Since we remove radiotap from skb data, clear all RX_FLAG_X related
info that indicate info on the skb data.
Also we need to do it only once so remove the clear from cooked_monitor.
Signed-off-by: Mordechay Goodstein <mordechay.goodstein@intel.com>
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Link: https://lore.kernel.org/r/20230301115906.74d3efe19eae.Ie17a35864d2e120f9858516a2e3d3047d83cf805@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Instead of the comment about holding RTNL, which is now wrong,
add a proper lockdep assertion for the wiphy mutex.
Fixes: a05829a722 ("cfg80211: avoid holding the RTNL when calling the driver")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Link: https://lore.kernel.org/r/20230301115906.84352e46f342.Id90fef8c581cebe19cb30274340cf43885d55c74@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Variables bcntime_cfg, bcn_cw and bcn_ifs are not effectively used, so
delete them.
drivers/net/wireless/realtek/rtlwifi/rtl8192se/hw.c:1555:6: warning: variable 'bcntime_cfg' set but not used.
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=4240
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Acked-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230302023911.59278-1-jiapeng.chong@linux.alibaba.com
rtw_mac_power_switch() calls rtw_pwr_seq_parser() which can return
-EINVAL, -EBUSY or 0. Propagate the original error code instead of
unconditionally returning -EINVAL in case of an error.
Fixes: e3037485c6 ("rtw88: new Realtek 802.11ac driver")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230226221004.138331-3-martin.blumenstingl@googlemail.com
rtw_pwr_seq_parser() calls rtw_sub_pwr_seq_parser() which can either
return -EBUSY, -EINVAL or 0. Propagate the original error code instead
of unconditionally returning -EBUSY in case of an error.
Fixes: e3037485c6 ("rtw88: new Realtek 802.11ac driver")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230226221004.138331-2-martin.blumenstingl@googlemail.com
Some phones from around 2016, as well as other random devices have
this chip called 43956 or 4359C0 or 43596A0, which is more or less
just a rev bump (v9) of the already-supported 4359. Add a corresponding
firmware definition to allow for choosing the correct blob.
Suggested-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230224-topic-brcm_tone-v1-1-333b0ac67934@linaro.org
When SER L1 triggered, driver need to stop Rx and clear RTW89_FLAG_RUNNING
flag. If track_work check RTW89_FLAG_RUNNING simultaneously, it will check
failed and never queue track_work again, and LPS won't enter either.
Therefore, we cancel delayed work when enter SER L1, and queue delayed work
when leave SER L1.
Signed-off-by: Chih-Kang Chang <gary.chang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230224082117.21241-1-pkshih@realtek.com
BCM4378B3 is a new silicon revision of BCM4378 present on the Apple M2
13" MacBook Pro "kyushu". Its PCI revision number is 5.
Signed-off-by: Hector Martin <marcan@marcan.st>
Reviewed-by: Julian Calaby <julian.calaby@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092838.17869-1-marcan@marcan.st
The calibration blob for a chip is normally stored in SROM and loaded
internally by the firmware. However, Apple ARM64 platforms instead store
it as part of platform configuration data, and provide it via the Apple
Device Tree. We forward this into the Linux DT in the bootloader.
Add support for taking this blob from the DT and loading it into the
dongle. The loading mechanism is the same as used for the CLM and TxCap
blobs.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-10-marcan@marcan.st
These blobs are named .txcap_blob, and exist alongside the existing
.clm_blob files. Use the existing firmware machinery to provide them to
the core.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-9-marcan@marcan.st
The TxCap blobs are additional data blobs used on Apple devices, and
are uploaded analogously to CLM blobs. Add core support for doing this.
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-8-marcan@marcan.st
Apparently the hex passphrase mechanism does not work on newer
chips/firmware (e.g. BCM4387). It seems there was a simple way of
passing it in binary all along, so use that and avoid the hexification.
OpenBSD has been doing it like this from the beginning, so this should
work on all chips.
Also clear the structure before setting the PMK. This was leaking
uninitialized stack contents to the device.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-6-marcan@marcan.st
Add support for the new PMKID_V3 API, which allows performing PMKID
mutations individually, instead of requiring the driver to keep track of
the full list. This new API is required by at least BCM4387.
