Age | Commit message (Collapse) | Author |
|
|
|
commit 393415203f5c916b5907e0a7c89f4c2c5a9c5505 upstream.
We need to increase TSO_HEADER_SIZE from 128 to 256.
Since otx2_sq_init() calls qmem_alloc() with TSO_HEADER_SIZE,
we need to change (struct qmem)->entry_sz to avoid truncation to 0.
Fixes: 7a37245ef23f ("octeontx2-af: NPA block admin queue init")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
|
|
commit 28b18e4eb515af7c6661c3995c6e3c34412c2874 upstream.
clang static analysis flags this garbage return
drivers/net/ethernet/marvell/sky2.c:208:2: warning: Undefined or garbage value returned to caller [core.uninitialized.UndefReturn]
return v;
^~~~~~~~
static inline u16 gm_phy_read( ...
{
u16 v;
__gm_phy_read(hw, port, reg, &v);
return v;
}
__gm_phy_read can return without setting v.
So handle similar to skge.c's gm_phy_read, initialize v.
Signed-off-by: Tom Rix <trix@redhat.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
|
|
commit d3d239dcb8aae6d7b10642d292b404e57604f7ea upstream.
In mvneta_config_interface() the RGMII modes are catched by the default
case which is an error return. The RGMII modes are valid modes for the
driver, so instead of returning an error add a break statement to return
successfully.
This avoids this warning for non comphy SoCs which use RGMII, like
SolidRun Clearfog:
WARNING: CPU: 0 PID: 268 at drivers/net/ethernet/marvell/mvneta.c:3512 mvneta_start_dev+0x220/0x23c
Fixes: b4748553f53f ("net: ethernet: mvneta: Fix Serdes configuration for SoCs without comphy")
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Reviewed-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
commit 41c2b6b4f0f807803bb49f65835d136941a70f85 upstream.
When writing the serdes configuration register was moved to
mvneta_config_interface() the whole code block was removed from
mvneta_port_power_up() in the assumption that its only purpose was to
write the serdes configuration register. As mentioned by Russell King
its purpose was also to check for valid interface modes early so that
later in the driver we do not have to care for unexpected interface
modes.
Add back the test to let the driver bail out early on unhandled
interface modes.
Fixes: b4748553f53f ("net: ethernet: mvneta: Fix Serdes configuration for SoCs without comphy")
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Reviewed-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
commit f2ca673d2cd5df9a76247b670e9ffd4d63682b3f upstream.
When support for short preambles was added, it incorrectly keyed its
decision off state->speed instead of state->interface. state->speed
is not guaranteed to be correct for in-band modes, which can lead to
short preambles being unexpectedly disabled.
Fix this by keying off the interface mode, which is the only way that
mvneta can operate at 2.5Gbps.
Fixes: da58a931f248 ("net: mvneta: Add support for 2500Mbps SGMII")
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
commit 1a642ca7f38992b086101fe204a1ae3c90ed8016 upstream.
The older SoCs like Armada XP support a 2500BaseX mode in the datasheets
referred to as DR-SGMII (Double rated SGMII) or HS-SGMII (High Speed
SGMII). This is an upclocked 1000BaseX mode, thus
PHY_INTERFACE_MODE_2500BASEX is the appropriate mode define for it.
adding support for it merely means writing the correct magic value into
the MVNETA_SERDES_CFG register.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
commit b4748553f53f2971e07d2619f13d461daac0f3bb upstream.
The MVNETA_SERDES_CFG register is only available on older SoCs like the
Armada XP. On newer SoCs like the Armada 38x the fields are moved to
comphy. This patch moves the writes to this register next to the comphy
initialization, so that depending on the SoC either comphy or
MVNETA_SERDES_CFG is configured.
With this we no longer write to the MVNETA_SERDES_CFG on SoCs where it
doesn't exist.
Suggested-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
|
|
commit 722c0f00d4feea77475a5dc943b53d60824a1e4e upstream.
The "info->fs.location" is a u32 that comes from the user via the
ethtool_set_rxnfc() function. We need to check for invalid values to
prevent a buffer overflow.
