openwrt/staging/blogic.git
5 years agomt76: mt7603: remove unnecessary mcu queue initialization
Lorenzo Bianconi [Fri, 9 Aug 2019 17:25:15 +0000 (19:25 +0200)]
mt76: mt7603: remove unnecessary mcu queue initialization

Remove unnecessary mcu queue initialization in mt7603_dma_init since it
has been already done in mt76_mmio_init

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: move mt76_tx_tasklet in mt76 module
Lorenzo Bianconi [Fri, 9 Aug 2019 17:06:02 +0000 (19:06 +0200)]
mt76: move mt76_tx_tasklet in mt76 module

Move mt76{15,03}_tx_tasklet in mt76_alloc_device in order to be used as
default tx_tasklet initialization. Remove duplicated code in
mt7603/mt7615 drivers

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: enable survey support
Lorenzo Bianconi [Mon, 5 Aug 2019 06:55:53 +0000 (08:55 +0200)]
mt76: mt7615: enable survey support

Introduce channel survey support for mt7615 driver

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7603: move survey_time in mt76_dev
Lorenzo Bianconi [Mon, 5 Aug 2019 06:55:52 +0000 (08:55 +0200)]
mt76: mt7603: move survey_time in mt76_dev

Move survey_time field in mt76_dev in order to be reused adding survey
support to mt7615 driver

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt76x02u: enable survey support
Lorenzo Bianconi [Mon, 5 Aug 2019 06:55:51 +0000 (08:55 +0200)]
mt76: mt76x02u: enable survey support

Introduce channel survey support for mt76x2u and mt76x0u drivers

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt76x02u: enable multi-vif support
Lorenzo Bianconi [Fri, 2 Aug 2019 14:36:20 +0000 (16:36 +0200)]
mt76: mt76x02u: enable multi-vif support

Enable multi-interface support for mt76x02u driver. For the moment
allow max two concurrent interfaces in order to preserve enough room
for ps traffic since we are using beacon slots for it.
I have successfully tested the following configuration:
- AP + STA
- AP0 + AP1

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Tested-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt76x02: do not copy beacon skb in mt76x02_mac_set_beacon_enable
Lorenzo Bianconi [Thu, 1 Aug 2019 08:26:23 +0000 (10:26 +0200)]
mt76: mt76x02: do not copy beacon skb in mt76x02_mac_set_beacon_enable

Do not copy beacon skb in mt76x02_mac_set_beacon_enable for usb devices
since it will be done in mt76x02_update_beacon_iter. Moreover squash
mt76x02_mac_set_beacon_enable and __mt76x02_mac_set_beacon_enable since
the latter is run just by mt76x02_mac_set_beacon_enable

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt76x02: introduce mt76x02_pre_tbtt_enable and mt76x02_beacon_enable macros
Lorenzo Bianconi [Thu, 1 Aug 2019 08:26:22 +0000 (10:26 +0200)]
mt76: mt76x02: introduce mt76x02_pre_tbtt_enable and mt76x02_beacon_enable macros

Improve code readability introducing mt76x02_pre_tbtt_enable and
mt76x02_beacon_enable utility macros

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: add Smart Carrier Sense support
Lorenzo Bianconi [Sun, 28 Jul 2019 19:03:18 +0000 (21:03 +0200)]
mt76: mt7615: add Smart Carrier Sense support

Introduce Smart Carrier Sense support in order to tune device
sensitivity according to RTS error rate and False CCA reported by the
radio

Tested-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: rework locking scheme for mt7615_set_channel
Lorenzo Bianconi [Sun, 28 Jul 2019 19:03:17 +0000 (21:03 +0200)]
mt76: mt7615: rework locking scheme for mt7615_set_channel

As already done for mt7603 driver, move mt76.mutex lock inside
mt7615_set_channel since we need to grab mt76.mutex in mt7615_mac_work.
This is a preliminary patch to add Smart Carrier Sense (SCS) support

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: switch to SPDX tag instead of verbose boilerplate text
Ryder Lee [Wed, 24 Jul 2019 08:58:20 +0000 (16:58 +0800)]
mt76: switch to SPDX tag instead of verbose boilerplate text

No functional change intended.

Add SPDX identifiers to all remaining files in /mt76.

Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt76x02: fix some checkpatch warnings
Ryder Lee [Wed, 24 Jul 2019 08:58:18 +0000 (16:58 +0800)]
mt76: mt76x02: fix some checkpatch warnings

This fixes the following checkpatch warnings:

ERROR: code indent should use tabs where possible
CHECK: Alignment should match open parenthesis
CHECK: No space is necessary after a cast
CHECK: Please don't use multiple blank lines
CHECK: Avoid precedence issues in macro
WARNING: Statements should start on a tabstop
WARNING: Unnecessary space before function pointer arguments

Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: fix some checkpatch warnings
Ryder Lee [Wed, 24 Jul 2019 08:58:17 +0000 (16:58 +0800)]
mt76: mt7615: fix some checkpatch warnings

This fixes the following checkpatch warnings:

WARNING: Improper SPDX comment style
Fix blank lines.

Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7603: fix some checkpatch warnings
Ryder Lee [Wed, 24 Jul 2019 08:58:16 +0000 (16:58 +0800)]
mt76: mt7603: fix some checkpatch warnings

This fixes the following checkpatch warnings:

WARNING: Improper SPDX comment style
CHECK: No space is necessary after a cast

Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: Add paragraphs to describe the config symbols fully
Ryder Lee [Wed, 24 Jul 2019 08:58:15 +0000 (16:58 +0800)]
mt76: Add paragraphs to describe the config symbols fully

Update the help text to fix a checkpatch warning:

WARNING: please write a paragraph that describes the config symbol fully

Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: update cw_min/max related settings
Ryder Lee [Mon, 22 Jul 2019 08:50:09 +0000 (16:50 +0800)]
mt76: mt7615: update cw_min/max related settings

Add default values of cw_min/max and use fls() for configuration.

Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: add 4 WMM sets support
Ryder Lee [Mon, 22 Jul 2019 08:50:08 +0000 (16:50 +0800)]
mt76: mt7615: add 4 WMM sets support

Hardware supports 4 sets of WMM that should be put to good use.
And fix incorrect queue mapping in mt7615_conf_tx().

Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7603: fix watchdog rescheduling in mt7603_set_channel
Lorenzo Bianconi [Thu, 18 Jul 2019 22:50:42 +0000 (00:50 +0200)]
mt76: mt7603: fix watchdog rescheduling in mt7603_set_channel

Convert MT7603_WATCHDOG_TIME in jiffies rescheduling watchdog delayed
work

Fixes: c8846e101502 ("mt76: add driver for MT7603E and MT7628/7688")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: fix MT7615_WATCHDOG_TIME definition
Lorenzo Bianconi [Thu, 18 Jul 2019 22:44:16 +0000 (00:44 +0200)]
mt76: mt7615: fix MT7615_WATCHDOG_TIME definition

Express watchdog timeout in jiffies since it is used directly in
ieee80211_queue_delayed_work

Fixes: 04b8e65922f6 ("mt76: add mac80211 driver for MT7615 PCIe-based chipsets")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: introduce mt76_mmio_read_copy routine
Lorenzo Bianconi [Sat, 13 Jul 2019 15:09:06 +0000 (17:09 +0200)]
mt76: introduce mt76_mmio_read_copy routine

Add mt76_mmio_read_copy routine and the related function pointer in
mt76_bus_ops data structure. mt76_mmio_read_copy will be used to add
BIP_CMAC_128 cipher hw support to mt7615 driver

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: add set_key_cmd and mt76_wcid to mt7615_mac_wtbl_set_key signature
Lorenzo Bianconi [Sat, 13 Jul 2019 15:09:05 +0000 (17:09 +0200)]
mt76: mt7615: add set_key_cmd and mt76_wcid to mt7615_mac_wtbl_set_key signature

Introduce set_key_cmd and mt76_wcid pointer to mt7615_mac_wtbl_set_key
signature and do not set key to NULL if cmd is DISABLE_KEY.
This is a preliminary patch to add BIP_CMAC_128 hw support to mt7615
driver

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: remove wtbl_sec_key definition
Lorenzo Bianconi [Sat, 13 Jul 2019 15:09:04 +0000 (17:09 +0200)]
mt76: mt7615: remove wtbl_sec_key definition

Get rid of wtbl_sec_key definition since it is no longer used

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: introduce mt7615_mac_wtbl_set_key routine
Lorenzo Bianconi [Sat, 13 Jul 2019 15:09:03 +0000 (17:09 +0200)]
mt76: mt7615: introduce mt7615_mac_wtbl_set_key routine

Add mt7615_mac_wtbl_set_key routine to configure wtbl key parameter
directly from host cpu. This is a preliminary patch to add BIP_CMAC_128
hw support. Moreover add static qualifier to mt7615_mac_get_key_info
routine

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: add mt7615_mac_wtbl_addr routine
Lorenzo Bianconi [Sat, 13 Jul 2019 15:09:02 +0000 (17:09 +0200)]
mt76: mt7615: add mt7615_mac_wtbl_addr routine

Introduce mt7615_mac_wtbl_addr rouinte to compute sta wtbl address.
This is a preliminary patch to update wtbl key directly from host
processor

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: move mt7615_mac_get_key_info in mac.c
Lorenzo Bianconi [Sat, 13 Jul 2019 15:09:01 +0000 (17:09 +0200)]
mt76: mt7615: move mt7615_mac_get_key_info in mac.c

This is a preliminary patch to update wtbl key directly from host
processor

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: add missing register initialization
Felix Fietkau [Tue, 16 Jul 2019 16:12:07 +0000 (18:12 +0200)]
mt76: mt7615: add missing register initialization

- initialize CCA signal source
- initialize clock for band 1 (7615D)
- initialize BAR rate

Reviewed-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt76x0u: add support to TP-Link T2UHP
Lorenzo Bianconi [Fri, 5 Jul 2019 19:01:04 +0000 (21:01 +0200)]
mt76: mt76x0u: add support to TP-Link T2UHP

Introduce support to TP-Link T2UHP
https://wikidevi.com/wiki/TP-LINK_Archer_T2UHP

Tested-by: Sid Hayn <sidhayn@gmail.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: use params->ssn value directly
Stanislaw Gruszka [Fri, 12 Jul 2019 12:08:00 +0000 (14:08 +0200)]
mt76: mt7615: use params->ssn value directly

There is no point to use pointer to params->ssn.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7603: use params->ssn value directly
Stanislaw Gruszka [Fri, 12 Jul 2019 12:07:59 +0000 (14:07 +0200)]
mt76: mt7603: use params->ssn value directly

There is no point to use pointer to params->ssn.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt76x02: use params->ssn value directly
Stanislaw Gruszka [Fri, 12 Jul 2019 12:07:58 +0000 (14:07 +0200)]
mt76: mt76x02: use params->ssn value directly

There is no point to use pointer to params->ssn.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: usb: remove unneeded {put,get}_unaligned
Stanislaw Gruszka [Tue, 9 Jul 2019 15:14:55 +0000 (17:14 +0200)]
mt76: usb: remove unneeded {put,get}_unaligned

Compiler give us guarantees on variables alignment, so use
an variable as buffer when read/write registers and remove
unneeded {put,get}_unaligned.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: usb: fix endian in mt76u_copy
Stanislaw Gruszka [Tue, 9 Jul 2019 15:14:54 +0000 (17:14 +0200)]
mt76: usb: fix endian in mt76u_copy

In contrast to mt76_wr() which we use to program registers,
on mt76_wr_copy() we should not change endian of the data.

Fixes: b40b15e1521f ("mt76: add usb support to mt76 layer")
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7603: fix invalid fallback rates
Felix Fietkau [Fri, 12 Jul 2019 06:19:35 +0000 (08:19 +0200)]
mt76: mt7603: fix invalid fallback rates

Only decrement the rate index on duplicate rates if it is not already 0

Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: fix invalid fallback rates
Felix Fietkau [Thu, 11 Jul 2019 19:34:25 +0000 (21:34 +0200)]
mt76: mt7615: fix invalid fallback rates

Only decrement the rate index on duplicate rates if it is not already 0

Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: fix PS buffering of action frames
Felix Fietkau [Thu, 11 Jul 2019 19:32:00 +0000 (21:32 +0200)]
mt76: mt7615: fix PS buffering of action frames

Bufferable management frames need to be put in the data queue, otherwise
they will not be buffered when the receiver is asleep.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: fix using VHT STBC rates
Felix Fietkau [Thu, 11 Jul 2019 19:17:06 +0000 (21:17 +0200)]
mt76: mt7615: fix using VHT STBC rates

The hardware expects MT_TX_RATE_NSS to be filled with the number of
space/time streams. For non-STBC rates, this is equal to nss.
For 1-stream STBC, this needs to be set to 2.
This is relevant for VHT rates only, on HT, the value is derived from MCS
internally.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt76u: fix typo in mt76u_fill_rx_sg
Lorenzo Bianconi [Wed, 3 Jul 2019 22:59:19 +0000 (00:59 +0200)]
mt76: mt76u: fix typo in mt76u_fill_rx_sg

Fix typo setting urb->transfer_buffer_length in mt76u_fill_rx_sg

Fixes: b40b15e1521f ("mt76: add usb support to mt76 layer")
Fixes: f8f527b16db5 ("mt76: usb: use EP max packet aligned buffer sizes for rx")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: sync with mt7603 rate control changes
Felix Fietkau [Wed, 3 Jul 2019 19:39:04 +0000 (21:39 +0200)]
mt76: mt7615: sync with mt7603 rate control changes

- Store the previous and current rate set in the driver + the TSF value
  at the time of the switch.
- Use the tx status TSF value to determine which rate set needs to be used
  as reference.
- Report only short or long GI rates for a single status event, not a mix.
- The hardware reports the last used rate index. Use it along with the
  retry count to figure out what rate was used for the first attempt.
- Use the same retry count value for all rate slots to make this calculation
  work.
- Derive the probe rate from the current rateset instead of the skb cb
- Do not wait for a status report for the probe frame before removing the
  probe rate from the rate table. Do it immediately after it was referenced
  in a tx status report.
- Use the first half of the first rate retry budget for the probe rate
  in order to avoid using too many retries on that rate
- Switch from lower rates to higher rates more conservatively
- enable hardware rate up/down selection

