- 31 May, 2017 7 commits
-
-
Jacob Keller authored
The i40e driver can only handle one Tx timestamp request at a time. This means it is possible for an application timestamp request to be ignored. There is no easy way for an administrator to determine if this occurred. Add a new statistic which tracks this, tx_hwtstamp_skipped. Signed-off-by:
Jacob Keller <jacob.e.keller@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com>
-
Jacob Keller authored
The i40e driver uses a bit lock to indicate when a Tx timestamp is in progress to avoid attempting to timestamp multiple packets at once. This is required because hardware only has registers to handle one request at a time. There is a corner case where we failed to cleanup the bit lock after a failed transmit. This can potentially result in a state bit being locked forever. Add some cleanup code to i40e_xmit_frame_ring to check and make sure we cleanup incase of these failures. We also modify i40e_tx_map to return an error code indication DMA failure. Reported-by:
Reported-by: David Mirabito <davidm@metamako.com> Signed-off-by:
-
Jacob Keller authored
Hardware related to the i40e driver has a limitation on Tx PTP packets. This requires us to limit the driver to timestamping a single packet at once. This is done using a state bitlock which enforces that only one timestamp request is honored at a time. Unfortunately this suffers from a race condition. The bit lock is not cleared until after skb_tstamp_tx() is called notifying applications of a new Tx timestamp. Even a well behaved application sending only one packet at a time and waiting for a response can wake up and send a new timestamped packet request before the bit lock is cleared. This results in needlessly dropping some Tx timestamp requests. We can fix this by unlocking the state bit as soon as we read the Timestamp register, as this is the first point at which it is safe to timestamp another packet. To avoid issues with the skb pointer, we'll use a copy of the pointer and set the global variable in the driver structure to NULL first. This ensures that the next timestamp request does not modify our local copy of the skb pointer. Now, a well behaved application which has at most one outstanding timestamp request will not accidentally race with the driver unlock bit. Obviously an application attempting to timestamp faster than one request at a time will have some timestamp requests skipped. Unfortunately there is nothing we can do about that. Reported-by:
-
Jesse Brandeburg authored
The i40evf hardware doesn't have any way to ever report FCoE enabled so just force the code to always report FCoE is disabled, remove the unused defines, and mark the OP as reserved. Signed-off-by:
Jesse Brandeburg <jesse.brandeburg@intel.com> Tested-by:
Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by:
-
Jesse Brandeburg authored
This patch fixes a missing line that was missed while merging, which results in a driver feature in the VF not working to enable RSS as a negotiated feature. Fixes: 43a3d9ba ("i40evf: Allow PF driver to configure RSS") Signed-off-by:
-
Jesse Brandeburg authored
This removes two duplicate lines that snuck into the code somehow. Signed-off-by:
-
Michael Chan authored
We need to write the doorbell if BQL has stopped the queue and skb->xmit_more is set. Otherwise it is possible for the tx queue to rot and cause tx timeout. Fixes: 4d172f21 ("bnxt_en: Implement xmit_more.") Suggested-by:
Yuval Mintz <yuval.mintz@cavium.com> Signed-off-by:
Michael Chan <michael.chan@broadcom.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
- 30 May, 2017 33 commits
-
-
David S. Miller authored
Michael Chan says: ==================== bnxt_en: Misc. updates for net-next. The 1st 2 patches add short firmware message support for new VF devices. The 3rd patch adds a pci shutdown callback for the RDMA driver for proper shutdown. The next 3 patches improve the doorbell operations by elimiating the double doorbell workaround on newer chips, and by adding xmit_more support. The last patch adds a parameter to bnxt_set_dflt_rings(). ==================== Signed-off-by: -
Michael Chan authored
In the existing code, the local variable sh is hardcoded to true to calculate default rings for shared ring configuration. It is better to have the caller determine the value of sh. Reported-by:
Gustavo A. R. Silva <garsilva@embeddedor.com> Signed-off-by:
-
Michael Chan authored
Do not write the TX doorbell if skb->xmit_more is set unless the TX queue is full. Signed-off-by:
-
Michael Chan authored
Older chips require the doorbells to be written twice, but newer chips do not. Add a new common function bnxt_db_write() to write all doorbells appropriately depending on the chip. Eliminating the extra doorbell on newer chips has a significant performance improvement on pktgen. Signed-off-by:
-
Michael Chan authored
Add additional chip definitions and macros for all supported chips. Add a new macro BNXT_CHIP_P4_PLUS for the newer generation of chips and use the macro to properly determine the features supported by these newer chips. Signed-off-by:
-
Michael Chan authored
When bnxt_en gets a PCI shutdown call, we need to have a new callback to inform the RDMA driver to do proper shutdown and removal. Signed-off-by:
-
Deepak Khungar authored
Signed-off-by:
Deepak Khungar <deepak.khungar@broadcom.com> Signed-off-by:
-
Deepak Khungar authored
The new short message format is used on the new BCM57454 VFs. Each firmware message is a fixed 16-byte message sent using the standard firmware communication channel. The short message has a DMA address pointing to the legacy long firmware message. Signed-off-by:
-
Vivien Didelot authored
The port net device passed to b53_fdb_copy is not used. Remove it. Signed-off-by:
Vivien Didelot <vivien.didelot@savoirfairelinux.com> Reviewed-by:
Florian Fainelli <f.fainelli@gmail.com> Signed-off-by:
-
Vivien Didelot authored