I copy and pasted this check from the mvpp2_ethtool_cls_rule_ins()
function.
Fixes: 90b509b39ac9 ("net: mvpp2: cls: Add Classification offload support")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
|
|
commit 065fd83e1be2e1ba0d446a257fd86a3cc7bddb51 upstream.
For the case where the last mvneta_poll did not process all
RX packets, we need to xor the pp->cause_rx_tx or port->cause_rx_tx
before claculating the rx_queue.
Fixes: 2dcf75e2793c ("net: mvneta: Associate RX queues with each CPU")
Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
|
|
commit c35947b8ff8acca33134ee39c31708233765c31a upstream.
Move rx_dropped and rx_errors counters in mvneta_pcpu_stats in order to
avoid possible races updating statistics
Fixes: 562e2f467e71 ("net: mvneta: Improve the buffer allocation method for SWBM")
Fixes: dc35a10f68d3 ("net: mvneta: bm: add support for hardware buffer management")
Fixes: c5aff18204da ("net: mvneta: driver for Marvell Armada 370/XP network unit")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
|
|
commit f3f2364ea14d1cf6bf966542f31eadcf178f1577 upstream.
phylink requires the MAC to report when its link status changes when
operating in inband modes. Failure to report link status changes
means that phylink has no idea when the link events happen, which
results in either the network interface's carrier remaining up or
remaining permanently down.
For example, with a fiber module, if the interface is brought up and
link is initially established, taking the link down at the far end
will cut the optical power. The SFP module's LOS asserts, we
deactivate the link, and the network interface reports no carrier.
When the far end is brought back up, the SFP module's LOS deasserts,
but the MAC may be slower to establish link. If this happens (which
in my tests is a certainty) then phylink never hears that the MAC
has established link with the far end, and the network interface is
stuck reporting no carrier. This means the interface is
non-functional.
Avoiding the link interrupt when we have phylink is basically not
an option, so remove the !port->phylink from the test.
Fixes: 4bb043262878 ("net: mvpp2: phylink support")
Tested-by: Sven Auhagen <sven.auhagen@voleatech.de>
Tested-by: Antoine Tenart <antoine.tenart@bootlin.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
|
|
commit 8e7f7faf9d90a84283093e02b90a9e9a12b061ae from
git@git.assembla.com:cavium/WindRiver.linux.git
A single interrupt line from VF to PF is used by
VF to send messages to PF and for replying to UP messages
sent by PF. Hence num_msgs in mbox header is cleared
explicitly in PF interrupt handler to avoid queuing a
work twice for an event. The same num_msgs which was
cleared in interrupt handler was set again in work
queue handler by mistake. This patch fixes it.
Also
- clearing interrupt in work queue handlers is not
necessary hence removed it.
- checking status in message response is not needed
hence removed it.
Change-Id: I1875d37816b61b4874eea1c7175c63dfea71bcb1
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/20142
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 46a0ce95e2db7ffd083cd2634b7037f32e27501f from
git@git.assembla.com:cavium/WindRiver.linux.git
Used mutex instead atomic variables for
mailbox lock. Also changed WARN_ON to
dev_err.
Change-Id: I18e1d3cc013bfc058db47a33ed5d123185d3c5ef
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/20141
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 7aff3a8035f71bfd3675067ea5b79e5ad9b7bd9d from
git@git.assembla.com:cavium/WindRiver.linux.git
iommu_domain can be NULL hence check for non
NULL pointer before dereferencing it.
Change-Id: Iff01276a67f35954f1556b731d43339d152e268d
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/20065
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 4ece484b7a0b62a79ab098f75c2a592839d4d7fb from
git@git.assembla.com:cavium/WindRiver.linux.git
MAC address field in vlan offload rule and interface mac address
needs to be same for VLAN offload functionality. So update mac
address field in vlan offload rule whenever user changes it.
Change-Id: I272ea3b94b7989fd15efd306c27dbae3dddfb2a6
Signed-off-by: hariprasad <hari1219@gmail.com>
Reviewed-on: https://sj1git1.cavium.com/19826
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit d4de909fb7de0de93d168d419ae5621ce0125ea2 from
git@git.assembla.com:cavium/WindRiver.linux.git
PF set mac address will be persistent across module reload.
