mt7921_mcu_sta_add will be used to add and remove wtbl entries.
Create broadcast wtbl entry after AP association
Co-developed-by: Deren Wu <deren.wu@mediatek.com> Signed-off-by: Deren Wu <deren.wu@mediatek.com> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Lorenzo Bianconi [Tue, 20 Apr 2021 21:05:31 +0000 (23:05 +0200)]
mt76: move token_lock, token and token_count in mt76_dev
Move token_lock, token and token_count data structures in mt76_dev.
This is a preliminary patch to move token management in mt76 common
module since it is shared by mt7615, mt7915 and mt7921 drivers.
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Sean Wang [Mon, 19 Apr 2021 15:58:05 +0000 (23:58 +0800)]
mt76: mt7921: fix possible invalid register access
Disable the interrupt and synchronze for the pending irq handlers to ensure
the irq tasklet is not being scheduled after the suspend to avoid the
possible invalid register access acts when the host pcie controller is
suspended.
[17932.910534] mt7921e 0000:01:00.0: pci_pm_suspend+0x0/0x22c returned 0 after 21375 usecs
[17932.910590] pcieport 0000:00:00.0: calling pci_pm_suspend+0x0/0x22c @ 18565, parent: pci0000:00
[17932.910602] pcieport 0000:00:00.0: pci_pm_suspend+0x0/0x22c returned 0 after 8 usecs
[17932.910671] mtk-pcie 11230000.pcie: calling platform_pm_suspend+0x0/0x60 @ 22783, parent: soc
[17932.910674] mtk-pcie 11230000.pcie: platform_pm_suspend+0x0/0x60 returned 0 after 0 usecs
Fixes: ffa1bf97425b ("mt76: mt7921: introduce PM support") Signed-off-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Introduce mt76_connac_mcu_set_deep_sleep to enable deep sleep mode
and will be activated immediately when the host returns the ownership
to the device.
Co-developed-by: Leon Yen <leon.yen@mediatek.com> Signed-off-by: Leon Yen <leon.yen@mediatek.com> Co-developed-by: YN Chen <YN.Chen@mediatek.com> Signed-off-by: YN Chen <YN.Chen@mediatek.com> Signed-off-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Add mt7921_wpdma_reinit_cond to check dummy reg if driver needs to
reinitialized WPDMA after driver_own operation
Co-developed-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Sean Wang <sean.wang@mediatek.com> Co-developed-by: Leon Yen <leon.yen@mediatek.com> Signed-off-by: Leon Yen <leon.yen@mediatek.com> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Introduce mt7921_dma_enable and mt7921_dma_disable utilities routine in
order for code reusing between mt7921_dma_reset and mt7921_dma_init.
This is a preliminary patch to reset dma during device driver_own
request.
Signed-off-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Lorenzo Bianconi [Mon, 19 Apr 2021 14:20:54 +0000 (22:20 +0800)]
mt76: mt7921: move mt7921_dma_reset in dma.c
Move mt7921_dma_reset routine in dma.c and make mt7921_dma_prefetch
static. Moreover add force parameter to mt7921_dma_reset signature.
This is a preliminary patch to reset dma mt7921_mcu_drv_pmctrl.
Signed-off-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Colin Ian King [Mon, 19 Apr 2021 13:20:32 +0000 (14:20 +0100)]
mt76: mt7615: Fix a dereference of pointer sta before it is null checked
Currently the assignment of idx dereferences pointer sta before
sta is null checked, leading to a potential null pointer dereference.
Fix this by assigning idx when it is required after the null check on
pointer sta.
Addresses-Coverity: ("Dereference before null check") Fixes: a4a5a430b076 ("mt76: mt7615: fix TSF configuration") Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Lorenzo Bianconi [Sun, 18 Apr 2021 16:45:37 +0000 (18:45 +0200)]
mt76: connac: remove MT76_STATE_PM in mac_tx_free
Get rid of MT76_STATE_PM chec in mt7615_mac_tx_free and
mt7921_mac_tx_free since we already rely on mt76_connac_pm_unref in the
NAPI callback.
