_TXTIME socket option) also requests the hardware transmit timestamp, the hardware timestamp is not delivered to the userspace. This is because the value in skb->tstamp is mistaken as the software timestamp. Applications, like ptp4l, request a hardware timestamp by setting the SOF_TIMESTAMPING_TX_HARDWARE socket option. Whenever a new timestamp is detected by the driver (this work is done in igb_ptp_tx_work() which calls igb_ptp_tx_hwtstamps() in igb_ptp.c[1]), it will queue the timestamp in the ERR_QUEUE for the userspace to read. When the userspace is ready, it will issue a recvmsg() call to collect this timestamp. The problem is in this recvmsg() call. If the skb->tstamp is not cleared out, it will be interpreted as a software timestamp and the hardware tx timestamp will not be successfully sent to the userspace. Look at skb_is_swtx_tstamp() and the callee function __sock_recv_timestamp() in net/socket.c for more details. Signed-off-by:
Aaron Brown <aaron.f.brown@intel.com> Signed-off-by:
-
David S. Miller authored
John Hurley says: ==================== Track recursive calls in TC act_mirred These patches aim to prevent act_mirred causing stack overflow events from recursively calling packet xmit or receive functions. Such events can occur with poor TC configuration that causes packets to travel in loops within the system. Florian Westphal advises that a recursion crash and packets looping are separate issues and should be treated as such. David Miller futher points out that pcpu counters cannot track the precise skb context required to detect loops. Hence these patches are not aimed at detecting packet loops, rather, preventing stack flows arising from such loops. ==================== Signed-off-by: -
John Hurley authored
TC hooks allow the application of filters and actions to packets at both ingress and egress of the network stack. It is possible, with poor configuration, that this can produce loops whereby an ingress hook calls a mirred egress action that has an egress hook that redirects back to the first ingress etc. The TC core classifier protects against loops when doing reclassifies but there is no protection against a packet looping between multiple hooks and recursively calling act_mirred. This can lead to stack overflow panics. Add a per CPU counter to act_mirred that is incremented for each recursive call of the action function when processing a packet. If a limit is passed then the packet is dropped and CPU counter reset. Note that this patch does not protect against loops in TC datapaths. Its aim is to prevent stack overflow kernel panics that can be a consequence of such loops. Signed-off-by:
John Hurley <john.hurley@netronome.com> Reviewed-by:
Simon Horman <simon.horman@netronome.com> Signed-off-by:
-
John Hurley authored
The TC_ACT_REINSERT return type was added as an in-kernel only option to allow a packet ingress or egress redirect. This is used to avoid unnecessary skb clones in situations where they are not required. If a TC hook returns this code then the packet is 'reinserted' and no skb consume is carried out as no clone took place. This return type is only used in act_mirred. Rather than have the reinsert called from the main datapath, call it directly in act_mirred. Instead of returning TC_ACT_REINSERT, change the type to the new TC_ACT_CONSUMED which tells the caller that the packet has been stolen by another process and that no consume call is required. Moving all redirect calls to the act_mirred code is in preparation for tracking recursion created by act_mirred. Signed-off-by:
-
Christian Brauner authored
Tools such as vpnc try to flush routes when run inside network namespaces by writing 1 into /proc/sys/net/ipv4/route/flush. This currently does not work because flush is not enabled in non-initial network namespaces. Since routes are per network namespace it is safe to enable /proc/sys/net/ipv4/route/flush in there. Link: https://github.com/lxc/lxd/issues/4257Signed-off-by:
Christian Brauner <christian.brauner@ubuntu.com> Signed-off-by:
-
git://git.open-mesh.org/linux-mergeDavid S. Miller authored
Simon Wunderlich says: ==================== This feature/cleanup patchset includes the following patches: - bump version strings, by Simon Wunderlich - fix includes for _MAX constants, atomic functions and fwdecls, by Sven Eckelmann (3 patches) - shorten multicast tt/tvlv worker spinlock section, by Linus Luessing - routeable multicast preparations: implement MAC multicast filtering, by Linus Luessing (2 patches, David Millers comments integrated) - remove return value checks for debugfs_create, by Greg Kroah-Hartman - add routable multicast optimizations, by Linus Luessing (2 patches) ==================== Signed-off-by: -
David S. Miller authored
Huazhong Tan says: ==================== net: hns3: some code optimizations & cleanups & bugfixes [patch 01/12] fixes a TX timeout issue. [patch 02/12 - 04/12] adds some patch related to TM module. [patch 05/12] fixes a compile warning. [patch 06/12] adds Asym Pause support for autoneg [patch 07/12] optimizes the error handler for VF reset. [patch 08/12] deals with the empty interrupt case. [patch 09/12 - 12/12] adds some cleanups & optimizations. ==================== Signed-off-by: -