Note that PMKID_V2 is not implemented yet.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-5-marcan@marcan.st
The "wlc_ver" iovar returns information on the WLC and EPI versions.
This can be used to determine whether the PMKID_V2 and _V3 features are
supported.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-4-marcan@marcan.st
This new API version is required for at least the BCM4387 firmware. Add
support for it, with a fallback to the v1 API.
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-3-marcan@marcan.st
BCM4387 has trailing odd-sized blocks as part of TCM which have
their size described as a multiple of 1024 instead of 8192. Handle this
so we can compute the TCM size properly.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-2-marcan@marcan.st
At least on BCM4387, the D11 cores are held in reset on cold startup and
firmware expects to release reset itself. Just assert reset here and let
firmware deassert it. Premature deassertion results in the firmware
failing to initialize properly some of the time, with strange AXI bus
errors.
Also, BCM4387 has 3 cores, up from 2. The logic for handling that is in
brcmf_chip_ai_resetcore(), but since we aren't using that any more, just
handle it here.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214092423.15175-1-marcan@marcan.st
Using the BCM4339 firmware from linux-firmware (version "BCM4339/2 wl0:
Sep 5 2019 11:05:52 version 6.37.39.113 (r722271 CY)" from
cypress/cyfmac4339-sdio.bin) the RSSI respose is only 4 bytes, which
results in an error being logged.
It seems that older devices send only the RSSI field and neither SNR nor
noise is included. Handle this by accepting a 4 byte message and
reading only the RSSI from it.
Fixes: 7dd56ea45a ("brcmfmac: add support for CQM RSSI notifications")
Signed-off-by: John Keeping <john@metanate.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230124104248.2917465-1-john@metanate.com
Newer Apple firmwares on chipsets without a hardware RNG require the
host to provide a buffer of 256 random bytes to the device on
initialization. This buffer is present immediately before NVRAM,
suffixed by a footer containing a magic number and the buffer length.
This won't affect chips/firmwares that do not use this feature, so do it
unconditionally for all Apple platforms (those with an Apple OTP).
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Hector Martin <marcan@marcan.st>
Reviewed-by: Julian Calaby <julian.calaby@gmail.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214080034.3828-3-marcan@marcan.st
On DT platforms, the module-instance and antenna-sku-info properties
are passed in the DT. On ACPI platforms, module-instance is passed via
the analogous Apple device property mechanism, while the antenna SKU
info is instead obtained via an ACPI method that grabs it from
non-volatile storage.
Add support for this, to allow proper firmware selection on Apple
platforms.
Signed-off-by: Hector Martin <marcan@marcan.st>
Reviewed-by: Julian Calaby <julian.calaby@gmail.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214080034.3828-2-marcan@marcan.st
The newer chips use the newer firmware branches, e.g. v027, v029.
And, those firmware branches are supposed to support packet drop
when they are just branched out.
The initial firmware branch used by each chip is as below.
* 8852A: v009
* 8852C: v027
* 8852B: v027
So, only 8852A may use firmware which doesn't support packet drop at
runtime. To save trivial positive listing in firmware feature table,
we change to reverse judgment.
Besides, rtw89_mac_ptk_drop_by_band_and_wait() missed to check firmware
feature before calling rtw89_fw_h2c_pkt_drop(). We also fix it.
Signed-off-by: Zong-Zhe Yang <kevin_yang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230220070202.29868-7-pkshih@realtek.com
To obtain correct packet frequency for hw_scan, 52b needs to decode
the channel index obtained from hardware. Change the driver related
set channel part as well to make driver/firmware compatible.
Signed-off-by: Po-Hao Huang <phhuang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230220070202.29868-5-pkshih@realtek.com
Since the range of channel table is identical among ICs. Make channel
encode/decode function common and not IC dependent. So all ICs with
matching firmware that needs this kind of coding can use it directly.
This patch doesn't change logic at all.
Signed-off-by: Po-Hao Huang <phhuang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230220070202.29868-4-pkshih@realtek.com
RTL8852B firmware supports CRASH_TRIGGER feature from v0.29.29.0.
After this is configured, debugfs fw_crash can support type 1 on
RTL8852B to trigger firmware crash and verify L2 recovery through
simulation.
