The 32bit counter can only count to 4GiB before wrapping.
Switching to the 64bit variant avoids this issue.
In practice some users are interpreting the counter values as signed
integer bringing down the usable range for 32bit values down to only
2GiB.
Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
This reverts to the earlier behaviour, because:
* phys can be renamed, which breaks hardcoding "phy%d"
* /sys/class/ieee80211/*/device can return networks of other phys,
since "device" is a symlink which can have multiple phys
The earlier behaviour fixes both points.
Fixes: 6194aaf0 "nl80211: simplify iterating over phy's devices"
Signed-off-by: Andre Heider <a.heider@gmail.com>
Tested-by: Olcay Korkmaz <nuke_mania@hotmail.com>
[ drop extra Fixes tag related to openwrt issue ]
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
There's no need to iterate over all devices in /sys/class/net and check
if the phy index matches, sysfs already provides per phy devices.
Signed-off-by: Andre Heider <a.heider@gmail.com>
Providing the channel alone isn't clear as there're overlapping channels
on e.g. band 2 and 6.
Note: This changes the ABI.
Signed-off-by: Andre Heider <a.heider@gmail.com>
So that consumers don't have to fiddle around with mapping frequencies
to bands, which everyone seems to do a little differently.
Note: This changes the ABI.
Signed-off-by: Andre Heider <a.heider@gmail.com>
Some device have embedded wifi card that are not connected with usb or
internall with pci. Such device have fake device_id and only the
vendor_id actually reflect something real but internally they don't have
any id and are just matched by the node compatible binding in DT.
We currently match this with a big if-else to match the single devices
but this can be improved and be matched directly in devices.txt
Rework this so that we can drop the big if-else and move the matching
to devices.txt
When a device is matched using compatible in iwinfo the hardware will be
flagged as embedded and won't print empty ids.
Update devices.txt by migrating all the compatible matching device from
fake id to compatible matching.
Tested-by: Christian Marangi <ansuelsmth@gmail.com> # ipq4019
Co-developed-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Tested-by: Robert Marko <robimarko@gmail.com> # ipq8074
Reviewed-by: Andre Heider <a.heider@gmail.com>
Repurpose the IWINFO_FREQ_NO_2160MHZ define for that, which was
introduced but never used. nl80211 doesn't have such a flag either.
Signed-off-by: Andre Heider <a.heider@gmail.com>
Just use the already provided info instead of mapping frequencies.
Fixes mapping lower frequencies of the 6G band, which starts at 5935MHz.
Signed-off-by: Andre Heider <a.heider@gmail.com>
Update using current linux sources, which this was apparently
based on.
Signed-off-by: Andre Heider <a.heider@gmail.com>
[don't fail if no ieee80211ax option is present]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
It's too slow to do it all over again and again, especially with e.g.
luci constantly polling.
Before:
$ time iwinfo phy0 info
real 0m 0.54s
$ time iwinfo radio0 info
real 0m 0.70s
$ time ubus call luci-rpc getWirelessDevices
real 0m 0.67s
After:
$ time iwinfo phy0 info
real 0m 0.04s
$ time iwinfo radio0 info
real 0m 0.09s
$ time ubus call luci-rpc getWirelessDevices
real 0m 0.17s
Signed-off-by: Andre Heider <a.heider@gmail.com>
Treat USB vendor and product IDs like PCI subsystem vendor and device IDs
respectively to allow describing USB devices as `0000 0000 $vid $pid` in
the hardware database.
As a side effect, this also skips potentially expensive MTD lookups on
embedded devices with plugged in USB radios.
With a 5.15 kernel, such MTD lookups even spam dmesg with:
mtdblock: MTD device 'factory' is NAND, please consider using UBI block devices instead.
This fix speeds up iwinfo considerably on affected systems:
Before:
sys 0m 3.56s
After:
sys 0m 0.09s
Suggested-by: Andre Heider <a.heider@gmail.com>
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
This isn't required nor used anymore, and it's been +8 years.