Peng Li authored
If CMDQ ring is full, hclge_cmd_send may return directly, but IMP still working and HW pointer changed, SW ring pointer do not match the HW pointer. This patch update the SW pointer every time when the space is full, so it can work normally next time if IMP and HW still working. Signed-off-by:
Peng Li <lipeng321@huawei.com> Signed-off-by:
Huazhong Tan <tanhuazhong@huawei.com> Signed-off-by:
-
Yunsheng Lin authored
The HNS3_RXD_VLD_B bit has already been checked in hns3_add_frag or hns3_handle_rx_bd before calling hns3_handle_bdinfo, so when hns3_handle_bdinfo is called, the HNS3_RXD_VLD_B bit is always set, which makes the checking in hns3_handle_bdinfo unnecessary. This patch removes the RXD_VLD_B checking in hns3_handle_bdinfo. Signed-off-by:
Yunsheng Lin <linyunsheng@huawei.com> Signed-off-by:
-
Jian Shen authored
This patch removes unused linkmode definition. Signed-off-by:
Jian Shen <shenjian15@huawei.com> Signed-off-by:
-
Yufeng Mo authored
The frame column is based on rx_crc_errors and rx_frame_errors. So l3l4 checksum error should not be counted by rx_crc_errors. Instead, l3l4 checksum error should be counted in ifconfig error column. Fixes: d3ec4ef6 ("net: hns3: refactor the statistics updating for netdev") Signed-off-by:
Yufeng Mo <moyufeng@huawei.com> Signed-off-by:
-
Huazhong Tan authored
Since some MSI-X interrupt's status may be cleared by hardware, so when the driver receives the interrupt, reading HCLGE_VECTOR0_PF_OTHER_INT_STS_REG register will get an empty unknown interrupt. For this case, the irq handler should enable vector0 interrupt. This patch also use dev_info() instead of dev_dbg() in the hclge_check_event_cause(), since this information will be useful for normal usage. Signed-off-by:
-
Huazhong Tan authored
The VF reset may fail for some probabilistic reasons, such as wait for hardware reset timeout, wait for mailbox response timeout, so this patch tries to re-schedule the reset task when the number of reset failing is under HCLGEVF_RESET_MAX_FAIL_CNT. This patch also add a function hclgevf_reset_err_handle() to handle the reset failing. Signed-off-by:
-
Yonglong Liu authored
Local device and link partner config auto-negotiation on both, local device config pause frame use as: rx on/tx off, link partner config pause frame use as: rx off/tx on. We except the result is: Local device: Autonegotiate: on RX: on TX: off RX negotiated: on TX negotiated: off Link partner: Autonegotiate: on RX: off TX: on RX negotiated: off TX negotiated: on But actually, the result of Local device and link partner is both: Autonegotiate: on RX: off TX: off RX negotiated: off TX negotiated: off The root cause is that the supported flag is has only Pause, reference to the function genphy_config_advert(): static int genphy_config_advert(struct phy_device *phydev) { ... linkmode_and(phydev->advertising, phydev->advertising, phydev->supported); ... } The pause frame use of link partner is rx off/tx on, so its advertising only set the bit Asym_Pause, and the supported is only set the bit Pause, so the result of linkmode_and(), is rx off/tx off. This patch adds Asym_Pause to the supported flag to fix it. Signed-off-by:Yonglong Liu <liuyonglong@huawei.com> Signed-off-by:
-
Yonglong Liu authored
When setting -Wformat=2, there is a compiler warning like this: hclge_main.c:xxx:x: warning: format not a string literal and no format arguments [-Wformat-nonliteral] strs[i].desc); ^~~~ This patch adds missing format parameter "%s" to snprintf() to fix it. Fixes: 46a3df9f ("Add HNS3 Acceleration Engine & Compatibility Layer Support") Signed-off-by:
-
Yunsheng Lin authored
When hdev->tx_sch_mode is HCLGE_FLAG_VNET_BASE_SCH_MODE, the hclge_tm_schd_mode_vnet_base_cfg calls hclge_tm_pri_schd_mode_cfg with vport->vport_id as pri_id, which is used as index for hdev->tm_info.tc_info, it will cause out of bound access issue if vport_id is equal to or larger than HNAE3_MAX_TC. Also hardware only support maximum speed of HCLGE_ETHER_MAX_RATE. So this patch adds two checks for above cases. Fixes: 84844054 ("net: hns3: Add support of TX Scheduler & Shaper to HNS3 driver") Signed-off-by:
-
Yunsheng Lin authored
Currently when there is share buffer in the SSU(storage switching unit), the low waterline for RX private buffer is too low to keep the hardware running. Hardware may have processed all the packet stored in the private buffer of the low waterline before the new packet comes, because hardware only tell the peer send packet again when the private buffer is under the low waterline. So this patch only allocate RX private buffer if there is enough buffer according to hardware user manual. This patch also reserve some buffer for reusing when TC num is less than or equal to 2, and change PAUSE_TRANS_GAP & HCLGE_NON_DCB_ADDITIONAL_BUF according to hardware user manual. Signed-off-by:
-
Yunsheng Lin authored