Reviewed-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: reset rate index/counters on rate table update
Felix Fietkau [Wed, 3 Jul 2019 19:32:10 +0000 (21:32 +0200)]
mt76: mt7615: reset rate index/counters on rate table update

These values must be initialized to zero, otherwise the hardware could
reuse previous values, especially the rate index

Reviewed-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: move mt7615_mcu_set_rates to mac.c
Felix Fietkau [Wed, 3 Jul 2019 18:08:35 +0000 (20:08 +0200)]
mt76: mt7615: move mt7615_mcu_set_rates to mac.c

It bypasses the MCU, so it does not belong in mcu.c
Also make mt7615_mac_tx_rate_val static

Reviewed-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7603: enable hardware rate up/down selection
Felix Fietkau [Wed, 3 Jul 2019 19:44:23 +0000 (21:44 +0200)]
mt76: mt7603: enable hardware rate up/down selection

Improves performance by switching away from bad rates faster

Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: introduce mt7615_mcu_send_ram_firmware routine
Lorenzo Bianconi [Tue, 2 Jul 2019 11:39:49 +0000 (13:39 +0200)]
mt76: mt7615: introduce mt7615_mcu_send_ram_firmware routine

Add mt7615_mcu_send_ram_firmware routine since mt7615_load_ram runs the
same code to send ram firmware to cr4 and n9 mcus. Moreover rename
gen_dl_mode in mt7615_mcu_gen_dl_mode.
This patch does not introduce any behaviour change, it is just code
refactor.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: always release sem in mt7615_load_patch
Lorenzo Bianconi [Tue, 2 Jul 2019 09:24:51 +0000 (11:24 +0200)]
mt76: mt7615: always release sem in mt7615_load_patch

Release patch semaphore even if request_firmware fails in
mt7615_load_patch

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: fall back to sw encryption for unsupported ciphers
Lorenzo Bianconi [Mon, 1 Jul 2019 13:04:39 +0000 (15:04 +0200)]
mt76: mt7615: fall back to sw encryption for unsupported ciphers

Fix following warning falling back to sw encryption for unsupported
ciphers

WARNING: CPU: 2 PID: 1495 at backports-4.19.32-1/net/mac80211/key.c:1023
mt76_wcid_key_setup+0x68/0xbc [mt76]
CPU: 2 PID: 1495 Comm: hostapd Not tainted 4.14.131 #0
Stack : 00000000 8f0f8bc0 00000000 8007ccec 805f0000 8058ec18 00000000 00000000
80559788 8dca79bc 8fefb10c 805c89c7 805545c8 00000001 8dca7960 53261662
00000000 00000000 80640000 00004668 00000000 000000e9 00000007 00000000
00000000 805d0000 00072537 00000000 80000000 00000000 805f0000 8f1e70d0
8e8fa098 000003ff 805c0000 8f0f8bc0 00000001 802d4340 00000008 80630008
[<800108d0>] show_stack+0x58/0x100
[<8049214c>] dump_stack+0x9c/0xe0
[<80033998>] __warn+0xe0/0x138
[<80033a80>] warn_slowpath_null+0x1c/0x2c
[<8e8fa098>] mt76_wcid_key_setup+0x68/0xbc [mt76]
[<8e889930>] mt7615_eeprom_init+0x7c0/0xe14 [mt7615e]

Suggested-by: Sebastian Gottschall <s.gottschall@newmedia-net.de>
Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: clean up FWDL TXQ during/after firmware upload
Felix Fietkau [Mon, 1 Jul 2019 18:17:40 +0000 (20:17 +0200)]
mt76: mt7615: clean up FWDL TXQ during/after firmware upload

Since we don't clean that tx queue from the tx tasklet, we need to do it
after the firmware upload is done. This patch also adds a cleanup step during
the upload, to help reclaim memory faster.

Fixes unprocessed queued frames eating up memory  long after the firmware
upload has already completed

Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: add radar pattern test knob to debugfs
Lorenzo Bianconi [Sat, 29 Jun 2019 10:36:11 +0000 (12:36 +0200)]
mt76: mt7615: add radar pattern test knob to debugfs

Introduce mt7615_mcu_rdd_send_pattern routine to trigger a radar pattern
detection. Moreover move debugfs related routines in a dedicated source
file.

Suggested-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: add csa support
Lorenzo Bianconi [Sat, 29 Jun 2019 10:36:10 +0000 (12:36 +0200)]
mt76: mt7615: add csa support

Add Channel Switch Announcement support to mt7615 driver updating beacon
template with CSA IE received from mac80211

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: do not perform txcalibration before cac is complited
Lorenzo Bianconi [Sat, 29 Jun 2019 10:36:08 +0000 (12:36 +0200)]
mt76: mt7615: do not perform txcalibration before cac is complited

Delay channel calibration after Channel Availability Check. Add some
code cleanup to mt7615_mcu_set_channel

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: add hw dfs pattern detector support
Lorenzo Bianconi [Sat, 29 Jun 2019 10:36:07 +0000 (12:36 +0200)]
mt76: mt7615: add hw dfs pattern detector support

Add hw radar detection support to mt7615 driver in order to
unlock dfs channels on 5GHz band

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: introduce mt7615_regd_notifier
Lorenzo Bianconi [Sat, 29 Jun 2019 10:36:06 +0000 (12:36 +0200)]
mt76: mt7615: introduce mt7615_regd_notifier

Introduce mt7615_regd_notifier callback. This is a preliminary patch to
add radar detection support to mt7615 driver

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: mt7615: fix sparse warnings: warning: restricted __le16 degrades to integer
Lorenzo Bianconi [Thu, 27 Jun 2019 14:49:44 +0000 (16:49 +0200)]
mt76: mt7615: fix sparse warnings: warning: restricted __le16 degrades to integer

Fix the following sparse warning in __mt7615_mcu_msg_send:
drivers/net/wireless/mediatek/mt76/mt7615/mcu.c:78:15: sparse: warning:
restricted __le16 degrades to integer
drivers/net/wireless/mediatek/mt76/mt7615/mcu.c:78:15: sparse: warning:
cast from restricted __le16

Fixes: 04b8e65922f6 ("mt76: add mac80211 driver for MT7615 PCIe-based chipsets")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agomt76: round up length on mt76_wr_copy
Felix Fietkau [Mon, 1 Jul 2019 11:15:07 +0000 (13:15 +0200)]
mt76: round up length on mt76_wr_copy

When beacon length is not a multiple of 4, the beacon could be sent with
the last 1-3 bytes corrupted. The skb data is guaranteed to have enough
room for reading beyond the end, because it is always followed by
skb_shared_info, so rounding up is safe.
All other callers of mt76_wr_copy have multiple-of-4 length already.