The helper is only used once and makes the code more complicated that it should. Remove it and reorganize the variables so that it fits on 80 columns. Signed-off-by:
-
Arjun Vynipadath authored
GRO is not supported by Chelsio HW when rx_csum is disabled. Update the netdev features flag when rx_csum is modified. Signed-off-by:
Arjun Vynipadath <arjun@chelsio.com> Signed-off-by:
Steve Wise <swise@opengridcomputing.com> Signed-off-by:
Ganesh Goudar <ganeshgr@chelsio.com> Signed-off-by:
-
Arjun Vynipadath authored
Disable FW_OK flag while flashing Firmware. This will help to fix any potential mailbox timeouts during Firmware flash. Grab new devlog parameters after Firmware restart. When we FLASH new Firmware onto an adapter, the new Firmware may have the Firmware Device Log located at a different memory address or have a different size for it. Signed-off-by:
Casey Leedom <leedom@chelsio.com> Signed-off-by:
-
Ganesh Goudar authored
Signed-off-by:
-
Surendra Mobiya authored
Mark carrier off before registering netdev to ensure that vlan device picks up the correct state of the carrier Signed-off-by:
Surendra Mobiya <surendra@chelsio.com> Signed-off-by:
-
David S. Miller authored
Stefan Wahren says: ==================== net: qualcomm: add QCA7000 UART driver The Qualcomm QCA7000 HomePlug GreenPHY supports two interfaces: UART and SPI. This patch series adds the missing support for UART. This driver based on the Qualcomm code [1], but contains some changes: * use random MAC address per default * use net_device_stats from device * share frame decoding between SPI and UART driver * improve error handling * reimplement tty_wakeup with work queue (based on slcan) * use new serial device bus instead of ldisc The patches 1 - 3 are just for clean up and are not related to the UART support. Patch 4 adds SET_NETDEV_DEV() to qca_spi. Patches 5 - 16 prepare the existing QCA7000 code for UART support. The last patch contains the new driver. The code itself has been tested on a Freescale i.MX28 board and a Raspberry Pi Zero. Changes in v8: * add necessary header includes to qca_7k.c in order to reflect dependencies Changes in v7: * fix race between tx workqueue and device deregistration (reported by Lino) ==================== Signed-off-by: -
Stefan Wahren authored
This patch adds the Ethernet over UART driver for the Qualcomm QCA7000 HomePlug GreenPHY. Signed-off-by:
Stefan Wahren <stefan.wahren@i2se.com> Reviewed-by:
Lino Sanfilippo <LinoSanfilippo@gmx.de> Signed-off-by:
-
Stefan Wahren authored
This merges the serdev binding for the QCA7000 UART driver (Ethernet over UART) into the existing document. Signed-off-by:
-
Stefan Wahren authored
This adds a new DT property to define the current baud rate of the slave device. Signed-off-by:
Rob Herring <robh@kernel.org> Signed-off-by:
-
Stefan Wahren authored
Before we can merge the QCA7000 UART binding the document needs to be renamed. Signed-off-by:
-
Stefan Wahren authored
In preparation for the QCA7000 UART binding rework the binding document. Signed-off-by:
-
Stefan Wahren authored
In order to share common functions between QCA7000 SPI and UART protocol driver the qca_7k_common needs to be a separate kernel module. Signed-off-by:
-
Stefan Wahren authored
Unfortunately the frame format is not exactly identical between SPI and UART. In case of SPI there is an additional HW length at the beginning. So store the initial state to make the decoding state machine more flexible and easy to extend for UART support. Signed-off-by:
-
Stefan Wahren authored
As preparation for the upcoming UART driver we need a module which contains common functions for both interfaces. The module qca_framing is a good candidate but renaming to qca_7k_common would make it clear. Signed-off-by:
-
Stefan Wahren authored
Since this driver is specific to the QCA7000, we should make the module description more precisely. Signed-off-by:
-
Stefan Wahren authored
The function qcaspi_tx_cmd() is only called from qca_spi.c. So we better move it there. Signed-off-by:
-
Stefan Wahren authored
There is no need for an additional MTU define. Signed-off-by:
-
Stefan Wahren authored
In order to avoid mixing things up, make the MTU and frame length defines easier to read. Signed-off-by:
-
Stefan Wahren authored
There is no need to export qcaspi_netdev_open and qcaspi_netdev_close because they are also accessible via the net_device_ops. Signed-off-by:
-
Stefan Wahren authored
Use SET_NETDEV_DEV() in qca_spi to create the "/sys/class/net/<if>/device" symlink. Signed-off-by:
-
Stefan Wahren authored
Use the BIT macro for the CONFIG and INT register values. Signed-off-by:
-
Stefan Wahren authored
It doesn't make sense to use a signed variable for offset here, so fix it up. Signed-off-by:
-
Stefan Wahren authored
Currently the includes doesn't reflect the dependencies. So fix this up by removing all unnecessary entries and add the necessary ones explicit. Signed-off-by:
-
David S. Miller authored
Bogdan Purcareata says: ==================== net: phy: Support managed Cortina phys So far, the Cortina family phys (CS4340 in this particular case) are only supported in fixed link mode (via fixed_phy_register). The generic 10G phy driver does not work well with the phylib state machine, when the phy is registered via of_phy_connect. This prohibits the user from describing the phy nodes in the device tree. In order to support this scenario, and to properly describe the board device tree, add a minimal Cortina driver that reads the status from the right register. With the generic 10G C45 driver, the kernel will print messages like: [ 0.226521] mdio_bus 8b96000: Error while reading PHY16 reg at 1.6 [ 0.232780] mdio_bus 8b96000: Error while reading PHY16 reg at 1.5 v3 -> v4: - Add trademark info. - Minor documentation entry consistency nit. v2 -> v3: - Add documentation entry. v1 -> v2: - Change approach for getting the phy_id from hacking get_phy_c45_ids to describing the device in the device tree via ethernet-phy-id. More patch version changes per individual patches. ==================== Signed-off-by:
-
