- 28 Dec, 2015 7 commits
-
-
David S. Miller authored
Igal Liberman says: ==================== Freescale DPAA FMan The Freescale Data Path Acceleration Architecture (DPAA) is a set of hardware components on specific QorIQ multicore processors. This architecture provides the infrastructure to support simplified sharing of networking interfaces and accelerators by multiple CPU cores and the accelerators. One of the DPAA accelerators is the Frame Manager (FMan) which contains a series of hardware blocks: ports, Ethernet MACs, a multi user RAM (MURAM) and Storage Profile (SP). This patch set introduce the FMan drivers. Each driver configures and initializes the corresponding FMan hardware module (described above). The MAC driver offers support for three different types of MACs (eTSEC, TGEC, MEMAC). v9 --> v10: - Addressed feedback from David Miller Remove private CRC implementation - Addressed feedback from Kenneth Klette Jonassen: - Use Kernel PHY API to configure dTSEC TBI - Use Kernel PHY API to configure mEMAC PCS This patchset requires device tree update: https://patchwork.ozlabs.org/patch/559501/ - Addressed feedback from Andy Fleming v8 --> v9: No changes v7 --> v8: - Addressed feedback from David Miller - Support for ARM: - Device tree parsing - IO Accessors - Addressed compilation issue on non-PPC targets v6 --> v7: - Addressed compilation issue on non-PPC targets - Removed B4860 rev 1 support v5 --> v6: - Addressed feedback from Scott: - Moved kernel doc to source files - Removed a series of configurable settings - Miscellaneous code updates v4 --> v5: - Addressed feedback from David Miller: - Removed driver layering - Reduce namespace pollution - Reduce code complexity and size v3 --> v4: - Remove device_initcall call in driver registration (redundant) - Remove hot/cold labels - Minor update in FMan Clock read from device-tree - Update fixed-link support - Addressed feedback from Stephen Hemminger - Remove bogus blank line v2 --> v3: - Addressed feedback from Scott: - Remove typedefs - Remove unnecessary memory barriers - Remove unnecessary casting - Remove KConfig options - Remove early_params - Remove Hungarian notation - Remove __packed__ attribute and padding from structures - Remove unlikely attribute (where it's not needed) - Use proper error codes and remove unnecessary prints - Use proper values for sleep routines - Replace complex Macros with functions - Improve device tree processing code - Use symbolic defines - Add time-out in busy-wait loops - Removed exit code (loadable module support will be added later) - Fixed "fixed-link" issue raised by Joakim Tjernlund v1 --> v2: - Addressed feedback from Paul Bolle: - General feedback of FMan Driver layer - Remove Errata defines - Aligned comments to Kernel Doc - Remove Loadable Module support (not yet supported) - Removed not needed KConfig dependencies - Addressed feedback from Scott Wood - Use Kernel ioread/iowrite services - Squash FLIB source and header patches together This submission is based on the prior Freescale DPAA FMan V3,RFC submission. Several issues addresses in this submission: - Reduced MAC layering and complexity - Reduced code base - T1024/T2080 10G best effort support ==================== Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Igal Liberman authored
This patch adds the Ethernet MAC driver supporting the three different types of MACs: dTSEC, tGEC and mEMAC. Signed-off-by:
Igal Liberman <igal.liberman@freescale.com> Signed-off-by:
-
Igal Liberman authored
Add the Data Path Acceleration Architecture Frame Manger Port Driver. The FMan driver uses a module called "Port" to represent the physical TX and RX ports. Each FMan version has different number of physical ports. This patch adds The FMan Port configuration, initialization and runtime control routines for both TX and RX. Signed-off-by:
-
Igal Liberman authored
The Storage Profiles contain parameters that are used by the FMan for frame reception and transmission. Signed-off-by:
-
Igal Liberman authored
Add the Data Path Acceleration Architecture Frame Manger MAC support. This patch adds The FMan MAC configuration, initialization and runtime control routines. This patch contains support for these types of MACs: - dTSEC: Three speed Ethernet controller (10/100/1000 Mbps) - tGEC: 10G Ethernet controller (10 Gbps) - mEMAC: Multi-rate Ethernet MAC (10/100/1000/10000 Mbps) Different FMan revisions have different type and number of MACs. Signed-off-by:
-
Igal Liberman authored
Add the Data Path Acceleration Architecture Frame Manger Driver. The FMan embeds a series of hardware blocks that implement a group of Ethernet interfaces. This patch adds The FMan configuration, initialization and runtime control routines. The FMan driver supports several hardware versions differentiated by things like: - Different type of MACs - Number of MAC and ports - Available resources - Different hardware errata Signed-off-by:
-
Igal Liberman authored
Add Frame Manager Multi-User RAM support. This internal FMan memory block is used by the FMan hardware modules, the management being made through the generic allocator. The FMan Internal memory, for example, is used for allocating transmit and receive FIFOs. Signed-off-by:
-
- 26 Dec, 2015 1 commit
-
-
Pravin B Shelar authored
By moving stats update into iptunnel_xmit(), we can simplify iptunnel_xmit() usage. With this change there is no need to call another function (iptunnel_xmit_stats()) to update stats in tunnel xmit code path. Signed-off-by:
Pravin B Shelar <pshelar@nicira.com> Signed-off-by:
-
- 24 Dec, 2015 21 commits
-
-
Hariprasad Shenai authored
Remove deprecated module parameters, and mark one parameter as deprecated. Signed-off-by:
Hariprasad Shenai <hariprasad@chelsio.com> Signed-off-by:
-
Hariprasad Shenai authored
CLIP is always enabled and hardware uses 2 TID entries instead of 4 for IPv6 in CLIP mode. Signed-off-by:
-
David S. Miller authored
Hariprasad Shenai says: ==================== Update support for T6 adapters This patch changes updates the various code changes related to register, stats and hardware related changes for T6 family of adapters. This patch series has been created against net-next tree and includes patches on cxgb4 and cxgb4vf driver. We have included all the maintainers of respective drivers. Kindly review the change and let us know in case of any review comments. ==================== Signed-off-by: -
Hariprasad Shenai authored
Signed-off-by:
-
Hariprasad Shenai authored
SGE context congestion map changed from 4 to 8 priority per port in T6 as there are only 2 channels. Signed-off-by:
-
Hariprasad Shenai authored
In T6, MPS classification has a 512 deep TCAM to do the match lookup. Each entry has 80x2b sets containing 48 bit MAC address, port number, VLAN Valid/ID, VNI, lookup type (outer or inner packet header). [71:48] bit locations are overloaded for outer vs. inner lookup types. Signed-off-by:
-
Hariprasad Shenai authored
Signed-off-by:
-
Hariprasad Shenai authored
Updating Congestion Channel/Priority Map in Congestion Manager Context for T6. In T6 port 0 is mapped to channel 0 and port 1 is mapped to channel 1. For 2 port T4/T5 adapter, port 0 is mapped to channel 0,1 and port 1 is mapped to channel 2,3 Signed-off-by:
-
Hariprasad Shenai authored
Signed-off-by:
-
Hariprasad Shenai authored
Ingress padding boundary values got changed for T6. T5: 0=32B 1=64B 2=128B 3=256B 4=512B 5=1024B 6=2048B 7=4096B T6: 0=8B 1=16B 2=32B 3=64B 4=128B 5=128B 6=256B 7=512B Updating the driver to set the correct boundary values in SGE_CONTROL to 32B. Also, need to take care of this fl alignment change when calculating the next packet offset. Signed-off-by: -
Hariprasad Shenai authored
Updated pm_stats code to display input FIFO wait (index 5) and read latency (index 7) counters for T6 adapters Signed-off-by:
-
Hariprasad Shenai authored
Signed-off-by:
-
Geliang Tang authored
kobj_to_dev has been defined in linux/device.h, so I replace to_dev with it. Signed-off-by:
Geliang Tang <geliangtang@163.com> Signed-off-by:
-
Geliang Tang authored
to_dev is not used anymore so drop it. Signed-off-by:
-
Andrew Lunn authored
The 6320 family of switch chips has a second bank for statistics, but is missing three statistics in the port registers. Generalise and extend the code: * adding a field to the statistics table indicating the bank/register set where each statistics is. * add a function indicating if an individual statistics is available on this device * calculate at run time the sset_count. * return strings based on the available statistics of the device * return statistics based on the available statistics of the device * Add support for reading from the second bank. Signed-off-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
David S. Miller authored
Bert Kenward says: ==================== sfc: additional virtual function support This introduces the client side of a mechanism to defer authorisation of operations, for example multicast subscription. Although primarily aimed at SRIOV VFs this can also apply to unprivileged PFs. Also handle reboot ordering corner cases better and reduce the level of some logging. v2: remove #ifdef DEBUG around new WARN_ON in mcdi.c. ==================== Signed-off-by: -
Bert Kenward authored
Depending on configuration the NIC may return errors for unprivileged functions and/or VFs. Where these are expected and handled, reduce the level of any output. Signed-off-by:
Bert Kenward <bkenward@solarflare.com> Signed-off-by:
-
Tomáš Pilař authored
When running in an unprivileged function we expect some MC commands to fail with permission errors. To avoid log spew downgrade these to debug only. Signed-off-by:
-
Bert Kenward authored
There are situations - mostly reset related - where our view of the filter table differs from the hardware. In this case we may try and remove filters that aren't actually installed. This isn't that interesting in most situations, so downgrade the logging. Signed-off-by:
-
Bert Kenward authored
For unprivileged functions operations can be authorised by an admin function. Extra steps are introduced to the MCDI protocol in this situation - the initial response from the MCDI tells us that the operation has been deferred, and we must retry when told. We then receive an event telling us to retry. Note that this provides only the functionality for the unprivileged functions, not the handling of the administrative side. Signed-off-by:
-
Bert Kenward authored
After reboot the vswitch configuration from the PF may not be complete before the VF attempts to restore filters. In that case we see NO_EVB_PORT errors from the MC. Retry up to a time limit or until a different result is seen. Signed-off-by:
-
- 23 Dec, 2015 5 commits
-
-
David S. Miller authored
Hariprasad Shenai says: ==================== Trivial enhancements for cxgb4 This series adds a debug message if adapter isn't inserted in right PCI slot. Changes naming conventions for iSCSI rx queues, use node info while allocating rx queue and use napi_complete_done() api in napi handler. This patch series has been created against net-next tree and includes patches on cxgb4 driver. We have included all the maintainers of respective drivers. Kindly review the change and let us know in case of any review comments. Thanks V2: Dropped 'dcb_info' debug entry patch, since the same can be achieved using lldp tool. Based on review comments by Or Gerlitz <gerlitz.or@gmail.com> and David Miller. ==================== Signed-off-by: -
Hariprasad Shenai authored
Signed-off-by:
-
Hariprasad Shenai authored
Signed-off-by:
-
Hariprasad Shenai authored
All the upper level protocols like rdma, iscsi have their own offload rx queues, so instead of using the generic naming convention be specific while naming them. Improves code readability Signed-off-by:
-
Hariprasad Shenai authored
Check if the device get enough bandwidth from the entire PCI chain to satisfy its capabilities. This patch determines the PCIe device's bandwidth capabilities by reading its PCIe Link Capabilities registers and then call the pcie_get_minimum_link function to ensure that the adapter is hooked into a slot which is capable of providing the necessary bandwidth capabilities. Signed-off-by:
-
- 22 Dec, 2015 6 commits
-
-
David S. Miller authored
Florian Westphal says: ==================== tcp: honour SO_BINDTODEVICE for TW_RST case too This is V2, this time as a small series since I followed Erics advice to split this into smaller chunks, I hope this makes it easier to review. First patch adds inet_sk_transparent helper. Second patch contains an if/else swap that I split from the original TW_RST v1 one. Third patch is the actual change without the superfluous sock_net change. ==================== Signed-off-by: -
Florian Westphal authored
Hannes points out that when we generate tcp reset for timewait sockets we pretend we found no socket and pass NULL sk to tcp_vX_send_reset(). Make it cope with inet tw sockets and then provide tw sk. This makes RSTs appear on correct interface when SO_BINDTODEVICE is used. Packetdrill test case: // want default route to be used, we rely on BINDTODEVICE `ip route del 192.0.2.0/24 via 192.168.0.2 dev tun0` 0.000 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3 // test case still works due to BINDTODEVICE 0.001 setsockopt(3, SOL_SOCKET, SO_BINDTODEVICE, "tun0", 4) = 0 0.100...0.200 connect(3, ..., ...) = 0 0.100 > S 0:0(0) <mss 1460,sackOK,nop,nop> 0.200 < S. 0:0(0) ack 1 win 32792 <mss 1460,sackOK,nop,nop> 0.200 > . 1:1(0) ack 1 0.210 close(3) = 0 0.210 > F. 1:1(0) ack 1 win 29200 0.300 < . 1:1(0) ack 2 win 46 // more data while in FIN_WAIT2, expect RST 1.300 < P. 1:1001(1000) ack 1 win 46 // fails without this change -- default route is used 1.301 > R 1:1(0) win 0 Reported-by:
Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by:
Florian Westphal <fw@strlen.de> Acked-by:
Eric Dumazet <edumazet@google.com> Acked-by:
-
Florian Westphal authored
tcp_md5_do_lookup requires a full socket, so once we extend _send_reset() to also accept timewait socket we would have to change if (!sk && hash_location) to something like if ((!sk || !sk_fullsock(sk)) && hash_location) { ... } else { (sk && sk_fullsock(sk)) tcp_md5_do_lookup() } Switch the two branches: check if we have a socket first, then fall back to a listener lookup if we saw a md5 option (hash_location). Signed-off-by: -
Florian Westphal authored
Avoids cluttering tcp_v4_send_reset when followup patch extends it to deal with timewait sockets. Suggested-by:
-
Jiri Pirko authored
KASan reported use-after-free for the hwmon structure. So fix this by using devm_kzalloc and let the core take care about freeing the memory during device dettach. Reported-by:
Ido Schimmel <idosch@mellanox.com> Fixes: 89309da3 ("mlxsw: core: Implement temperature hwmon interface") Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Signed-off-by:
-
Lorenzo Colitti authored
When closing a listen socket, tcp_abort currently calls tcp_done without clearing the request queue. If the socket has a child socket that is established but not yet accepted, the child socket is then left without a parent, causing a leak. Fix this by setting the socket state to TCP_CLOSE and calling inet_csk_listen_stop with the socket lock held, like tcp_close does. Tested using net_test. With this patch, calling SOCK_DESTROY on a listen socket that has an established but not yet accepted child socket results in the parent and the child being closed, such that they no longer appear in sock_diag dumps. Reported-by:
Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by:
Lorenzo Colitti <lorenzo@google.com> Acked-by:
-
