- 02 May, 2014 2 commits
-
-
Grygorii Strashko authored
Add a resource managed devm_mdiobus_alloc[_size]()/devm_mdiobus_free() to automatically clean up MDIO bus alocations made by MDIO drivers, thus leading to simplified MDIO drivers code. Cc: Florian Fainelli <f.fainelli@gmail.com> Cc: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Acked-and-tested-by:
Lad, Prabhakar <prabhakar.csengg@gmail.com> Signed-off-by:
Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Alexey Charkov authored
This adds two new flags to quirks and thus removes the need to carry revision in rhine_private. As a result, the init logic is simplified a bit. This also fixes a compiler warning in OF code on 64bit due to pointer casting: drivers/net/ethernet/via/via-rhine.c: In function ‘rhine_init_one_platform’: drivers/net/ethernet/via/via-rhine.c:1132:13: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast] revision = (u32)match->data; ^ That code was added in commit 2d283862 ("net: via-rhine: add OF bus binding"). Tested in platform configuration on a VIA WM8950 APC Rock board. Reported-by:Jan Moskyto Matejka <mq@suse.cz> Signed-off-by:
Alexey Charkov <alchark@gmail.com> Signed-off-by:
-
- 30 Apr, 2014 7 commits
-
-
KY Srinivasan authored
We send packets using a copy-free mechanism (this is the Guest to Host transport via VMBUS). While this is obviously optimal for large packets, it may not be optimal for small packets. Hyper-V host supports a second mechanism for sending packets that is "copy based". We implement that mechanism in this patch. In this version of the patch I have addressed a comment from David Miller. With this patch (and all of the other offload and VRSS patches), we are now able to almost saturate a 10G interface between Linux VMs on Hyper-V on different hosts - close to 9 Gbps as measured via iperf. Signed-off-by:
K. Y. Srinivasan <kys@microsoft.com> Reviewed-by:
Haiyang Zhang <haiyangz@microsoft.com> Signed-off-by:
-
Dinh Nguyen authored
mii_irq[] array is never initialized anywhere in the driver, thus mii_irq[] will always equate to zero. So, for the case where the PHY does not have an irq, we should use PHY_POLL for that situation. Signed-off-by:
Dinh Nguyen <dinguyen@altera.com> Tested-by:
Vince Bridgers <vbridger@altera.com> Signed-off-by:
-
Zhangjie \(HZ\) authored
This is a small supplement for commit e7428e95 ("virtio-net: put virtio-net header inline with data"). TCP packages have enough room to put virtio-net header in, but UDP packages do not. By setting dev->needed_headroom for virtio-net device, UDP packages could have enough room. For UDP packages, sk_buff is alloced in fun __ip_append_data. The size is "alloclen + hh_len + 15", and "hh_len = LL_RESERVED_SPACE(rt-dst.dev);". The Macro is defined as follows: #define LL_RESERVED_SPACE(dev) \ ((((dev)->hard_header_len+(dev)->needed_headroom)\ &~(HH_DATA_MOD - 1)) + HH_DATA_MOD) By default, for UDP packages, after skb is allocated, only 16 bytes reserved. And 2 bytes remained after mac header is set. That is not enough to put virtio-net header in. If we set dev->needed_headroom to 12 or 10 (according to mergeable_rx_bufs is on or off ), more room can be reserved. Then there is enough room for UDP packages to put the header in. test result list as below: guest and host: suse11sp3, netperf, intel 2.4GHz +-------+---------+---------+---------+---------+ | | old | new | +-------+---------+---------+---------+---------+ | UDP | Gbit/s | pps | Gbit/s | pps | | 64 | 0.57 | 692232 | 0.61 | 742420 | | 256 | 1.60 | 686860 | 1.71 | 733331 | | 512 | 2.92 | 674576 | 3.07 | 710446 | | 1024 | 4.99 | 598977 | 5.17 | 620821 | | 1460 | 5.68 | 483757 | 7.16 | 610519 | | 4096 | 6.98 | 637468 | 7.21 | 658471 | +-------+---------+---------+---------+---------+ Signed-off-by:
Zhang Jie <zhangjie14@huawei.com> Acked-by:
Rusty Russell <rusty@rustcorp.com.au> Acked-by:
Jason Wang <jasowang@redhat.com> Acked-by:
Michael S. Tsirkin <mst@redhat.com> Signed-off-by:
-
Ying Xue authored
Commit 1bb8dce5 ("tipc: fix memory leak during module removal") introduced a memory leak issue: when name table is stopped, it's forgotten that publication instances are freed properly. Additionally the useless "continue" statement in tipc_nametbl_stop() is removed as well. Reported-by:
Jason <huzhijiang@gmail.com> Signed-off-by:
Ying Xue <ying.xue@windriver.com> Acked-by:
Erik Hugne <erik.hugne@ericsson.com> Signed-off-by:
-
Lorenzo Colitti authored