Cc: stable@vger.kernel.org
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5 years agozd1211rw: zd_usb: Use struct_size() helper
Gustavo A. R. Silva [Fri, 30 Aug 2019 18:57:16 +0000 (13:57 -0500)]
zd1211rw: zd_usb: Use struct_size() helper

One of the more common cases of allocation size calculations is finding
the size of a structure that has a zero-sized array at the end, along
with memory for some number of elements for that array. For example:

struct usb_int_regs {
...
        struct reg_data regs[0];
} __packed;

Make use of the struct_size() helper instead of an open-coded version
in order to avoid any potential type mistakes.

So, replace the following function:

static int usb_int_regs_length(unsigned int count)
{
       return sizeof(struct usb_int_regs) + count * sizeof(struct reg_data);
}

with:

struct_size(regs, regs, count)

This code was detected with the help of Coccinelle.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agobrcmfmac: get chip's default RAM info during PCIe setup
Rafał Miłecki [Thu, 29 Aug 2019 08:27:01 +0000 (10:27 +0200)]
brcmfmac: get chip's default RAM info during PCIe setup

Getting RAM info just once per driver's lifetime (during chip
recognition) is not enough as it may get adjusted later (depending on
the used firmware). Subsequent inits may load different firmwares so a
full RAM recognition is required on every PCIe setup. This is especially
important since implementing hardware reset on a firmware crash.

Moreover calling brcmf_chip_get_raminfo() makes sure that RAM core is
up. It's important as having BCMA_CORE_SYS_MEM down on BCM4366 was
resulting in firmware failing to initialize and following error:
[   65.657546] brcmfmac 0000:01:00.0: brcmf_pcie_download_fw_nvram: Invalid shared RAM address 0x04000001

This change makes brcmf_chip_get_raminfo() call during chip recognition
redundant for PCIe devices but SDIO and USB still need it and it's a
very small overhead anyway.

Fixes: 4684997d9eea ("brcmfmac: reset PCIe bus on a firmware crash")
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agobcma: fix incorrect update of BCMA_CORE_PCI_MDIO_DATA
Colin Ian King [Tue, 27 Aug 2019 08:16:20 +0000 (09:16 +0100)]
bcma: fix incorrect update of BCMA_CORE_PCI_MDIO_DATA

An earlier commit re-worked the setting of the bitmask and is now
assigning v with some bit flags rather than bitwise or-ing them
into v, consequently the earlier bit-settings of v are being lost.
Fix this by replacing an assignment with the bitwise or instead.

Addresses-Coverity: ("Unused value")
Fixes: 2be25cac8402 ("bcma: add constants for PCI and use them")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agobrcmfmac: replace strncpy() by strscpy()
Xulin Sun [Fri, 23 Aug 2019 07:47:08 +0000 (15:47 +0800)]
brcmfmac: replace strncpy() by strscpy()

The strncpy() may truncate the copied string,
replace it by the safer strscpy().

To avoid below compile warning with gcc 8.2:

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c:In function 'brcmf_vndr_ie':
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c:4227:2:
warning: 'strncpy' output truncated before terminating nul copying 3 bytes from a string of the same length [-Wstringop-truncation]
  strncpy(iebuf, add_del_cmd, VNDR_IE_CMD_LEN - 1);
  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Signed-off-by: Xulin Sun <xulin.sun@windriver.com>
Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agort2x00: do not set IEEE80211_TX_STAT_AMPDU_NO_BACK on tx status
Stanislaw Gruszka [Fri, 23 Aug 2019 07:09:56 +0000 (09:09 +0200)]
rt2x00: do not set IEEE80211_TX_STAT_AMPDU_NO_BACK on tx status

According to documentation IEEE80211_TX_STAT_AMPDU_NO_BACK is suppose
to be used when we do not recive BA (BlockAck). However on rt2x00 we
use it when remote station fail to decode one or more subframes within
AMPDU (some bits are not set in BlockAck bitmap). Setting the flag result
in sent of BAR (BlockAck Request) frame and this might result of abuse
of BA session, since remote station can sent BA with incorrect
sequence numbers after receiving BAR. This problem is visible especially
when connecting two rt2800 devices.

Previously I observed some performance benefits when using the flag
when connecting with iwlwifi devices. But currently possibly due
to reacent changes in rt2x00 removing the flag has no effect on
those test cases.

So remove the IEEE80211_TX_STAT_AMPDU_NO_BACK.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agoipw2x00: fix spelling mistake "initializationg" -> "initialization"
Colin Ian King [Thu, 22 Aug 2019 22:00:25 +0000 (23:00 +0100)]
ipw2x00: fix spelling mistake "initializationg" -> "initialization"

There is a spelling mistake in an IPW_DEBUG_INFO message. Fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agoairo: fix memory leaks
Wenwen Wang [Fri, 16 Aug 2019 03:50:02 +0000 (22:50 -0500)]
airo: fix memory leaks

In proc_BSSList_open(), 'file->private_data' is allocated through kzalloc()
and 'data->rbuffer' is allocated through kmalloc(). In the following
execution, if an error occurs, they are not deallocated, leading to memory
leaks. To fix this issue, free the allocated memory regions before
returning the error.

Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agozd1211rw: remove false assertion from zd_mac_clear()
Oliver Neukum [Tue, 13 Aug 2019 12:04:11 +0000 (14:04 +0200)]
zd1211rw: remove false assertion from zd_mac_clear()

The function is called before the lock which is asserted was ever used.
Just remove it.

Reported-by: syzbot+74c65761783d66a9c97c@syzkaller.appspotmail.com
Signed-off-by: Oliver Neukum <oneukum@suse.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agobrcmfmac: remove redundant assignment to pointer hash
Colin Ian King [Fri, 9 Aug 2019 17:22:17 +0000 (18:22 +0100)]
brcmfmac: remove redundant assignment to pointer hash

The pointer hash is being initialized with a value that is never read
and is being re-assigned a little later on. The assignment is
redundant and hence can be removed.

Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agobcma: remove two unused variables
YueHaibing [Fri, 9 Aug 2019 08:53:08 +0000 (16:53 +0800)]
bcma: remove two unused variables

drivers/bcma/driver_mips.c:70:18: warning:
 ipsflag_irq_shift defined but not used [-Wunused-const-variable=]
drivers/bcma/driver_mips.c:62:18: warning:
 ipsflag_irq_mask defined but not used [-Wunused-const-variable=]

They are never used, so can be removed.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agobrcm80211: Avoid possible null-pointer dereferences in wlc_phy_radio_init_2056()
Jia-Ju Bai [Mon, 29 Jul 2019 09:56:52 +0000 (17:56 +0800)]
brcm80211: Avoid possible null-pointer dereferences in wlc_phy_radio_init_2056()

In wlc_phy_radio_init_2056(), regs_SYN_2056_ptr, regs_TX_2056_ptr and
regs_RX_2056_ptr may be not assigned, and thus they are still NULL.
Then, they are used on lines 20042-20050:
    wlc_phy_init_radio_regs(pi, regs_SYN_2056_ptr, (u16) RADIO_2056_SYN);
wlc_phy_init_radio_regs(pi, regs_TX_2056_ptr, (u16) RADIO_2056_TX0);
wlc_phy_init_radio_regs(pi, regs_TX_2056_ptr, (u16) RADIO_2056_TX1);
wlc_phy_init_radio_regs(pi, regs_RX_2056_ptr, (u16) RADIO_2056_RX0);
wlc_phy_init_radio_regs(pi, regs_RX_2056_ptr, (u16) RADIO_2056_RX1);

Thus, possible null-pointer dereferences may occur.

