linux-toradex.git/drivers/net/ethernet/socionext, branch v5.0

cross-tree: phase out dma_zalloc_coherent()

2019-01-08T12:58:37+00:00

We already need to zero out memory for dma_alloc_coherent(), as such
using dma_zalloc_coherent() is superflous. Phase it out.

This change was generated with the following Coccinelle SmPL patch:

@ replace_dma_zalloc_coherent @
expression dev, size, data, handle, flags;
@@

-dma_zalloc_coherent(dev, size, handle, flags)
+dma_alloc_coherent(dev, size, handle, flags)

Suggested-by: Christoph Hellwig 
Signed-off-by: Luis Chamberlain 
[hch: re-ran the script on the latest tree]
Signed-off-by: Christoph Hellwig

net: socionext: remove mmio reads on Tx

2018-12-15T21:15:19+00:00

Currently the driver issues 2 mmio reads to figure out the number of
transmitted packets and clean them. We can get rid of the expensive
reads since BIT 31 of the Tx descriptor can be used for that.
We can also remove the budget counting of Tx completions since all of
the descriptors are not deliberately processed.

Performance numbers using pktgen are:
size  pre-patch(pps)  post-patch(pps)
64       362483           427916
128      358315           411686
256      352725           389683
512      215675           216464
1024     113812           114442

Signed-off-by: Ilias Apalodimas 
Signed-off-by: David S. Miller

net: socionext: correctly recover txq after being full

2018-12-15T21:15:19+00:00

Running pktgen with packets sizes > 512b ends up in the interface Txq
getting stuck.
"netsec 522d0000.ethernet eth0: netsec_netdev_start_xmit: TxQFull!"
appears on dmesg but the interface never recovers. It requires an
ifconfig down/up to make the interface usable again.

The reason that triggers this, is a race condition between
.ndo_start_xmit and the napi completion. The available budget is
calculated first and indicates the queue is full. Due to a costly
netif_err() the queue is not stopped in time while the napi completion
runs, clears the irq and frees up descriptors, thus the queue never wakes
up again.

Fix this by moving the print after stopping the queue, make the print
ratelimited, add barriers and check for cleaned descriptors..

Signed-off-by: Ilias Apalodimas 
Signed-off-by: David S. Miller

Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

2018-12-10T05:43:31+00:00

Several conflicts, seemingly all over the place.

I used Stephen Rothwell's sample resolutions for many of these, if not
just to double check my own work, so definitely the credit largely
goes to him.

The NFP conflict consisted of a bug fix (moving operations
past the rhashtable operation) while chaning the initial
argument in the function call in the moved code.

The net/dsa/master.c conflict had to do with a bug fix intermixing of
making dsa_master_set_mtu() static with the fixing of the tagging
attribute location.

cls_flower had a conflict because the dup reject fix from Or
overlapped with the addition of port range classifiction.

__set_phy_supported()'s conflict was relatively easy to resolve
because Andrew fixed it in both trees, so it was just a matter
of taking the net-next copy.  Or at least I think it was :-)

Joe Stringer's fix to the handling of netns id 0 in bpf_sk_lookup()
intermixed with changes on how the sdif and caller_net are calculated
in these code paths in net-next.

The remaining BPF conflicts were largely about the addition of the
__bpf_md_ptr stuff in 'net' overlapping with adjustments and additions
to the relevant data structure where the MD pointer macros are used.

Signed-off-by: David S. Miller

net: ethernet: ave: Add MODULE_AUTHOR and MAINTAINERS entry

2018-11-29T18:38:39+00:00

Add missing MODULE_AUTHOR of ave driver and an entry to MAINTAINERS.

Signed-off-by: Kunihiko Hayashi 
Signed-off-by: David S. Miller

net: ethernet: ave: Replace NET_IP_ALIGN with AVE_FRAME_HEADROOM

2018-11-29T18:38:39+00:00

In commit 26a4676faa1a ("arm64: mm: define NET_IP_ALIGN to 0"),
AVE controller affects this modification because the controller forces
to ignore lower 2bits of buffer start address, and make 2-byte headroom,
that is, data reception starts from (buffer + 2).

This patch defines AVE_FRAME_HEADROOM macro as hardware-specific value,
and replaces NET_IP_ALIGN with it.

Signed-off-by: Kunihiko Hayashi 
Signed-off-by: David S. Miller

net: ethernet: ave: Increase descriptors to improve performance

2018-11-29T18:38:39+00:00

To improve performance, this increases Rx descriptor to 256, Tx descriptor
to 64, and adjusts NAPI weight to NAPI_POLL_WEIGHT.

Signed-off-by: Kunihiko Hayashi 
Signed-off-by: David S. Miller

net: ethernet: ave: Preserve wol state in suspend/resume sequence

2018-11-29T18:34:45+00:00

Since the wol state forces to be initialized after reset, the state should
be preserved in suspend/resume sequence.

Signed-off-by: Kunihiko Hayashi 
Signed-off-by: David S. Miller

net: ethernet: ave: Set initial wol state to disabled

2018-11-29T18:34:45+00:00

If wol state of phy hardware is enabled after reset, phy_ethtool_get_wol()
returns that wol.wolopts is true.

However, since net_device.wol_enabled is zero and this doesn't apply wol
state until calling ethtool_set_wol(), so mdio_bus_phy_may_suspend()
returns true, that is, it's in a state where phy can suspend even though
wol state is enabled.

In this inconsistency, phy_suspend() returns -EBUSY, and at last,
suspend sequence fails with the following message:

    dpm_run_callback(): mdio_bus_phy_suspend+0x0/0x58 returns -16
    PM: Device 65000000.ethernet-ffffffff:01 failed to suspend: error -16
    PM: Some devices failed to suspend, or early wake event detected

In order to fix the above issue, this patch forces to set initial wol state
to disabled as default.

Signed-off-by: Kunihiko Hayashi 
Signed-off-by: David S. Miller

net: ethernet: ave: Add suspend/resume support

2018-11-29T18:34:45+00:00

This patch introduces suspend and resume functions to ave driver.

Signed-off-by: Kunihiko Hayashi 
Signed-off-by: David S. Miller