Signed-off-by: Zong-Zhe Yang <kevin_yang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230220070202.29868-3-pkshih@realtek.com
8852B has the same issue: management frames get stuck when wifi
chip enters low ps mode, so we alse add notify wake function to
trigger wifi chip wake before forwarding management frames.
Signed-off-by: Chin-Yen Lee <timlee@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230220070202.29868-2-pkshih@realtek.com
The efuse of the SDIO RTL8822CS chip has only one known member: the mac
address is at offset 0x16a. Add a struct rtw8822cs_efuse describing this
and use it for copying the mac address when the SDIO bus is used.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230218152944.48842-6-martin.blumenstingl@googlemail.com
The efuse of the SDIO RTL8822BS chip has only one known member: the mac
address is at offset 0x11a. Add a struct rtw8822bs_efuse describing this
and use it for copying the mac address when the SDIO bus is used.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230218152944.48842-5-martin.blumenstingl@googlemail.com
The efuse of the SDIO RTL8821CS chip has only one known member: the mac
address is at offset 0x11a. Add a struct rtw8821cs_efuse describing this
and use it for copying the mac address when the SDIO bus is used.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230218152944.48842-4-martin.blumenstingl@googlemail.com
txdma_queue_mapping() and priority_queue_cfg() can use the first entry
of each chip's rqpn_table and page_table. Add this mapping so data
transmission is possible on SDIO based chipsets.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230218152944.48842-3-martin.blumenstingl@googlemail.com
rtw_pwr_seq_parser() needs to know about the HCI bus interface mask for
the SDIO bus so it can parse the chip state change sequences.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230218152944.48842-2-martin.blumenstingl@googlemail.com
Since 6 GHz band has around 60 channels and more strict rules for
active probing. Reduced neighbor report can be used to reduce the
channels we scan and get specific target BSS info to probe for.
Declare flag WIPHY_FLAG_SPLIT_SCAN_6GHZ so the scan request could be
divided into two portions: legacy bands and 6 GHz bands. So RNR
information from legacy bands could later be used when 6 GHz scan.
When the scan flag NL80211_SCAN_FLAG_COLOCATED_6GHZ is set, cfg80211
will pass down a reduced channel set which contains PSCs and non-PSC
with RNR info received in the 2 GHz/5 GHz band. This reduces the
scan duration by allowing us to only scan for channels in which APs
are currently operating.
Signed-off-by: Po-Hao Huang <phhuang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230217120007.8835-1-pkshih@realtek.com
Intentionally parsing single one element of RA report by breaking loop
causes a smatch warning:
drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8188e.c:1678 rtl8188e_handle_ra_tx_report2() warn:
ignoring unreachable code.
With existing comments, it intends to process single one element for
station mode, but it will parse more elements in AP mode if it's
implemented. Implement program logic according to existing comment to avoid
smatch warning, and also be usable for both AP and stations modes.
Compile test only.
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <error27@gmail.com>
Link: https://lore.kernel.org/r/202302142135.LCqUTVGY-lkp@intel.com/
Cc: Bitterblue Smith <rtl8821cerfe2@gmail.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Tested-by: Bitterblue Smith <rtl8821cerfe2@gmail.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230216004654.4642-1-pkshih@realtek.com
wait_for_completion_timeout() can not return a <0 value.
So simplify the logic and remove dead code.
-ERESTARTSYS can not be returned by do_wait_for_common() for tasks with
TASK_UNINTERRUPTIBLE, which is the case for wait_for_completion_timeout()
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/809c4a645c8d1306c0d256345515865c40ec731c.1676464422.git.christophe.jaillet@wanadoo.fr
On OpenWRT platform, RTL8192CE could be soldered on board, but not standard PCI
module. In this case, some EEPROM values aren't programmed and left 0xff.
Load default values when the EEPROM values are empty to avoid problems.
Signed-off-by: Lu jicong <jiconglu58@gmail.com>
Acked-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230214063602.2257263-1-jiconglu58@gmail.com
Hardware requires an alignment to 64 bytes to return ASO data. Missing
this alignment caused to unpredictable results while ASO events were
generated.