To avoid these bugs, when these variables are not assigned,
wlc_phy_radio_init_2056() directly returns.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortw88: pci: enable MSI interrupt
Yu-Yen Ting [Tue, 3 Sep 2019 09:14:07 +0000 (17:14 +0800)]
rtw88: pci: enable MSI interrupt

MSI interrupt should be enabled on certain platform.

Add a module parameter disable_msi to disable MSI interrupt,
driver will then use legacy interrupt instead.

One could rebind the PCI device, probe() will pick up the
new value of the module parameter. Such as:

    echo '0000:01:00.0' > /sys/bus/pci/drivers/rtw_pci/unbind
    echo '0000:01:00.0' > /sys/bus/pci/drivers/rtw_pci/bind

Tested-by: Ján Veselý <jano.vesely@gmail.com>
Reviewed-by: Brian Norris <briannorris@chromium.org>
Reviewed-by: Daniel Drake <drake@endlessm.com>
Signed-off-by: Yu-Yen Ting <steventing@realtek.com>
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortw88: pci: Move a mass of jobs in hw IRQ to soft IRQ
Jian-Hong Pan [Tue, 3 Sep 2019 09:14:06 +0000 (17:14 +0800)]
rtw88: pci: Move a mass of jobs in hw IRQ to soft IRQ

There is a mass of jobs between spin lock and unlock in the hardware
IRQ which will occupy much time originally. To make system work more
efficiently, this patch moves the jobs to the soft IRQ (bottom half) to
reduce the time in hardware IRQ.

Signed-off-by: Jian-Hong Pan <jian-hong@endlessm.com>
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortw88: remove redundant assignment to pointer debugfs_topdir
Colin Ian King [Thu, 22 Aug 2019 11:37:28 +0000 (12:37 +0100)]
rtw88: remove redundant assignment to pointer debugfs_topdir

Pointer debugfs_topdir is initialized to a value that is never read
and it is re-assigned later. 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: Kalle Valo <kvalo@codeaurora.org>
5 years agortw88: drop unused rtw_coex_coex_dm_reset()
Guenter Roeck [Mon, 19 Aug 2019 18:17:57 +0000 (11:17 -0700)]
rtw88: drop unused rtw_coex_coex_dm_reset()

0day reports:

sparse warnings:

drivers/net/wireless/realtek/rtw88/coex.c:2457:6: sparse:
symbol 'rtw_coex_coex_dm_reset' was not declared. Should it be static?

rtw_coex_coex_dm_reset() is not called. Remove it.

Fixes: 4136214f7c46 ("rtw88: add BT co-existence support")
Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortw88: Fix an error message
Dan Carpenter [Wed, 14 Aug 2019 10:46:42 +0000 (13:46 +0300)]
rtw88: Fix an error message

The WARN_ON() macro takes a condition, not a warning message.  I've
changed this to use WARN() instead.

Fixes: 4136214f7c46 ("rtw88: add BT co-existence support")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192ce: Fix incorrect returned values
Larry Finger [Tue, 3 Sep 2019 01:18:13 +0000 (20:18 -0500)]
rtlwifi: rtl8192ce: Fix incorrect returned values

In commit 98fd8db59a00 ("rtlwifi: rtl8192ce: Convert macros that set
descriptor"), all the routines that get fields from a descriptor
were changed to return signed integer values. This is incorrect for the
routines that get the entire 32-bit word. In this case, an unsigned
quantity is required.

Fixes: 98fd8db59a00 ("rtlwifi: rtl8192ce: Convert macros that set descriptor")
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8188ee: Fix incorrect returned values
Larry Finger [Tue, 3 Sep 2019 01:18:12 +0000 (20:18 -0500)]
rtlwifi: rtl8188ee: Fix incorrect returned values

In commit 36eda7568f2e ("rtlwifi: rtl8188ee: Convert macros that set
descriptor"), all the routines that get fields from a descriptor were
changed to return signed integer values. This is incorrect for the
routines that get the entire 32-bit word. In this case, an unsigned
quantity is required.

Fixes: 36eda7568f2e ("rtlwifi: rtl8188ee: Convert macros that set descriptor")
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8821ae: Fix incorrect returned values
Larry Finger [Tue, 3 Sep 2019 01:18:11 +0000 (20:18 -0500)]
rtlwifi: rtl8821ae: Fix incorrect returned values

In commit bd421dab7515 ("rtlwifi: rtl8821ae: Convert macros that set
descriptor"), all the routines that get fields from a descriptor
were changed to return signed integer values. This is incorrect for the
routines that get the entire 32-bit word. In this case, an unsigned
quantity is required.

Fixes: bd421dab7515 ("rtlwifi: rtl8821ae: Convert macros that set descriptor")
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192cu: Convert inline routines to little-endian words
Larry Finger [Sun, 1 Sep 2019 15:47:06 +0000 (10:47 -0500)]
rtlwifi: rtl8192cu: Convert inline routines to little-endian words

In this step, the read/write routines for the descriptors are converted
to use __le32 quantities, thus a lot of casts can be removed. Callback
routines still use the 8-bit arrays, but these are changed within the
specified routine.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192cu: Convert macros that set descriptor
Larry Finger [Sun, 1 Sep 2019 15:47:05 +0000 (10:47 -0500)]
rtlwifi: rtl8192cu: Convert macros that set descriptor

As a first step in the conversion, the macros that set the RX and TX
descriptors are converted to static inline routines, and the names are
changed from upper to lower case. To minimize the changes in a given
step, the input descriptor information is left as as a byte array
(u8 *), even though it should be a little-endian word array (__le32 *).
That will be changed in the next patch.

Several places where checkpatch.pl complains about a space after a cast
and other warnings are fixed.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192cu: Replace local bit manipulation macros
Larry Finger [Sun, 1 Sep 2019 15:47:04 +0000 (10:47 -0500)]
rtlwifi: rtl8192cu: Replace local bit manipulation macros

This driver uses a set of local macros to manipulate the TX and RX
descriptors, which are all little-endian quantities. These macros
are replaced by the bitfield macros le32p_replace_bits() and
le32_get_bits(). In several places, the macros operated on an entire
32-bit word. In these cases, a direct read or replacement is used.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192cu: Remove unused GET_XXX and SET_XXX
Larry Finger [Sun, 1 Sep 2019 15:47:03 +0000 (10:47 -0500)]
rtlwifi: rtl8192cu: Remove unused GET_XXX and SET_XXX

As the first step in converting from macros that get/set information
in the RX and TX descriptors, unused macros are being removed.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192ee: Remove some variable initializations
Larry Finger [Sat, 31 Aug 2019 18:06:44 +0000 (13:06 -0500)]
rtlwifi: rtl8192ee: Remove some variable initializations

A number of variables are initialized when declared that set later in the
routine, thus the initialization can be removed.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192ee: Convert inline routines to little-endian words
Larry Finger [Sat, 31 Aug 2019 18:06:43 +0000 (13:06 -0500)]
rtlwifi: rtl8192ee: Convert inline routines to little-endian words

In this step, the read/write routines for the descriptors are converted
to use __le32 quantities, thus a lot of casts can be removed. Callback
routines still use the 8-bit arrays, but these are changed within the
specified routine.