Remove mt76_connac_power_save_sched calls in mt7615_mac_tx_free and
mt7921_mac_tx_free
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Lorenzo Bianconi [Sun, 18 Apr 2021 16:45:36 +0000 (18:45 +0200)]
mt76: connac: check wake refcount in mcu_fw_pmctrl
In order to avoid synchronization races between tx and rx path, rely on
mt76_connac_skip_fw_pmctrl putting the chip in sleep mode for mt7921 and
mt7663 devices
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Lorenzo Bianconi [Sun, 18 Apr 2021 16:45:31 +0000 (18:45 +0200)]
mt76: mt7663: rely on mt76_connac_pm_ref/mt76_connac_pm_unref in tx path
Introduce mt7615_tx_worker routine as mt76 tx worker callback for
mt7663.
Rely on mt76_connac_pm_ref/mt76_connac_pm_unref to check PM state and
increment/decrement wake counter
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Lorenzo Bianconi [Sun, 18 Apr 2021 16:45:30 +0000 (18:45 +0200)]
mt76: mt7921: rely on mt76_connac_pm_ref/mt76_connac_pm_unref in tx path
Introduce mt7921_tx_worker routine as mt76 tx worker callback for
mt7921.
Rely on mt76_connac_pm_ref/mt76_connac_pm_unref to check PM state and
increment/decrement wake counter
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Lorenzo Bianconi [Sun, 18 Apr 2021 16:45:28 +0000 (18:45 +0200)]
mt76: mt7663: fix a race between mt7615_mcu_drv_pmctrl and mt7615_mcu_fw_pmctrl
Introduce a mutex in order to avoid a race between mt7615_mcu_lp_drv_pmctrl and
mt7615_mcu_fw_pmctrl routines since they are run two independent works
Fixes: 1f549009b5b2 ("mt76: mt7615: do not request {driver,fw}_own if already granted") Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Clean up the flow of per-rate txpower limit setting to get rid of
duplicate work since it has already been handled by firmware, and set
proper max_power based on different channels and regdomains.
Signed-off-by: Shayne Chen <shayne.chen@mediatek.com> Signed-off-by: Ryder Lee <ryder.lee@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Sean Wang [Fri, 16 Apr 2021 15:30:36 +0000 (23:30 +0800)]
mt76: mt7921: abort uncompleted scan by wifi reset
Scan abort should be required for the uncompleted hardware scan
interrupted by wifi reset. Otherwise, it is possible that the scan
request after wifi reset gets error code -EBUSY from mac80211 and
then blocks the reconnectting to the access point.
Fixes: 0c1ce9884607 ("mt76: mt7921: add wifi reset support") Signed-off-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Sean Wang [Fri, 16 Apr 2021 15:30:35 +0000 (23:30 +0800)]
mt76: mt7921: add wifisys reset support in debugfs
Introduce chip_reset knob in mt7921 debugfs to export a way to users
able to trigger wifi reset, and group the similar operations
previously defined in chip_config in the same knob.
Signed-off-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Felix Fietkau [Fri, 16 Apr 2021 08:24:46 +0000 (10:24 +0200)]
mt76: mt7615: load ROM patch before checking patch semaphore status
For MT7663, the availability of the patch files is used to detect, which
corresponding firmware is going to be used (AP firmware or STA offload
firmware). If the ROM patch was already applied, it could attempt to
load the wrong firmware (without considering the alternative).
Felix Fietkau [Tue, 13 Apr 2021 09:08:36 +0000 (11:08 +0200)]
mt76: add functions for parsing rate power limits from DT
This subnode can be used to set per-rate tx power limits either per
country code / regdomain or globally.
These limits are typically provided by the device manufacturers and are
used to limit sideband emissions and stay within regulatory limits
Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Add support to send larger packet in testmode to meet requirements
of some test cases.
The limit of max packet size is determined based on tx rate mode setting.
Signed-off-by: Shayne Chen <shayne.chen@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name>
rtlwifi: implement set_tim by update beacon content
Once beacon content is changed, we update the content to wifi card by
send_beacon_frame(). Then, STA with PS can wake up properly to receive its
packets.
Since we update beacon content to PCI wifi devices every beacon interval,
the only one usb device, 8192CU, needs to update beacon content when
mac80211 calling set_tim.
Reported-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name> Signed-off-by: Ping-Ke Shih <pkshih@realtek.com> Tested-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Link: https://lore.kernel.org/r/20210419065956.6085-1-pkshih@realtek.com
We used to stop napi before disabling irqs. And it turns out
to cause some problem when we try to stop device while interrupt arrives.
To safely stop pci, we do three steps:
1. disable interrupt
2. synchronize_irq
3. stop_napi
Since step 2 and 3 may not finish as expected when interrupt is enabled,
use rtwpci->running to decide whether interrupt should be re-enabled at
the time.