PF set mac address cannot be overwritten by VF
VF set mac address will be preserved until module unload.
Change-Id: Ie228a0af9e10ff2fd5e2b0363afd9a24be5f2035
Signed-off-by: hariprasad <hkelam@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/19827
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit bbcdd24b2067f23ce553f57bd5153a3e3793b651 from
git@git.assembla.com:cavium/WindRiver.linux.git
Set RR_QUANTUM value high for transmit levels to
prevent negative accumulations of the deficit count.
Change-Id: I015ed6cfb27f853d653ebd299492a11bf3990abe
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/19885
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 721cf946546df138bb6087999d2a693891bfe7d1 from
git@git.assembla.com:cavium/WindRiver.linux.git
Current pause frame time is too high for a 40G
interface. 0xFFFF times 512bits will result in 780us
pause time. Reduced pause time config to 0x7FF.
Change-Id: I20055d578835926d9027f1a1b32885f91fd52cbe
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/19694
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 94cef32782641d0e0d89a93db273c583d8fb5dd6 from
git@git.assembla.com:cavium/WindRiver.linux.git
HW takes care of data coherency at LLC level, to save some
CPU cycles skip calling dma_map and dma_unmap apis when
IOMMU is bypassed. This improves IP forwarding performance
by ~5%. These changes are not for upstream.
Change-Id: Ie1be6982d62ffe4bbb4434092c0c5ce3575d9339
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/19664
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 66a0f365f4618d9528abcdff121a9e57f7bbb1a2 from
git@git.assembla.com:cavium/WindRiver.linux.git
Add likely and unliley to conditional statements for
generation of optimised code by compiler in terms of
branching.
Change-Id: I7a80fd14999e187732dd8b8b2037544b97bc8800
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/19653
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 02237bdd58d65c17d5b502bb027290dd7a281bf2 from
git@git.assembla.com:cavium/WindRiver.linux.git
Process transmitted pkt's CQEs first so that the freed skbs
are recycled for received packets.
Change-Id: Ia399659851c4b8181619ee212d1c9ee26711efc6
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/19652
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit f5b4312caee99567741fcb75470673d69a771a18 from
git@git.assembla.com:cavium/WindRiver.linux.git
For platform with two NIX blocks, LF may be
allocated from either of NIX blocks. Hence discover
the assigned block address from RVUM register.
Change-Id: Id34b7ad968c6b91ea41ef833fad6bdebfef61c51
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/19496
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 8d48b6712f60b9bf29c482c090f3aee1166ac37b from
git@git.assembla.com:cavium/WindRiver.linux.git
Change-Id: I0bf0176bf6cd32f3a0b4c7963836b848747d626e
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/19129
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit b1b525868f846e2599dd96ac881d8e483d162c0d from
git@git.assembla.com:cavium/WindRiver.linux.git
The stats are put in the shared data struct fwdata. A PHY driver indicates
that it has FEC stats by setting the flag fwdata.phy.misc.has_fec_stats
(for now, only the LIO3 card's 6141 PHY does this; all other PHYs clear
that flag).
On LIO3, FEC can be enabled at the PHY line-side only, never at the PHY
host-side. So for LIO3, the ethtool FEC stats come from the PHY (not from
CGX).
Besides CGX_CMD_GET_PHY_FEC_STATS, also add CGX_CMD_PRBS and
CGX_CMD_DISPLAY_EYE to enum cgx_cmd_id so that Linux's enum list is in sync
with ATF's enum list.
Change-Id: Ib0cb8e9315b88e3c3f1ddd16cae709f8837d4f1d
Signed-off-by: Felix Manlunas <fmanlunas@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18814
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 56e5f30e794e51e3075a18791198d01cacbfbbc1 from
git@git.assembla.com:cavium/WindRiver.linux.git
Disable NPC configuration (shift data off by 8 bytes) done
for PTP in FLR.