This replaces 6 identical code snippets with a call to a new static inline function. Signed-off-by:
Lorenzo Colitti <lorenzo@google.com> Signed-off-by:
-
Florian Fainelli authored
Use the helper function to retrieve the driver private context instead of using (void *)(ds + 1). Signed-off-by:
Florian Fainelli <f.fainelli@gmail.com> Signed-off-by:
-
Florian Fainelli authored
DSA drivers have a trick which consists in allocating "priv_size" more bytes to account for the DSA driver private context. Add a helper function to access that private context instead of open-coding it in drivers with (void *)(ds + 1). Signed-off-by:
-
- 28 Apr, 2014 31 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-nextDavid S. Miller authored
Jeff Kirsher says: ==================== Intel Wired LAN Driver Updates This series contains updates to i40e and i40evf. Jacob provides a i40e patch to remove ptp_tx_work() because it conflicts with the interrupt-based handler and will cause spurious warning messages to be displayed in the kernel log. Since the hardware can properly trigger an interrupt, we do not need to poll for an available timestamp in a work queue any more. Greg provides two patches for i40e, first to retain the MAC filter list when changing a port VLAN. Second fixes an issue where the function to set the VF MAC address was not taking the port VLAN filter into account when setting/clearing/resetting the VF's host administered MAC address. Mitch provides three patches for i40evf, first adds support for getting and setting RSS hashing options, RSS LUT entries, and getting the number of device channels from ethtool. Fixes a panic that woudl occur in the VF if the PF driver failed or was removed from the host kernel. Shannon adds driver version string to the driver version command so that can be passed down through the firmware to low-level NC-SI functions. Neerav provides several patches, first redistributes queue vectors after DCB reconfiguration because the total number of queues that are configured for a VSI may change based on a change in the total number of TCs enabled as a result of a change in the DCB configuration based on DCBX. Second, allows the driver probe to continue when querying DCB config fails. Lastly, provides changes to retrieve and store missing link config information that was not being cached in the struct i40e_link_status. Kevin provides a couple of patches, first to change a int variable type to u16 to avoid a typecheck failure. Second to update a function parameters to remove an argument which is no longer used. ==================== Signed-off-by: -
Ying Xue authored
Commit a89778d8 ("tipc: add support for link state subscriptions") introduced below possible deadlock scenario: CPU0 CPU1 T0: tipc_publish() link_timeout() T1: tipc_nametbl_publish() [grab node lock]* T2: [grab nametbl write lock]* link_state_event() T3: named_cluster_distribute() link_activate() T4: [grab node lock]* tipc_node_link_up() T5: tipc_nametbl_publish() T6: [grab nametble write lock]* The opposite order of holding nametbl write lock and node lock on above two different paths may result in a deadlock. If we move the the delivery of named messages via link out of name nametbl lock, the reverse order of holding locks will be eliminated, as a result, the deadlock will be killed as well. Signed-off-by:
-
Erik Hugne authored
Commit 78acb1f9 ("tipc: add ioctl to fetch link names") introduced a buffer overflow bug where specially crafted ioctl requests could cause out-of-bounds indexing of the node->links array. This was caused by an incorrect check vs MAX_BEARERS, and the static code checker complaint is: net/tipc/node.c:459 tipc_node_get_linkname() error: buffer overflow 'node->links' 2 <= 2 Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
-
Frank Blaschka authored
PTR_RET is deprecated. Do not recommend its usage anymore. Use PTR_ERR_OR_ZERO instead. Signed-off-by:
Frank Blaschka <frank.blaschka@de.ibm.com> Signed-off-by:
-
Frank Blaschka authored
PTR_RET is deprecated. Do not recommend its usage anymore. Use PTR_ERR_OR_ZERO instead. Signed-off-by:
-
Stefan Raspl authored
The existing options for priority queueing within OSA devices were based on the now partially defunct TOS field. This patch adds two new options as follows: - prio_queueing_skb: bases priority queueing on skb-priority - prio_queueing_vlan: uses the priority code point in the 802.1Q header Signed-off-by:
Stefan Raspl <raspl@linux.vnet.ibm.com> Signed-off-by:
Ursula Braun <ursula.braun@de.ibm.com> Signed-off-by:
-
Stefan Raspl authored
Make the current priority queueing logic apply to IPv6 traffic. Signed-off-by:
-
Duan Jiong authored
PTR_RET is deprecated. Do not recommend its usage anymore. Use PTR_ERR_OR_ZERO instead. Signed-off-by:
Duan Jiong <duanj.fnst@cn.fujitsu.com> Signed-off-by:
Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by:
-
Duan Jiong authored
PTR_RET is deprecated. Do not recommend its usage anymore. Use PTR_ERR_OR_ZERO instead. Signed-off-by:
-
Stefan Raspl authored
The default queue for queueing modes prio_queueing_prec and prio_queueing_tos is supposed to be QETH_DEFAULT_QUEUE. However, neither mode will reset the default default_out_queue value when enabled, leaving a previously set value activate as the default. E.g. enabling one of the fixed queueing modes, e.g. no_prio_queueing:0, and successively switching to any of the prio_queueing_* modes will leave default_out_queue at the previous (wrong) value 0. Signed-off-by:
-
Richard Cochran authored
PTP Hardware Clock drivers specify a maximum frequency adjustment that their clocks can accommodate. Normally, user space programs will want to respect the advertised limits. However, no kernel or driver code checks that the dialed frequency offset is within the bounds, and out of range values can lead to surprising results. This patch fixes the issue by rejecting bad values. Signed-off-by:
Richard Cochran <richardcochran@gmail.com> Signed-off-by:
-
Hisao Tanabe authored
Signed-off-by:
Hisao Tanabe <xtanabe@gmail.com> Signed-off-by:
-
Jean Sacren authored
The commit 3de0b592 ("ethtool: Support for configurable RSS hash key") introduced a new function ethtool_copy_validate_indir() with full iteration of the loop to validate the ring indices, which could be an overkill. To minimize the impact, we ought to exit the loop as soon as the invalid index occurs for the very first time. The remaining loop simply doesn't serve any more purpose. Signed-off-by:
Jean Sacren <sakiwit@gmail.com> Cc: Venkata Duvvuru <VenkatKumar.Duvvuru@Emulex.Com> Signed-off-by:
-
Catherine Sullivan authored
Bump i40e to 0.3.46 and i40evf to 0.9.23. Change-ID: Ia604ae6d513d9aaa8bfdac79665d9a3a72507df7 Signed-off-by:
Catherine Sullivan <catherine.sullivan@intel.com> Tested-by:
Kavindya Deegala <kavindya.s.deegala@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com>
-
Neerav Parikh authored
Some information returned via "Get Link Status" command was not being cached in the struct i40e_link_status. Add this so the driver can utilize this information as needed. Change-ID: If084a0ae2a63b16b97572196b993742af2c67772 Signed-off-by:
Neerav Parikh <neerav.parikh@intel.com> Acked-by:
Greg Rose <gregory.v.rose@intel.com> Tested-by:
-
Kevin Scott authored
Change the formal parameters to remove argument which is no longer used. Change-ID: I493b56eba74a1f396fa1b7d24c0a4acbe536b5bf Signed-off-by:
Kevin Scott <kevin.c.scott@intel.com> Acked-by:
Shannon Nelson <shannon.nelson@intel.com> Tested-by:
-
Anjali Singhai Jain authored
Ethtool -k/-K can handle this so we do not need it in debugfs. Change-ID: I9df692f10a60b71805f8f48d3b87c7da3820b2aa Signed-off-by:
Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by:
-
Anjali Singhai Jain authored
Although NTUPLE feature control is disabled by default, do not allow user to be able to change it. Change-ID: I4c3f95fb6dbc56b1b6cc47116ea1f72de02bc99c Signed-off-by:
-
Kevin Scott authored
Change the variable type to avoid compiler warning about int to u16 possible data truncation. Change-ID: I5eb3b578c86513c9625ca32d2f0b57cc01d7dc98 Signed-off-by:
-
Mitch Williams authored
If the PF driver fails or is removed from the host, the VF driver will fill up its log with this message. Change-ID: I67045f987f7c0d444d21ded403adc509343cdb8f Signed-off-by:
Mitch Williams <mitch.a.williams@intel.com> Signed-off-by:
Sibai Li <sibai.li@intel.com> Signed-off-by:
-
Mitch Williams authored
Fix a panic that would occur in the VF if the PF driver failed or was removed from the host kernel. In this case, the VF driver calls i40evf_close(), but this function does nothing because the driver is in the resetting state. Because of this, the driver doesn't free its irqs and causes a kernel panic when it tries to disable MSI-X. Change-ID: If95644a89e554b4d7be0dca1b6add26f63047129 Signed-off-by:
-
Neerav Parikh authored