Fixes: 9e2fd29864c5 ("rtw88: add napi support") Signed-off-by: Po-Hao Huang <phhuang@realtek.com> Signed-off-by: Ping-Ke Shih <pkshih@realtek.com> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Link: https://lore.kernel.org/r/20210415084703.27255-4-pkshih@realtek.com
If there are lots of packets to be transmitted, the driver would check
whether the available descriptors are sufficient according the read/write
point of tx queue. Once the available descriptor is not enough,
ieee80211_stop_queue is called.
TX ISR, meanwhile, is releasing the tx resources after the packets are
transmitted. This routine may call ieee80211_wake_queue by checking the
available descriptor.
The potential queue stop problem would occur when the tx queue is
stopped due to the heavy traffic. Then thare is no chance to wake the
queue up because the read point is not updated immediately, as a result,
no more packets coulde be transmitted in this queue.
This patch makes sure the ieee80211_wake_queue could be called properly
and avoids the race condition when ring->r.rp, ring->queue_stopped are
updated.
Guo-Feng Fan [Thu, 15 Apr 2021 08:47:01 +0000 (16:47 +0800)]
rtw88: 8821c: Don't set RX_FLAG_DECRYPTED if packet has no encryption
The value of GET_RX_DESC_SWDEC() indicates that if this RX
packet requires software decryption or not. And software
decryption is required when the packet was encrypted and the
hardware failed to decrypt it.
So, GET_RX_DESC_SWDEC() is negative does not mean that this
packet is decrypted, it might just have no encryption at all.
To actually see if the packet is decrypted, driver needs to
further check if the hardware has successfully decrypted it,
with a specific type of encryption algorithm.
The mhi_wwan_rx_budget_dec function is supposed to return true if
RX buffer budget has been successfully decremented, allowing to queue
a new RX buffer for transfer. However the current implementation is
broken when RX budget is '1', in which case budget is decremented but
false is returned, preventing to requeue one buffer, and leading to
RX buffer starvation.
Fixes: fa588eba632d ("net: Add Qcom WWAN control driver") Signed-off-by: Loic Poulain <loic.poulain@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Walle [Tue, 20 Apr 2021 10:29:29 +0000 (12:29 +0200)]
net: phy: at803x: fix probe error if copper page is selected
The commit c329e5afb42f ("net: phy: at803x: select correct page on
config init") selects the copper page during probe. This fails if the
copper page was already selected. In this case, the value of the copper
page (which is 1) is propagated through phy_restore_page() and is
finally returned for at803x_probe(). Fix it, by just using the
at803x_page_write() directly.
Also in case of an error, the regulator is not disabled and leads to a
WARN_ON() when the probe fails. This couldn't happen before, because
at803x_parse_dt() was the last call in at803x_probe(). It is hard to
see, that the parse_dt() actually enables the regulator. Thus move the
regulator_enable() to the probe function and undo it in case of an
error.
Fixes: c329e5afb42f ("net: phy: at803x: select correct page on config init") Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: David Bauer <mail@david-bauer.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Colin Ian King [Tue, 20 Apr 2021 12:27:30 +0000 (13:27 +0100)]
net: mana: remove redundant initialization of variable err
The variable err is being initialized with a value that is
never read and it is being updated later with a new value. The
initialization is redundant and can be removed
Addresses-Coverity: ("Unused value") Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Tue, 20 Apr 2021 09:43:41 +0000 (02:43 -0700)]
virtio-net: fix use-after-free in page_to_skb()
KASAN/syzbot had 4 reports, one of them being:
BUG: KASAN: slab-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline]
BUG: KASAN: slab-out-of-bounds in page_to_skb+0x5cf/0xb70 drivers/net/virtio_net.c:480
Read of size 12 at addr ffff888014a5f800 by task systemd-udevd/8445
Fixes: fb32856b16ad ("virtio-net: page_to_skb() use build_skb when there's sufficient tailroom") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Reported-by: Guenter Roeck <linux@roeck-us.net> Reported-by: Mat Martineau <mathew.j.martineau@linux.intel.com> Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Cc: Jason Wang <jasowang@redhat.com> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: virtualization@lists.linux-foundation.org Acked-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Walle [Tue, 20 Apr 2021 14:28:21 +0000 (16:28 +0200)]
net: enetc: automatically select IERB module
Now that enetc supports flow control we have to make sure the settings in
the IERB are correct. Therefore, we actually depend on the enetc-ierb
module. Previously it was possible that this module was disabled while the
enetc was enabled. Fix it by automatically select the enetc-ierb module.