Change-Id: I8e6d245ae13601841fbaa03c15da6d9bfc0595d2
Signed-off-by: Harman Kalra <hkalra@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18456
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit a9d863bf6de65580c54c2424bdc0d09dce663b67 from
git@git.assembla.com:cavium/WindRiver.linux.git
Generic arm64 atomic APIs for "ldadd" and "ldeor" instructions can't be
used if CONFIG_LSE_ATOMIC is disabled.
Hence, using asm instructions to remove dependency on kernel config.
Change-Id: I40fd44b6246b3fad7479bc3cd3115c93331f80b6
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18269
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 162b9f43e31611ecfac36820e1b70a2397133624 from
git@git.assembla.com:cavium/WindRiver.linux.git
Change-Id: I5743e37aa0e3d51602d6658de94c358d2cb6339f
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18164
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
(cherry picked from commit f81d7acae881eabc3aeeb9c47a97ee13ecf5ee6e)
Reviewed-on: https://sj1git1.cavium.com/18169
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
The LBKs are not allowed to set the parse mode. It will return
an error if we try to do so, then cause a failure when opening
these ports.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit d91c0634ae5d5e9432f054919a1952dfd91507db from
git@git.assembla.com:cavium/WindRiver.linux.git
To install NPC MCAM rules to support VF MAC and VF VLAN settings
PF needs to know ingress hw channel number. This info is sent to
PF by AF in response to NIXLF_ALLOC mbox ie NIXLF init request.
So skip 'set vf vlan' and 'set vf mac' if PF interface is not UP.
Change-Id: Ib8d0c5c84296aa22090d39038055e769668ac07f
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18079
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 556600410db1e2490ebf8d46b8f4419b580d8e21 from
git@git.assembla.com:cavium/WindRiver.linux.git
Use 2048 byte buffers for better last level cache
utilization or data distribution across regions.
Change-Id: I726bc8919afb8f52bca7da1ff0facbd3cfc378df
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18078
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 39c012db6c348c4f0d02d4a2724607a4e3fe25b4 from
git@git.assembla.com:cavium/WindRiver.linux.git
Usage of NETIF_F_LLTX is deprecated upstream hence this
patch cannot be upstreamed. But still enabling this locally
as it improves IP forwarding performance especially when all
24 cores are transmitting pkts.
Change-Id: I33a58c0e71d63ee0937ba50c47c5476ba62be094
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18065
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit fc87b5d941e7357f1da2cd4fab5a71f6e230017d from
git@git.assembla.com:cavium/WindRiver.linux.git
Fields aura, qidx, pnc, df in SQE_HDR doesn't change for each
SQE (or packet). Hence avoid repopulating them for every
SQE, this saves few cycles.
Change-Id: Ia968853def360aaa360888ec2328c97022e30436
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18066
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit b17de4c302b0e2faad7bbf4e970538713f35775e from
git@git.assembla.com:cavium/WindRiver.linux.git
Move HW related settings info fields to otx2_hw struct
and cleanup both otx2_nic and otx2_hw structs.
Change-Id: Ic142c188b310fe5d4833fe1a85f211deb436a2bc
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18007
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 97137802696ed80c4125a40ad97f6cd43ec86240 from
git@git.assembla.com:cavium/WindRiver.linux.git
Cleaned up multiple boolean fields and flags in otx2_nic struct into
a single flags field.
Change-Id: Ic090c8b06b7cc2c19a0121c913f674330486ad4c
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/18006
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 4926267192697d0b61940b42dad702094091c1fe from
git@git.assembla.com:cavium/WindRiver.linux.git
Right now SDP interface support is limited to single
interface/channel and SDP PF(RVU PF) device, No VF support.
Change-Id: I4ecce3a42b4537ca34c864ef84bf785482966f81
Signed-off-by: Subrahmanyam Nilla <snilla@marvell.com>
Signed-off-by: Venkateshwarlu Nalla <venkatn@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17982
Reviewed-by: Yan Markman <ymarkman@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit c94525c3a9b0e992be718aa5dc214bdda0ebd3b8 from
git@git.assembla.com:cavium/WindRiver.linux.git
Each of the CGX mapped RVU PFVF has 210 MSIX vectors and kernel
globally support only 8192 IRQs. So enabling 210 vectors for all PF,
their VFs and LBKVFs results in failure of pci_alloc_vectors for some
of the devices. Increasing the Linux IRQ count from 8192 has upstream
depencency and for reasons unknown resulted in performance regression
on Armada silicon.
