The implementation is not efficient on ar8xxx switches. It triggers high
CPU load and degrades device performance.
The high CPU load has been traced down to the ar8xxx_reg_wait() call in
ar8xxx_mib_op(), which has to usleep_range() till the MIB busy flag set
by the request to update the MIB counter is cleared.
This commit removes the get_port_stats() code introduced in 4d8a66d and
leaves a note for future hacker's beware.
Fixes: FS#1004
Signed-off-by: Thibaut VARENE <hacks@slashdirt.org>
This patch provides a generic switch_dev_ops 'get_port_stats()' callback by
taping into the relevant port MIB counters.
The implementation uses a generic callback that select the correct MIB counter
index based on chip version.
This callback is used by swconfig_leds led trigger to blink LEDs with port
network traffic.
Signed-off-by: Thibaut VARENE <hacks@slashdirt.org>
commit 33b72b8e0f
"ar8216: adjust ATU flushing in case of link changes"
introduced portwise flushing on link down events. Now the ARL table could
be in a chaotic state after boot where ar8xxx_sw_get_arl_table looped
forever (depending on the entries collected while booting).
Signed-off-by: Günther Kelleter <guenther.kelleter@devolo.de>
The priv->vlan_id member is of size AR8X16_MAX_VLANS, not AR8X16_MAX_PORTS,
so check for the proper maximum value in order to avoid capping valid VLAN IDs
to 7 (AR8X16_MAX_PORTS - 1).
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Imported from e1aaf7ec00%5E%21/#F0
Signed-off-by: Pavel Kubelun <be.dissent@gmail.com>
CHROMIUM: net: ar8216: address security vulnerabilities in swconfig & ar8216
This patch does the following changes:
*address the security vulnerabilities in both swconfig framework and in
ar8216 driver (many bound check additions, and turned swconfig structure
signed element into unsigned when applicable)
*address a couple of whitespaces and indendation issues
BUG=chrome-os-partner:33096
TEST=none
Change-Id: I94ea78fcce8c1932cc584d1508c6e3b5dfb93ce9
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
Reviewed-on: https://chromium-review.googlesource.com/236490
Reviewed-by: Toshi Kikuchi <toshik@chromium.org>
Commit-Queue: Toshi Kikuchi <toshik@chromium.org>
Tested-by: Toshi Kikuchi <toshik@chromium.org>
Import from fd7b89dd46%5E%21/#F0
Signed-off-by: Pavel Kubelun <be.dissent@gmail.com>
CHROMIUM: drivers: ar8216: prevent device duplication in ar8xxx_dev_list
If probe is called twice, once for PHY0 and a second time for PHY4,
the same switch device will be added twice to ar8xxx_dev_list, while
supposedly this list should have one element per hardware switch present
in the system.
While no negative impact have been observed, it does happen if a
platform instanciates these two PHYs from device-tree, as an example.
Change-Id: Iddcbdf7d4adacb0af01975b73f8e56b4582e894e
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
Reviewed-on: https://chromium-review.googlesource.com/234790
Reviewed-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Toshi Kikuchi <toshik@chromium.org>
Tested-by: Toshi Kikuchi <toshik@chromium.org>
Import from c3fd96a7b8%5E%21/#F0
Signed-off-by: Pavel Kubelun <be.dissent@gmail.com>
CHROMIUM: drivers: ar8216: hold ar8xxx_dev_list_lock during use_count--
It is possible for the remove() callback to run twice in parallel, which
could result into --use_count returning only 1 in both cases and the
rest of the unregistration path to never be reached.
This case has never been observed in practice, but we will fix
preventively to make the code more robust.
BUG=chrome-os-partner:33096
TEST=none
Change-Id: If09abe27fdb2037f514f8674418bafaab3cbdef6
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
Reviewed-on: https://chromium-review.googlesource.com/232870
Reviewed-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Toshi Kikuchi <toshik@chromium.org>
Tested-by: Toshi Kikuchi <toshik@chromium.org>
Import from c05af20272
Signed-off-by: Pavel Kubelun <be.dissent@gmail.com>
CHROMIUM: drivers: ar8216: sync mib_work cancellation
ar8xxx_mib_stop() is called from ar8xxx_phy_remove(), so we want to make
sure the work doesn't run after priv is freed / the device ceases to
exist.