Fixes: e7d48e5fbf30 ("net: enetc: add a mini driver for the Integrated Endpoint Register Block") Signed-off-by: Michael Walle <michael@walle.cc> Acked-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Tue, 20 Apr 2021 20:01:44 +0000 (13:01 -0700)]
virtio-net: restrict build_skb() use to some arches
build_skb() is supposed to be followed by
skb_reserve(skb, NET_IP_ALIGN), so that IP headers are word-aligned.
(Best practice is to reserve NET_IP_ALIGN+NET_SKB_PAD, but the NET_SKB_PAD
part is only a performance optimization if tunnel encaps are added.)
Unfortunately virtio_net has not provisioned this reserve.
We can only use build_skb() for arches where NET_IP_ALIGN == 0
We might refine this later, with enough testing.
Fixes: fb32856b16ad ("virtio-net: page_to_skb() use build_skb when there's sufficient tailroom") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Guenter Roeck <linux@roeck-us.net> Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Cc: Jason Wang <jasowang@redhat.com> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: virtualization@lists.linux-foundation.org Signed-off-by: David S. Miller <davem@davemloft.net>
bit operation helpers such as test_bit, clear_bit, etc take bit
position as parameter and not value. Current usage causes double
shift => BIT(BIT(0)). Fix that in wwan_core and mhi_wwan_ctrl.
Fixes: 9a44c1cc6388 ("net: Add a WWAN subsystem") Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Loic Poulain <loic.poulain@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Tue, 20 Apr 2021 23:51:20 +0000 (16:51 -0700)]
Merge branch 'dsa-tag-override'
Tobias Waldekranz says:
====================
net: dsa: Allow default tag protocol to be overridden from DT
This is a continuation of the work started in this patch:
https://lore.kernel.org/netdev/20210323102326.3677940-1-tobias@waldekranz.com/
In addition to the mv88e6xxx support to dynamically change the
protocol, it is now possible to override the protocol from the device
tree. This means that when a board vendor finds an incompatibility,
they can specify a working protocol in the DT, and users will not have
to worry about it.
Some background information:
In a system using an NXP T1023 SoC connected to a 6390X switch, we
noticed that TO_CPU frames where not reaching the CPU. This only
happened on hardware port 8. Looking at the DSA master interface
(dpaa-ethernet) we could see that an Rx error counter was bumped at
the same rate. The logs indicated a parser error.
It just so happens that a TO_CPU coming in on device 0, port 8, will
result in the first two bytes of the DSA tag being one of:
00 40
00 44
00 46
My guess was that since these values looked like 802.3 length fields,
the controller's parser would signal an error if the frame length did
not match what was in the header.
This was later confirmed using two different workarounds provided by
Vladimir. Unfortunately these either bypass or ignore the hardware
parser and thus robs working combinations of the ability to do RSS and
other nifty things. It was therefore decided to go with the option of
a DT override.
v1 -> v2:
- Fail if the device does not support changing protocols instead of
falling back to the default. (Andrew)
- Only call change_tag_protocol on CPU ports. (Andrew/Vladimir)
- Only allow changing the protocol on chips that have at least
"undocumented" level of support for EDSA. (Andrew).
- List the supported protocols in the binding documentation. I opted
for only listing the protocols that I have tested. As more people
test their drivers, they can add them. (Rob)
v2 -> v3:
- Rename "dsa,tag-protocol" -> "dsa-tag-protocol". (Rob)
- Some cleanups to 4/5. (Vladimir)
- Add a comment detailing how tree/driver agreement on the tag
protocol is enforced. (Vladimir).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
The 'dsa-tag-protocol' is used to force a switch tree to use a
particular tag protocol, typically because the Ethernet controller
that it is connected to is not compatible with the default one.
Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com> Signed-off-by: David S. Miller <davem@davemloft.net>
net: dsa: Allow default tag protocol to be overridden from DT
Some combinations of tag protocols and Ethernet controllers are
incompatible, and it is hard for the driver to keep track of these.
Therefore, allow the device tree author (typically the board vendor)
to inform the driver of this fact by selecting an alternate protocol
that is known to work.
Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com> Reviewed-by: Vladimir Oltean <olteanv@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
net: dsa: Only notify CPU ports of changes to the tag protocol
Previously DSA ports were also included, on the assumption that the
protocol used by the CPU port had to the matched throughout the entire
tree.