Fixes: 8518d05b8f ("net/mlx5e: Create Advanced Steering Operation object for IPsec")
Reported-by: Emeel Hakim <ehakim@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Link: https://lore.kernel.org/r/de0302c572b90c9224a72868d4e0d657b6313c4b.1676797613.git.leon@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
1) From Gal Tariq and Parav, Few cleanups for mlx5 driver.
2) From Vlad: Allow offloading of ct 'new' match based on [1]
[1] https://lore.kernel.org/netdev/20230201163100.1001180-1-vladbu@nvidia.com/
-----BEGIN PGP SIGNATURE-----
iQEzBAABCAAdFiEEGhZs6bAKwk/OTgTpSD+KveBX+j4FAmPwk+4ACgkQSD+KveBX
+j7IiQf+JbjSOqKvQqAQLOjNwtp+ejs9SeMR/KXMSXfmS62Uca8O3ropNazFE4xS
oRcoGHGSCvvQpcm9j+o2xOnzhi6lTmJpmBfRDllLFw3MjXGPMkuIo0BJnIkHCbPw
zL9hL+OSpVy0gA5KOzc4zwUDOLaWAXwSTb22JifJpHSoRCu9NmBummDDtnLS5lKf
CQ+Vc6xd+p2uNelgl4SVcNhIBKVNjNgaf2q1psI/qGf5iS0c3UBJhr0BJLFRBQYY
7S33fgaesysdOvUgjM/j0D5vUCYqhfuAVwCLTLcrrazv4B/G4z2QBHm2HQ3Tq1Qd
jVZF+Fkvfe+Ck9s7Xk0YeB4kyGFcWA==
=saOV
-----END PGP SIGNATURE-----
Merge tag 'mlx5-updates-2023-02-15' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux
Saeed Mahameed says:
====================
mlx5-updates-2023-02-15
1) From Gal Tariq and Parav, Few cleanups for mlx5 driver.
2) From Vlad: Allow offloading of ct 'new' match based on [1]
[1] https://lore.kernel.org/netdev/20230201163100.1001180-1-vladbu@nvidia.com/
* tag 'mlx5-updates-2023-02-15' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux:
net/mlx5e: RX, Remove doubtful unlikely call
net/mlx5e: Fix outdated TLS comment
net/mlx5e: Remove unused function mlx5e_sq_xmit_simple
net/mlx5e: Allow offloading of ct 'new' match
net/mlx5e: Implement CT entry update
net/mlx5: Simplify eq list traversal
net/mlx5e: Remove redundant page argument in mlx5e_xdp_handle()
net/mlx5e: Remove redundant page argument in mlx5e_xmit_xdp_buff()
net/mlx5e: Switch to using napi_build_skb()
====================
Link: https://lore.kernel.org/r/20230218090513.284718-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Paul Blakey says:
====================
net/sched: cls_api: Support hardware miss to tc action
This series adds support for hardware miss to instruct tc to continue execution
in a specific tc action instance on a filter's action list. The mlx5 driver patch
(besides the refactors) shows its usage instead of using just chain restore.
Currently a filter's action list must be executed all together or
not at all as driver are only able to tell tc to continue executing from a
specific tc chain, and not a specific filter/action.
This is troublesome with regards to action CT, where new connections should
be sent to software (via tc chain restore), and established connections can
be handled in hardware.
Checking for new connections is done when executing the ct action in hardware
(by checking the packet's tuple against known established tuples).
But if there is a packet modification (pedit) action before action CT and the
checked tuple is a new connection, hardware will need to revert the previous
packet modifications before sending it back to software so it can
re-match the same tc filter in software and re-execute its CT action.