Currently when TC num is one, the DCB will be disabled no matter if pfc_en is non-zero or not. This patch enables the DCB if pfc_en is non-zero, even when TC num is one. Signed-off-by:
-
Huazhong Tan authored
When change MTU or other operations, which just calling .reset_notify to do HNAE3_DOWN_CLIENT and HNAE3_UP_CLIENT, then the netdev_tx_reset_queue() in the hns3_clear_all_ring() will be ignored. So the dev_watchdog() may misdiagnose a TX timeout. This patch separates netdev_tx_reset_queue() from hns3_clear_all_ring(), and unifies hns3_clear_all_ring() and hns3_force_clear_all_ring into one, since they are doing similar things. Fixes: 3a30964a ("net: hns3: delay ring buffer clearing during reset") Signed-off-by:
-
David S. Miller authored
Vladimir Oltean says: ==================== Better PHYLINK compliance for SJA1105 DSA After discussing with Russell King, it appears this driver is making a few confusions and not performing some checks for consistent operation. Changes in v2: - Removed redundant print in the phylink_validate callback (in 2/3). ==================== Acked-by:
Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by:
-
Vladimir Oltean authored
We need a better way to signal this, perhaps in phylink_validate, but for now just print this error message as guidance for other people looking at this driver's code while trying to rework PHYLINK. Cc: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by:
Vladimir Oltean <olteanv@gmail.com> Signed-off-by:
-
Vladimir Oltean authored
PHYLINK being designed with PHYs in mind that can change MII protocol, for correct operation it is necessary to ensure that the PHY interface mode stays the same (otherwise clear the supported bit mask, as required). Because this is just a hypothetical situation for now, we don't bother to check whether we could actually support the new PHY interface mode. Actually we could modify the xMII table, reset the switch and send an updated static configuration, but adding that would just be dead code. Cc: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by:
-
Vladimir Oltean authored
It has been pointed out that PHYLINK can call mac_config only to update the phy_interface_type and without knowing what the AN results are. Experimentally, when this was observed to happen, state->link was also unset, and therefore was used as a proxy to ignore this call. However it is also suggested that state->link is undefined for this callback and should not be relied upon. So let the previously-dead codepath for SPEED_UNKNOWN be called, and update the comment to make sure the MAC's behavior is sane. Cc: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by:
-
Arnd Bergmann authored
On 32-bit architectures, putting an array of 256 u32 values on the stack uses more space than the warning limit: drivers/net/ethernet/huawei/hinic/hinic_main.c: In function 'hinic_rss_init': drivers/net/ethernet/huawei/hinic/hinic_main.c:286:1: error: the frame size of 1068 bytes is larger than 1024 bytes [-Werror=frame-larger-than=] I considered changing the code to use u8 values here, since that's all the hardware supports, but dynamically allocating the array is a more isolated fix here. Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
David S. Miller authored
Jose Abreu says: ==================== net: stmmac: 10GbE using XGMAC Support for 10Gb Link using XGMAC core plus some performance tweaks. Tested in a PCI based setup. iperf3 TCP results: TSO ON, MTU=1500, TX Queues = 1, RX Queues = 1, Flow Control ON Pinned CPU (-A), Zero-Copy (-Z) [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-600.00 sec 643 GBytes 9.21 Gbits/sec 1 sender [ 5] 0.00-600.00 sec 643 GBytes 9.21 Gbits/sec receiver ==================== Signed-off-by: -
Jose Abreu authored
We support more speeds now. Update the Kconfig entry. Signed-off-by:
Jose Abreu <joabreu@synopsys.com> Cc: Joao Pinto <jpinto@synopsys.com> Cc: David S. Miller <davem@davemloft.net> Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com> Cc: Alexandre Torgue <alexandre.torgue@st.com> Signed-off-by:
-
Jose Abreu authored
Only disable the interrupts if RX NAPI gets to be scheduled. Also, schedule the TX NAPI only when the interrupts are disabled. Signed-off-by:
-
Jose Abreu authored
Update the RX Tail Pointer to the last available SKB entry. Signed-off-by:
-
Jose Abreu authored
Currently, stmmac only supports 32 bits addressing for SKB. Enable the support for upto 48 bits addressing in XGMAC core. This avoids the use of bounce buffers and increases performance. Changes from v1: - Fallback to 32 bits in failure (Andrew) Signed-off-by:
-
Jose Abreu authored
This is a performance killer and anyways the interrupts are being disabled by RX NAPI so no need to disable them again. Signed-off-by:
-
Jose Abreu authored
XGMAC supports following speeds: - 10G XGMII - 5G XGMII - 2.5G XGMII - 2.5G GMII - 1G GMII - 100M MII - 10M MII Add them to the stmmac driver. Signed-off-by:
-
Jose Abreu authored
Undefined burst shall only be set if pdata asks to. Signed-off-by:
-
Jose Abreu authored
For performance reasons decrease the default RX Watchdog value for the minimum allowed. Signed-off-by:
-