In case of any AQ command to query port's DCB configuration fails during driver's probe time; the probe fails and returns an error. This patch prevents this issue by continuing the driver probe even when an error is returned. Also, added an error message to dump the AQ error status to show what error caused the failure to get the DCB configuration from firmware. Change-ID: Ifd5663512588bca684069bb7d4fb586dd72221af Signed-off-by:
-
Neerav Parikh authored
The queue vectors are originally distributed among the total number of queues that are configured for a given VSI. The number of queues that are configured for a VSI depends on the number of TCs that are enabled for the VSI at VSI setup/creation time. But, the total number of queues that are configured for a VSI may change based on change in the total number of TCs enabled as a result of change in DCB configuration based on DCBX. Hence, it is required for the queue vectors to be redistributed when the total queues that are configured for a given VSI changes. Without redistributing the vectors the queues that are configured for the newly enabled TCs may not be able to do any Tx/Rx. Change-ID: I4e780903db019e6199c7ce43627cef22c916acfe Signed-off-by:
-
Shannon Nelson authored
The driver version string was added to this struct to be passed down through the firmware to low-level NC-SI functions. We tell the firmware about the length of the ASCII string not counting any terminating null. Change-ID: I09ac98ff9b869e8661c55fc6a5c98808fc280c91 Signed-off-by:
-
Greg Rose authored
The function to set the VF MAC address was not taking the port VLAN filter into account when setting/clearing/resetting the VF's host administered MAC address. Be sure to use the port VLAN for VF MAC filtering configurations. Change-ID: I12595331981c79529738f041dcdbdb667ef8f5e6 Signed-off-by:
-
Mitch A Williams authored
Add support for getting and setting RSS hashing options, RSS LUT entries, and getting the number of device channels from ethtool. Because the LUT is so small for the VFs, we just read and write the device registers directly instead of maintaining a shadow copy. Signed-off-by:
-
Greg Rose authored
We were already taking care to retain the MAC filter list when deleting a port VLAN. Take some additional care to retain the MAC filter list when changing a port VLAN. Change-ID: Iacf9599ea24ecb4dca8e419aacaf4b58ca361a9c Signed-off-by:
-
Jacob Keller authored
This patch removes the unnecessary ptp_tx_work item. It conflicts with the interrupt-based handler, and will cause spurious warning messages to be displayed in the kernel log. Since the hardware can properly trigger an interrupt, we do not need to poll for an available timestamp in a work queue any more. Signed-off-by:
Jacob Keller <jacob.e.keller@intel.com> Acked-by:
Jesse Brandeburg <jesse.brandeburg@intel.com> Tested-by:
-
Cong Wang authored
When we change the list of action on a given filter, currently we don't change it to empty. This is a bug, we should allow to change to whatever users given. Cc: Jamal Hadi Salim <jhs@mojatatu.com> Cc: David S. Miller <davem@davemloft.net> Signed-off-by:
Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by:
Cong Wang <cwang@twopensource.com> Signed-off-by:
Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by:
-
Cong Wang authored
When actions are attached to a filter, they are a part of the filter itself, so when changing a filter we should allow to overwrite the actions inside as well. In my specific case, when I tried to _append_ a new action to an existing filter which already has an action, I got EEXIST since kernel refused to overwrite the existing one in kernel. This patch checks if we are changing the filter checking NLM_F_CREATE flag (Sigh, filters don't use NLM_F_REPLACE...) and then passes the boolean down to actions. This fixes the problem above. Cc: Jamal Hadi Salim <jhs@mojatatu.com> Cc: David S. Miller <davem@davemloft.net> Signed-off-by:
-
David S. Miller authored
Jitendra Kalsaria says: ==================== qlcnic driver updates This patch series containes following changes - * Optimize MAC learning code. * Export board temperature using hwmon-sysfs interface. * Add support for configuring Tx interrupt coalescing parameters from VF. * Logging error messages in error path. * Restrict extended vNIC support in legacy interrupt mode and log appropriate message. Please apply this series to net-next. ==================== Signed-off-by:
-
