- 29 Oct, 2016 37 commits
-
-
Tariq Toukan authored
Following the previous patch, as an optimization, the slave will not even bother sending the DUMP_ETH_STATS command over the comm channel. Signed-off-by:
Tariq Toukan <tariqt@mellanox.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Jack Morgenstein authored
mlx4_en_DUMP_ETH_STATS took the *counter mutex* and then called the FW command, with WRAPPED attribute. As a result, the fw command is wrapped on the Hypervisor when it calls mlx4_en_DUMP_ETH_STATS. The FW command wrapper flow on the hypervisor takes the *slave_cmd_mutex* during processing. At the same time, a VF could be in the process of coming up, and could call mlx4_QUERY_FUNC_CAP. On the hypervisor, the command flow takes the *slave_cmd_mutex*, then executes mlx4_QUERY_FUNC_CAP_wrapper. mlx4_QUERY_FUNC_CAP wrapper calls mlx4_get_default_counter_index(), which takes the *counter mutex*. DEADLOCK. The fix is that the DUMP_ETH_STATS fw command should be called with the NATIVE attribute, so that on the hypervisor, this command does not enter the wrapper flow. Since the Hypervisor no longer goes through the wrapper code, we also simply return 0 in mlx4_DUMP_ETH_STATS_wrapper (i.e.the function succeeds, but the returned data will be all zeroes). No need to test if it is the Hypervisor going through the wrapper. Fixes: f9baff50 ("mlx4_core: Add "native" argument to mlx4_cmd ...") Signed-off-by:
Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by:
-
Eugenia Emantayev authored
Currently interrupt test that is part of ethtool selftest runs the check over all interrupt vectors of the device. In mlx4_en package part of interrupt vectors are uninitialized since mlx4_ib doesn't exist. This causes NOP FW command to time out. Change logic to test current port interrupt vectors only. Signed-off-by:
Eugenia Emantayev <eugenia@mellanox.com> Signed-off-by:
-
Jack Morgenstein authored
In the Hypervisor, there are several FW commands which are invoked before the comm channel is initialized (in mlx4_multi_func_init). These include MOD_STAT_CONFIG, QUERY_DEV_CAP, INIT_HCA, and others. If any of these commands fails, say with a timeout, the Hypervisor driver enters the internal error reset flow. In this flow, the driver attempts to notify all slaves via the comm channel that an internal error has occurred. Since the comm channel has not yet been initialized (i.e., mapped via ioremap), this will cause dereferencing a NULL pointer. To fix this, do not access the comm channel in the internal error flow if it has not yet been initialized. Fixes: 55ad3592 ("net/mlx4_core: Enable device recovery flow with SRIOV") Fixes: ab9c17a0 ("mlx4_core: Modify driver initialization flow to accommodate SRIOV for Ethernet") Signed-off-by:
-
Eugenia Emantayev authored
Fix a kernel panic that occurs as a result of an asynchronous event handled in roce_gid_mgmt: mlx4_en_get_drvinfo is called and accesses freed resources. This happens in a shutdown flow only, since pci device is destroyed while netdevice is still alive. Fixes: c27a02cd ("mlx4_en: Add driver for Mellanox ConnectX 10GbE NIC") Signed-off-by:
-
Erez Shitrit authored
Currently there is a race between incoming traffic and initialization flow. HW is able to receive the packets after INIT_PORT is done and unicast steering is configured. Before we set priv->port_up NAPI is not scheduled and receive queues become full. Therefore we never get new interrupts about the completions. This issue could happen if running heavy traffic during bringing port up. The resolution is to schedule NAPI once port_up is set. If receive queues were full this will process all cqes and release them. Fixes: c27a02cd ("mlx4_en: Add driver for Mellanox ConnectX 10GbE NIC") Signed-off-by:
Erez Shitrit <erezsh@mellanox.com> Signed-off-by:
-
Eugenia Emantayev authored
When doing roundup_pow_of_two for large enough number with bit 31, an overflow will occur and a value equal to 1 will be returned. In this case 1 will be subtracted from the return value and division by zero will be reached. Fixes: 31c128b6 ("net/mlx4_en: Choose time-stamping shift value according to HW frequency") Signed-off-by:
-
Moshe Lazer authored
Change the default status of quality of service back to disabled, as it hurts performance in some cases. Fixes: 38438f7c ("net/mlx4: Set enhanced QoS support by default when ...") Signed-off-by:
Moshe Lazer <moshel@mellanox.com> Signed-off-by:
-
Maor Gottlieb authored
When only one port type is supported, it should be read only. We reject changing requests, even to the auto sense mode. Fixes: 27bf91d6 ("mlx4_core: Add link type autosensing") Signed-off-by:
Maor Gottlieb <maorg@mellanox.com> Signed-off-by:
Moshe Shemesh <moshe@mellanox.com> Signed-off-by:
-
Jack Morgenstein authored
The resource type enum in the resource tracker was incorrect. RES_EQ was put in the position of RES_NPORT_ID (a FC resource). Since the remaining resources maintain their current values, and RES_EQ is not passed from slaves to the hypervisor in any FW command, this change affects only the hypervisor. Therefore, there is no backwards-compatibility issue. Fixes: 623ed84b ("mlx4_core: initial header-file changes for SRIOV support") Signed-off-by:
-
shamir rabinovitch authored
rds use Kconfig option called "RDS_DEBUG" to enable rds debug messages. This option cause the rds Makefile to add -DDEBUG to the rds gcc command line. When CONFIG_DYNAMIC_DEBUG is enabled, the "DEBUG" macro is used by include/linux/dynamic_debug.h to decide if dynamic debug prints should be sent by default to the kernel log. rds should not enable this macro for production builds. rds dynamic debug work as expected follow this fix. Signed-off-by:
Shamir Rabinovitch <shamir.rabinovitch@oracle.com> Acked-by:
Santosh Shilimkar <santosh.shilimkar@oracle.com> Reviewed-by:
Wengang Wang <wen.gang.wang@oracle.com> Signed-off-by:
-
David S. Miller authored
Merge tag 'mac80211-for-davem-2016-10-27' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211 Johannes Berg says: ==================== Just two fixes: * a fix to process all events while suspending, so any potential calls into the driver are done before it is suspended * small markup fixes for the sphinx documentation conversion that's coming into the tree via the doc tree ==================== Signed-off-by:
-
Thomas Falcon authored
Schedule these XPORT event tasks in the shared workqueue so that IRQs are not freed in an interrupt context when sub-CRQs are released. Signed-off-by:
Thomas Falcon <tlfalcon@linux.vnet.ibm.com> Signed-off-by:
-
Jason Gunthorpe authored
The MAC is capable of RGMII mode and that is probably a more typical connection type than GMII today (eg it is used by Marvell Reference designs for several SOCs). Let DT users specify the standard phy-connection-type = "rgmii-id"; On a phy node. Signed-off-by:
Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Reviewed-by:
Florian Fainelli <f.fainelli@gmail.com> Signed-off-by:
-
git://git.open-mesh.org/linux-mergeDavid S. Miller authored
Simon Wunderlich says: ==================== Here are three batman-adv bugfix patches: - Fix RCU usage for neighbor list, by Sven Eckelmann - Fix BATADV_DBG_ALL loglevel to include TP Meter messages, by Sven Eckelmann - Fix possible splat when disabling an interface, by Linus Luessing ==================== Signed-off-by: -
Stephen Hemminger authored
This reverts commit e3f74b84 ("hv_netvsc: report vmbus name in ethtool")' because of problem introduced by commit f9a56e5d6a0ba ("Drivers: hv: make VMBus bus ids persistent"). This changed the format of the vmbus name and this new format is too long to fit in the bus_info field of ethtool. Signed-off-by:
Stephen Hemminger <sthemmin@microsoft.com> Signed-off-by:
-
Willem de Bruijn authored