BUG=chrome-os-partner:33096
TEST=none
Change-Id: Iafb44ce93a87433adc4576e5fea5fda58d1f43a9
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
Reviewed-on: https://chromium-review.googlesource.com/232827
Reviewed-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Toshi Kikuchi <toshik@chromium.org>
Reviewed-by: Grant Grundler <grundler@chromium.org>
Tested-by: Toshi Kikuchi <toshik@chromium.org>
The commit "generic: ar8216: add sanity check to ar8216_probe"
(774da6c7a4) stated that PHY IDs
should be checked at address 0-4. However, the PHY 4 was
never check by the loop. This patch extends the check to be
similar to the Atheors SDK. It tries all 4 ports and skips
unconnected PHYs if necessary. If it cannot find any familiar
PHYs, it will prevent the phy driver from initializing.
This patch is necessary for the C-60. It doesn't have a
PHY at port 3, so this caused the check in ar8xxx_is_possible
to fail. As a result, the ethernet ports on the C-60 didn't
work.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Combine all bus operations for one MMD access in one function.
Protecting all these bus operations with one lock also helps
to avoid potential issues due to bus operations intercepting
the register and data write.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48914
The default TTL for address resolution table entries is 5 minutes
for all members of the AR8216 family. This can cause issues if
e.g. Wifi clients roam to another AP and their MAC appears on
another switch port suddenly. Then the client may not be reachable
until the old ARL entry expires.
I would have expected the switch to invalidate old entries if it
detects the same MAC on another port. But that's not the case.
Therefore make the TTL for ARL entries configurable.
The effective TTL will always be a multiple of 7 seconds.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48913
The line before includes the port number anyway so there's no need
to duplicate the port number in the MIB info header.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48912
The decimal values especially for TxByte and RxGoodByte are hard to read
once bigger amounts of data have been transferred.
Therefore complement the decimal values with info in GiB / MiB / KiB.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48911
For unused switch ports all MIB values are zero. Displaying ~40 empty
MIB counters is just confusing and makes it hard to read the output of
swconfig dev <dev> show.
Therefore, if all MIB counters for a port are zero, just display
an info that the MIB counters are empty.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48910
If a link goes down, don't flush the complete ARL table.
Only flush the entries for the respective port.
Don't touch ARL table if a link goes up.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 46381
Adds functions for flushing ARL table entries per port.
Successfully tested on AR8327. Implementation for AR8216/AR8236/AR8316
is based on the AR8236 datasheet and assumes that the three chips
share a common ATU register layout.
Compile-tested only for AR8216/AR8236/AR8316.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 46380
Adds the chip-specific part of reading ARL table for AR8216/AR8236/AR8316.
It's based on the AR8236 datasheet and compile-tested only as I couldn't
find datasheets for AR8216/AR8316 and don't own devices with these chips.
The existing ar8216_atu_flush implementation was used for all three
chip types, therefore I guess they share a common ATU register layout.
More testing would be appreciated.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 46379
Add global read-only swconfig attribute "arl_table" to display the
address resolution table.
So far the chip-specific part is implemented for AR8327/AR8337 only
as I don't have the datasheets for the other AR8XXX chips.
Successfully tested on TL-WDR4300 (AR8327rev2)
and TL-WDR4900 (AR8327rev4).
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 44104
Until a few years ago the page switch wait time was set to msleep(1)
what was changed to usleep_range(1000, 2000) later.
I can not imagine that a low-level operation like switching page
on register level takes so much time.
Most likely the value of 1ms was initially set to check whether
it fixes an issue and then remained w/o further checking whether
also a smaller value would be sufficient.
