- 10 Nov, 2018 2 commits
-
-
배석진 authored
Only first fragment has the sport/dport information, not the following ones. If we want consistent hash for all fragments, we need to ignore ports even for first fragment. This bug is visible for IPv6 traffic, if incoming fragments do not have a flow label, since skb_get_hash() will give different results for first fragment and following ones. It is also visible if any routing rule wants dissection and sport or dport. See commit 5e5d6fed ("ipv6: route: dissect flow in input path if fib rules need it") for details. [edumazet] rewrote the changelog completely. Fixes: 06635a35 ("flow_dissect: use programable dissector in skb_flow_dissect and friends") Signed-off-by:
배석진 <soukjin.bae@samsung.com> Signed-off-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Subash Abhinov Kasiviswanathan authored
A null dereference was observed when a sysctl was being set from userspace and rmnet was stuck trying to complete some actions in the NETDEV_REGISTER callback. This is because the real_dev is set only after the device registration handler completes. sysctl call stack - <6> Unable to handle kernel NULL pointer dereference at virtual address 00000108 <2> pc : rmnet_vnd_get_iflink+0x1c/0x28 <2> lr : dev_get_iflink+0x2c/0x40 <2> rmnet_vnd_get_iflink+0x1c/0x28 <2> inet6_fill_ifinfo+0x15c/0x234 <2> inet6_ifinfo_notify+0x68/0xd4 <2> ndisc_ifinfo_sysctl_change+0x1b8/0x234 <2> proc_sys_call_handler+0xac/0x100 <2> proc_sys_write+0x3c/0x4c <2> __vfs_write+0x54/0x14c <2> vfs_write+0xcc/0x188 <2> SyS_write+0x60/0xc0 <2> el0_svc_naked+0x34/0x38 device register call stack - <2> notifier_call_chain+0x84/0xbc <2> raw_notifier_call_chain+0x38/0x48 <2> call_netdevice_notifiers_info+0x40/0x70 <2> call_netdevice_notifiers+0x38/0x60 <2> register_netdevice+0x29c/0x3d8 <2> rmnet_vnd_newlink+0x68/0xe8 <2> rmnet_newlink+0xa0/0x160 <2> rtnl_newlink+0x57c/0x6c8 <2> rtnetlink_rcv_msg+0x1dc/0x328 <2> netlink_rcv_skb+0xac/0x118 <2> rtnetlink_rcv+0x24/0x30 <2> netlink_unicast+0x158/0x1f0 <2> netlink_sendmsg+0x32c/0x338 <2> sock_sendmsg+0x44/0x60 <2> SyS_sendto+0x150/0x1ac <2> el0_svc_naked+0x34/0x38 Fixes: b752eff5 ("net: qualcomm: rmnet: Implement ndo_get_iflink") Signed-off-by:Sean Tranchetti <stranche@codeaurora.org> Signed-off-by:
Subash Abhinov Kasiviswanathan <subashab@codeaurora.org> Signed-off-by:
-
- 09 Nov, 2018 15 commits
-
-
David S. Miller authored
Igor Russkikh says: ==================== net: aquantia: 2018-11 bugfixes The patchset fixes a number of bugs found in various areas after driver validation. ==================== Signed-off-by: -
Dmitry Bogdanov authored
RX Checksum offloads could not be configured and ignored netdev features flag for checksumming. Signed-off-by:
Igor Russkikh <igor.russkikh@aquantia.com> Signed-off-by:
Dmitry Bogdanov <dmitry.bogdanov@aquantia.com> Signed-off-by:
-
Dmitry Bogdanov authored
Packets with marked invalid IP/UDP/TCP checksums were considered as good by the driver. The error was in a logic, processing offload bits in RX descriptor. Signed-off-by:
-
Igor Russkikh authored
Fixed a condition mistake due to which macvlans unicast item number 32 was not added in the unicast filter. The consequence is that when exactly 32 macvlans are created on NIC, the last created macvlan receives no traffic because its MAC was not registered in HW. Fixes: 94b3b542 ("net: aquantia: vlan unicast address list correct handling") Signed-off-by:
Nikita Danilov <nikita.danilov@aquantia.com> Signed-off-by:
-
Dmitry Bogdanov authored
IOMMU fault may occurr on unbind/bind or if_down/if_up sequence. Although driver disables the rings on down, this is not enough. Due to internal HW design, during subsequent initialization NIC sometimes may reuse RX descriptors cache and write to the host memory from the descriptor cache. That's get catched by IOMMU on host. This patch invalidates the descriptor cache in NIC on interface down to prevent writing to the cached descriptors and to the memory pointed in those descriptors. Signed-off-by:
-
Igor Russkikh authored
Flow control statuses were not synchronized between blocks, that caused packets/link drop on some corner cases, when MAC sent PFC although Phy was not expecting these to come. Driver should readout the negotiated FC from phy and configure RX block accordigly. This is done on each link change event with information from FW. Fixes: 288551de ("net: aquantia: Implement rx/tx flow control ethtools callback") Signed-off-by:
-
Stefan Wahren authored
The commit f77f0aee ("net: use core MTU range checking in USB NIC drivers") introduce a common MTU handling for usbnet. But it's missing the necessary changes for smsc95xx. So set the MTU range accordingly. This patch has been tested on a Raspberry Pi 3. Fixes: f77f0aee ("net: use core MTU range checking in USB NIC drivers") Signed-off-by:
Stefan Wahren <stefan.wahren@i2se.com> Signed-off-by:
-
Thor Thayer authored
Ping problems with packets > 8191 as shown: PING 192.168.1.99 (192.168.1.99) 8150(8178) bytes of data. 8158 bytes from 192.168.1.99: icmp_seq=1 ttl=64 time=0.669 ms wrong data byte 8144 should be 0xd0 but was 0x0 16 10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f 20 21 22 23 24 25 26 27 28 29 2a 2b 2c 2d 2e 2f %< ---------------snip-------------------------------------- 8112 b0 b1 b2 b3 b4 b5 b6 b7 b8 b9 ba bb bc bd be bf c0 c1 c2 c3 c4 c5 c6 c7 c8 c9 ca cb cc cd ce cf 8144 0 0 0 0 d0 d1 ^^^^^^^ Notice the 4 bytes of 0 before the expected byte of d0. Databook notes that the RX buffer must be a multiple of 4/8/16 bytes [1]. Update the DMA Buffer size define to 8188 instead of 8192. Remove the -1 from the RX buffer size allocations and use the new DMA Buffer size directly. [1] Synopsys DesignWare Cores Ethernet MAC Universal v3.70a [section 8.4.2 - Table 8-24] Tested on SoCFPGA Stratix10 with ping sweep from 100 to 8300 byte packets. Fixes: 286a8372 ("stmmac: add CHAINED descriptor mode support (V4)") Suggested-by:Jose Abreu <jose.abreu@synopsys.com> Signed-off-by:
Thor Thayer <thor.thayer@linux.intel.com> Signed-off-by:
-
David S. Miller authored
Denis Bolotin says: ==================== qed: Slowpath Queue bug fixes This patch series fixes several bugs in the SPQ mechanism. It deals with SPQ entries management, preventing resource leaks, memory corruptions and handles error cases throughout the driver. Please consider applying to net. ==================== Signed-off-by: -
Sagiv Ozeri authored
A stuck ramrod should be deleted from the completion_pending list, otherwise it will be added again in the future and corrupt the list. Return error value to inform that ramrod is stuck and should be deleted. Signed-off-by:
Sagiv Ozeri <sagiv.ozeri@cavium.com> Signed-off-by:
Denis Bolotin <denis.bolotin@cavium.com> Signed-off-by:
-
Denis Bolotin authored
qed_sp_destroy_request() API was added for SPQ users that need to free/return the entry they acquired in their error flows. Signed-off-by:
Michal Kalderon <michal.kalderon@cavium.com> Signed-off-by:
-
Denis Bolotin authored
When there are no SPQ entries left in the free_pool, new entries are allocated and are added to the unlimited list. When an entry in the pool is available, the content is copied from the original entry, and the new entry is sent to the device. qed_spq_post() is not aware of that, so the additional entry is stored in the original entry as p_post_ent, which can later be returned to the pool. Signed-off-by:
-
Denis Bolotin authored
Free the allocated SPQ entry or return the acquired SPQ entry to the free list in error flows. Signed-off-by:
-
Eric Dumazet authored
Multiple cpus might attempt to insert a new fragment in rhashtable, if for example RPS is buggy, as reported by 배석진 in https://patchwork.ozlabs.org/patch/994601/ We use rhashtable_lookup_get_insert_key() instead of rhashtable_insert_fast() to let cpus losing the race free their own inet_frag_queue and use the one that was inserted by another cpu. Fixes: 648700f7 ("inet: frags: use rhashtables for reassembly units") Signed-off-by:
-
Huazhong Tan authored
hns3_reset_notify_init_enet() only return error early if the return value of hns3_restore_vlan() is not 0. This patch adds checking for the return value of hns3_restore_vlan. Fixes: 7fa6be4f ("net: hns3: fix incorrect return value/type of some functions") Signed-off-by:
Huazhong Tan <tanhuazhong@huawei.com> Signed-off-by:
-
- 08 Nov, 2018 8 commits
-
-
Michał Mirosław authored
VLAN.TCI == 0 is perfectly valid (802.1p), so allow it to be accelerated. Signed-off-by:
Michał Mirosław <mirq-linux@rere.qmqm.pl> Signed-off-by:
-
Michał Mirosław authored
Don't request tag insertion when it isn't present in outgoing skb. Signed-off-by:
-
David S. Miller authored
Maciej W. Rozycki says: ==================== FDDI: defza: Fix a bunch of small issues Here is a bunch of small fixes addressing issues that I missed in my final round of testing. None of these affect run-time behaviour. One was actually found by the kbuild bot, which turned out to be more pedantic than my compiler. See individual change descriptions for details. ==================== Signed-off-by: -
Maciej W. Rozycki authored
The driver version string is obviously not meant to be changed at run time, so mark it `const'. Signed-off-by:
Maciej W. Rozycki <macro@linux-mips.org> Signed-off-by:
-
Maciej W. Rozycki authored
Move the temporary data buffer used when tapping into the SMT Tx queue from the outer function level into the conditional block it's actually used in and its containing skb is also declared, making the structure of code better. Signed-off-by:
-
Maciej W. Rozycki authored
Fix: drivers/net/fddi/defza.h:238:1: warning: "/*" within comment [-Wcomment] by adding a missing comment closing. Signed-off-by:
-
Maciej W. Rozycki authored
The SPDX annotation for this driver does not match the license text, which specifies GNU GPL 2 or later. Make the two match by correcting the SPDX tag. Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-queueDavid S. Miller authored
Jeff Kirsher says: ==================== Intel Wired LAN Driver Updates 2018-11-07 This series contains fixes to igb, i40e and ice drivers. Anirudh fixes an issue during rebuild of the ice driver, where we need to set the carrier state, as well as start or stop the queues all based on the link status. Removed functions that were duplicating current functionality in the VSI rebuild/replay framework. Dave fixes a potential resource collision during the remove path, so add a check to see if we are in the middle of a reset. Fixed the remove path to ensure we call netif_napi_del() to free vectors before we set vsi->netdev to NULL. Akeem fixes an issue when the receive or transmit pause parameter is set, results in link loss on the interface. Fixed the spelling of "Enabling" in error message. Victor fixes potential memory leak by also freeing the related VSI contexts in the unload path. Md Fahad fixes a flag during port VLAN insertion, which was not being set properly. Brett fixes a transmit timeout during stress due to the hardware tail and software tail were incorrectly out of sync. Miroslav Lichvar fixes the igb PHC timecounter update interval to be sure the timecounter is updated in time. Chinh fixes the req_speeds variable to be u16 instead of u8 so that it can handle all the link speeds. Jake fixes i40e to add back the missing feature flags, which was causing IP-in-IP offloads to be reported as not supported. ==================== Signed-off-by:
-
- 07 Nov, 2018 3 commits
-
-
Jacob Keller authored
The assignment of the feature flag NETIF_F_NTUPLE and NETIF_F_HW_TC occurs prior to the initial setup of the local hw_features variable. This means the features are set as user-changeable, but are not set in the currently active feature list. This results in the features being disabled at the driver's initial load. Move the assignment after the initial assignment of hw_features, and assign to the local variable. This ensures that NETIF_F_NTUPLE and NETIF_F_HW_TC are marked as user-changeable, and also enables them by default when the driver loads. Signed-off-by:
Jacob Keller <jacob.e.keller@intel.com> Tested-by:
Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com>
-
Jacob Keller authored
Since commit bacd75cf ("i40e/i40evf: Add capability exchange for outer checksum", 2017-04-06) the i40e driver has not reported support for IP-in-IP offloads. This likely occurred due to a bad rebase, as the commit extracts hw_enc_features into its own variable. As part of this change, it dropped the NETIF_F_FSO_IPXIP flags from the netdev->hw_enc_features. This was unfortunately not caught during code review. Fix this by adding back the missing feature flags. For reference, NETIF_F_GSO_IPXIP4 was added in commit 7e13318d ("net: define gso types for IPx over IPv4 and IPv6", 2016-05-20), replacing NETIF_F_GSO_IPIP and NETIF_F_GSO_SIT. NETIF_F_GSO_IPXIP6 was added in commit bf2d1df3 ("intel: Add support for IPv6 IP-in-IP offload", 2016-05-20). Signed-off-by:
-
Chinh T Cao authored
Since the req_speeds field in struct ice_link_status is a u8, req_speeds & ICE_AQ_LINK_SPEED_40GB always returns 0. This was caught by a coverity scan. Fix this by changing req_speeds to be u16. Reported-by:
Bruce Allan <bruce.w.allan@intel.com> Signed-off-by:
Chinh T Cao <chinh.t.cao@intel.com> Signed-off-by:
Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by:
-
- 06 Nov, 2018 12 commits
-
-
Miroslav Lichvar authored
The timecounter needs to be updated at least once per ~550 seconds in order to avoid a 40-bit SYSTIM timestamp to be misinterpreted as an old timestamp. Since commit 500462a9 ("timers: Switch to a non-cascading wheel"), scheduling of delayed work seems to be less accurate and a requested delay of 540 seconds may actually be longer than 550 seconds. Also, the PHC may be adjusted to run up to 6% faster than real time and the system clock up to 10% slower. Shorten the delay to 360 seconds to be sure the timecounter is updated in time. This fixes an issue with HW timestamps on 82580/I350/I354 being off by ~1100 seconds for few seconds every ~9 minutes. Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by:
Miroslav Lichvar <mlichvar@redhat.com> Acked-by:
Richard Cochran <richardcochran@gmail.com> Tested-by:
Aaron Brown <aaron.f.brown@intel.com> Signed-off-by:
-
Brett Creeley authored
Currently if the driver does a TSO offload the bytecount sent to netdev_tx_sent_queue will be incorrect. This is because in ice_tso we overwrite the initial value that we set in ice_tx_map. This creates a mismatch between the Tx and Tx clean flow. In the Tx clean flow we calculate the bytecount (called total_bytes) as we clean the descriptors so the value used in the Tx clean path is correct. Fix this by using += in ice_tso instead of =. This fixes the mismatch in bytecount mentioned above. Signed-off-by:
Brett Creeley <brett.creeley@intel.com> Signed-off-by:
-
Brett Creeley authored
Prior to this commit the driver was running into tx_timeouts when a queue was stressed enough. This was happening because the HW tail and SW tail (NTU) were incorrectly out of sync. Consequently this was causing the HW head to collide with the HW tail, which to the hardware means that all descriptors posted for Tx have been processed. Due to the Tx logic used in the driver SW tail and HW tail are allowed to be out of sync. This is done as an optimization because it allows the driver to write HW tail as infrequently as possible, while still updating the SW tail index to keep track. However, there are situations where this results in the tail never getting updated, resulting in Tx timeouts. Tx HW tail write condition: if (netif_xmit_stopped(txring_txq(tx_ring) || !skb->xmit_more) writel(sw_tail, tx_ring->tail); An issue was found in the Tx logic that was causing the afore mentioned condition for updating HW tail to never happen, causing tx_timeouts. In ice_xmit_frame_ring we calculate how many descriptors we need for the Tx transaction based on the skb the kernel hands us. This is then passed into ice_maybe_stop_tx along with some extra padding to determine if we have enough descriptors available for this transaction. If we don't then we return -EBUSY to the stack, otherwise we move on and eventually prepare the Tx descriptors accordingly in ice_tx_map and set next_to_watch. In ice_tx_map we make another call to ice_maybe_stop_tx with a value of MAX_SKB_FRAGS + 4. The key here is that this value is possibly less than the value we sent in the first call to ice_maybe_stop_tx in ice_xmit_frame_ring. Now, if the number of unused descriptors is between MAX_SKB_FRAGS + 4 and the value used in the first call to ice_maybe_stop_tx in ice_xmit_frame_ring then we do not update the HW tail because of the "Tx HW tail write condition" above. This is because in ice_maybe_stop_tx we return success from ice_maybe_stop_tx instead of calling __ice_maybe_stop_tx and subsequently calling netif_stop_subqueue, which sets the __QUEUE_STATE_DEV_XOFF bit. This bit is then checked in the "Tx HW tail write condition" by calling netif_xmit_stopped and subsequently