The macro that cleared a descriptor has now been converted into an inline
routine.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192ee: Convert macros that set descriptor
Larry Finger [Sat, 31 Aug 2019 18:06:42 +0000 (13:06 -0500)]
rtlwifi: rtl8192ee: Convert macros that set descriptor

As a first step in the conversion, the macros that set the RX and TX
descriptors are converted to static inline routines, and the names are
changed from upper to lower case. To minimize the changes in a given
step, the input descriptor information is left as as a byte array
(u8 *), even though it should be a little-endian word array (__le32 *).
That will be changed in the next patch.

Several places where checkpatch.pl reports lines too long are fixed.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192ee: Replace local bit manipulation macros
Larry Finger [Sat, 31 Aug 2019 18:06:41 +0000 (13:06 -0500)]
rtlwifi: rtl8192ee: Replace local bit manipulation macros

This driver uses a set of local macros to manipulate the RX and TX
descriptors, which are all little-endian quantities. These macros
are replaced by the bitfield macros le32p_replace_bits() and
le32_get_bits(). In several places, the macros operated on an entire
32-bit word. In these cases, a direct read or replacement is used.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192ee: Remove unused GET_XXX and SET_XXX
Larry Finger [Sat, 31 Aug 2019 18:06:40 +0000 (13:06 -0500)]
rtlwifi: rtl8192ee: Remove unused GET_XXX and SET_XXX

As the first step in converting from macros that get/set information
in the RX and TX descriptors, unused macros are being removed.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl_pci: Fix memory leak when hardware init fails
Larry Finger [Mon, 26 Aug 2019 22:03:44 +0000 (17:03 -0500)]
rtlwifi: rtl_pci: Fix memory leak when hardware init fails

If the call to hw_init() fails for any of the drivers, the driver will
leak memory that was allocated in BT coexistence setup. Technically, each
of the drivers should have done this free; however placing it in rtl_pci
fixes all the drivers with only a single patch.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: remove unused variables 'RTL8712_SDIO_EFUSE_TABLE' and 'MAX_PGPKT_SIZE'
YueHaibing [Fri, 16 Aug 2019 14:05:13 +0000 (22:05 +0800)]
rtlwifi: remove unused variables 'RTL8712_SDIO_EFUSE_TABLE' and 'MAX_PGPKT_SIZE'

drivers/net/wireless/realtek/rtlwifi/efuse.c:16:31:
 warning: RTL8712_SDIO_EFUSE_TABLE defined but not used [-Wunused-const-variable=]
drivers/net/wireless/realtek/rtlwifi/efuse.c:9:17:
 warning: MAX_PGPKT_SIZE defined but not used [-Wunused-const-variable=]

They are never used, so can be removed.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: rtl8192cu: Fix value set in descriptor
Larry Finger [Mon, 12 Aug 2019 19:27:41 +0000 (14:27 -0500)]
rtlwifi: rtl8192cu: Fix value set in descriptor

In the process of converting the bit manipulation macros were converted
to use GENMASK(), the compiler reported a value too big for the field.
The offending statement was trying to write 0x100 into a 5-bit field.
An accompaning comment says to set bit 3, thus the code is changed
appropriately.

This error has been in the driver since its initial submission.

Fixes: 29d00a3e46bb ("rtlwifi: rtl8192cu: Add routine trx")
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agortlwifi: fix non-kerneldoc comment in usb.c
Valdis Klētnieks [Thu, 8 Aug 2019 01:51:58 +0000 (21:51 -0400)]
rtlwifi: fix non-kerneldoc comment in usb.c

Fix spurious warning message when building with W=1:

  CC [M]  drivers/net/wireless/realtek/rtlwifi/usb.o
drivers/net/wireless/realtek/rtlwifi/usb.c:243: warning: Cannot understand  * on line 243 - I thought it was a doc line
drivers/net/wireless/realtek/rtlwifi/usb.c:760: warning: Cannot understand  * on line 760 - I thought it was a doc line
drivers/net/wireless/realtek/rtlwifi/usb.c:790: warning: Cannot understand  * on line 790 - I thought it was a doc line

Clean up the comment format.

Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agoMerge branch 'mvpp2-per-cpu-buffers'
David S. Miller [Mon, 2 Sep 2019 19:07:46 +0000 (12:07 -0700)]
Merge branch 'mvpp2-per-cpu-buffers'

Matteo Croce says:

====================
mvpp2: per-cpu buffers

This patchset workarounds an PP2 HW limitation which prevents to use
per-cpu rx buffers.
The first patch is just a refactor to prepare for the second one.
The second one allocates percpu buffers if the following conditions are met:
- CPU number is less or equal 4
- no port is using jumbo frames

If the following conditions are not met at load time, of jumbo frame is enabled
later on, the shared allocation is reverted.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomvpp2: percpu buffers
Matteo Croce [Mon, 2 Sep 2019 10:21:37 +0000 (12:21 +0200)]
mvpp2: percpu buffers

Every mvpp2 unit can use up to 8 buffers mapped by the BM (the HW buffer
manager). The HW will place the frames in the buffer pool depending on the
frame size: short (< 128 bytes), long (< 1664) or jumbo (up to 9856).

As any unit can have up to 4 ports, the driver allocates only 2 pools,
one for small and one long frames, and share them between ports.
When the first port MTU is set higher than 1664 bytes, a third pool is
allocated for jumbo frames.

This shared allocation makes impossible to use percpu allocators,
and creates contention between HW queues.

If possible, i.e. if the number of possible CPU are less than 8 and jumbo
frames are not used, switch to a new scheme: allocate 8 per-cpu pools for
short and long frames and bind every pool to an RXQ.

When the first port MTU is set higher than 1664 bytes, the allocation
scheme is reverted to the old behaviour (3 shared pools), and when all
ports MTU are lowered, the per-cpu buffers are allocated again.

Signed-off-by: Matteo Croce <mcroce@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomvpp2: refactor BM pool functions
Matteo Croce [Mon, 2 Sep 2019 10:21:36 +0000 (12:21 +0200)]
mvpp2: refactor BM pool functions

Refactor mvpp2_bm_pool_create(), mvpp2_bm_pool_destroy() and
mvpp2_bm_pools_init() so that they accept a struct device instead
of a struct platform_device, as they just need platform_device->dev.

Removing such dependency makes the BM code more reusable in context
where we don't have a pointer to the platform_device.