Now the wait time is set to 5us and I successfully tested this on
AR8327. IMHO 5us should be plenty of time for all supported chips.
However I couldn't test this due to missing hardware.
If other chips should need a longer wait time we can add the
wait time as a parameter to the ar8xxx_chip struct.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 44103
Check for switch port link changes and
- flush ATU in case of a change
- report link change via syslog
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 44102
The functionality to flush the address translation table contains two bugs
which luckily compensate each other.
1. Just setting the operation is not sufficient to perform the flushing.
The "active" bit needs to be set to actually trigger an action.
For the vtu operations this is implemented correctly.
2. ar8xxx_phy_read_status is called every 2s by the phy state machine
to check for link changes. This would have caused an ATU flush
every 2s.
Fix the chip-specific ATU flush functions and remove the ATU flush call
from ar8xxx_phy_read_status.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 44101
AR8327/AR8337 allow to read the result of EEE autonegotiation.
If EEE is autonegotiated between the link partners, display
this as part of the swconfig get_link attribute.
eee100: 100MBit EEE supported by both link partners
eee1000: 1GBit EEE supported by both link partners
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 44022
Users reported network issues with AR8327 which turned out to be caused
by EEE not working correctly with certain link partners (ticket 14597).
The workaround was to disable EEE on all ports (changeset 41577).
The issue was with certain link partners only, therefore this patch
allows to control usage of EEE per port via swconfig.
Still the default is to initially disable EEE on all ports.
Successfully tested on a TL-WDR4900 (AR8327 rev.4)
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 44021
Move all AR8327/AR8337-specific driver code into a separate source file
ar8327.c and adjust patches so that ar8327.c is compiled if
CONFIG_AR8216_PHY is set.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43845
Move several structure definitions and #defines from ar8216.c
to ar8216.h and move AR8327/AR8337 header stuff into a new
header file ar8327.h.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43844
Remove read/write/rmw member functions from ar8xxx_priv
There seems to be no real benefit of the ar8xxx_priv member functions
read/write/rmw as one implementation exists for each of them only.
Especially ar8xxx_mii_rmw is assigned to priv->rmw first and then
mapped to ar8xxx_rmw.
Rename the ar8xxx_mii_.. functions to ar8xxx_.. and use them directly.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43742
Create helpers mii_read32 / mii_write32 for 32 bit MII ops.
Rename r3 variable to page in ar8xxx_mii_write to make it consistent
with the other ar8xxx_mii_xxxx functions.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43741
Factor out chip-specific parameters from ar8xxx_probe_switch.
Move the ar8xxx_chip definitions after the swops definitions.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43740
Factor out set_mirror_regs to ar8xxx_chip.
Remove related chip_is_... checks.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43468
Factor out info whether switch should be configured at probe stage
to ar8xxx_chip. Remove related chip_is_... checks.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43466
Patch reverts 43332 which seems to cause issues with VLAN functionality.
Add a specific check to check whether ANEG is still enabled and re-enable
it if necessary. Disable generic phy soft reset for kernel >=3.16.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43356
We should make sure that also for ar8216 hw gets initialized.
For ar8216 hw_init is a dummy currently. The hw_init used for ar8236
should be generic enough to be usable with ar8216 too.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43334
Move the PHY fixup call to the PHY init loop.
Use ar8xxx_has_gige in the PHY init instead of passing the gigE
capability via function parameter.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43333
Kernel 3.14 introduced a switch reset in phy_init_hw in drivers/net/phy
causing BMCR_ANENABLE to get cleared.
Due to the fact that ar8xxx_phy_config_aneg does nothing for
PHY 0 autonegatiation support remains disabled.
This can cause ports to operate at 10MBit/half-duplex only.
Fix this by calling genphy_config_aneg for PHY 0 too as
genphy_config_aneg sets BMCR_ANENABLE if it's not yet set.
Fixes: ticket 17800
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43332
PHY init code in the switch-specific hw_init functions is mainly
identical. Factor it out into a generic ar8xxx_phy_init function.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43331