See
"735a03be: iwinfo: adjust for changed wpa_supplicant control socket path"
on the main repository.
Signed-off-by: Andre Heider <a.heider@gmail.com>
No data at all can currently be read from the control socket.
Set up the permission exactly like hostapd's wpa_cli utility to fix it.
Among other this fixes a mesh's encryption being shown as "None".
Signed-off-by: Andre Heider <a.heider@gmail.com>
Increase buffer sizes to get rid of e.g.:
iwinfo_nl80211.c:797:41: note: 'snprintf' output between 24 and 279 bytes into a destination of size 64
Signed-off-by: Andre Heider <a.heider@gmail.com>
Remove the assumption that phy names follow the format phy%d and instead look
up the phy index from sysfs properly
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Don't assume a fixed naming pattern. Check all options in this order:
- netdev
- phy name
- wifi device in uci
Use the first one that matches
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Don't add the hwmode A for radios which do not support 5GHz operation.
Before, this hwmode was added to all non-VHT radios regardless of their
supported bands.
Fixes commit 562d015326 ("iwinfo: nl80211: fix hwmode parsing for multi-band NICs")
Signed-off-by: David Bauer <mail@david-bauer.net>
In case a NIC supports multiple frequency bands, the supported
hw-modelist might not contain all valid hwmodes, as B/G/AD hwmodes are
only included based on the last parsed supported frequency.
In case a radio supports multiple bands, this might result in these
hwmodes not being flagged as supported.
Circumvent this by tracking all seen frequency bands using a bitmask
which later determined which HW modes are listed as supported.
Signed-off-by: David Bauer <mail@david-bauer.net>
The previous shell script implementation accepted shorter path values
that omitted initial parts before the pcie bus node by only checking if
the configured path is included in the determined path
Add support for doing the same thing here
Signed-off-by: Felix Fietkau <nbd@nbd.name>
This adds 802.11ax HE specific rate information to iwinfo.
Add fields for HE status of a STA as well as DCM and guard interval
fields specific to HE operation.
Signed-off-by: David Bauer <mail@david-bauer.net>
Perform a split wiphy dump when acquiring the hw modelist of a mac80211
device. Otherwise HE capabilities are missing from the message.
This is necessary since upstream commit f8d504caa ("nl80211: reduce
non-split wiphy dump size") (Kernel 5.9).
Signed-off-by: David Bauer <mail@david-bauer.net>
This adds basic support for IEEE 802.11ax when requesting HW or HT
Modelist for a PHY from iwinfo. This way, applications using iwinfo can
detect HE phys.
Signed-off-by: David Bauer <mail@david-bauer.net>
Add description strings and detection logic for all 6
types of WiFi NICs integrated in QCA MIPS WiSoCs.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Extend support for WPA ciphers by GCMP which is required for 802.11ad.
Breaks ABI as ciphers now needs to be a field of 16 bits instead of 8.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Iwinfo already export the htmode but there is no way to know where the
channel expan in case a 40Mhz+ channel width is used. Export the center
channels used by the driver to better know the channel utilizzation of
the wifi.
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Export ht and vht operation data in scan results. These additional data
can be usefull to check wifi channel utilizzation by neraby stations.
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
As it's the most reliable and straight forward way on modern platforms,
detect the hardware ID by checking the 'compatible'-string of non-PCI
(ie. built-in) devices.
As this is only relevant for WiSoCs, there won't be that many records,
so it's not worth making any larger structural changes. Hence just
match the compatible string and translate it into fake PCI IDs.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This reverts commit a6914dc0dc.
iwinfo currently misses ABI version tracking in OpenWrt, potentially
breaking other packages unintentionally.
Revert this commit for now until this is implemented.
Otherwise, we are not able to safely bump iwinfo at the moment.
Signed-off-by: David Bauer <mail@david-bauer.net>