When transmitting on a packet socket with PACKET_VNET_HDR and PACKET_QDISC_BYPASS, validate device support for features requested in vnet_hdr. Drop TSO packets sent to devices that do not support TSO or have the feature disabled. Note that the latter currently do process those packets correctly, regardless of not advertising the feature. Because of SKB_GSO_DODGY, it is not sufficient to test device features with netif_needs_gso. Full validate_xmit_skb is needed. Switch to software checksum for non-TSO packets that request checksum offload if that device feature is unsupported or disabled. Note that similar to the TSO case, device drivers may perform checksum offload correctly even when not advertising it. When switching to software checksum, packets hit skb_checksum_help, which has two BUG_ON checksum not in linear segment. Packet sockets always allocate at least up to csum_start + csum_off + 2 as linear. Tested by running github.com/wdebruij/kerneltools/psock_txring_vnet.c ethtool -K eth0 tso off tx on psock_txring_vnet -d $dst -s $src -i eth0 -l 2000 -n 1 -q -v psock_txring_vnet -d $dst -s $src -i eth0 -l 2000 -n 1 -q -v -N ethtool -K eth0 tx off psock_txring_vnet -d $dst -s $src -i eth0 -l 1000 -n 1 -q -v -G psock_txring_vnet -d $dst -s $src -i eth0 -l 1000 -n 1 -q -v -G -N v2: - add EXPORT_SYMBOL_GPL(validate_xmit_skb_list) Fixes: d346a3fa ("packet: introduce PACKET_QDISC_BYPASS socket option") Signed-off-by:
Willem de Bruijn <willemb@google.com> Acked-by:
Eric Dumazet <edumazet@google.com> Acked-by:
Daniel Borkmann <daniel@iogearbox.net> Signed-off-by:
-
Johannes Berg authored
Use nla_parse_nested instead of open-coding the call to nla_parse() with the attribute data/len. Signed-off-by:
Johannes Berg <johannes.berg@intel.com> Acked-by:
Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by:
-
Ganesh Goudar authored
Fix to release resources properly in error handling path of alloc_uld_rxqs(), This patch also removes unwanted arguments and avoids calling the same function twice. Fixes: 94cdb8bb (cxgb4: Add support for dynamic allocation of resources for ULD Signed-off-by:
Ganesh Goudar <ganeshgr@chelsio.com> Signed-off-by:
-
Arnd Bergmann authored
There is an old warning about mlx4_SW2HW_EQ_wrapper on x86: ethernet/mellanox/mlx4/resource_tracker.c: In function ‘mlx4_SW2HW_EQ_wrapper’: ethernet/mellanox/mlx4/resource_tracker.c:3071:10: error: ‘eq’ may be used uninitialized in this function [-Werror=maybe-uninitialized] The problem here is that gcc won't track the state of the variable across a spin_unlock. Moving the assignment out of the lock is safe here and avoids the warning. Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Reviewed-by:
Yishai Hadas <yishaih@mellanox.com> Signed-off-by:
-
Eli Cooper authored
This patch updates skb->protocol to ETH_P_IPV6 in ip6_tnl_xmit() when an IPv6 header is installed to a socket buffer. This is not a cosmetic change. Without updating this value, GSO packets transmitted through an ipip6 tunnel have the protocol of ETH_P_IP and skb_mac_gso_segment() will attempt to call gso_segment() for IPv4, which results in the packets being dropped. Fixes: b8921ca8 ("ip4ip6: Support for GSO/GRO") Signed-off-by:
Eli Cooper <elicooper@gmx.com> Signed-off-by:
-
Daniel Borkmann authored
Some of the sample files are causing issues when they are loaded with tc and cls_bpf, meaning tc bails out while trying to parse the resulting ELF file as program/map/etc sections are not present, which can be easily spotted with readelf(1). Currently, BPF samples are including some of the kernel headers and mid term we should change them to refrain from this, really. When dynamic debugging is enabled, we bail out due to undeclared KBUILD_MODNAME, which is easily overlooked in the build as clang spills this along with other noisy warnings from various header includes, and llc still generates an ELF file with mentioned characteristics. For just playing around with BPF examples, this can be a bit of a hurdle to take. Just add a fake KBUILD_MODNAME as a band-aid to fix the issue, same is done in xdp*_kern samples already. Fixes: 65d472fb ("samples/bpf: add 'pointer to packet' tests") Fixes: 6afb1e28 ("samples/bpf: Add tunnel set/get tests.") Fixes: a3f74617 ("cgroup: bpf: Add an example to do cgroup checking in BPF") Reported-by:
Chandrasekar Kannan <ckannan@console.to> Signed-off-by:
Alexei Starovoitov <ast@kernel.org> Signed-off-by:
-
Craig Gallek authored
As part of a series to implement faster SO_REUSEPORT lookups, commit 086c653f ("sock: struct proto hash function may error") added return values to protocol hash functions and commit 496611d7 ("inet: create IPv6-equivalent inet_hash function") implemented a new hash function for IPv6. However, the latter does not respect the former's convention. This properly propagates the hash errors in the IPv6 case. Fixes: 496611d7 ("inet: create IPv6-equivalent inet_hash function") Reported-by:
Soheil Hassas Yeganeh <soheil@google.com> Signed-off-by:
Craig Gallek <kraig@google.com> Acked-by:
-
David S. Miller authored
Saeed Mahameed says: ==================== Mellanox 100G mlx5 fixes 2016-10-25 This series contains some bug fixes for the mlx5 core and mlx5e driver. From Daniel: - Cache line size determination at runtime, instead of using L1_CACHE_BYTES hard coded value, use cache_line_size() - Always Query HCA caps after setting them even on reset flow From Mohamad: - Reorder netdev cleanup to uregister netdev before detaching it for the kernel to not complain about open resources such as vlans - Change the acl enable prototype to return status, for better error resiliency - Clear health sick bit when starting health poll after reset flow - Fix race between PCI error handlers and health work - PCI error recovery health care simulation, in case when the kernel PCI error handlers are not triggered for some internal firmware errors From Noa: - Avoid passing dma address 0 to firmware when mapping system pages to the firmware From Paul: Some straight forward flow steering fixes - Keep autogroups list ordered - Fix autogroups groups num not decreasing - Correctly initialize last use of flow counters From Saeed: - Choose the nearest LRO timeout to the wanted one instead of blindly choosing "dev_cap.lro_timeout[2]" This series has no conflict with the for-next pull request posted earlier today ("Mellanox mlx5 core driver updates 2016-10-25"). ==================== Signed-off-by: -
Noa Osherovich authored
Currently the firmware can't work with a page with dma address 0. Passing such an address to the firmware will cause the give_pages command to fail. To avoid this, in case we get a 0 dma address of a page from the dma engine, we avoid passing it to FW by remapping to get an address other than 0. Fixes: bf0bf77f ('mlx5: Support communicating arbitrary host...') Signed-off-by:
Noa Osherovich <noaos@mellanox.com> Signed-off-by:
Saeed Mahameed <saeedm@mellanox.com> Signed-off-by:
-
Mohamad Haj Yahia authored
In case that the kernel PCI error handlers are not called, we will trigger our own recovery flow. The health work will give priority to the kernel pci error handlers to recover the PCI by waiting for a small period, if the pci error handlers are not triggered the manual recovery flow will be executed. We don't save pci state in case of manual recovery because it will ruin the pci configuration space and we will lose dma sync. Fixes: 89d44f0a ('net/mlx5_core: Add pci error handlers to mlx5_core driver') Signed-off-by:
Mohamad Haj Yahia <mohamad@mellanox.com> Signed-off-by:
-
Mohamad Haj Yahia authored
Currently there is a race between the health care work and the kernel pci error handlers because both of them detect the error, the first one to be called will do the error handling. There is a chance that health care will disable the pci after resuming pci slot. Also create a separate WQ because now we will have two types of health works, one for the error detection and one for the recovery. Fixes: 89d44f0a ('net/mlx5_core: Add pci error handlers to mlx5_core driver') Signed-off-by:
-
Mohamad Haj Yahia authored
The health sick status should be cleared when we start the health poll. This is crucial for driver reload (unload + load) in order to behave right in case of health issue. Fixes: fd76ee4d ('net/mlx5_core: Fix internal error detection conditions') Signed-off-by:
-
Mohamad Haj Yahia authored