The following is an example configuration of stateless nat
on mlx5 driver that isn't supported before this patchet:
#Setup corrosponding mlx5 VFs in namespaces
$ ip netns add ns0
$ ip netns add ns1
$ ip link set dev enp8s0f0v0 netns ns0
$ ip netns exec ns0 ifconfig enp8s0f0v0 1.1.1.1/24 up
$ ip link set dev enp8s0f0v1 netns ns1
$ ip netns exec ns1 ifconfig enp8s0f0v1 1.1.1.2/24 up
#Setup tc arp and ct rules on mxl5 VF representors
$ tc qdisc add dev enp8s0f0_0 ingress
$ tc qdisc add dev enp8s0f0_1 ingress
$ ifconfig enp8s0f0_0 up
$ ifconfig enp8s0f0_1 up
#Original side
$ tc filter add dev enp8s0f0_0 ingress chain 0 proto ip flower \
ct_state -trk ip_proto tcp dst_port 8888 \
action pedit ex munge tcp dport set 5001 pipe \
action csum ip tcp pipe \
action ct pipe \
action goto chain 1
$ tc filter add dev enp8s0f0_0 ingress chain 1 proto ip flower \
ct_state +trk+est \
action mirred egress redirect dev enp8s0f0_1
$ tc filter add dev enp8s0f0_0 ingress chain 1 proto ip flower \
ct_state +trk+new \
action ct commit pipe \
action mirred egress redirect dev enp8s0f0_1
$ tc filter add dev enp8s0f0_0 ingress chain 0 proto arp flower \
action mirred egress redirect dev enp8s0f0_1
#Reply side
$ tc filter add dev enp8s0f0_1 ingress chain 0 proto arp flower \
action mirred egress redirect dev enp8s0f0_0
$ tc filter add dev enp8s0f0_1 ingress chain 0 proto ip flower \
ct_state -trk ip_proto tcp \
action ct pipe \
action pedit ex munge tcp sport set 8888 pipe \
action csum ip tcp pipe \
action mirred egress redirect dev enp8s0f0_0
#Run traffic
$ ip netns exec ns1 iperf -s -p 5001&
$ sleep 2 #wait for iperf to fully open
$ ip netns exec ns0 iperf -c 1.1.1.2 -p 8888
#dump tc filter stats on enp8s0f0_0 chain 0 rule and see hardware packets:
$ tc -s filter show dev enp8s0f0_0 ingress chain 0 proto ip | grep "hardware.*pkt"
Sent hardware 9310116832 bytes 6149672 pkt
Sent hardware 9310116832 bytes 6149672 pkt
Sent hardware 9310116832 bytes 6149672 pkt
A new connection executing the first filter in hardware will first rewrite
the dst port to the new port, and then the ct action is executed,
because this is a new connection, hardware will need to be send this back
to software, on chain 0, to execute the first filter again in software.
The dst port needs to be reverted otherwise it won't re-match the old
dst port in the first filter. Because of that, currently mlx5 driver will
reject offloading the above action ct rule.
This series adds support for hardware partially executing a filter's action list,
and letting tc software continue processing in the specific action instance
where hardware left off (in the above case after the "action pedit ex munge tcp
dport... of the first rule") allowing support for scenarios such as the above.
====================
Link: https://lore.kernel.org/r/20230217223620.28508-1-paulb@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently, CT misses restore the missed chain on the tc skb extension so
tc will continue from the relevant chain. Instead, restore the CT action's
miss cookie on the extension, which will instruct tc to continue from the
this specific CT action instance on the relevant filter's action list.
Map the CT action's miss_cookie to a new miss object (ACT_MISS), and use
this miss mapping instead of the current chain miss object (CHAIN_MISS)
for CT action misses.
To restore this new miss mapping value, add a RX restore rule for each
such mapping value.
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Oz Sholmo <ozsh@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This reg usage is always a mapped object, not necessarily
containing chain info.
Rename to properly convey what it stores.
This patch doesn't change any functionality.
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Move tc miss handling code to en_tc.c, and remove
duplicate code.
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tc skb extension is a basic requirement for using tc
offload to support correct restoration on action miss.
Depend on it.
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
To support hardware miss to tc action in actions on the flower
classifier, implement the required getting of filter actions,
and setup filter exts (actions) miss by giving it the filter's
handle and actions.
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
To support miss to action during hardware offload the filter's
handle is needed when setting up the actions (tcf_exts_init()),
and before offloading.
Move filter handle initialization earlier.
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
For drivers to support partial offload of a filter's action list,
add support for action miss to specify an action instance to
continue from in sw.
CT action in particular can't be fully offloaded, as new connections
need to be handled in software. This imposes other limitations on
the actions that can be offloaded together with the CT action, such
as packet modifications.
Assign each action on a filter's action list a unique miss_cookie
which drivers can then use to fill action_miss part of the tc skb
extension. On getting back this miss_cookie, find the action
instance with relevant cookie and continue classifying from there.
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