Signed-off-by: Matteo Croce <mcroce@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: Fix off-by-one number of calls to devlink_port_unregister
Vladimir Oltean [Sat, 31 Aug 2019 12:46:19 +0000 (15:46 +0300)]
net: dsa: Fix off-by-one number of calls to devlink_port_unregister

When a function such as dsa_slave_create fails, currently the following
stack trace can be seen:

[    2.038342] sja1105 spi0.1: Probed switch chip: SJA1105T
[    2.054556] sja1105 spi0.1: Reset switch and programmed static config
[    2.063837] sja1105 spi0.1: Enabled switch tagging
[    2.068706] fsl-gianfar soc:ethernet@2d90000 eth2: error -19 setting up slave phy
[    2.076371] ------------[ cut here ]------------
[    2.080973] WARNING: CPU: 1 PID: 21 at net/core/devlink.c:6184 devlink_free+0x1b4/0x1c0
[    2.088954] Modules linked in:
[    2.092005] CPU: 1 PID: 21 Comm: kworker/1:1 Not tainted 5.3.0-rc6-01360-g41b52e38d2b6-dirty #1746
[    2.100912] Hardware name: Freescale LS1021A
[    2.105162] Workqueue: events deferred_probe_work_func
[    2.110287] [<c03133a4>] (unwind_backtrace) from [<c030d8cc>] (show_stack+0x10/0x14)
[    2.117992] [<c030d8cc>] (show_stack) from [<c10b08d8>] (dump_stack+0xb4/0xc8)
[    2.125180] [<c10b08d8>] (dump_stack) from [<c0349d04>] (__warn+0xe0/0xf8)
[    2.132018] [<c0349d04>] (__warn) from [<c0349e34>] (warn_slowpath_null+0x40/0x48)
[    2.139549] [<c0349e34>] (warn_slowpath_null) from [<c0f19d74>] (devlink_free+0x1b4/0x1c0)
[    2.147772] [<c0f19d74>] (devlink_free) from [<c1064fc0>] (dsa_switch_teardown+0x60/0x6c)
[    2.155907] [<c1064fc0>] (dsa_switch_teardown) from [<c1065950>] (dsa_register_switch+0x8e4/0xaa8)
[    2.164821] [<c1065950>] (dsa_register_switch) from [<c0ba7fe4>] (sja1105_probe+0x21c/0x2ec)
[    2.173216] [<c0ba7fe4>] (sja1105_probe) from [<c0b35948>] (spi_drv_probe+0x80/0xa4)
[    2.180920] [<c0b35948>] (spi_drv_probe) from [<c0a4c1cc>] (really_probe+0x108/0x400)
[    2.188711] [<c0a4c1cc>] (really_probe) from [<c0a4c694>] (driver_probe_device+0x78/0x1bc)
[    2.196933] [<c0a4c694>] (driver_probe_device) from [<c0a4a3dc>] (bus_for_each_drv+0x58/0xb8)
[    2.205414] [<c0a4a3dc>] (bus_for_each_drv) from [<c0a4c024>] (__device_attach+0xd0/0x168)
[    2.213637] [<c0a4c024>] (__device_attach) from [<c0a4b1d0>] (bus_probe_device+0x84/0x8c)
[    2.221772] [<c0a4b1d0>] (bus_probe_device) from [<c0a4b72c>] (deferred_probe_work_func+0x84/0xc4)
[    2.230686] [<c0a4b72c>] (deferred_probe_work_func) from [<c03650a4>] (process_one_work+0x218/0x510)
[    2.239772] [<c03650a4>] (process_one_work) from [<c03660d8>] (worker_thread+0x2a8/0x5c0)
[    2.247908] [<c03660d8>] (worker_thread) from [<c036b348>] (kthread+0x148/0x150)
[    2.255265] [<c036b348>] (kthread) from [<c03010e8>] (ret_from_fork+0x14/0x2c)
[    2.262444] Exception stack(0xea965fb0 to 0xea965ff8)
[    2.267466] 5fa0:                                     00000000 00000000 00000000 00000000
[    2.275598] 5fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
[    2.283729] 5fe0: 00000000 00000000 00000000 00000000 00000013 00000000
[    2.290333] ---[ end trace ca5d506728a0581a ]---

devlink_free is complaining right here:

WARN_ON(!list_empty(&devlink->port_list));

This happens because devlink_port_unregister is no longer done right
away in dsa_port_setup when a DSA_PORT_TYPE_USER has failed.
Vivien said about this change that:

    Also no need to call devlink_port_unregister from within dsa_port_setup
    as this step is inconditionally handled by dsa_port_teardown on error.

which is not really true. The devlink_port_unregister function _is_
being called unconditionally from within dsa_port_setup, but not for
this port that just failed, just for the previous ones which were set
up.

ports_teardown:
for (i = 0; i < port; i++)
dsa_port_teardown(&ds->ports[i]);

Initially I was tempted to fix this by extending the "for" loop to also
cover the port that failed during setup. But this could have potentially
unforeseen consequences unrelated to devlink_port or even other types of
ports than user ports, which I can't really test for. For example, if
for some reason devlink_port_register itself would fail, then
unconditionally unregistering it in dsa_port_teardown would not be a
smart idea. The list might go on.

So just make dsa_port_setup undo the setup it had done upon failure, and
let the for loop undo the work of setting up the previous ports, which
are guaranteed to be brought up to a consistent state.

Fixes: 955222ca5281 ("net: dsa: use a single switch statement for port setup")
Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomlx5: Add missing init_net check in FIB notifier
Jiri Pirko [Fri, 30 Aug 2019 08:25:30 +0000 (10:25 +0200)]
mlx5: Add missing init_net check in FIB notifier

Take only FIB events that are happening in init_net into account. No other
namespaces are supported.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Acked-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
David S. Miller [Mon, 2 Sep 2019 18:20:17 +0000 (11:20 -0700)]
Merge git://git./linux/kernel/git/netdev/net

r8152 conflicts are the NAPI fixes in 'net' overlapping with
some tasklet stuff in net-next

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoLinux 5.3-rc7
Linus Torvalds [Mon, 2 Sep 2019 16:57:40 +0000 (09:57 -0700)]
Linux 5.3-rc7

5 years agoMerge tag 'char-misc-5.3-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Mon, 2 Sep 2019 16:30:34 +0000 (09:30 -0700)]
Merge tag 'char-misc-5.3-rc7' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are some small char and misc driver fixes for reported issues for
  5.3-rc7

  Also included in here is the documentation for how we are handling
  hardware issues under embargo that everyone has finally agreed on, as
  well as a MAINTAINERS update for the suckers who agreed to handle the
  LICENSES/ files.

  All of these have been in linux-next last week with no reported
  issues"

* tag 'char-misc-5.3-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  fsi: scom: Don't abort operations for minor errors
  vmw_balloon: Fix offline page marking with compaction
  VMCI: Release resource if the work is already queued
  Documentation/process: Embargoed hardware security issues
  lkdtm/bugs: fix build error in lkdtm_EXHAUST_STACK
  mei: me: add Tiger Lake point LP device ID
  intel_th: pci: Add Tiger Lake support
  intel_th: pci: Add support for another Lewisburg PCH
  stm class: Fix a double free of stm_source_device
  MAINTAINERS: add entry for LICENSES and SPDX stuff
  fpga: altera-ps-spi: Fix getting of optional confd gpio

5 years agoMerge tag 'usb-5.3-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Mon, 2 Sep 2019 16:15:30 +0000 (09:15 -0700)]
Merge tag 'usb-5.3-rc7' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some small USB fixes that have been in linux-next this past
  week for 5.3-rc7

  They fix the usual xhci, syzbot reports, and other small issues that
  have come up last week.