The Ingress/Egress ACL enable function may fail and it should return status to its caller to avoid NULL pointer dereference. Fixes: f942380c ('net/mlx5: E-Switch, Vport ingress/egress ACLs rules for spoofchk') Signed-off-by:
-
Mohamad Haj Yahia authored
Detaching the netdev before unregistering it cause some netdev cleanup ndos to fail because they check presence of the netdev, so we need to unregister the netdev first. Fixes: 26e59d80 ('net/mlx5e: Implement mlx5e interface attach/detach callbacks') Signed-off-by:
-
Saeed Mahameed authored
Instead of predicting the index of the wanted LRO timeout value from hardware capabilities, look for the nearest LRO timeout value. Fixes: 5c50368f ('net/mlx5e: Light-weight netdev open/stop') Signed-off-by:
-
Paul Blakey authored
Currently, last use timestamp is initialized to zero. This is not the expected value by higher layers such as when we do TC action offloading. To fix that, set it to the current time, e.g when the counter/rule is offloaded. This is the same behaviour of non-offloaded TC actions. Fixes: 43a335e0 ('mlx5_core: Flow counters infrastructure') Signed-off-by:
Paul Blakey <paulb@mellanox.com> Signed-off-by:
-
Paul Blakey authored
Autogroups groups num is increased when creating a new flow group, but is never decreased. Now decreasing it when deleting a flow group. Fixes: f0d22d18 ('net/mlx5_core: Introduce flow steering autogrouped flow table') Signed-off-by:
-
Paul Blakey authored
Finding a new autogroup range is done by going over a group list sorted by each group start index. The search is stopped after finding the first free range. Adding the newly created group to the list is wrongly added to the end of the list regardless of its start index as the parameter of where to insert it is ignored. This commit makes sure to use that unused parameter to insert it where requested. Fixes: f0d22d18 ('net/mlx5_core: Introduce flow steering autogrouped flow table') Signed-off-by:
-
Daniel Jurgens authored
Always query the HCA caps after setting them to update the capablities data structures. Not doing so results in incorrect capabilities being reported including max_dc, max_qp and several others. Fixes: 59211bd3 ("net/mlx5: Split the load/unload flow into hardware and software flows") Signed-off-by:
Daniel Jurgens <danielj@mellanox.com> Signed-off-by:
-
Daniel Jurgens authored
ARM 64B cache line systems have L1_CACHE_BYTES set to 128. cache_line_size() will return the correct size. Fixes: cf50b5efa2fe('net/mlx5_core/ib: New device capabilities handling.') Signed-off-by: -
Marcelo Ricardo Leitner authored
Andrey Konovalov reported that KASAN detected that SCTP was using a slab beyond the boundaries. It was caused because when handling out of the blue packets in function sctp_sf_ootb() it was checking the chunk len only after already processing the first chunk, validating only for the 2nd and subsequent ones. The fix is to just move the check upwards so it's also validated for the 1st chunk. Reported-by:
Andrey Konovalov <andreyknvl@google.com> Tested-by:
Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Reviewed-by:
Xin Long <lucien.xin@gmail.com> Acked-by:
Neil Horman <nhorman@tuxdriver.com> Signed-off-by:
-
- 28 Oct, 2016 3 commits
-
-
David S. Miller authored
Jiri Pirko says: ==================== mlxsw: Couple of fixes Couple of LPM tree management fixes. ==================== Signed-off-by: -
Jiri Pirko authored
Only trees which are in use should be compared to requested prefix usage. Fixes: 53342023 ("mlxsw: spectrum_router: Implement LPM trees management") Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Reviewed-by:
Ido Schimmel <idosch@mellanox.com> Signed-off-by:
-
Jiri Pirko authored
Currently, the prefix bitlist is not saved for LPM trees, causing the compare to always fail which causes the tree to be destroyed and created for every inserted and removed FIB entry. So fix this by saving the bitlist as it should have been done from the very beginning. Fixes: 53342023 ("mlxsw: spectrum_router: Implement LPM trees management") Signed-off-by:
-