Since netdev drivers are not registering NPA block IRQs, reverse the
NPA and NIX MSIX vectors allocation. This way we will end up using
vectors less than half of what we are using currently. Vectors
allocated are PF_INT + NIX_QINT + NIX_CINT (max queues/cores).
Change-Id: I495c2d43caf0f9a8b12407ef44cfbe13bd01c25e
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17958
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit b1eb5ae22500bf8984c4fe4facd0a55fc659a819 from
git@git.assembla.com:cavium/WindRiver.linux.git
Use napi_complete_done to allow for the use of gro_flush_timeout.
Change-Id: Ic9456cbcd147c5613d5aa99301a80084ab23dd57
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17957
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 2e161e8b861ab07e73257e6c0abf29ffb042ada8 from
git@git.assembla.com:cavium/WindRiver.linux.git
For GRO'ble packets napi_get_frags helps in recycling
SKBs instead of allocating new SKB for every packet.
This improves performance of GRO'ble frames but effects
IP forwarding performance.
Change-Id: I191194cb5feefb1e617a4c7fa152f0e325d9aa1a
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17956
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit e6c9a391e46edc478b6ac0b7bf57493a969979f0 from
git@git.assembla.com:cavium/WindRiver.linux.git
Current cleanup is done using napi structs which when used
in non-napi context leads to issues. This patch fixes this
issue.
Change-Id: I3b13e5c4ec54e4a6a4f902e28170367f88a9ef67
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17955
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit d8a3a2d822c226650d7f362c291d3c4ad0a7031e from
git@git.assembla.com:cavium/WindRiver.linux.git
Major changes done
- Segregate Rx and Tx CQ handlers
- Removed costly CQ_OP_STATUS reads, instead check cqe_type
in cqe_hdr to determine valid CQE.
Change-Id: I14bfcf2dae39178681464b10174fb269a89c46a0
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17954
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit e768cc7824fd22df72cdef19f517697408bd33d3 from
git@git.assembla.com:cavium/WindRiver.linux.git
When switch headers like EDSA, Higig2 etc are present in ingress
or egress pkts the pkt parsing done by NPC needs to take additional
headers into account. KPU profile handles these using specific PKINDs
(the iKPU index) to start parsing pkts differently.
This patch enables user to configure these PKINDs into HW for proper
pkt parsing. Patch also handles changes to max frame size due to
additional headers in pkt.
higig2:
ethtool --set-priv-flags eth0 higig2 on/off
edsa:
ethtool --set-priv-flags eth0 edsa on/off
Change-Id: If27c198a555b480434a5d56e15f0f7cff3800144
Signed-off-by: hariprasad <hkelam@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17912
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit ec5575a4f4505c09b75c2d40b71371ef88c9be15 from
git@git.assembla.com:cavium/WindRiver.linux.git
Drop packets with length more than configured FRS to avoid
SQ Meta-descriptor error.
This patch also include some cleanup changes.
Change-Id: I4c28134e004151094de125189f0f904d6a29a7a2
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17865
Tested-by: sa_ip-sw-jenkins <sa_ip-sw-jenkins@marvell.com>
Reviewed-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
(cherry picked from commit f910cf67294b5f9ff3e902154e51afb218219997)
Reviewed-on: https://sj1git1.cavium.com/17873
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|
|
commit 51edab760bd32c81bd807e61061b92cf105617e4 from
git@git.assembla.com:cavium/WindRiver.linux.git
Adding the default ldata extraction profile for higig2 header.
Ingress side we will extract classification and vid and for
egress side we will extract classification. These fields has been
placed in place of DMAC.
Change-Id: Id439af938b359ded470a3690e2d90dfa7a39c0a0
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Reviewed-on: https://sj1git1.cavium.com/17480
Reviewed-by: Subbaraya Sundeep Bhatta <Sundeep.Bhatta@cavium.com>
Tested-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
|