As there is not yet any consumer in need of this, drop the call.
Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com> Reviewed-by: Vladimir Oltean <olteanv@gmail.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
net: dsa: mv88e6xxx: Allow dynamic reconfiguration of tag protocol
For devices that supports both regular and Ethertyped DSA tags, allow
the user to change the protocol.
Additionally, because there are ethernet controllers that do not
handle regular DSA tags in all cases, also allow the protocol to be
changed on devices with undocumented support for EDSA. But, in those
cases, make sure to log the fact that an undocumented feature has been
enabled.
Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com> Reviewed-by: Vladimir Oltean <olteanv@gmail.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
net: dsa: mv88e6xxx: Mark chips with undocumented EDSA tag support
All devices are capable of using regular DSA tags. Support for
Ethertyped DSA tags sort into three categories:
1. No support. Older chips fall into this category.
2. Full support. Datasheet explicitly supports configuring the CPU
port to receive FORWARDs with a DSA tag.
3. Undocumented support. Datasheet lists the configuration from
category 2 as "reserved for future use", but does empirically
behave like a category 2 device.
So, instead of listing the one true protocol that should be used by a
particular chip, specify the level of support for EDSA (support for
regular DSA is implicit on all chips). As before, we use EDSA for all
chips that fully supports it.
In upcoming changes, we will use this information to support
dynamically changing the tag protocol.
Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com> Reviewed-by: Vladimir Oltean <olteanv@gmail.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Tue, 20 Apr 2021 23:44:04 +0000 (16:44 -0700)]
Merge tag 'mac80211-next-for-net-next-2021-04-20' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211-next
Johannes Berg says:
====================
Another set of updates, all over the map:
* set sk_pacing_shift for 802.3->802.11 encap offload
* some monitor support for 802.11->802.3 decap offload
* HE (802.11ax) spec updates
* userspace API for TDLS HE support
* along with various other small features, cleanups and
fixups
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, mlxsw admits for offload a suitable root qdisc, and its
children. Thus up to two levels of hierarchy are offloaded. Often, this is
enough: one can configure TCs with RED and TCs with a shaper, and can even
see counters for each TC by looking at a qdisc at a sufficiently shallow
position.
While simple, the system has obvious shortcomings. It is not possible to
configure both RED and shaping on one TC. It is not possible to place a
PRIO below root TBF, which would then be offloaded as port shaper. FIFOs
are only offloaded at root or directly below, which is confusing to users,
because RED and TBF of course have their own FIFO.
This patchset is a step towards the end goal of allowing more comprehensive
qdisc tree offload and cleans up the qdisc offload code.
- Patches #1-#4 contain small cleanups.
- Up until now, since mlxsw offloaded only a very simple qdisc
configurations, basically all bookkeeping was done using one container
for the root qdisc, and 8 containers for its children. Patches #5, #6, #8
and #9 gradually introduce a more dynamic structure, where parent-child
relationships are tracked directly at qdiscs, instead of being implicit.
- This tree management assumes only one qdisc is created at a time. In FIFO
handlers, this condition was enforced simply by asserting RTNL lock. But
instead of furthering this RTNL dependence, patch #7 converts the whole
qdisc offload logic to a per-port mutex.
- Patch #10 adds a selftest.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Petr Machata [Tue, 20 Apr 2021 14:53:48 +0000 (16:53 +0200)]
selftests: mlxsw: sch_red_ets: Test proper counter cleaning in ETS
There was a bug introduced during the rework which cause non-zero backlog
being stuck at ETS. Introduce a selftest that would have caught the issue
earlier.
Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Petr Machata [Tue, 20 Apr 2021 14:53:47 +0000 (16:53 +0200)]
mlxsw: spectrum_qdisc: Index future FIFOs by band number
mlxsw used to hold an array of qdiscs indexed by the TC number. In the
previous patch, it was changed to allocate child qdiscs dynamically, and
they are now indexed by band number. Follow suit with the array of future
FIFOs.
Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of keeping qdiscs in globally-preallocated arrays, introduce a
per-qdisc-kind value num_classes, and then allocate the necessary child
qdiscs (if any) based on that value. Since now dynamic allocation is
involved, mlxsw_sp_qdisc_replace() gets messy enough that it is worth it to
split it to two cases: a new qdisc allocation and a change of existing
qdisc. (Note that the change also includes what TC formally calls replace,
if the qdisc kind is the same.)
Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / linux-platform-drivers-x86.git / log