  All have been in linux-next with no reported issues"

* tag 'usb-5.3-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  USB: cdc-wdm: fix race between write and disconnect due to flag abuse
  usb: host: xhci: rcar: Fix typo in compatible string matching
  usb: host: xhci-tegra: Set DMA mask correctly
  USB: storage: ums-realtek: Whitelist auto-delink support
  USB: storage: ums-realtek: Update module parameter description for auto_delink_en
  usb: host: ohci: fix a race condition between shutdown and irq
  usb: hcd: use managed device resources
  typec: tcpm: fix a typo in the comparison of pdo_max_voltage
  usb-storage: Add new JMS567 revision to unusual_devs
  usb: chipidea: udc: don't do hardware access if gadget has stopped
  usbtmc: more sanity checking for packet size
  usb: udc: lpc32xx: silence fall-through warning

5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Mon, 2 Sep 2019 01:45:28 +0000 (18:45 -0700)]
Merge git://git./linux/kernel/git/netdev/net

Pull networking fixes from David Miller:

 1) Fix some length checks during OGM processing in batman-adv, from
    Sven Eckelmann.

 2) Fix regression that caused netfilter conntrack sysctls to not be
    per-netns any more. From Florian Westphal.

 3) Use after free in netpoll, from Feng Sun.

 4) Guard destruction of pfifo_fast per-cpu qdisc stats with
    qdisc_is_percpu_stats(), from Davide Caratti. Similar bug is fixed
    in pfifo_fast_enqueue().

 5) Fix memory leak in mld_del_delrec(), from Eric Dumazet.

 6) Handle neigh events on internal ports correctly in nfp, from John
    Hurley.

 7) Clear SKB timestamp in NF flow table code so that it does not
    confuse fq scheduler. From Florian Westphal.

 8) taprio destroy can crash if it is invoked in a failure path of
    taprio_init(), because the list head isn't setup properly yet and
    the list del is unconditional. Perform the list add earlier to
    address this. From Vladimir Oltean.

 9) Make sure to reapply vlan filters on device up, in aquantia driver.
    From Dmitry Bogdanov.

10) sgiseeq driver releases DMA memory using free_page() instead of
    dma_free_attrs(). From Christophe JAILLET.

* git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (58 commits)
  net: seeq: Fix the function used to release some memory in an error handling path
  enetc: Add missing call to 'pci_free_irq_vectors()' in probe and remove functions
  net: bcmgenet: use ethtool_op_get_ts_info()
  tc-testing: don't hardcode 'ip' in nsPlugin.py
  net: dsa: microchip: add KSZ8563 compatibility string
  dt-bindings: net: dsa: document additional Microchip KSZ8563 switch
  net: aquantia: fix out of memory condition on rx side
  net: aquantia: linkstate irq should be oneshot
  net: aquantia: reapply vlan filters on up
  net: aquantia: fix limit of vlan filters
  net: aquantia: fix removal of vlan 0
  net/sched: cbs: Set default link speed to 10 Mbps in cbs_set_port_rate
  taprio: Set default link speed to 10 Mbps in taprio_set_picos_per_byte
  taprio: Fix kernel panic in taprio_destroy
  net: dsa: microchip: fill regmap_config name
  rxrpc: Fix lack of conn cleanup when local endpoint is cleaned up [ver #2]
  net: stmmac: dwmac-rk: Don't fail if phy regulator is absent
  amd-xgbe: Fix error path in xgbe_mod_init()
  netfilter: nft_meta_bridge: Fix get NFT_META_BRI_IIFVPROTO in network byteorder
  mac80211: Correctly set noencrypt for PAE frames
  ...

5 years agoMerge branch 'net-dsa-mv88e6xxx-centralize-SERDES-IRQ-handling'
David S. Miller [Sun, 1 Sep 2019 19:16:38 +0000 (12:16 -0700)]
Merge branch 'net-dsa-mv88e6xxx-centralize-SERDES-IRQ-handling'

Vivien Didelot says:

====================
net: dsa: mv88e6xxx: centralize SERDES IRQ handling

Following Marek's work on the abstraction of the SERDES lanes mapping, this
series trades the .serdes_irq_setup and .serdes_irq_free callbacks for new
.serdes_irq_mapping, .serdes_irq_enable and .serdes_irq_status operations.

This has the benefit to limit the various SERDES implementations to simple
register accesses only; centralize the IRQ handling and mutex locking logic;
as well as reducing boilerplate in the driver.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: mv88e6xxx: centralize SERDES IRQ handling
Vivien Didelot [Sat, 31 Aug 2019 20:18:36 +0000 (16:18 -0400)]
net: dsa: mv88e6xxx: centralize SERDES IRQ handling

The .serdes_irq_setup are all following the same steps: get the SERDES
lane, get the IRQ mapping, request the IRQ, then enable it. So do
the .serdes_irq_free implementations: get the SERDES lane, disable
the IRQ, then free it.

This patch removes these operations in favor of generic functions.

Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: mv88e6xxx: introduce .serdes_irq_status
Vivien Didelot [Sat, 31 Aug 2019 20:18:35 +0000 (16:18 -0400)]
net: dsa: mv88e6xxx: introduce .serdes_irq_status

Introduce a new .serdes_irq_status operation to prepare the abstraction
of IRQ thread from the SERDES IRQ setup code.

Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: mv88e6xxx: introduce .serdes_irq_enable
Vivien Didelot [Sat, 31 Aug 2019 20:18:34 +0000 (16:18 -0400)]
net: dsa: mv88e6xxx: introduce .serdes_irq_enable

Introduce a new .serdes_irq_enable operation to prepare the abstraction
of IRQ enabling from the SERDES IRQ setup code.

Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: mv88e6xxx: pass lane to .serdes_power
Vivien Didelot [Sat, 31 Aug 2019 20:18:33 +0000 (16:18 -0400)]
net: dsa: mv88e6xxx: pass lane to .serdes_power

Now the first step of all .serdes_power implementations is getting
the lane mapping. Since we have an operation for that, call it in
the wrapper and pass the lane down to the .serdes_power operation.

This also allows to avoid querying the SERDES lane twice in
mv88e6xxx_port_set_cmode.

At the same time provide mv88e6xxx_serdes_power_{up,down} helpers
and prefer up/down instead of on/off as in the documentation.

Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: mv88e6xxx: merge mv88e6352_serdes_power_set
Vivien Didelot [Sat, 31 Aug 2019 20:18:32 +0000 (16:18 -0400)]
net: dsa: mv88e6xxx: merge mv88e6352_serdes_power_set

The mv88e6352_serdes_power_set helper is only used at one place, in
mv88e6352_serdes_power. Keep it simple and merge the two functions
together.

Use mv88e6xxx_serdes_get_lane instead of mv88e6352_port_has_serdes
to avoid moving code. No functional changes.

Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>