updating HW tail if the afore mentioned bit is set. In ice_clean_tx_irq, if next_to_watch is not NULL, we end up cleaning the descriptors that HW sets the DD bit on and we have the budget. The HW head will eventually run into the HW tail in response to the description in the paragraph above. The next time through ice_xmit_frame_ring we make the initial call to ice_maybe_stop_tx with another skb from the stack. This time we do not have enough descriptors available and we return NETDEV_TX_BUSY to the stack and end up setting next_to_watch to NULL. This is where we are stuck. In ice_clean_tx_irq we never clean anything because next_to_watch is always NULL and in ice_xmit_frame_ring we never update HW tail because we already return NETDEV_TX_BUSY to the stack and eventually we hit a tx_timeout. This issue was fixed by making sure that the second call to ice_maybe_stop_tx in ice_tx_map is passed a value that is >= the value that was used on the initial call to ice_maybe_stop_tx in ice_xmit_frame_ring. This was done by adding the following defines to make the logic more clear and to reduce the chance of mucking this up again: ICE_CACHE_LINE_BYTES 64 ICE_DESCS_PER_CACHE_LINE (ICE_CACHE_LINE_BYTES / \ sizeof(struct ice_tx_desc)) ICE_DESCS_FOR_CTX_DESC 1 ICE_DESCS_FOR_SKB_DATA_PTR 1 The ICE_CACHE_LINE_BYTES being 64 is an assumption being made so we don't have to figure this out on every pass through the Tx path. Instead I added a sanity check in ice_probe to verify cache line size and print a message if it's not 64 Bytes. This will make it easier to file issues if they are seen when the cache line size is not 64 Bytes when reading from the GLPCI_CNF2 register. Signed-off-by:
-
Dave Ertman authored
In the remove path, the vsi->netdev is being set to NULL before the call to free vectors. This is causing the netif_napi_del call to never be made. Add a call to ice_napi_del to the same location as the calls to unregister_netdev and just prior to them. This will use the reverse flow as the register and netif_napi_add calls. Signed-off-by:
Dave Ertman <david.m.ertman@intel.com> Signed-off-by:
-
Anirudh Venkataramanan authored
Print should say "Enabling" instead of "Enaabling" Signed-off-by:
Akeem G Abodunrin <akeem.g.abodunrin@intel.com> Signed-off-by:
-
Md Fahad Iqbal Polash authored
According to the spec, whenever insert PVID field is set, the VLAN driver insertion mode should be set to 01b which isn't done currently. Fix it. Signed-off-by:
Md Fahad Iqbal Polash <md.fahad.iqbal.polash@intel.com> Signed-off-by:
-
Anirudh Venkataramanan authored
ice_restore_vlan and active_vlans were originally put in place to reprogram VLAN filters in the replay path. This is now done as part of the much broader VSI rebuild/replay framework. So remove both ice_restore_vlan and active_vlans Signed-off-by:
-
Victor Raj authored
In the unload path, all VSIs are freed. Also free the related VSI contexts to prevent memory leaks. Signed-off-by:
Victor Raj <victor.raj@intel.com> Signed-off-by:
-
Akeem G Abodunrin authored
Setting Rx or Tx pause parameter currently results in link loss on the interface, requiring the platform/host to be cold power cycled. Fix it. Signed-off-by:
-
Anirudh Venkataramanan authored
The remove path does not currently check to see if a reset is in progress before proceeding. This can cause a resource collision resulting in various types of errors. Check for reset in progress and wait for a reasonable amount of time before allowing the remove to progress. Signed-off-by:
-
Anirudh Venkataramanan authored
Set the carrier state post rebuild by querying the link status. Also start/stop queues based on link status. Signed-off-by:
-
Tao Ren authored
This patch allows users to enable/disable internal TX and/or RX clock delay for BCM54616S PHYs so as to satisfy RGMII timing specifications. On a particular platform, whether TX and/or RX clock delay is required depends on how PHY connected to the MAC IP. This requirement can be specified through "phy-mode" property in the platform device tree. The patch is inspired by commit 73333626 ("net: phy: Allow BCM5481x PHYs to setup internal TX/RX clock delay"). Signed-off-by:
Tao Ren <taoren@fb.com> Reviewed-by:
Florian Fainelli <f.fainelli@gmail.com> Signed-off-by:
-
